[House Hearing, 116 Congress]
[From the U.S. Government Publishing Office]
EXPOSURE NOTIFICATION AND CONTACT
TRACING: HOW AI HELPS LOCALITIES REOPEN
SAFELY AND RESEARCHERS FIND A CURE
=======================================================================
VIRTUAL HEARING
BEFORE THE
TASK FORCE ON ARTIFICIAL INTELLIGENCE
OF THE
COMMITTEE ON FINANCIAL SERVICES
U.S. HOUSE OF REPRESENTATIVES
ONE HUNDRED SIXTEENTH CONGRESS
SECOND SESSION
__________
JULY 8, 2020
__________
Printed for the use of the Committee on Financial Services
Serial No. 116-101
[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]
______
U.S. GOVERNMENT PUBLISHING OFFICE
43-194 PDF WASHINGTON : 2021
HOUSE COMMITTEE ON FINANCIAL SERVICES
MAXINE WATERS, California, Chairwoman
CAROLYN B. MALONEY, New York PATRICK McHENRY, North Carolina,
NYDIA M. VELAZQUEZ, New York Ranking Member
BRAD SHERMAN, California ANN WAGNER, Missouri
GREGORY W. MEEKS, New York FRANK D. LUCAS, Oklahoma
WM. LACY CLAY, Missouri BILL POSEY, Florida
DAVID SCOTT, Georgia BLAINE LUETKEMEYER, Missouri
AL GREEN, Texas BILL HUIZENGA, Michigan
EMANUEL CLEAVER, Missouri SEAN P. DUFFY, Wisconsin
ED PERLMUTTER, Colorado STEVE STIVERS, Ohio
JIM A. HIMES, Connecticut ANDY BARR, Kentucky
BILL FOSTER, Illinois SCOTT TIPTON, Colorado
JOYCE BEATTY, Ohio ROGER WILLIAMS, Texas
DENNY HECK, Washington FRENCH HILL, Arkansas
JUAN VARGAS, California TOM EMMER, Minnesota
JOSH GOTTHEIMER, New Jersey LEE M. ZELDIN, New York
VICENTE GONZALEZ, Texas BARRY LOUDERMILK, Georgia
AL LAWSON, Florida ALEXANDER X. MOONEY, West Virginia
MICHAEL SAN NICOLAS, Guam WARREN DAVIDSON, Ohio
RASHIDA TLAIB, Michigan TED BUDD, North Carolina
KATIE PORTER, California DAVID KUSTOFF, Tennessee
CINDY AXNE, Iowa TREY HOLLINGSWORTH, Indiana
SEAN CASTEN, Illinois ANTHONY GONZALEZ, Ohio
AYANNA PRESSLEY, Massachusetts JOHN ROSE, Tennessee
BEN McADAMS, Utah BRYAN STEIL, Wisconsin
ALEXANDRIA OCASIO-CORTEZ, New York LANCE GOODEN, Texas
JENNIFER WEXTON, Virginia DENVER RIGGLEMAN, Virginia
STEPHEN F. LYNCH, Massachusetts WILLIAM TIMMONS, South Carolina
TULSI GABBARD, Hawaii VAN TAYLOR, Texas
ALMA ADAMS, North Carolina
MADELEINE DEAN, Pennsylvania
JESUS ``CHUY'' GARCIA, Illinois
SYLVIA GARCIA, Texas
DEAN PHILLIPS, Minnesota
Charla Ouertatani, Staff Director
TASK FORCE ON ARTIFICIAL INTELLIGENCE
BILL FOSTER, Illinois, Chairman
EMANUEL CLEAVER, Missouri BARRY LOUDERMILK, Georgia, Ranking
KATIE PORTER, California Member
SEAN CASTEN, Illinois TED BUDD, North Carolina
ALMA ADAMS, North Carolina TREY HOLLINGSWORTH, Indiana
SYLVIA GARCIA, Texas DENVER RIGGLEMAN, Virginia
DEAN PHILLIPS, Minnesota VAN TAYLOR, Texas
ANTHONY GONZALEZ, Ohio
C O N T E N T S
----------
Page
Hearing held on:
July 8, 2020................................................. 1
Appendix:
July 8, 2020................................................. 29
WITNESSES
Wednesday, July 8, 2020
Kuppalli, Krutika, Infectious Diseases and Global Health
Physician...................................................... 7
McClendon, Brian, CEO, CVKey Project............................. 5
Perry, Andre M., Fellow, Metropolitan Policy Program, the
Brookings Institution.......................................... 8
Raskar, Ramesh, Associate Professor, MIT, and Founder and Chief
Scientist, PathCheck Foundation................................ 10
APPENDIX
Prepared statements:
Kuppalli, Krutika............................................ 30
McClendon, Brian............................................. 49
Perry, Andre M............................................... 53
Raskar, Ramesh............................................... 61
EXPOSURE NOTIFICATION AND
CONTACT TRACING: HOW AI
HELPS LOCALITIES REOPEN
SAFELY AND RESEARCHERS
FIND A CURE
----------
Wednesday, July 8, 2020
U.S. House of Representatives,
Task Force on Artificial Intelligence,
Committee on Financial Services,
Washington, D.C.
The task force met, pursuant to notice, at 12 p.m., via
Webex, Hon. Bill Foster [chairman of the task force] presiding.
Members present: Representatives Foster, Porter, Casten,
Adams, Garcia of Texas; Loudermilk, Budd, Hollingsworth,
Taylor, and Gonzalez of Ohio.
Ex officio present: Representative Waters.
Also present: Representative Himes.
Chairman Foster. The Task Force on Artificial Intelligence
will now come to order. Without objection, the Chair is
authorized to declare a recess of the task force at any time.
Also, without objection, members of the full Financial
Services Committee who are not members of this task force are
authorized to participate in today's hearing.
Members are reminded to keep their video function on at all
times, even when they are not being recognized by the Chair.
Members are also responsible for muting and unmuting
themselves, and for muting themselves after they have finished
speaking.
Consistent with regulations accompanying H. Res 965, staff
will only mute members and witnesses as appropriate when not
being recognized by the Chair to avoid inadvertent background
noise. Also, all House Rules related to order and decorum apply
to this remote hearing.
Today's hearing is entitled, ``Exposure Notification and
Contact Tracing: How AI Helps Localities Reopen Safely and
Researchers Find a Cure.''
I will now recognize myself for 4 minutes to give an
opening statement.
Good morning. Today's hearing will focus on the essential
tradeoff incumbent upon us to make between life, liberty,
privacy, and the pursuit of happiness. And the role that
technology, financial transaction data, contact tracing apps,
and artificial intelligence, in particular, may be able to play
to improve that essential tradeoff. And there will be a
tradeoff.
Americans have come to see privacy as an essential part of
liberty, and the tradeoffs between privacy and State interests
are something that we struggle with continually on the
Financial Services Committee. If one's only concern were life
and the pursuit of happiness, and crushing the COVID-19 virus
as quickly as possible, one can imagine setting up a dystopian
surveillance state where every credit card purchase, every web
search query, and the exact locations of every cell phone were
continuously monitored, and where AI-powered facial recognition
cameras would record one's every move, and not only when you
are buying groceries at Whole Foods, but everywhere you went.
This would be an epidemiologist's dream and would likely
allow whatever country implemented that to very rapidly crush
the coronavirus pandemic. Whenever someone tested positive,
their credit card purchases, their cell phone data, all of
their financial data, and other data would be immediately
analyzed. And, in fact, some press reports indicate that Korea,
an otherwise apparently free country, finds credit card data to
be a key component in effective contact tracing.
This aggressive approach would allow for faster scientific
understanding of the spreading mechanisms as well. AI learning
algorithms might rapidly learn that you are, for example, much
more likely to spread the disease if you happen to buy three
beers at the bar than if you bought only one. And cell phone
location data would indicate to whom you were likely to have
spread it. Such an AI-enabled pandemic surveillance state would
correctly identify in real-time, for example, that
superspreader individuals almost never wore masks in public.
And that churches and sports events were especially dangerous
places to congregate. But a surveillance state like that would
automatically warn Black Lives Matter protesters to disburse,
and there are a huge number of worries along those lines.
And this is not science fiction. It is essentially the same
technology that has protesters in Hong Kong fighting for their
liberty. Such surveillance societies might well save lives and
maximize life, and arguably, even part of the pursuit of
happiness, and perhaps it would even maximize economic liberty
by opening our economy sooner, but this would come at a
tremendous cost to individual privacy and liberty.
The tradeoffs that we make responding to COVID-19 have real
financial and economic impacts. A recent study by Goldman Sachs
quantified the tradeoff between mandatory masking policies,
which are unquestionably an infringement on personal liberty,
and economic growth. They found, unsurprisingly to scientists,
that mandatory masking policies have a negative impact on the
spread of COVID-19 but a positive impact on economic growth.
