ai
更新於 2021/07/30閱讀時間約 169 分鐘
Introduction
Artificial intelligence can help transform healthcare by improving diagnosis, treatment, and the delivery of
patient care. Researchers in academia, the private sector, and government have gained increasing access to
large amounts of health data and high-powered Al-ready computing systems. These powerful tools can greatly
improve doctors’ abilities to diagnose their patients’ medical issues, classify risk at a patient level by drawing on
the power of population data, and provide much-needed support to clinics and hospitals in under-resourced
areas. Al can also expand the operational capacity of different organizations, identify potentially fraudulent
health claims, and streamline manual tasks to boost productivity.
Much of this progress depends on sharing and utilizing large amounts of health data, which informs the
development of algorithms and machine learning. While the private sector has driven much of the innovation in
this field, the federal government and its partners can play a major role by both sharing their own data and
addressing challenges across the sector. HHS, private sector stakeholders, and academic and clinical
researchers can support this transformation by collaborating to apply Al both inside and outside of
government.
Researchers and practitioners now face multiple challenges in using Al to improve healthcare. These challenges
include limited access to data, poor data quality, concerns over data governance, and the ethical use of data,
including accountability and liability for data applications. Multiple stakeholders will need to work together to
address these challenges as new technical applications emerge.
The HHS Office of the CTO is now exploring the potential for a department-wide Al strategy to help realize the
potential of Al, and to establish policies and practices for facilitating Al development. This strategy comes in
tandem with the February 2019 “Executive Order on Maintaining American Leadership in Artificial
Intelligence” and The State of Data Sharing at the U.S. Department of Health and Human Services report, published
by the HHS Office of the CTO in September 2018."
The Roundtable on Sharing and Utilizing Health Data for Al Applications was designed to bring together HHS
leaders, and experts in Al and health data from other federal and state government agencies, industry,
academia, and patient-centered research organizations. The Roundtable began with calls to action by Mona
Siddiqui, the HHS Chief Data Officer, and Ed Simcox, the HHS Chief Technology Officer. Following these
keynotes, several speakers gave lightning talks on high-priority use cases for Al in healthcare, including
representatives from Verily, Amazon Web Services, Health Catalyst, and the Michael J. Fox Foundation. The
second part of the Roundtable transitioned into possibilities for Al strategies, where representatives from
Pfizer, the Center for Medicare and Medicaid Innovation (CMMI), the Government Accountability Office
(GAO), and the Assistant Secretary for Preparedness and Response at HHS outlined possible paths forward.
The day also featured a keynote address from Eric Hargan, the Deputy Secretary of HHS, who spoke about how
Al is being deployed within the department.
Throughout the day, Roundtable participants engaged in three in-depth breakout sessions. These sessions
focused on the following topics: (1) Identifying high-priority Al applications, (2) Improving and using data for Al
applications, and (3) Outlining key issues and objectives for an HHS Al strategy. The day concluded with a
presentation of highlights and actionable recommendations for HHS to advance its own Al strategy across the
department.
? The White House, “Executive Order on Maintaining American Leadership in Artificial Intelligence,” February 11, 2019,
Retrieved from
bttps: bitehouse.gov/presidential-actions/executive-order-maintaining-american-leadership-artificial-intelligence
3 U.S. Department of Health and Human Services Office of the Chief Technology Officer, ‘The State of Data Sharing at the
U.S. Department of Health and Human Services,” September 2018, Retrieved from
bttps: bhs. gov/sites/default/files/HHS StateofDataSharing 0915 pdf
4 of 20
Background and Key Concepts
Artificial intelligence has gained significant attention in recent years, particularly in the context of improving
health and well-being. The following section presents an overview of key concepts and terminology related to
artificial intelligence. For more information on the use of Al in healthcare, please refer to the briefing paper that
CODE developed in preparation for the Roundtable on Sharing and Utilizing Health Data for Al Applications*
At the core of artificial intelligence is the need for high-quality, clean and accurate data to fuel the development
of algorithms. Researchers emphasize the need for large, multifaceted datasets that allow machine learning
processes to incorporate as many factors as possible into analysis.° Artificial intelligence also demands clear,
accountable data governance with defined data elements and processes for ensuring data quality and access.®
Researchers are now attempting to tap large troves of health data - from electronic health records (EHRs) to
data collected from wearable devices and sensors - to improve diagnostics and predictive analytics. More
connected and interoperable data in the healthcare system will enable more transformative Al applications in
the future.
Most Al applications depend on algorithms, which describe a logical process that follows a set of rules.
Computers can be taught a series of steps in order to process large amounts of data to produce a desired
outcome. There are two forms of algorithm:
1. Supervised algorithms use ‘training datasets’ in which the input factors and output are known in
advance. Supervised processes can produce highly accurate algorithms because the ‘right answers’ are
already known. For example, scientists may feed a dataset of retina images into the algorithm in which
board-certified physicians have already identified and agreed upon diagnoses for each image.
2. Unsupervised algorithms are developed through a process whereby data is fed into the algorithm and
the computer has to ‘learn’ what to look for. Unlike the training datasets fed into supervised
algorithms, the data fed into unsupervised algorithms does not necessarily include the ‘right answers.’
Unsupervised algorithms are adept at finding clusters of relationships between observations in the
data, but may identify erroneous relationships because they are not instructed what to look for.
Machine learning is the process by which computers are trained to ‘learn’ by exposing them to data. Machine
learning is a subset of Al, and deep learning is a further subset of machine learning. Deep learning is the process
by which algorithms can learn to identify hierarchies within data that allow for truly complex understandings of
data. Natural language processing (NLP) refers to the subfield of machine learning designed to allow computers
to examine, extract, and interpret data that is structured within a language.
Augmented Intelligence is a form of Al that enhances human capabilities rather than replacing physicians and
healthcare providers. Augmented Intelligence has been embraced as a concept by physician organizations to
underscore that emerging Al systems are designed to aid humans in clinical decision-making, implementation,
*The Center for Open Data Enterprise, Briefing Paper: Roundtable on Sharing and Utilizing Health Data for Al Applications, April
2019, Retrieved from bttp://reports opendataenterprise ore/RI 1470 Briefing 2420 Paper pdf
> Nick Ismail, “The Success of Artificial Intelligence Depends on Data,” Information Age, April 23, 2018. Retrieved from
bttps: information-age.com/success-artificial-intelligence-data-17347160
® Rob Thomas, “The Road to Al Leads through Information Architecture.” VentureBeat, January 12, 2018. Retrieved from
bttps:-//venturebeatcom/?018/01/17 he-road-to-ai-leads-through-information-architecture
5 of 20
and administration to scale healthcare.’ In a 2019 white paper, Intel framed augmented intelligence as the Al
tools that perform specific tasks and are designed to support users, rather than replacing human experts.®
Stages of Al Development
Existing and potential Al applications vary in their level of sophistication, ranging from simple augmentation of
common tasks to full automation of systems and processes. Experts have begun categorizing these stages of Al
development. Among them, venture capitalist and author Kai-Fu Lee has characterized four “waves” of Al
applications:’
Wave 1 | Internet Al
Wave 2 | Business Al
Wave 3 | Perception Al
Wave 4 | Autonomous Al
Figure 1. “The Four Waves of Al”
Adapted from Kai-Fu Lee (2018)
According to Lee, the first wave of Al applications uses data generated on the Internet to better understand
the habits, interests, and desires of an individual or population.*° The second wave of Al applications uses
algorithms to inform and improve decision making. Clinical researchers, for example, can construct treatment
plans by using algorithms “to digest enormous quantities of data on patient diagnoses, genomic profiles,
resultant therapies, and subsequent health outcomes.”** The third wave of Al applications relates to the
proliferation of sensors and devices that collect data about the physical world such as smart watches and
virtual assistants. The fourth wave of Al applications integrates all previous waves and gives machines the
ability to make decisions without human intervention.?? This includes technologies such as automated vehicles.
7ACT | The App Association, “Appendix: Key Terminology for Al in Health,” Connected Health, Retrieved from
bitp://actonline org/wp-content/uploads/Artificial- Intelligence-in-Health-Appendix pdf
® Intel, “Intel's Recommendations for the U.S. National Strategy on Artificial Intelligence,” Retrieved from
bttps://newsroom,intelL.com/wp-content/uploads/sites/11/2019/03/intel-ai-white-paper pdf.
; er iu Lee, “The Four Waves of Al.” “Fortune, October 22, 2018, Retrieved from
bttp: fortune com/?018/10/22, artificial- intelligence- ak deep- learning- kai-fu-lee/
1 Kai-Fu Lee, “The Four Waves of Al.” Fortune, October 22, 2018, Retrieved from
bttp.//fortune.com/7018/10/9 2 /artificial-intelligence-ai-deep-learming-kai-fi-lee,
? kai-Fu Lee, “The Four Waves of Al.” Fortune, October 22, 2018, Retrieved from
bttp.//fortune.com/7018/10/9 2 /artificial-intelligence-ai-deep-learming-kai-fi-lee,
18 Kai-Fu Lee, “The Four Waves of Al.” Fortune, October 22, 2018, Retrieved from
bttp.//fortune.com/7018/10/9 2 /artificial-intelligence-ai-deep-learming-kai-fi-lee,
6 of 20
Al Applications in Healthcare
The following section outlines five general uses of Al in healthcare, including examples of existing and
near-term applications. These applications, while not mutually exclusive, are examples of cross-cutting themes
that emerged from the Roundtable discussions.
Roundtable participants emphasized the value of Al in improving clinician and administrative workflows.
Through NLP and other Al tools, machines can rapidly process EHRs and automatically transcribe medical
notes. Automation can free up time and reduce costs by eliminating manual data entry. Moreover, participants
noted that Al helps reduce administrative burden by correcting human errors in billing processes.
e Interpreting handwritten medical records. Amazon Web Services has used NLP to extract and
interpret handwritten notes and text from medical records.’* NLP is particularly well-suited to
deciphering physician input since EHRs do not follow a single, unified structure, yet contain important
information for understanding diagnostic trends and risk profiles of individuals.’
e Detecting fraud and improper payments. The Centers for Medicare and Medicaid Services (CMS)
uses statistical analysis to identify fraudulent and improper payments made to healthcare providers. In
2018, CMS determined that 8.12 percent of all Medicare payments were improper. In order to
address this problem, CMS employs a testing methodology called Comprehensive Error Rate Testing
(CERT) and uses Al to engage in predictive analysis of fraudulent and improper healthcare payments.
This process has saved the government approximately $42 billion, according to CMS.*6
Roundtable participants emphasized the value of using Al to connect patients with available resources and
care, especially in rural areas. Examples include:
e Providing patients with personalized healthcare recommendations. Sage Bionetworks launched
mPower as a study using surveys and phone sensors to track symptoms of Parkinson's Disease.” The
results can help patients, doctors, and caregivers better understand changes over time and the impact
of exercise or medication. Using artificial intelligence, data from mPower could also be used to develop
specific healthcare recommendations for patients.
e Creating virtual care programs for patients with chronic conditions. Verily Health's Onduo project,
which combines a smart device and mobile application, offers virtual care for people with type 2
diabetes. Onduo can measure blood glucose levels as well as provide information on nutrition and
medication management. The app also offers a coaching dimension that identifies lifestyle patterns and
gives patients feedback to improve their health.
* Rachel Arndt, “Amazon Technology Deciphers Text in Electronic Health Record,” Modern Healthcare, November 27, 2018,
Retrieved from
https: modernhealthcare.com/article/?0181127/TRANSFORMATIONO1/18112995 1/amazon-technology-deciph
4 Mike Miliard, “EHR Natural Language Processing Isn't Perfect, but It’s Really Useful,” Healthcare IT News, May 18, 2017.