This allowed countries with mandatory masking policies to
reopen their economies sooner, while maintaining acceptable
COVID-19 infection rates, thereby successfully, in that case,
trading off the loss of liberty for mandatory masking for an
increase in economic liberty from early reopening of their
retail economy.
Contact tracing apps together with back-end AI that
combines the raw data from cell phone tracking in various forms
with other data sources, including financial data that COVID-
positive patients might opt into on a voluntary basis, has the
potential to capture some of the health and economic benefits
of much more intrusive [inaudible] Privacy for those who opt
in.
At this hearing, we have expert practitioners experienced
in the development of privacy-preserving contact tracing
software in real-life tracing in pandemic situations and
reducing discrimination in technology.
So, I look forward to hearing from our witnesses, and I now
recognize the ranking member of the subcommittee, Mr.
Loudermilk, for 5 minutes.
Mr. Loudermilk. Thank you, Mr. Chairman. And a special
thank you to our witnesses for being part of this hearing
today.
Today, we will discuss how local, State, and Federal
Governments, as well as businesses, schools, and other
organizations can take advantage of modern technology to help
[inaudible] Technology could be useful is in contact tracing.
The manual method of contact tracing involves public health
officials speaking with individuals who have the virus to
identify people with whom they had close contact during the
time they have been infectious, notifying those people of their
exposure and referring them for testing.
In a nation with about 330 million people, it is difficult
to do manual contact tracing on a mass scale. However, there
are encouraging technologies that can enable contact tracing to
be done with smartphones, including Bluetooth or other location
data to track peoples' movements and match them to the
movements of others who have been exposed to the virus.
In order for digital contact tracing to work, it must have
strong participation from citizens. Experts estimate that 40 to
60 percent of the population in a given area would need to
participate in a contact tracing system for it to be effective.
This means in the United States, where participation and
contact tracing is voluntary, and should be, and is handled
primarily by State and local public health officials, citizens
will need to trust the apps that they are using and feel
confident their privacy will not be violated.
In order to have that trust, it is critical that citizens
understand what data will be collected, who will have access to
the data, and how the data will be used. [inaudible] For
digital contact tracing have been low in the United States and
in other parts of the world. Recent polling indicates that
about two-thirds of Americans would not trust a contact tracing
app developed by a Big Tech company or the Federal Government.
And in some European countries where their federal government
has released a contact tracing app, less than 3 percent of the
population is using the app.
Some authoritarian countries, like China, have increased
participation by making contact tracing apps mandatory and have
used digital payments and social media platforms to track
citizens' movements. Needless to say, that will not and should
not happen in the United States.
Given that privacy will be paramount, a major limiting
factor in our ability to implement nationwide digital contact
tracing is that we do not have a Federal consumer data privacy
law, nor do we have a national data security and breach
notification standard. Contact tracing is operating under a
patchwork of State laws, and the circumstances we are dealing
with today only increase the need for a national standard.
Committee Republicans have been working diligently to
develop a consumer data privacy proposal without a national
consumer data privacy protection law. This issue is best left
up to the States. Another limiting factor is that the accuracy
of digital contact tracing apps has been questionable and could
result in false positives or false negatives.
The CDC stated in its guidelines that more data is needed
to assess the true public health value of digital contact
tracing. Criminals have also been posing as contact tracers to
ask consumers to share their Social Security numbers or bank
account numbers, which can subject consumers to fraud and
further reduce the public's trust in contact tracing.
Another challenge is that low-income and homeless
individuals who have been disproportionately affected by the
virus often do not have a smartphone, which limits the
effectiveness of digital contact tracing in those populations.
While these are important topics to address, they generally are
not Financial Services Committee issues. It is best to have the
appropriate committee address them, and the Energy and Commerce
Committee will have a hearing on some of these issues tomorrow.
I look forward to hearing from our witnesses, and I yield
back.
Chairman Foster. Thank you. The Chair will now recognize
the Chair of the Full Committee, Chairwoman Waters.
Chairwoman Waters. Thank you very much, Chairman Foster. I
thank you for holding this hearing. Containing and preventing
the further spread of the coronavirus is essential to
protecting our economy.
While last week's employment figures demonstrated a slight
rebound, the number of coronavirus cases is sharply increasing,
and State reopenings have halted or reversed. Contact tracing
and limiting the period of the coronavirus is critical for all
aspects of our economy, including financial institutions that
may bring workers back to the workplace.
I look forward to hearing our expert panel discuss how
artificial intelligence and other technologies can be used in
an inclusive manner to help contain the coronavirus, and
ultimately, keep people safe.
I want to thank you, and I yield back the balance of my
time.
Chairman Foster. Thank you.
Today, we welcome the testimony of, first, Brian McClendon,
the CEO and co-founder of the CVKey project, which is building
a suite of privacy-first open-source applications to help
communities reopen responsibly without compromising privacy.
Previously, he was vice president and co-founder of Google
Earth and Street View and vice president of mapping at Uber.
Second, Dr. Krutika Kuppalli, an infectious diseases
physician, who has previously worked on the front lines of
various humanitarian responses, including the 2014 West Africa
Ebola outbreak and the current DRC Ebola outbreak and novel
coronavirus outbreak.
Third, Andre M. Perry, a fellow at the Metropolitan Policy
Program of the Brookings Institution, whose research is focused
on race, structural inequality, education, economic inclusion,
and, since the onset of the COVID-19 pandemic, the underlying
causes of the outsized number of coronavirus-related deaths in
Black communities.
And finally, Ramesh Raskar, a professor at MIT Media Lab,
where he focuses on machine learning and imaging for health and
sustainability. He is also the founder of the PathCheck
Foundation, a global nonprofit, building open-source software
and standards for digital contact tracing and exposure
notification to stop COVID-19.
Witnesses are reminded that your oral testimony will be
limited to 5 minutes. A chime will go off at the end of your
time, and I ask that you respect the Members' and other
witnesses' time by wrapping up your oral testimony as
scheduled.
And without objection, your written statements will be made
a part of the record.
Mr. McClendon, you are now recognized for 5 minutes to give
an oral presentation of your testimony.
STATEMENT OF BRIAN MCCLENDON, CEO, CVKEY PROJECT
Mr. McClendon. Chairman Foster, Ranking Member Loudermilk,
and members of the task force, my name is Brian McClendon, and
I am the CEO of CVKey Project, and also a research professor at
the University of Kansas. Previously, I spent 10 years leading
the teams that built Google Maps and Google Earth, Street View,
and many other georelated services. Thank you for giving me the
chance to speak before the task force today.
In my testimony I will describe how privacy, disclosure,
and opt-in data collection impact our ability to identify,
locate, and isolate those who have been exposed to or infected
by COVID-19.
CVKey Project is a 501(c)(3) focused on helping America
reopen responsibly given the ongoing pandemic. Together with my
team of world-class engineers and product developers, we built
that user data privacy into a key app called CVKey, and we are
focused on getting that adopted around the country today.
It focuses on individual symptom checking, policy
communications across communities, and access control into
venues such as universities, workplaces, and schools. As we
have recently seen in the media, there is a significant need to
improve communication of policies if we hope to operate at
anywhere near close to normal for the next 18 months.
Artificial intelligence and machine learning require large
amounts of data collection for training purposes. This ground
truth data helps algorithms figure out how to make better
predictions. The most valuable data to combat COVID-19 can
actually be found in the contact tracing interviews of infected
and exposed people. Early detection, reporting, and testing
leads to quick self-isolation and quarantine, and can shut down
the spread of the virus faster than any other method, but only
if it is resourced sufficiently and executed well.
The phone calls that contact tracers make are often not
answered because most people don't pick up from unknown numbers
these days. When the contact tracer does reach a potential
case, they have to make a decision about whether to ask the
person to quarantine based on what they learn [inaudible]
Information from the infected or exposed, the virus is quietly
spreading.
What exposure notification apps provide is a way to use
cell phone data to detect after the fact whether you are near
someone who later tests positive for COVID-19 by notifying you
and informing you of next steps, which are usually to contact
the contact tracing team or get tested or quarantined.
These new apps offer a way to help contact tracing scale
with less effort, more accuracy, and more coverage than is
otherwise possible. There are two ways, at least, to do this.
The first is using GPS information logs to compare where the
infected person was relative to everyone else.
The first problem with this is that GPS locations are not
accurate enough [inaudible] For 10 minutes. The other, more
serious problem with location data is it contains personally
identifiable information that can be impossible to
algorithmically remove. If I can guess where someone works and
lives, I can easily figure out who that person is, usually by
pulling additional data from commercially available data
sources like Equifax.
Naive implementations of this method were deployed in the
U.K., North Dakota, and Utah, and did not meet with success for
multiple reasons. The biggest one was fear that either Big Gov
or Big Tech was tracking anyone who installed it. Next, always-
on GPS location has a material impact on phone battery life, so
folks turned it off or uninstalled it.