Retrieved from ttps: bealthcareitnews. com/news/ebr-natural- language-processing-isnt-perfect-its-really-useful
46 Jack McCarthy, “CMS Snags $42 Billion in Medicare and Medicaid Fraud with Predictive Analytics.” Healthcare IT News,
July 22, 2016, Retrieved from
bttps: bealthcareitnews.com/news/cms-snags-47-pillion-medicare-and-medicaid-fraud-predictive-analytic
1” Sage Bionetworks, “mPower Parkinson Study and Public Researcher Portal,” Retrieved from
bttp.//sagebionetworks ore/research-projects/mpower-researcher-portal
7 of 20
e Expanding treatment access for rural populations. Through voice assistants and chatbots, Al has the
potential to improve and increase access to treatment in rural and other resource-constrained
environments. There is increasing evidence that Al-driven chatbots can address routine patient
questions and help doctors communicate with patients about their diagnosis and risk evaluations.®
Informing Population Health Management
Population health management (PHM) involves using population-level data to identify broad health risks and
treatment opportunities for a group of individuals or community. Al can contribute to PHM by combining,
synthesizing, and analyzing datasets from third parties with clinical or patient-generated data. For example,
researchers and health providers can use Al to aggregate longitudinal patient-generated data into larger
datasets that tell better stories about the incidence and prevalence of disease.
e Identifying at-risk populations. Al can be used to identify populations at risk for opioid abuse or
overdose. One population health management company, for example, integrates data on social
determinants of health and pharmacy claims to better understand the diverse “spectrum of opioid
abuse cases.”?
Improving Diagnosis and Early Detection
Diagnostic errors are a major problem in the healthcare system, with most patients experiencing at least one
diagnostic error in their lifetime.2° Al promises to help physicians accurately diagnose medical conditions in
their patients and treat disease at an early stage. Al algorithms draw upon large datasets on medical and social
determinants of health to better identify patterns and assist physicians in making diagnoses and developing
treatment plans.*?
Al can deploy technologies like image recognition, NLP, and deep learning to quickly detect life-threatening
conditions and assess risk for diseases like brain cancer or heart disease. Roundtable participants noted that it
may be more accurate to think of these applications as “augmented intelligence” rather than artificial
intelligence. The goal is not to replace the doctor’s clinical judgment, but to help physicians rapidly prioritize
patient symptoms and assess a range of diagnostic possibilities rather than ask patients a standard slate of
questions. Examples from the Roundtable include:
e Diagnosing diabetic retinopathy through image recognition. Al can help doctors diagnose diabetic
retinopathy, the world’s leading cause of blindness, by using image recognition. Researchers at Google
have trained algorithms to analyze images of retinas and diagnose this disease with over 90 percent
accuracy.**
18 Rachel Arndt, “Healthcare providers are teaming with chatbots to assist patients,” Modern Healthcare, December 8, 2018,
Retrieved from
bttps: modernhealthcare.com/article/70181708/TRANSFORMATIONO1/181709977//healthcare-providers-are-te
aming-with-chatbots-to-assist-patient
ZeOmega, “ZeOmega Launches Jiva Opioid Al to Stem Drug Abuse Crisis,” May 28, 2019, Retrieved from
bttp: eomega,com/zeomega-media-center/press-releases/?019-press-releases/7zeomega-launches-jiva-opioid-ait
22 The National Academies of Sciences, Engineering, and Medicine (2015) Improving Diagnosis in Health Care, Washington,
DC: The National Academies Press, Retrieved from bttps://daotore/10.1/276/91/94
1 | aura Landro, “The Key to Reducing Doctors’ Misdiagnoses,” The Wall Street Journal, September 13, 2017, Retrieved from
bttps: jcom/articles/the-key-to-reducing-doctors-misdiagnoses-1505276691.
22 Rohit Varma, “Using Artificial Intelligence to Automate Screening for Diabetic Retinopathy,” Ophthalmology Times,
October 22, 2018, Retrieved from
bttps: opbthalmologytimes.com/article/using-artificial intelligence-automate-screening-diabetic-retinapath
8 of 20
e Predicting brain deterioration using advanced machine learning. Al can be used to analyze a
diverse array of datasets and identify potential biomarkers that can indicate the onset of deterioration
in cases that range from concussion to coma”?
Drug development is a costly and time-consuming process. Al can help improve drug development through the
entire development lifecycle, from identifying gaps in current therapeutics to bringing new products to market.
Pharmaceutical researchers can use Al to sort through huge numbers of research papers and patents, as well as
comprehensive lists of chemical compounds and their properties, to suggest opportunities for drug
development. By analyzing the growing databases of biomarker data, they can then work to target different
treatments to different types of patients. And when drugs or other treatments reach the clinical trial stage, Al
can help match ideal patients to the right trials. Examples include:
e Improving clinical trial participation. HHS recently completed a “tech sprint” engaging external
experts, such as TrialX and Intel, to develop Al applications to match patients to appropriate clinical
trials. This kind of matching can help researchers find appropriate subjects for their studies and help
patients find potentially valuable treatments at the same time. “
e Supporting precision medicine. The National Institutes of Health (NIH) defines precision medicine as
“an emerging approach for disease treatment and prevention that takes into account individual
variability in genes, environment and lifestyle for each person.””? Researchers at startups like Lam
Therapeutics and Lantern Pharma are using supervised machine learning strategies to generate new
correlations between genomic biomarkers and drug activity to pilot individualized cancer treatments.
3 Suzanne Leigh, “Artificial Intelligence Aids Scientists in Uncovering Hallmarks of Mystery Concussion,” University of
California San Francisco Blog. Retrieved from
bttps: ucsfedu/news/2017/03/405297 6/artificial-intelligence-aids-scientists-uncovering-hallmarks-mystery-concu
200.
* Gil Alterovitz and Kristen Honey, “TOP Health’ Tech Sprint Unleashes the Power of Open Data and Al,” U.S. Department
of Health and Human Services Office of the Chief Technology Officer Blog, January 17, 2019, Retrieved from
https: bhs.gov/cto/blog/?019/1/17/top-health-tech-s print-unleashes-the-power-of-open-data-and-ai html
25 National Institutes of Health, “All of Us Research Program,” Retrieved from bttps://allofus nihov/
2 of 20
Health Data for Al Applications
Data is the foundation of all Al applications. During the Roundtable, participants identified a number of
high-value health data types that can be used for Al development. Building on the expert feedback gathered at
the Roundtable and subsequent research, this section provides a summary of six major health data types and
the challenges associated with their use.
High-Value Health Data Types
Clinical Trials Data
EHR Data
lol Data
Social Media Data
Registry Data
Survey Data
Vitals Data
10 of 20
Administrative and Claims Data generally comes from federal, state, and local government agencies as well as
healthcare providers and insurers. This can range from hospital discharge summaries to payment records
between insured patients and the healthcare system.
Clinical Data is a broad term that encompasses different kinds of data generated “in a clinical setting and
controlled by aclinician, as opposed to a patient or caregiver.’*”
e Clinical Trials Data includes registries and results from publicly and privately funded clinical studies.
Large amounts of data, including sensitive information about participants, are generated over the
course of a clinical trial. Researchers must obtain regulatory approval to collect and use this data.
e EHR Data is focused on individual patients, and can include information on routine checkups,
prescriptions, and medical procedures. Physicians can draw upon EHR data to develop individual
treatment plans and diagnose conditions. This data can also be combined with social determinants of
health to develop rich longitudinal profiles of individual patients and populations.
Genomic Data can include many different characteristics, ranging from full DNA sequences to individual DNA
variants.2° Recent advances have made it possible to analyze and store data on a person's entire genome
sequence. According to the National Institutes of Health, “Genome-based research is already enabling medical
researchers to develop improved diagnostics, more effective therapeutic strategies, evidence-based
approaches for demonstrating clinical efficacy, and better decision-making tools for patients and providers.”@”
Genomic data is considered highly sensitive and must be shared and used under carefully controlled conditions.
Patient-Generated Data includes “health-related data created and recorded by or from patients outside of the
clinical setting to help address a health concern.”*° This data type is becoming increasingly prevalent through
the creation of mobile health applications and wearable health devices.
e = loT Data includes data from mobile software applications, voice assistants, and wearable devices such
as smart watches. These technologies are part of the “internet of things,” or loT, which refers to the
growing system of machines and devices connected to the internet. This data is generally collected
under “terms of service” agreements and has the potential to provide important information on a
variety of critical health indicators, such as heart rate, sleep cycles, and diet.
e Social Media Data includes interactions on social media platforms such as Facebook and Twitter.
Researchers have noted that “Social media may offer insight into the relationship between an
individual's health and their everyday life, as well as attitudes towards health and the perceived quality
of healthcare services,” among other opportunities.*! Like loT data, social media data is collected under
“terms of service” agreements.
76 University of Washington Health Sciences Library, “Data Resources in the Health Sciences,” Retrieved from .
bttp://guides lib uw.edu/hsl/data/findclin
27 Office of the National Coordinator for Health Information Technology, Conceptualizing a Data Infrastructure for the
Capture, Use, and Sharing of Patient-Generated Health Data in Care Delivery and Research through 2024, January 2018,
Retrieved from bttps: healthit gov/sites/default/files/onc pghd final white paperpdf
78 BHG Foundation at the University of Cambridge, {dentification and genomic data, December 2017, Retrieved from
bttp: phgfoundationore/documents/PHGF-|dentification-and-genomic-data pdt
?? National Institutes of Health National Human Genome Research Institute, “A Brief Guide to Genomics,” Retrieved from
https: genome.gov/about-genomics/fact-sheets/A-Brief-Guide-to-Genamic
%° Office of the National Coordinator for Health Information Technology, Conceptualizing a Data Infrastructure for the
Capture, Use, and Sharing of Patient-Generated Health Data in Care Delivery and Research through 2024, January 2018,
Retrieved from bttps: healthit gov/sites/default/files/onc pghd final white paperpdf
3? Kevin Padrez et al, “Linking social media and medical record data: a study of adults presenting to an academic, urban
emergency department,” BMJ Quality & Safety, 2016, Retrieved from bttp//dxdoiore/10.1136/bmiqs-7015-004489
11 of 20
Social Determinants of Health Data represent “conditions in the environments in which people are born, live,
learn, [and] work...that affect a wide range of health, functioning, and quality-of-life outcomes and risks.”°4
Examples of these social determinants include access to transportation, education, and job opportunities as
well as the availability of food and housing options. Social determinants of health data can come from many
sources inside and outside of government, and can be used to better understand population health.
Surveillance Data is a broad term that encompasses the “ongoing, systematic collection, analysis, and
interpretation of health-related data essential to planning, implementation, and evaluation of public health
practice.”*?
e Registry Data includes data shared voluntarily by individuals that is generally focused around a
specific diagnosis or condition such as cancer or cystic fibrosis. This data can be used to track trends
and better understand conditions over time. According to the NIH, this data “belongs to the sponsor of
the registry and..may be shared with the participants and their families, and approved health care
professionals and researchers. However, personal, identifying information is kept private.”™*
e Survey Data includes the results of surveys and studies conducted to assess population health. This
data can help stakeholders monitor the spread of disease, track health insurance coverage across
regions, and assess trends in nutrition and exercise, among other uses.°°
e Vitals Data is generally collected and exchanged between local jurisdictions and the federal
government. This data represents “vital events,” such as births, deaths, marriages, divorces, and fetal
deaths.**
Roundtable participants identified numerous legal, cultural, and technical challenges associated with sharing
and utilizing health data for Al applications. While some of these challenges are specific to Al development,
others are general issues that impact all applications of health data.
Legal challenges
e Inconsistent restrictions on data use. Among the legal challenges, participants noted that health data
types have different legal and regulatory constraints on their use. For example, administrative and
claims data, clinical data, and certain types of surveillance data, such as survey data, can include
sensitive, individual-level information.*” The use of these data types is often restricted under existing
privacy frameworks such as HIPAA. Patient-generated data, such as data collected from mobile
applications and wearable devices, can also contain sensitive information about individuals ranging
from fertility treatments to mental health conditions. However, there are relatively few legal guidelines
that protect this emerging data type from misuse.