Finally, early implementations did upload everyone's
location to the government, and without sufficient protections,
that data can then be exposed/stolen by others even if it
wasn't misused by their government. A better solution is to use
low-energy Bluetooth signals to allow phones to record when
they are near other phones. A naive implementation can still
allow tracking, but luckily Apple and Google have come to
agreement on a privacy-protecting method called the Google/
Apple Exposure Notification method, and it was developed to
work on both Android and IOS.
They have asked countries' health agencies to build and
release apps using this tech. In the United States, they
weren't delusional enough to assume that a single Federal
institution could be adopted, so, they are working with one
group per State.
Like Dr. Raskar's Safepaths group, my team is building the
app using this tech, and we believe that it preserves privacy
better and works better than GPS solutions. The goal would be
to get as many people as possible in the State to download this
app, but the challenge, of course, is that even if we could
only get 40 percent installed uniform across the group, we will
only have 16 percent of exposures detected, so we need to find
better ways to market the app.
The privacy is much better than GPS because the data that
eventually gets uploaded in a rare case where you have been
infected is a set of random numbers that have no personally
identifiable information. And then, when the other phones
download this random number set, they notify their users
without doing any cloud work at all and tell the user that they
have been exposed and recommend that they call the contact
tracing team.
I believe that privacy is critical to marketing these apps
to Americans, because I think they are very worried about what
these apps could do, and we need to get and install a base for
these apps.
[The prepared statement of Mr. McClendon can be found on
page 49 of the appendix.]
Chairman Foster. Thank you.
Dr. Kuppalli, you are now recognized for 5 minutes.
STATEMENT OF KRUTIKA KUPPALLI, INFECTIOUS DISEASES AND GLOBAL
HEALTH PHYSICIAN
Dr. Kuppalli. Thank you, Chairman Foster, Ranking Member
Loudermilk, and distinguished members of the task force for the
opportunity to testify before you today.
I am extremely grateful for your interest and commitment
towards helping the support of the novel coronavirus disease
efforts in the United States as we attempt to control the
deadly pandemic.
I am an infectious disease physician by training and have
extensive experience in global health and health security. My
clinical and research interests focus on health systems
strengthening in resource-limited settings, outbreak
preparedness and response, research and clinical care of
emerging infections, and healthcare policy.
As the COVID-19 pandemic continues to spread across the
United States, I hope to give you a greater understanding of
the threat this has to the global and economic security of the
United States. I will discuss challenges facing public health
experts and researchers leading COVID-19 efforts domestically,
and will explain how we may better leverage available resources
and tools to improve our response capacity and provide examples
of other countries who have successfully implemented policies
to control COVID-19.
During this unprecedented time, it is vital for the United
States to collaborate with the global community so we can learn
from each other and develop best practices to ensure that
science informs policy.
In early 2020, as the world watched COVID-19 sweep over
Wuhan, and spread through China, other countries recognized
that COVID-19 posed an unprecedented risk to the physical and
economic health of their countries and activated pandemic
preparedness plans. The three most common components of
successful plans are: one, the development of a comprehensive
national plan led by science; two, rapid scaling up of testing
capacity; and three, prioritization and implementation of
contact tracing.
The United States accounts for approximately 4 percent of
the global population, but 25 percent of COVID-19 cases and
fatalities globally. More concerning is that, as lockdowns have
been lifted over the past month, there has been a record
increase of COVID-19 cases, with over 45,000 cases per day
since July 1st, and a record 60,000 cases in the last 24 hours,
leading to the halting or reversal of business openings across
the country.
With coronavirus cases surging across the United States, it
is imperative we adopt essential elements of an outbreak
response, which should include a national cohesive plan led by
science. The cornerstone of such a plan should be to test,
isolate, and contact trace cases so we can contain this
pandemic.
Contact tracing, often called the linchpin of an outbreak
response, is critical to identifying those who have potentially
been exposed. The hiring of contact tracers, and additional
Federal resources, infrastructure, and training will be
required to ensure that we have a sufficient and well-
coordinated workforce to provide this vital task.
Based on current population and data, experts estimate each
case requires 10 to 25 contacts to be traced. That is requiring
at least 300,000 contact tracers. Another vital component of
successful contact tracing is leveraging new and innovative
technologies to promote efficient and broad implementation
strategies.
Although technologies are increasingly used, it is
important to remember the ethics of public health information:
data protection and data privacy when using any of these
technologies. South Korea is an example of a country that has
successfully contained COVID-19 and used technology to assist
in their contact tracing efforts.
In the aftermath of the 2015 MERS outbreak, South Korea
invested resources in training healthcare personnel, and
developing response systems, infrastructure, and laboratory
capacity for future infectious disease outbreaks. When the 2019
outbreak took off in South Korea, in addition to using standard
methods of contact tracing, they also used medical records, GPS
systems, credit card transactions, and closed circuit
transactions to utilize a more robust contact tracing system.
It was thought that this information provided more accurate
information on individuals' location, duration [inaudible]
Exposure might not be able to call or confirm. Considerations
and protocols for privacy were taken into account to ensure
only information related to infectious diseases is [inaudible]
This social, health, and economic threat of our generation, and
how we choose to manage it will be our legacy. Until we have a
vaccine, control will rely on using surveillance, testing,
contact tracing, and isolation to prevent transmission.
Given the United States' prominent global role in
technology, it can leverage this expertise as an opportunity to
take a leadership role in developing a world-class contact
tracing system to contain this outbreak, using information
technologies to develop integrated systems to enhance our
ability to prevent, detect, and respond to future outbreaks
before they become national or global epidemics.
Thank you, again, for the opportunity to testify before
you, and I look forward to your questions.
[The prepared statement of Dr. Kuppalli can be found on
page 30 of the appendix.]
Chairman Foster. Mr. Perry, you are now recognized for 5
minutes.
STATEMENT OF ANDRE M. PERRY, FELLOW, METROPOLITAN POLICY
PROGRAM, THE BROOKINGS INSTITUTION
Mr. Perry. Chairman Foster, Ranking Member Loudermilk, and
members of the task force, thank you for inviting me to testify
today on this extremely important issue affecting millions of
people across the country.
The COVID-19 pandemic will continue to take significantly
more lives than the approximately 130,000 it has already
claimed if the United States Government does not invest in
tools that are proven to combat the spread of the virus,
including strategies that aim to dismantle structural racism.
According to an analysis published July 5th by the New York
Times, Latino or Black or Hispanic and Black residents in the
U.S. are 3 times as likely to become infected as their White
neighbors. Higher COVID-19 mortality rates among Black and
Brown communities reflect the historic devaluation and
disenfranchisement of their lives, properties, and communities.
Structural racism is the preexisting condition that must be
accounted for in our battle with the coronavirus. It is hard to
calculate the damage that the lack of coordinated comprehensive
Federal response has caused families in terms of lives, jobs,
and businesses. Those losses will even be more severe if there
is not sizeable investment in infection testing, social
distancing, extended paid leave, supplemental employment
insurance, hazard pay, and contact tracing.
However, the universal application of these preventive
tools won't eradicate the substandard housing, poverty, limited
job opportunities, and other conditions of structural racism
that underlie the racial health inequities. My written
testimony presents three general concerns regarding artificial
intelligence in contact tracing as it pertains to structural
racism and racial bias.
First, contact tracing and exposure notification are not
necessarily remedies for structural racism. Contact tracing and
other public health tools are not neutral. They can exacerbate
or mitigate the impacts of structural racism.
A second concern taken up by my testimony revolves around
representation. Contact tracing systems should include the
people from the communities that have historically been
excluded from other systems that generate better health and
economic outcome. From the tech tools that are developed to the
contact [inaudible] Be included in any effort to expand contact
tracing. Manual contact tracers rely on the skills of
interpersonal communication and empathy in order to build trust
and receive and interpret information.
If we hire an army of White tracers to track the spread of
the virus, we should expect unequal or even negative results in
Black communities. In addition, if we do not hire local Black
and Brown people to serve those neighborhoods, we exacerbate
the racial wealth gap which also serves as a barrier of
protection against infection. Hiring Black and Brown manual
tracers offers an opportunity to add jobs to the neighborhoods
that are experiencing higher levels of unemployment.
Third, AI tools in health pose the same risk that they do
in other fields. AI is only as good as [inaudible] And their
biases can ultimately lead to flaws in the technology and
amplify biases in the real world. Our expedition into digital
tools must demand greater recruitment and investment in Black
and Brown tech firms, rigorous reviews and testing for racial
biases, and more engagement and involvement from local
communities before we deploy AI technology to communities en
masse.
Systemic racism and discrimination are already embedded in
our health, housing, and educational systems. Developers must
intentionally build AI systems through a lens of racial equity
if the technology is not going to generate outcomes that
reflect the biases of the developers.
As more Black and Brown people are exposed to these racial
inequities, more will die from COVID. The proliferation of the
coronavirus forces us to see our inherent connections in a way
that our public policy has not always recognized. Individual
recovery is contingent upon how much we collectively live by
the principle of all being in this together.