*2 Office of Disease Prevention, and Health Promotion, “Social Determinants of Health,” Retrieved from
bttps: bealthypeaple gow/7070/topics-objectives/tapic/social-determinants-of-health
33 World Health Organization, “Public health surveillance,” Retrieved from
bttps: bojnt/topics/public health surveillance/en
* National Institutes of Health, “List of Registries,” Retrieved from
bttps: nibgov/health-information/nih-clinical-research-trials-you/list-registrie
°° National Center for Health Statistics, “Resources for Survey Participants,” Retrieved from
https: cdcgov/nchs/nchs for you/survey participants htm
%¢ National Center for Health Statistics, “National Vital Statistics System,” Retrieved from
bttps: cdcogov/nchs/n Index btm
3” Amy Bernstein and Marie Haring Sweeney, Marie Haring, “Public Health Surveillance Data: Legal, Policy, Ethical,
Regulatory, and Practical Issues,” Morbidity and Mortality Weekly Report. Retrieved from
https: cdcgov/mmwr/preview/mmwrhtml/sué 10337 btm
12 of 20
Concerns about intellectual property. Roundtable participants also discussed the challenges of using
and sharing proprietary data and algorithms. Data collected in drug development trials, through
private-sector health surveys, or in other ways could benefit researchers and organizations in the
health sector developing Al applications, and proprietary Al models could be developed for greater
accuracy if the algorithms they use were shared. But while all parties stand to benefit from sharing
data and algorithms, it is difficult to balance that benefit against companies’ need to protect their
intellectual property for competitive advantage.
Cultural challenges
Underlying bias in health data. Some Roundtable participants highlighted concerns about bias and
lack of diversity in health data, which can have serious consequences when utilized for Al development.
As one expert notes, “If the data are flawed, missing pieces, or don’t accurately represent a population
of patients, then any algorithm relying on the data is at a higher risk of making a mistake.”?
Data silos and administrative hurdles. While HHS is developing more efficient ways for its operating
agencies to share data - for example, by developing common data use agreements (DUAs) - it is still
difficult for HHS agencies to share data with each other, and can be difficult for organizations outside
of government to obtain data from HHS. Roundtable participants said that tt can take 12 to 18 months
to get access to data from various agencies and offices within HHS. Culture changes are needed to
reduce the administrative hurdles that prevent timely data sharing.
Overly restrictive interpretations of HIPAA. Some Roundtable participants noted that fears about
violating HIPAA have created a risk-averse environment for data sharing. While HIPAA is intended to
protect patient privacy, it does allow data sharing and use under specific conditions.°*’ Participants
suggested that HHS could provide more guidance on what is and is not permissible under HIPAA in
different contexts.
Technical challenges
Limited technical capacity for data management and analysis. Roundtable participants inside and
outside of government noted the need for more staff with data science training. In particular, both
government and the private sector need more experts in Al and its application to health data and
issues.
Inadequate IT infrastructure for hosting and analyzing large datasets. Al applications require large
quantities of data, and large computational capacity, to train and test algorithms. The increasing
demand for real-time data adds to these technical requirements. Both HHS and the stakeholders that
work with the department may need to upgrade their infrastructure to meet these challenges.
Poor data interoperability. Roundtable participants flagged a number of challenges related to joining
and combining health datasets. Across the healthcare system, large amounts of data are structured in
different ways, preventing stakeholders from easily exchanging and integrating this information.
Participants attributed these challenges to a lack of common data standards and issues with
enforcement where standards do exist.
38 Dave Gershorn, “If Al is going to be the world’s doctor, it needs better textbooks,” Quartz, September 6, 2018. Retrieved
from bttps://q7.com/author/dgershgorna
% Office of the National Coordinator for Health Information Technology, “How HIPAA Supports Data Sharing,” Retrieved
from https: bealthit gov/topic/interoperability/how-bipaa-supparts-data-sharing
13 of 20
Recommendations and Actionable Opportunities
The February 2019 “Executive Order on Maintaining American Leadership in Artificial Intelligence” outlined a
number of strategic objectives for developing Al.° These include (text bolded for emphasis):
Executive Order Objectives
A. “Promote sustained investment in Al R&D in collaboration with industry,
academia, international partners and allies, and other non-federal entities to
generate technological breakthroughs in Al and related technologies and to rapidly
transition those breakthroughs into capabilities that contribute to our economic and
national security.
B. Enhance access to high-quality and fully traceable federal data, models, and
computing resources to increase the value of such resources for Al R&D, while
maintaining safety, security, privacy, and confidentiality protections consistent
with applicable laws and policies.
C. Reduce barriers to the use of Al technologies to promote their innovative
application while protecting American technology, economic and national security, civil
liberties, privacy, and values.
D. Ensure that technical standards minimize vulnerability to attacks from malicious
actors and reflect federal priorities for innovation, public trust, and public confidence
in systems that use Al technologies; and develop international standards to promote
and protect those priorities.
E. Train the next generation of American Al researchers and users through
apprenticeships; skills programs; and education in science, technology, engineering,
and mathematics (STEM), with an emphasis on computer science, to ensure that
American workers, including federal workers, are capable of taking full advantage of
the opportunities of Al.
F. Develop and implement an action plan, in accordance with the National Security
Presidential Memorandum of February 11, 2019 (Protecting the United States
Advantage in Artificial Intelligence and Related Critical Technologies) (the NSPM) to
protect the advantage of the United States in Al and technology critical to United
States economic and national security interests against strategic competitors and
foreign adversaries.”
Under the Executive Order, agencies funding and deploying Al are expected to use these government-wide
objectives to inform their work.
4° The White House, “Executive Order on Maintaining American Leadership in Artificial Intelligence,” February 11, 2019,
Retrieved from
bttps: bitehouse.gov/presidential-actions/executive-order-maintaining-american-leadership-artificial-intelligence
14 of 20
The HHS Office of the CTO is exploring the potential for a department-wide Al strategy to help realize the
value of Al within government, and to establish policies and practices for facilitating its development across the
health sector. Over the course of the Roundtable, participants outlined a number of recommendations for HHS
and other stakeholders that align with the objectives of the Executive Order. These recommendations and
related actionable opportunities are summarized below:
Al demands a robust information technology (IT) infrastructure, including data infrastructure, and staff
with the skills to apply it. Both infrastructure and expertise must be able to manage the large amounts of
data needed to support Al as well as the development of advanced Al applications.