If undocumented residents are sick, the country's citizens
will be as well. If Black and Latino or Hispanic people suffer
from COVID-19's effects, so will Asian American and White
people. But being aware of our vulnerability is not the main
problem. The trap of racism and structural inequality is.
Thank you.
[The prepared statement of Mr. Perry can be found on page
53 of the appendix.]
Chairman Foster. Thank you.
And, Mr. Raskar, you are now recognized for 5 minutes.
STATEMENT OF RAMESH RASKAR, ASSOCIATE PROFESSOR, MIT, AND
FOUNDER AND CHIEF SCIENTIST, PATHCHECK FOUNDATION
Mr. Raskar. Thank you, Chairman Foster and Ranking Member
Loudermilk, and to the members of the task force and the Full
Committee for the opportunity to testify today. It is an honor
and a privilege to be here today.
Let us compare how we use AI and technology for weather
forecasting versus a pandemic. The National Weather Service can
analyze and predict by collecting data every day. However,
responding to the dynamics of a hurricane is different from
responding to a pandemic, as pandemics are about people being
at risk because of other innocent people.
We need real-time participation from people, and that is
why smartphone apps can play a big role. Given this is a social
problem, I want to emphasize three things: first, how to
augment top-down manual contact tracing with bottom-up apps
used by people; second, how to make sure that apps are
trustworthy by obviously preserving inclusive and built by open
source and nonprofits; and third, how we may create a
[inaudible] National weather--we have this national AI platform
that leverages this privacy-preserving apps.
I come to this question having spent much of my last decade
working at MIT in AI digital health and algorithms for privacy.
We built privacy-preserving global AI, a distributed machine
learning method called split learning that can build AI without
accessing any raw, identifiable data from individuals. And
other teams have built wonderful techniques, such as federated
learning and differential privacy that is going to be used in
the U.S. Census 2020.
In March, our team at MIT created one of the first privacy-
preserving smartphone apps, which led to the creation of
PathCheck Foundation, a charitable nonprofit organization
dedicated to building free, open-source and industry standards
that assist U.S. States, nations, and private sector
organizations.
And our nonprofit is already building exposure notification
and case management apps back in servers and dashboards for
various U.S. States and nations. And, frankly, it has been a
humbling experience to work on this complex challenge over the
last few months. And based on our learning so far, let me
address concerns and also present some recommendations.
First, Congress should require that public health
authorities augment their manual contact tracing solution with
digital apps for solutions, and this should be privacy first,
because assuring trust is key to public acceptance, and public
acceptance of these digital apps will let the country solve the
tracing problem cheaply, quickly, and at scale. At PathCheck
Foundations, we are working very actively on this hybrid
approach by combining exposure notification, personal guidance,
and case management, and we hope such apps can augment and
simplify manual contact tracing.
Second, Congress should set out a series of requirements to
ensure that contact tracing apps are built transparently as
open source, ideally by nonprofits, and just like other public
utilities, are open for scrutiny by the public.
And, obviously, this app should be inclusive across
socioeconomic disparity, and not exclude those who don't have
the latest technology.
So, while at PathCheck Foundation, we are very proud to
deploy Google, Apple Bluetooth API, we also want to support
local solutions, such as QR codes, WiFi, and so on, and we
continue to work with communities and employers and schools to
improve the inclusion and innovation.
And then, finally, the National Pandemic Response Service,
based on a smartphone app. We need a national pandemic response
that will allow us to take micro and macro aggregation of
analyses and predictions. Instead of creating a surveillance
state and a top-down system, let's build a bottom-up
smartphone-based solution. We need a new AI that relies on the
information stored on peoples' phones, and what we have learned
through our work at MIT privacy-preserving AI is that we can
truly create this decentralized AI and orchestrate the
socioeconomic interactions between governments, businesses,
individuals, and even vulnerable communities without creating a
surveillance state.
Our analysis shows that a modest budget of $100 million to
$150 million that creates a public-private partnership can
provide a data-ready national service for a national pandemic
response. At the same time, this national service can help us
be ready for a nasty flu season in the future, boost data
solutions for public health, and be prepared for the next
pandemic.
To conclude, the need of the hour is augmentation of
[inaudible] Apps inclusive, open source, and trustworthy, and
[inaudible] Built on aggregation of data from these privacy-
preserving apps.
Thank you.
[The prepared statement of Mr. Raskar can be found on page
61 of the appendix.]
Chairman Foster. Thank you.
And I will now recognize myself for 5 minutes for
questions.
Mr. McClendon, you mentioned the necessity of maximizing
adoption of contact tracing apps in order to maximize their
effectiveness. Businesses may well have a role here. According
to a PricewaterhouseCoopers' survey conducted in April, 32
percent of businesses use contact tracing to help protect their
workforce. This raises a number of questions as financial
services and other companies continue to bring back workers to
offices over the next few months. How can contact tracing
effectively fit into their plans? Are there problems in making
the use of contact tracing apps a condition of employment?
And, in particular, does contact tracing offer benefits
over, for example, performing frequent COVID tests and
temperature checks?
Mr. McClendon. I don't know if I can speak about whether
frequent checks will do better or worse, but I do believe that
digital contact tracing, deployed widely, will help.
So, I was speaking about the coverage problem: 40 percent
of uniform population jobs, only 16 percent of exposures are
notified. So, we need a bigger percentage. We need 60, 70
percent adoption. And one of the best ways to do that is to
work with employers, universities, schools, and populations
that feel a sense of community and want to protect each other
by installing this app. It is really a marketing challenge, and
similar to how the Waze app was deployed in cities, and how
Uber installed its app and got sufficient penetration to be
interesting.
If there isn't enough penetration in a community, these
apps will not be successful. So having employers do it, I
think, is a valid way to increase the penetration, but it
should not be mandatory. It just should be strongly
recommended, and you should use the fact that if we don't get
these exposures detected, a company might then become a hot
spot, have an outbreak, and it might have to shut down again.
And I think that everybody is now very aware of what
happens during a shutdown, and so preventing the shutdown by
keeping your community safe is a pretty strong motivation.
Chairman Foster. Dr. Kuppalli, do you have any thoughts
from your perspective on this?
Dr. Kuppalli. Yes, thank you. Going to your question about
temperature checks versus testing, I would say that one of the
challenges with doing things like temperature checks is that we
know that up to, at least right now, 40 percent of
transmissions occur in the asymptomatic or pre-symptomatic
state, so transmissions occur when patients don't even know
that they are infected.
Relying on things like that is really challenging in this
particular infection in terms of trying to understand, maybe
who you have been in contact with, or who might be infected.
And I think that is something we need to think about as we are
thinking about employing these types of applications.
Chairman Foster. Thank you.
Mr. Raskar, you note in your testimony that you have been
working on how to create machine learning methods that can
build AI without accessing the raw data. This sounds a lot like
the homomorphic encryption that we had a previous hearing on,
which is, to me, just a fascinating technology, but I had a
simple question of how this can work when the number of
contacts is small.
For example, if you are by yourself all day for a whole
week, and then you go out and you meet one person and then get
notified that you have been exposed, you pretty much know which
one person that is. So, what are the limits to these privacy-
preserving techniques that may be important for people to know
about?
Mr. Raskar. Congressman, given your background in physics,
I would love to have a discussion about encryption with you,
but to the question of privacy, we ought to, first of all,
think about how data production has different layers. Those
three things are very different.
In a public health scenario, actually today we only care
about confidentiality, not as much anonymity or privacy,
because eventually, you do have to meet a doctor. It is not
like a cryptocurrency where [inaudible]. The challenge that if
you met only one person and you got an exposure notification,
you would know where that person is, is a classic example that
our goal here is to serve the public health purpose as well.
So, in that sense, although exposure notification based on
Bluetooth serves some purpose, which is whether you got an
exposure notification or not, a bigger challenge here for
public health is understanding the context of that encounter.
And for the context--
Chairman Foster. I will attempt to treat myself fairly and
gavel myself quiet here, and recognize the ranking member for 5
minutes for questions.
Mr. Raskar. Sure.
Mr. Loudermilk. Thank you, Mr. Chairman. The information
that we have received so far has been very interesting.
My first question is going to be to Mr. McClendon.
I do appreciate your focus on data privacy and security
when it comes to the contact tracing, and coming from an IT
background using the Bluetooth technology and your cooperation
with different businesses is, I think, a unique approach and
may have some validity there, but if we were to implement a
Federal contact tracing system, how would you be able to
reassure the citizens that they are able to know exactly what
data is going to be used or collected and who has access to
that data and how that data is going to be used?
I think those are three critical areas that if you are
going to get buy-in by the general public and get their
confidence in this, that they do need to know those three
things: one, exactly what data is going to be collected; two,
who is going to have access to that data; and three, how it is
going to be used?
Mr. McClendon. Right. As I said, I feel like it is a
marketing challenge. It is communicating how things work and
explaining it to users. And so, for the data that is being
collected in the Bluetooth Apple/Google gain solution, that
data is random and completely stored on the phone until the
user who is infected opts in to uploading those random numbers
to the server.
It requires a second opt-in. So, from a control
perspective, you are putting your control in the hands of the
user. Their data is never going to be pulled out from under
them. It is always going to be in their control, and I think
that is a very strong message for folks.
As far as what happens when you get an exposure
notification, the other innovation in this technology is that
the notification determination is made entirely on the phone
and then the communication of what message is sent to the user
is provided by the State or, in some cases, the country.
And so you have the opportunity to communicate to an
exposed user, ``Here are the things you should do,'' but we
don't know who that user is until, again, they opt-in by
calling in to the manual contact tracing team. And these
inbound phone calls are going to be much easier to handle
because the contact tracing team won't spend a ton of time
making dead phone calls.
These will be inbound, and they will be users who are
interested in learning more about what this exposure means.
Mr. Loudermilk. Let me make sure I understand the
situation. So, for instance, I may have tested positive. I have
this app. I go to a store. I am wearing my mask, doing
everything. So, if someone else gets near me, my phone, through
the app, captures their number, and then I choose whether to
upload that? Is that the scenario? Or does it automatically
notify the person that, ``Hey, you just got in contact or in
close proximity to somebody who has it?''
Mr. McClendon. The general system is that, after you test
positive, you are staying at home, let's hope. The goal,
though, is that prior to you testing, for 2 weeks of data,
prior to you testing positive, your phone has been broadcasting
these random numbers and has been collecting the data on your
own phone of who else you have seen. And on the day that you
test positive, the contact tracing team calls you up on the
phone to talk about your positive test because they were
notified, and this is all part of the standard process. They
ask you to upload your data, and give you a unique one-time
password to upload that data into the cloud, but the data that
is uploaded is random numbers, no identifying information at
all.
And at that time, all of the other phones download the
data, and see if in the last 2 weeks, they were near you. And
if they were, they get an exposure notification, but it is only
that they were exposed and they have to call the manual contact
tracing team to learn more.
Mr. Loudermilk. So the only numbers that are uploaded are
numbers of people who are using the app, or all numbers?
Mr. McClendon. The only numbers that are uploaded are of
the infected person if they were using the app during the last
2 weeks.
Mr. Loudermilk. Does that create a problem where the person
with the app is then determining if the information--let's say,
if I was around someone who had been infected and got tested,
then that person is determining if my phone number is going to
be uploaded or not?
Mr. McClendon. No. The data that you upload, uploads only
those random numbers that they receive out, not receive in.
Mr. Loudermilk. Okay.
Mr. McClendon. So, literally, only your information.
Everybody else compares whether they ever saw your number in
the phone. And so, again, it is only if each of the parties is
participating in the app.
If somebody's app is not--if somebody's phone does not have
the app installed, then they are not part of this situation at
all. They won't get notified. And if they are infected, there
is nothing for them to upload.
Mr. Loudermilk. Okay. Thank you.
I yield back.
Chairman Foster. Thank you for that, Mr. Ranking Member.
And I would now like to recognize the gentlewoman from
California, the Chair of the Full Committee, Chairwoman Waters.
Chairwoman Waters. Thank you very much. I want to first
turn to Mr. Perry. You have reminded us through your research
that historically, Black and Brown people have already been
socially distanced through segregated housing policy, systemic
racism, limited and inadequate and unsafe living areas, more
polluted neighborhoods, and fewer job opportunities. So, in
addition to strengthening our response to the pandemic,
policymakers must make sure that our response does not
perpetuate these structural inequalities.
Mr. Perry, what consideration should a digital contract
tracing program take to ensure that minority communities are
served well by these app-based solutions? Do you have other
suggestions on how technology can be utilized in the
coronavirus crisis response to ensure that communities of color
are not further disadvantaged?
Mr. Perry. I have a couple of responses to that. One,
whenever we have a crisis, it presents an opportunity to build
capacity in areas that we currently don't have. One of the
weaknesses of many AI-enabled technologies is that they have
not been tested, vetted, or created by founders of color,
leading to all kinds of gaps and lapses, but it is clear that
we do need, in Black communities, levels of a technology
because we are disproportionately impacted.
The problem is that Black people, in particular, and Brown
people, are caught between a rock and a hard place. When they
do find out that they may have been exposed, they are
disproportionately in jobs and in occupations where they can't
take off 2 weeks from work, and they are unable to socially
distance for periods of time.
As I indicated in my written testimony, we see how
intergenerational housing concentrations in Cities like New
Orleans, Detroit, Birmingham, and all of these places where
Black people are concentrated, we know that there will be
exposure because of the underlying conditions that lead to the
rapid spread of the virus.
And so, for me, I am looking at this from a structural
perspective. Yes, we still need to maximize manual tracing and
use technological tools to help them along, but manual tracing
also relies on the quality of the tracing mechanisms; there is
a lot to be gained from still leaning on the manual tracing.
But I have to emphasize that when we are ultimately investing
in tech companies, firms, innovations, we need to use this as
an opportunity to build wealth, because that is the reason why
or the lack thereof of those things are the reason that we see
rapid spread in Black communities.
So, for me, it is not to say that it is too late to deploy
technology in a way that makes sense for Black communities, but
it is to say that if we don't advance manual tracing and if we
don't build capacity, then in the next inevitable crisis, we
will be right back in the same situation we are in today where
we can't respond to technology in a way that the alerts want us
to, and we won't build capacity in areas where we need it
built.
Chairwoman Waters. Thank you very much.
I yield back.
Chairman Foster. Thank you.
The gentleman from North Carolina, Mr. Budd, is recognized
for 5 minutes.
Mr. Budd. This hearing is fascinating, and very timely.
Professor Raskar, this question is for you.
We have now seen several countries use contact tracing as a
means to combat the spread of COVID-19. Is there any credible
evidence to show that these methods have had a practical impact
on mitigating the spread of the disease?
Mr. Raskar. Yes. We understand this is still very early
data, inventing this whole field of digital contact tracing. We
do know that manual contact tracing works for small outbreaks,
but for an outbreak of this scale, many experts have shown that
manual contact tracing simply doesn't scale. In fact, it
disproportionately affects people who are vulnerable because
they cannot be reached.
Digital solutions has some hope that they can get--I think
in some countries, we have seen an option of 10 to 15 percent
in just a matter of weeks, which is good news, and as Brian
McClendon explained very beautifully, that if these
technologies are truly deployed by tech players as tools and
APIs, in fact, they could be out there in many ways.
So my assumption is that, if we use a bottom-up as well as
a top-down solution--a top-down solution that will be like how
we respond to hurricanes. So, manual contact tracing would be
like a top-down solution. We tried to do the best we can, but
this storm is really about people interacting with people, so
we need a bottom-up solution as well where the participants, in
this case, people who are innocent, but could be spreading
diseases to others are participating in the system, and if the
infected and exposed individuals are willing to participate in
the system, then we can, in fact, combine this top-down and
bottom-up approach.
And right now in the U.S., we have 70 percent smartphone
penetration in the age group of 18 to 64. If 90 percent cell-
phone penetration [inaudible] Estimates up to 70 percent of the
world [inaudible] Manual contact tracing could be off-loaded to
the people who have smartphones.
Mr. Budd. Thank you. Continuing on with you, Professor,
these contact tracing apps pose an interesting dilemma. Digital
contact tracing involves a level of surveillance that can make
a lot of people uncomfortable, especially given the involvement
of large technology companies with spotty records on privacy.
What steps are developers taking to ensure these privacy
concerns are being addressed? I think some of you mentioned
earlier the low-energy Bluetooth, the Apple/Google protocols
and standards. I just wanted to see what is being done to
address these privacy concerns?
Mr. Raskar. That is a very good question. I think it is a
fantastic development, and I am glad the technology company
knows they should not be in the business of creating solutions
that could interfere with their for-profit business. So, as
great as that Apple/Google protocol is, if another technology
company tries to provide a solution which is already for-
profit, even if the exposure notification is not-for-profit,
simply the IP address of the same user can be used by the for-
profit company in multiple ways.
That comes into conflict with the original intention that
Apple and Google and PathCheck Foundation have. So we believe
that to make it trustworthy and inclusive, it should definitely
be open source available for everybody to scrutinize and also
collaborate. And ideally, it should be built by companies that
are nonprofit or have no other related for-profit businesses.
I think if we do that and create a national grid of this
app because we are going to see a mushrooming of these apps,
but at PathCheck Foundation we believe that we can all work
together in creation of a national grid, and that will also be
part of this National Pandemic Response System.
Mr. Budd. Thank you.
Just following up, I have been supportive of a robust
Federal data privacy standard to replace the confusing
patchwork of States' regulations and laws. Would the
implementation of a national standard be beneficial for the
States and for the consumers alike, who are trying to protect
their own data privacy?