Actionable Opportunities:
e Develop comprehensive technology investment plans to support organizational Al
strategies. Within HHS, this can include improving legacy systems for managing the data
that will fuel Al applications as well as IT modernization overall. It may also include
public-private collaboration to reduce the cost to government of technical
improvements.
e Build expertise in designing and implementing Al applications. Most federal
departments, including HHS, have limited organizational knowledge and experience
needed to develop Al applications with their data. HHS can bridge this gap through
hiring programs, public-private collaborations, or fellowship programs to bring Al
experts into government on a temporary basis.
e Create national testbeds for Al development. |ndustry has led the development of Al
applications, since commercial companies collect massive amounts of data and have the
resources, expertise, and technical capacity to apply tt. HHS and other agencies can help
remove these barriers to entry by creating collaborative environments where data and
code for Al applications can be tested, stored, and shared.
Concerns about privacy are paramount in the application of individual health data. While the use of
health data in EHRs and other medical records is governed by federal and state legislation, other data
types like loT data are only regulated through “terms of service” agreements developed by the private
sector. HHS and its partners will need to ensure that sensitive information is not disclosed or misused
when these data sources are applied. At the same time, researchers need to be able to use sensitive
data appropriately to develop new insights, diagnostic methods, and treatments. (The challenge of
balancing privacy with health data access will be the subject of the next Roundtable in this series.)
Actionable Opportunities:
e Provide guidance for de-identifying sensitive data. In order to protect privacy, health
data can be de-identified in different ways before researchers analyze It. Data scientists,
for example, have utilized codes that make it possible to link data about an individual
from different sources without revealing the person’s identity. HHS could provide
additional guidance on de-identification methods to protect data privacy and security
while encouraging its use for Al applications. This guidance may include updating
HIPAA’s rules around de-identification to meet modern demands.
15 of 20
e Develop credentialing systems for controlled access to sensitive health data. Some
sensitive data maintained by the federal government, such as collections of genomic
data, are now available only to qualified researchers only, under agreements that
prohibit them from sharing the data more widely. HHS could apply this model more
broadly and develop credentialing systems to determine who should have access to what
kinds of data and under what conditions.
Data for Al applications should be clean, timely, accurate, and standardized. Roundtable participants
identified numerous challenges related to integrating data and metadata from multiple sources.
Common standards for data collection and management can ensure that data and metadata are
accurate and consistent across healthcare applications, using a shared library of variables that are
applied across datasets. Standardization also ensures that datasets will be interoperable between
agencies within HHS or between HHS and its external partners. The current lack of interoperability is
amajor obstacle to applying data for Al development.
Actionable Opportunity:
e Adopt and expand existing common data models. Many participants noted the value
of adopting existing common data models for data and metadata, wherever possible.
Common data models standardize the way information is structured and make it easier
to use in combination with other data. Examples of existing common data models include
Patient Centered Outcome Research Network (PCORNet) model and the Observational
Medical Outcomes Partnership (OMOP) model. Participants also mentioned ICD-10,
which is a widely used coding system that could be expanded to improve health data
quality and interoperability nationwide.
Al applications are most effective when they can integrate large amounts of data about diverse facets
of health. However, researchers inside and outside of HHS often have difficulty accessing the data
they need. To share data from other sources, researchers must have DUAs that abide by HIPAA
regulations, whether HHS is sharing data with outside researchers or whether different operating
divisions within HHS are sharing data with each other. Drawing up and approving separate DUAs can
take time and administrative resources that are a burden on researchers and slow down the research
process.
Actionable Opportunity:
e Update and standardize data use agreements. A set of standard DUAs, using common
terms and conditions, could accelerate and simplify data sharing between operating
agencies within HHS. Revised DUAs could substantially reduce the time it takes for HHS
researchers to request and receive important, time-sensitive data: Finalizing DUAs can
now take up to 12 months. Standard DUAs for internal use within HHS could also
become a model for agreements between HHS and outside partners.
Participants also highlighted two areas for action that go beyond the Executive Order:
16 of 20
Increasingly, patients are generating data about themselves that can complement research and
clinical data? Patient-generated data includes data collected through sensors and wearables, and
through social media and mobile applications. Large amounts of this data are collected under “terms
of service” agreements and are being used by entities that are not covered by HIPAA.” As interest in
patient-generated data increases, there is a need for clearer rules around its appropriate use,
particularly in the context of Al development.
Actionable Opportunities:
e Develop specific guidelines for entities not covered by HIPAA. HIPAA applies to
traditional entities, such as health plans and healthcare providers, but does not apply to
software development and social media companies that may be collecting
patient-generated data with sensitive health information? While the HHS Office for
Civil Rights has developed several informational resources for health app developers,
entities that are not covered by HIPAA, and the individuals whose data they collect,
would benefit from further guidance and best practices on appropriate uses of
patient-generated data.
Many Al applications that use health data are being developed as a “black box” without clear
information about the algorithms and data being used to make decisions. Al strategies should include
steps to address concerns about accountability, bias, and oversight. This will require improved
transparency for both Al algorithms and the data that supports them.
Actionable Opportunities:
e Develop guidelines for mitigating bias in health-related Al applications. Some
Roundtable participants expressed interest in having HHS and its partners develop
guidance for identifying and reducing bias inAl applications. Participants also suggested
including an internal HHS review function to enforce such guidelines and help increase
transparency.
e Pilot and implement an FDA regulation for health-related Al applications.
Roundtable participants expressed similar concerns about a lack of quality assurance
and oversight for Al development in healthcare. The Food and Drug Administration
411) S. Department of Health and Human Services, “Patient-Generated Health Data,” Retrieved from
https: bealthit gov/topic/scientific-initiatives/patient-generated-health-data
#2 U.S. Department of Health and Human Services, Examining Oversight of the Privacy & Security of Health Data Collected by
Entities Not Regulated by HIPAA, June 2016, Retrieved from
bttps: healthit. gov/sites/default/files/non-covered entities report june 17 7016 pdf
#3 U.S. Department of Health and Human Services, Examining Oversight of the Privacy & Security of Health Data Collected by
Entities Not Regulated by HIPAA, June 2016, Retrieved from
bttps: bealthit gov/sites/default/files/non-covered entities report june 17 7016 pdf
* US. Department of Health and Human Services Office for Civil Rights, “Health app developers, what are your questions
about HIPAA?” Retrieved from bttps://hipaaqsportalbhs gov/
17 of 20
(FDA) has established a set of iterative, agile guidelines to precertify the rapid
development of Software as a Medical Device (SaMD).*? The FDA should continue its
efforts to adopt a revised regulatory framework for Al applications in which proposed
changes to algorithms must be disclosed to the FDA prior to market release. This
framework should take into account the ability of Al applications to adopt in real time
and provide ways to assess any risks from those changes.
e Publish metadata about data sources. Metadata provides information about the
structure of a dataset, the meaning of each variable within the data, the method of
collection, and other important characteristics. Metadata can also provide information
about the source of the data, the way it was collected, and other factors that may
indicate potential causes of bias. Publishing metadata will make it easier to assess
whether the data and the algorithms it supports are at risk of being biased in any way.
“5 US. Food and Drug Administration, “Developing Software Precertification Program: A Working Model,” June 2018,
Retrieved from
https: fdagov/downloads/MedicalDevices/DigitalHealth/DigitalHealthPreCertProgram/UCM611103. pdf
46 US. Food and Drug Administration, “Proposed Regulatory Framework for Modifications to Artificial
Intelligence/Machine Learning (AI/ML)-Based Software as a Medical Device,” April 2019, Retrieved from
https: fda gov/media/177535/download
18 of 20
Conclusion
While the promise of Al in healthcare is significant, a number of challenges can impede its successful
implementation. The Roundtable on Sharing and Utilizing Health Data for Al Applications was a first step to finding
solutions by identifying innovative examples of Al applications, high-value data types, and ways that all
stakeholders can contribute to the successful and appropriate use of Al.
In the two months since the Roundtable, HHS has demonstrated its commitment to exploring the use of Al
inside and outside of government. For example, HHS is moving forward with its “Reimagined — Buy Smarter”
program designed to use Al to conduct strategic comparative analysis of industry pricing to ensure that HHS is
saving taxpayers as much money as possible.4”7 HHS and CMS are also working to expand their cloud capacity to
manage the growing data assets that are critical to their daily operations and future Al applications.”
The Executive Office of the President has also advanced government-wide Al initiatives. In addition to the
February 2019 “Executive Order on Maintaining American Leadership in Artificial Intelligence,” the National
Science and Technology Council updated The National Artificial Intelligence Research and Development Strategic
Plan in June 2019. The Plan recommends that the federal government develop a coordinated approach to
maximize the impact of Al technology as it grows in scope. The Plan also proposes eight different strategies to
bolster Al development such as understanding the ethical, legal, and societal implications of Al, adopting
effective strategies for Al-human collaboration, and supporting the safety and security of Al systems.
This summary report presents research and diverse stakeholder input from the Roundtable on Sharing and
Utilizing Health Data for Al Applications that can inform the development of an HHS Al strategy. The report
outlines a number of ways that HHS can take action that align with the Executive Order and other
government-wide Al initiatives.
The same kinds of recommendations and actionable opportunities may be useful to the growing number of
stakeholders outside of government who are working to develop applications based on health data.
Private-sector companies, patients and their advocates, academic researchers, healthcare providers, and other
stakeholders will all play critical roles in the development of health-related Al in the months and years ahead.
CODE hopes that this report will provide context, perspective, and the beginnings of a framework for the
important work to come.
47 Amelia Brust, “HHS wants to use Al to help it ‘Buy Smarter,” Federal News Network, May 17, 2019, Retrieved from
hitps://federalnewsnetwork.com/ai-machine-learning-month/?019/05/hhs-wants-to-use-ai-to-help-it-buy-smarter,
8 Joseph Goedert, “CMS works with MarkLogic to expand cloud platforms,” Health Data Management, April 30, 2019,
Retrieved from https: bealthdatamanagementcom/news/cms-warks-with-marklogic-to-expand-claud-platform
*? The National Science and Technology Council, The National Artificial Intelligence Research and Development Strategic Plan,
June 21, 2019, Retrieved from
https: hitehouse.gov/wp-content/uploads/?2019/06/National-Al-Research-and-Development:-Strategic-Plan-2019
19 of 20
Acknowledgements and Appendices
The Roundtable on Sharing and Utilizing Health Data for Al Applications was
funded through a Patient-Centered Outcomes Research Institute (PCORI) ry
Engagement Award Initiative (12667-CODE). This Roundtable is part of the Corl
Open Data Roundtable Series: Sharing and Utilizing Data to Enhance and Protect °
Health and Well-Being funded through this award.
PATIENT-CENTERED OUTCOMES RESEARCH INSTITUTE
CODE would like to thank the HHS Office of the Chief Technology Officer for their partnership in co-hosting
this Roundtable series. We also thank the Multi-Stakeholder Advisory Committee for this series:
Lisa Bari, CMS Innovation Center, Centers for Medicare and Medicaid Services
Sohini Chowdhury, Michael J. Fox Foundation
James Craver, National Center for Health Statistics, Centers for Disease Control and Prevention
Gwen Darien, National Patient Advocate Foundation
Stephanie Devaney, Al! of Us Research Program, National Institutes of Health
Natalie Evans-Harris, BrightHive
Jason Gerson, Patient-Centered Outcomes Research Institute
Joel Gurin, Center for Open Data Enterprise
William Hoffman, World Economic Forum
Charles Keckler, Associate Deputy Secretary, HHS
Lisa Khorey, Allscripts Healthcare Solutions
Michael Seres, 11 Health
Mona Siddiqui, Chief Data Officer, HHS
Paul Tarini, Robert Wood Johnson Foundation
John Wilbanks, Sage Bionetworks
The following resources are included as appendices to this report:
e = List of Participating Organizations
e Roundtable Agenda
20 of 20
Roundtable on Sharing and Utilizing Health Data for Al Applications
PARTICIPATING ORGANIZATIONS
Academia
Dell Medical School, University of Texas at Austin is the graduate medical school of The University of
Texas at Austin in Austin, Texas. The school opened to the inaugural class of 50 students in the summer of
2016 as the newest of 18 colleges and schools on the UT Austin campus.