Mr. Raskar. Definitely. I think that is necessary, but just
to be sure, we have to go well beyond GDPR or CCPU, one level
above that for privacy, which means that even the app company
doesn't know anything about the user.
Mr. Budd. Very good.
And I have a few seconds left, but I will yield back to the
Chair.
Thank you.
Chairman Foster. Thank you.
And the Chair will now recognize the gentleman from
Illinois, Mr. Casten, for 5 minutes.
Mr. Casten. Thank you, Mr. Chairman. And thank you so much
to all of our witnesses.
Really building on Mr. Budd's comments. I would love to
hear from Dr. Kuppalli. We have, globally, such a wide range of
how contact tracing is being done. Is it being done by cities?
Is it being done by counties? Is it being done by States? Is it
being done by countries? People, of course, move across
borders, but there is a logistical challenge with scaling this
up at the highest level.
Dr. Kuppalli, purely from an epidemiological perspective,
where do you think is the ideal level in government to
coordinate a contact tracing strategy?
Dr. Kuppalli. That is a wonderful question. Thank you very
much. So, as I said in both my written testimony and in my oral
testimony, I really think that we need a national plan to help
coordinate this entire response, and the linchpin of that is
contact tracing.
And when it comes to contact tracing, I think the important
thing would be to be have some sort of Federal plan for that,
that is then farmed down to the State and local governments.
And as we have all talked about, and especially in an
epidemic like this where the transmissibility of the disease is
so much and, like I said, for every contact, there are about 10
to 25 contacts that need to be traced, and that is just not
something a manual contact tracer can do, it is important that
we rely on our technologies to be able to augment what manual
contact tracers are doing. And I think that it is important
that it is a coordinated effort between our local and State and
Federal Governments, while leveraging our technology.
Mr. Casten. Thank you,
Shifting to Mr. McClendon, if I am following right, all of
the apps that we are talking about developing are essentially
backwards-looking. We are going to know who you were in contact
with. If we find out you had a positive test, we are going to
go back, and we are going to look through this historical
record.
But in the context of artificial intelligence, so many of
the really cool things we do with AI right now are predictive,
whether it is Waze predicting where the traffic is going to be
at 4:00 today as I tried to get to downtown Chicago, or the
stuff that we spend a lot of time with on this committee of
algorithmic trading, predicting what stock is going to be
surging or where there is going to be a short sale next week.
Is anybody thinking about what could be done on a
predictive level if you had these contact tracing tools? Could
you start predicting that this is a high-risk place to go,
where we should be going? Is anybody thinking there, or are
these purely backwards-looking apps at this point in the scope
of people's ambition?
Mr. McClendon. I think the challenge right now is that the
only way to really predict the future is to understand the past
pretty well. And so, manual contact tracing interviews, where
the contact tracing team interviews the infected or exposed
person, where that person has been, what they have done, that
is the kind of data that would be great to send into a machine
learning algorithm to then predict where hot spots might be in
the future.
But right now that data is obviously siloed due to privacy
issues, siloed due to county tools and State tools that are not
compatible with each other, and it is also not normalized.
To Dr. Kuppalli's comment, I think that there does need to
be a Federal guidance on how we should do things. I don't think
the Federal Government should own this problem, but I think
there should be clear leadership about, here are the tools and
processes that we recommend that States do and that counties do
and that contact tracing teams do to collect data effectively,
and then using data analysis that retains privacy and
differential privacy, we can start to generate these predictive
models.
Large-scale data collection is being used for predictive
things. If you look at the University of Washington's Institute
for Health Metrics and Evaluation (IHME) models, they are using
a large collection of data, including things like mobility
reports and when different States and counties have implemented
mask rules or lockdown or some of the other aspects.
They are able to see the effect of different policies and
are now predicting that if a State implements a mask
requirement today, they can save this many lives. For example,
in Kansas, if we all follow a mask mandate, we will save 250
lives by November 1st. That is the projection, and that
projection is made using data that has been collected over the
last several months.
Mr. Casten. Thank you. I have a lot more questions, but I
see I am out of time, so I yield back. Thank you so much for
your time.
Chairman Foster. Thank you. And I now recognize the
gentleman from Indiana, Mr. Hollingsworth, for 5 minutes.
Mr. Hollingsworth. Good afternoon, everybody. I am really
excited about the discussion that we have had already on this
panel, and I look forward to talking more about it and hearing
more about it.
My questions are for Dr. Kuppalli. Just more on the
biological, epidemiological level, here is what I want to know:
Given where we are today, what does a realistic success,
achievable success look like in the United States pre-vaccine?
And then I am going to ask you what it looks like post-vaccine.
What is achievable for where we are, given the number of
infections that we have today?
Dr. Kuppalli. That is a great question. Thank you for that.
Given where we are right now, we are not in a great place,
right? As I said, yesterday, we had 60,000 new infections in
the United States.
And I think one of the things that has been challenging
with this particular infection in the United States is that we
have what we like to call a patchwork type of system. Every
State, every local municipality is doing its own thing, and
that is making things very challenging in terms of trying to
contain this outbreak.
Additionally, we are still having challenges with things
that we were having challenges with 5 months ago. Testing is
still a problem. Contact tracing is still a problem. Isolation
and quarantine is still a problem. If we don't get those things
fixed, we are not going to get this outbreak under control. So,
we need to fix those things.
In terms of where we are today and where we are going, we
need to work on fixing those things. We need to work on really
advocating for the public health interventions for which we
have been advocating.
Mr. Hollingsworth. To be more specific, sorry, are we to
the point, given the number of infections that we have had,
where we are basically managing down the number of infections
and the rate of infections to the hospital capacity system
until we get to the point of a vaccine, or is there a realistic
hope, even given where we are, that we can bring infections to
a grinding, very slow, very small number?
That is what I am trying to understand. I read a lot of
epidemiological studies that say, given where we are today, the
best we can hope for is managing infections down, managing
hospital capacity, until we get to a vaccine, not trying to
stop an outbreak in and of itself.
Dr. Kuppalli. I think, again, it is going to depend on how
well we uptake these public health interventions. We know these
public health interventions, what we call the nonpharmaceutical
interventions, we know they work. It depends on how well we can
get communities to buy into them in terms of how well the
outbreak is going to be contained.
And then, even going forward, once we have a vaccine, it is
going to not just depend on the efficacy of the vaccine, it is
also going to depend on how well we can get people to uptake
the vaccine. Studies I have been involved with have shown that
there is a lot of work to be done in trying to convince people
to take the vaccine. So, we can't say we are going to have the
vaccine, and this will be over.
Mr. Hollingsworth. I agree. I think that is right. And I
think that, frankly, at least in Indiana--I am not sure what is
happening elsewhere--people's desire to take significant steps
based on COVID-19 is declining, not increasing, given the fear
that they felt in March and April. And at least, we haven't
seen a significant resurgence, so there is a less palpable
fear.
I wanted to get to two other quick questions. I only have a
minute-and-a-half left. One, there is an obsession with kind of
the total number of infections since January 20th, this 2
million infections number. That to me seems strange, Dr.
Kuppalli, because I think what we care about are current
symptomatic or communicable infections at the present point,
right?
What matters is how many people have this today, either
know about it or don't know about it, either are symptomatic or
asymptomatic, but can transmit the disease to others, not those
who got it on January 20th and recovered February 15th and no
longer transmit the disease.
Are we tracking the right number in thinking about total
number of cases since genesis, or should we be thinking more
about, as you articulated, the number of infections per day and
the number of recoveries per day, to better understand the
evolution at present?
Dr. Kuppalli. I think all of those numbers are important. I
think we need to understand the total number of infections that
we see in our society, but we definitely need to understand
what is going on, on a day-to-day basis, right?
And so now particularly, we are not just seeing the numbers
go up, but we are seeing the percent positivity in the number
of cases going up. And that is really important, and also
looking at what is going on in the hospitals and the hospitals
being at capacity.
Mr. Hollingsworth. Perfect. Can you clearly tell me, what
is the clinical definition of ``recovered'' in this infection
with COVID-19? Is it the lack of communicability? Is it not
showing symptoms? What does ``recovered'' mean, explicitly?
Dr. Kuppalli. According to the current guidelines and
criteria, ``recovered'' means that you no longer have symptoms.
Mr. Hollingsworth. Okay. Thank you so much.
Dr. Kuppalli. Thank you.
Chairman Foster. Thank you.
The gentlewoman from North Carolina, Ms. Adams, is
recognized for 5 minutes.
Ms. Adams. Thank you, Mr. Chairman. And thank you to our
witnesses today.
According to a 2019 Pew Research study, 81 percent of
Americans owned a smartphone, up significantly from just 35
percent in 2011. The same study found that while Whites,
Blacks, and Hispanics own smartphones at roughly the same rate
overall, smartphone ownership rates are significantly lower for
older and low-income Americans. These communities of color have
also been disproportionately impacted by the pandemic, and they
are at greater risk of contracting COVID-19 and have a much
higher likelihood of being unable to recover.
So, Mr. McClendon, what is the benefit of a voluntary app-
based digital contact tracing system when the most vulnerable
at-risk populations are also the least likely to be able to
engage with the technology?
Mr. McClendon. That is a very fair question. With some of
the apps in question, you have a non-tech participation
capability, but this digital contact detection is one where you
really do need a phone, preferably a smartphone, to
participate.
And I think the only argument that can be made is that if
we deploy these widely, the infection rate overall will go
down, and there will be much less opportunity for anyone to get
infected.
To the question about--we are in the second wave right now,
and contact tracing is going to be very hard to solve the
problems that we are seeing as in many States it ramps up quite
high. But if we can get the disease down to a slow enough
level, we can actually drive it down so, effectively, there is
little to no exposure for anyone, and digital contact tracing
can really help with that.
And I think it is important to understand that, to the
prior question, wearing a mask can greatly reduce the
transmission. And by itself, a single app can, I think, reduce
the R-factor from well above one to below one and eventually
negate the disease. Masks are something that everybody can
participate in, and I am a strong believer that we should be
promoting masks more effectively.
Ms. Adams. With the amount of data being collected and
potentially analyzed by AI, what type of privacy and exposure
notification and contact notification is there?
Mr. McClendon. In the case of this specific Google/Apple
implementation, it is extremely private. There is no
information exposed at all unless you are infected and unless
you choose then to provide data to a service. And even then,
that is not part of the analysis.
The only interesting data to analyze is who then talks to
the contact tracers, what that interview yields as far as
number of contacts they have made, what kind of symptoms are
they showing, what is the progress of their disease, and
monitoring those people.
So, the manual contact tracing team has a lot of work to
do. And just to reiterate what everybody else has said, we need
more and better staffing and we need people in the local
community to work, because it is an empathetic, communicative
experience, and we need the best data we can collect, and for
that we need people of the community.
Ms. Adams. Okay. Thank you.
Mr. Perry, do you foresee any risk for misuse of the mass
collection of health data on low-income people of color in the
future, and how important is it for us to limit data
collection, if you think that?
Mr. Perry. Oh, yes, we should be prepared for racial bias
that will negatively impact Black and Brown communities. We
have already seen the risk of use-biased algorithms in
healthcare. United Health, using a medical algorithm, steered
Black patients away from getting higher-quality care. And
criminal justice software used to forecast the risk of
reoffending incorrectly marks Black defendants as future
criminals at twice the rate of White defendants.
So, again, what I said early in my verbal testimony was
that the AI for contact tracing runs the same risk if we do not
thoroughly vet these products that are going out. And so we
don't know necessarily, but the way I hedge that not knowing is
through inclusion.
We have to bring in, one, contact tracers, as was
mentioned, from the community, because if you don't establish
trust, these products will not work well. But more importantly,
we have to invest in Black and Brown firms, because they are
just more sensitive to the algorithms and functions that often
lead to a bad outcome.
Ms. Adams. Thank you.
Dr. Kuppalli, what does meeting the mitigation rate versus
meeting the suppression goal rate for testing mean in terms of
empowering States, municipalities, and localities to safely
reopen? What does meeting those rates mean in terms of opening?
Dr. Kuppalli. That is a great question. Meeting the
mitigation rates--
Chairman Foster. Go quickly, please. We are at the time
limit.
Dr. Kuppalli. Okay. So, basically, there are different
rates that we look at for when we are in different phases of
the epidemic, and we need to scale up testing quite a bit to
meet those rates.
Ms. Adams. I yield back, Mr. Chairman.
Chairman Foster. Thank you.
The gentleman from the Lone Star State, Mr. Taylor, is
recognized for 5 minutes.
Mr. Taylor. Thank you, Mr. Chairman. I appreciate that. I
appreciate being here with all of you. I think this is an
important conversation. And I think this epidemic has certainly
brought up to me the importance of having consistency in our
vital statistics accounting. We have had--I see some heads
nodding.
I served in the State legislature for 8 years before I came
to Congress, and we came into problems around counting data. We
would have different counties counting flu deaths at different
rates. And when we kind of dug down and started figuring it
out, we realized, well, one JP was doing it one way; another JP
was doing it another way. And so, we started working on trying
to come up with a State standard.
If you can believe it, in 2020, there is not a Federal
national standard for collecting vital statistics. And that, in
turn, is leading to disparate answers in terms of the way
localities and then, in turn, States are reporting.
And then we, as Federal policymakers, are trying to figure
out, how do I deploy assets? Where are my problems? And because
the data is being collected differently, you might not realize
you have a problem somewhere where you do have a problem. You
might think you have a problem somewhere that you don't have a
problem.
And so I am working, on a bipartisan basis, with
Congresswoman Shalala, my colleague from Florida, to work on
the beginnings of a study--an appropriations bill, so we will
see this, members, hopefully later this month--to start to
study a national standard for vital statistics collection.
And I will just share one issue that I will just take it
home to Collin County, where I live. We were not counting--we
were actually originally counting people who were serologically
positive as a COVID case, and then we were told, don't count
that as a COVID case when you report your accounting.
Mr. Perry, I see you nodding and smiling, so maybe you have
a reaction to what I am talking about or maybe you have seen
this in the field yourself.
Mr. Perry. No. I will just say, as a person who works at a
think tank, we need consistent, reliable data, and that often
is not had because there is a lack of standards.
So I would fully support any kind of legislation that led
to standardizing these vital records because, without it, we
really don't have the information to address the problem well.
I can't really add much more, but we need consistent reporting
on these issues.
Mr. Raskar. If I could add to that, Mr. Taylor, I think
data is so important, as Andre said. At the same time, data is
valuable only if it can be triangulated from multiple sources
so that it is really valid.
Like you said, if you just do the collection in a top-down
manner, you don't know if it is going in the right way. So, if
it is also bottom-up, citizens are participating, counties are
participating, communities are participating in it in some
other way, then the data becomes more valuable and more
accurate.
Mr. Taylor. Sure. Another example: At home, we originally,
if we had a husband and wife and the husband tested positive
and the wife had symptoms, we would assume that they were COVID
positive, but we wouldn't report them as COVID-positive because
we didn't want to use a test on them unnecessarily unless they
needed hospitalization.
The County was then told by the State of Texas, actually,
start counting that person as a COVID-positive, report that as
a positive case. So, my County then showed a spike in cases
because they took a bunch of people who had previously not been
reported as positive and then reported them as positive.
But, again, this was a County decision that was then
overridden by a State decision. And what I would like to see is
a consistency in the reporting of data.
And I am glad to see so many people nodding as a sign that
it is--when I talked to Congresswoman Shalala, Secretary of
Health and Human Services, she immediately understood that,
wow, I need consistent vital statistics. We need to have a
national standard so that my county is not guessing on what to
do; every county knows what to do, and it is a standard.
Anybody else want to comment on that?
Mr. Raskar. The point you are making about a national
standard is useful not just for hindsight, what has happened in
the past, but also for insight, what is going on right now, and
to the point made earlier about the prediction, about the
foresight as well. And again, for that, we need a triangulation
of data.
Because if we do have the development of a national weather
service here, it will look at not just the past hurricanes,
what is going on today, but use models to start predicting them
going forward.
For this virus, now we have heard that different mutations
are virulent in different ways. And responding to them in old-
fashioned ways, like the mechanisms we used a week ago or a
month ago are not as useful as what we would see them a month
or even 6 months from now.
So, whether it applies to the common cold or other public
health challenges or dealing with this pandemic, I totally
agree that a national standard of that kind would be extremely
valuable.
Mr. Perry. May I just add one very brief--Ramesh is
absolutely correct. This is an opportunity for us to really
rally behind the data. If you really want to get people
involved in a very mass way, you have them participate in the
collection. And so this is just an opportunity to really
exercise that principle that we are all in this together.
Mr. Taylor. Sure.
Thank you, Mr. Chairman. I yield back.
Chairman Foster. Thank you.
And I will abuse my chairman's prerogative to point out
that also in the upcoming appropriations will be to once again
have the House hopefully vote to repeal the ban on a unique
patient identifier. This is simply a database key that would
allow you to uniquely identify medical records, which is a very
important issue. And I was very proud we got a strong
bipartisan vote to repeal that ban in the House. And, of
course, the Senate did what the Senate does, which is nothing.
But I will just flag that for Members. It is an important
thing, really, the starting point in getting consistent
national records is to just be able to identify who this person
was.
And now, the Chair will recognize Representative Garcia of
Texas for 5 minutes.
Ms. Garcia. Also, from the Lone Star State, Mr. Chairman.
Chairman Foster. Also from--I am sorry to have missed that.
Well, you don't have the flag behind you, right?
Ms. Garcia of Texas. I guess I have that set up downstairs
when I do videos, not when I am here on calls.
But thank you for this very important hearing at such a
critical time because, as some people have noted, it is really
not just about what we need to do today, but in terms of
standards and requirements or guidance, depending on which we
choose to operate under. It is about future pandemics, future
events, so that we can be better prepared. So, thank you very
much.
And I want to start with Mr. McClendon. I just want to make
sure I understood when you were responding to the ranking
member on his question about who uses the app and what
information is picked up, everybody else has to be using the
app for anything to talk to from your app, correct?
Mr. McClendon. Yes, that is correct. Only data from
participants who have installed the app and opted in will have
their data recorded within the phones or, in the case of an
infected person, uploaded. If they don't have the app
installed, they are completely outside the system.
Ms. Garcia of Texas. So, for this to really work, sort of
like masks, everybody has to wear a mask, everybody has to use
the app, correct?
Mr. McClendon. I would like for that to happen, and I will
endeavor very hard to market that message to convince
communities and States to do that.
Ms. Garcia of Texas. I just wanted to make sure that I was
clear on that because now I want to switch to the good doctor.
Dr. Kuppalli, it has just been so hard to get everybody to
wear a mask. I have some of my own colleagues on one of my
committees who all refuse to wear a mask, except two who are
now wearing them. It has become politicized.
How are we going to convince people to sign up to an app
when we have not been able to convince them to wear a mask?
That really is the reality.
Dr. Kuppalli. I think that is a great question. I think
that, first off, we need to stop politicizing this pandemic,
and we really need to let the science lead the way. We are
talking about human lives.
Second, I think that we need to really be clear about the
privacy safety that we are going to have when it comes to using
these apps. I think that is really important. And I think that,
as we have seen in other places that have been able to
successfully contain COVID, one of those things is also people
having good trust with not just the people who are doing the
work, but also in terms of the governments and the people who
are working on this work.
I think that is all really important. One of the things
that is extremely important that we need to continue to work on
is also engagement, right? And so, engagement is going to be
important. Trust is going to be important, and making sure we
protect people's privacy is going to be important in getting
people to uptake these types of things.
Ms. Garcia of Texas. Mr. Raskar, I wanted to ask you a
question related to that. I have been struck that you have used
the word several times that it should be a ``nonprofit'' or
that it should--and also, it should be available inclusively to
all communities.
Tell me about that. Why would you suggest that it needs to
be an app that would be distributed through a nonprofit rather
than a for-profit?
Mr. Raskar. I think that we have great companies, Big
Technology companies doing great work. I worked at Facebook for
2 years. I was on an Apple privacy team last year. And all of
these companies have great intentions, but derivative companies
out of that use the metadata in nefarious ways. You may not
realize that every time you go to--
Ms. Garcia of Texas. No, I realize that. I just want to
hear you say it for the rest of the world to hear it.
Mr. Raskar. Exactly.
Ms. Garcia of Texas. But I just want to focus on this
particular app. Do you think it is critical that it be done
through distribution of sorts through a nonprofit?
Mr. Raskar. Yes, exactly.
Ms. Garcia of Texas. Make sure everybody can get it?
Mr. Raskar. Yes. It should be done by a nonprofit. It
should be open source. And just like public utilities, it
should be audited any time in a trustworthy manner. And make
sure it is inclusive.
And the challenge is that if it is not done by a nonprofit,
if it is done by a for-profit, they can use the metadata. Even
with the Google/Apple exposure notification app, they can use
the metadata from that app on their servers and they could be
tempted to use that in other ways.
So, if the same company has two apps, one app that is used
for contact tracing, which is supposedly private, and another
app that is doing something else, those two apps collectively
can create a much bigger picture of the user. And that is why
we think a nonprofit, which is committed to the mission,
committed to being open source, committed to collaboration,
committed to being audited, should be the only ones deploying
such solutions.
Ms. Garcia of Texas. And it has to be just to augment the
use of community people, or in my community, people who are
used to doing the community health work neighborhood by
neighborhood.
Mr. Raskar. Exactly. And the important part of that is the
development of the app, not just how the app works right now.
It should also be driven by the mission, which is, as Brian
McClendon said, getting the exposure notification is just the
beginning.
The case management, the interviews, the monitoring, the
support is what we do at PathCheck Foundation. That is why we
don't use a hybrid approach where we can offload a lot of the
work for manual contact tracing through the app that is 70
percent of the population, but then the rest of the process
still requires manual contact tracing operations.
Ms. Garcia of Texas. Great. Mr. Chairman, I see that I have
run out of time. I did have a question for Mr. Perry, but I
will submit it in writing. Thank you so much. I yield back.
Chairman Foster. Thank you.
The Chair now recognizes the gentleman from the Buckeye
State, Mr. Gonzalez, for 5 minutes.
Mr. Gonzalez of Ohio. Thank you, Mr. Chairman. And thank
you for recognizing the great Buckeye State. And thanks for
this important hearing, on a very interesting topic certainly,
and one that I think we are all still trying to wrestle with,
frankly.
I want to start my questions with Mr. McClendon. You have
stated--and I agree--that it is a marketing challenge, but I
would argue that it is an enormous marketing challenge, asking
sort of everyday Americans to trust two entities, Big Tech and
Government, that naturally we don't trust, and I think for good
reason, unfortunately.
My first question is, and you have sort of alluded to it,
but what is the minimum amount of adoption you could get and
still have an impact, knowing I don't believe you can get
anywhere near perfect. You are not going to get 100 percent.
People are just not going to opt in.
So, tell me a little bit about that, please.
Mr. McClendon. Right. So, if it was perfect, according to
the statistic I just heard, 18- to 65-year-olds have 90 percent
penetration. So, in that category, even if it were perfect, we
would only discover 81 percent of exposures, because both
parties have to have it installed.
The square problem is a big deal. So, 40 percent only gets
you 16 percent of exposures. I am hoping for 50 to 60 percent
where we discover 25 to 36 percent of exposures, and those
become those inbound phone calls.
And, as far as how to do that, I think we really need to
get communities' buy-in. Whether it is the State that is the
official agency that publishes the app, we need every town and
university and church and synagogue to convince their members
to participate in this process, because if you get densities of
installed base, that works, right?
If you could get 80 percent of a university community to
install this app, then that works within the community. All of
the exposures that occur within the university will be
discovered much more quickly, and the ones that happen in the
rest of the town, maybe not so much.
But I think there is an opportunity for individual
communities to bring their members in and convince them to
install the app. We really need this to be--like wearing masks,
it should be a community response.
Mr. Gonzalez of Ohio. I think you are right, it is going to
be community by community. I can just tell you for a fact that
my most rural counties are not in. They are not going to go for
contact tracing, signing up for the app. And I don't blame
them, frankly. So, I think a different strategy would probably
make sense for them.
Mr. Perry, I want to shift to you. You mentioned something
that I think is really important, which is, look, we talk about
who is most vulnerable here. My fear from the beginning with
this virus is that it would accentuate our biggest
inequalities, and I think it has. It has shined a light on the
fact that those who are most vulnerable are having the worst
outcomes with the virus, I think.
And when I think of how we can get ahead of that as the
outbreak continues, I think of sort of how do we protect our
low-income wage workers, our minority populations, those who
can't socially distance, either at work or in the home.
I am Hispanic. My family was born in Cuba. This beautiful
lady here over my shoulder was my grandmother, who lived with
us basically until she passed away. So, multigenerational
households don't distance as well, just by nature of the living
conditions.
With that in mind, one thing that I have started working on
that we are going to drop here soon would be a grant program
which would allow hotels, on a voluntary basis, to convert to
sort of COVID housing places, so that if you then get the virus
and you can't distance at home, you have an option of checking
into a hotel, which would be picked up by the government.
Just as a general concept, I would love to just get your
thoughts on that as a way to help our communities who can't
distance for a variety of reasons, giving them an option if
they pick up the virus to self-quarantine in a way that is safe
and won't affect the whole family.
Mr. Perry. As was mentioned, contact tracing must be paired
with other social supports. And if they are not, you are really
minimizing the impact of them.
And so, I agree that there are a couple of areas that are
critical. We need a housing response, and we need an employer
response.
Mr. Gonzalez of Ohio. Yes.
Mr. Perry. And if we are not talking about unemployment
insurance, if we are not talking about some type of workforce
housing, housing for the sick, housing for the vulnerable, then
the contact tracing mechanisms really will be limited in scope.
Mr. Gonzalez of Ohio. Thank you, and I yield back.
Chairman Foster. Thank you.
And I would like to thank all of our witnesses for their
testimony today.
The Chair notes that some Members may have additional
questions for this panel, which they may wish to submit in
writing. Without objection, the hearing record will remain open
for 5 legislative days for Members to submit written questions
to these witnesses and to place their responses in the record.
Also, without objection, Members will have 5 legislative days
to submit extraneous materials to the Chair for inclusion in
the record.
This hearing is now adjourned.
[Whereupon, at 1:27 p.m., the hearing was adjourned.]
[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]