Duke Margolis Center for Health Policy’s mission is to improve health and the value of health care through
practical, innovative, and evidence-based policy solutions. Duke-Margolis catalyzes Duke University's
leading capabilities to inform policy making and implementation for better health and health care.
Harvard Law School, Petrie-Flom Center founding mission is to promote interdisciplinary analysis and legal
scholarship in the fields of Health Law Policy, Biotechnology, and Bioethics.
University of Maryland School of Public Health offers an unusual breadth of expertise to explore complex
issues through public health disciplines, and lenses as varied as engineering, arts and humanities, business
and public policy.
Stanford University, one of the world's leading teaching and research universities, dedicated to finding
solutions to big challenges and to preparing students for leadership in a complex world.
Hall Center for Law and Health, Indiana University Robert H. McKinney School of Law was established in
1987 to expand the curriculum and teaching of health law and provide opportunities for students.
Civil Society
ACT | The App Association represents more than 5,000 app companies and information technology firms
across the mobile economy. ACT advocates for an environment that inspires and rewards innovation, while
providing the necessary resources to help its members leverage their intellectual assets to raise capital and
create jobs.
American Medical Association (AMC) promotes the art and science of medicine and the betterment of
public health. AMC provides timely, essential resources to empower physicians, residents and medical
students to succeed at every phase of their medical lives.
Center for Data Innovation educates policymakers and the public about the opportunities and challenges
associated with data, as well as technology trends such as predictive analytics, open data, cloud computing,
and the Internet of Things.
Center for Open Data Enterprise (CODE) is an independent nonprofit organization based in Washington,
D.C. whose mission is to maximize the value of open government data for the public good.
Healthcare Leadership Council (HLC)}, a coalition of chief executives from all disciplines within American
healthcare, is the exclusive forum for the nation’s healthcare leaders to jointly develop policies, plans, and
programs to achieve their vision of a 21st century system that makes affordable, high-quality care
accessible to all Americans.
World Economic Forum is the International Organization for Public-Private Cooperation. The Forum
engages the foremost political, business and other leaders of society to shape global, regional and industry
agendas.
Government
The U.S. Department of Health and Human Services is a cabinet-level department of the U.S. federal
government with the goal of protecting the health of all Americans and providing essential human services.
The Agency for Healthcare Research and Quality's (AHRQ) mission is to produce evidence to make
health care safer, higher quality, more accessible, equitable, and affordable, and to work within the
U.S. Department of Health and Human Services and with other partners to make sure that the
evidence is understood and used.
The Centers for Disease Control and Prevention Center (CDC) works to protect America from
health, safety and security threats, both foreign and in the U.S. Whether diseases start at home or
abroad, are chronic or acute, curable or preventable, human error or deliberate attack, CDC fights
disease and supports communities and citizens to do the same.
The National Center for Health Statistics (NCHS), part of the CDC, compiles statistical
information to help guide policies to improve the health of Americans. Holds a biennial data
user conference; consult the NCHS website for date and location. NCHS disseminates data
and statistics online and in print.
Center for Medicare and Medicaid Services (CMS) administers the Medicare program and works in
partnership with state governments to administer Medicaid, the Children's Health Insurance
Program (CHIP), and health insurance portability standards.
The Innovation Center with CMS supports the development and testing of innovative
health care payment and service delivery models.
Food and Drug Administration (FDA) is responsible for protecting the public health by ensuring
the safety, efficacy, and security of human and veterinary drugs, biological products, and medical
devices; and by ensuring the safety of our nation's food supply, cosmetics, and products that emit
radiation.
Health Resources and Services Administration (HRSA) is the primary federal agency for improving
health care to people who are geographically isolated, economically or medically vulnerable.
The Immediate Office of the Secretary (IOS) is responsible for operations and coordination of the
work of the Secretary.
The National Institutes of Health (NIH) seeks fundamental knowledge about the nature and
behavior of living systems and the application of that knowledge to enhance health, lengthen life,
and reduce illness and disability.
NIH Clinical Center mission is to provide hope through pioneering clinical research to
improve human health. The center has an Individual and collective passion for high
reliability in the safe delivery of patient-centric care in a clinical research environment.
The Office of the Assistant Secretary for Preparedness and Response leads the nation’s medical
and public health preparedness for, response to, and recovery from disasters and public health
emergencies.
The Office of the Chief Technology Officer (CTO) provides leadership and direction on data,
technology, innovation and strategy across the HHS. Areas of focus include promoting open data
and its use to create value, driving more efficient operations through technology utilization, and
coordinating innovation strategy across the Department to improve the lives of the American
people and the performance of the Department.
Office of Inspector General (OIG) mission is to protect the integrity of Department of Health &
Human Services (HHS) programs as well as the health and welfare of program beneficiaries.
Office of the National Coordinator Improve the health and well-being of individuals and
communities through the use of technology and health information that is accessible when and
where it matters most.
U.S. Government Accountability Office (GAO) examines how taxpayer dollars are spent and provides
Congress and federal agencies with objective, reliable information to help the government save money and
work more efficiently.
Healthcare Insurers and Providers
HCA Healthcare is committed to the care and improvement of human life. HCA follows a vision of
healthcare the way it should be: patient-centered, constantly evolving and constantly improving, practiced
with integrity and compassion.
UnitedHealth has a mission to help people live healthier lives and make the health system work better for
everyone and are working to create a system that delivers high quality care, responsive to the needs of each
person and the communities in which they live.
New York Presbyterian Hospital is one of the nation’s most comprehensive, integrated academic health
care delivery systems, dedicated to providing the highest quality, most compassionate care and service to
patients in the New York metropolitan area, nationally, and throughout the globe.
Nonprofit & Philanthropic Organizations
Michael J. Fox Foundation is dedicated to finding a cure for Parkinson's disease through an aggressively
funded research agenda and to ensuring the development of improved therapies for those living with
Parkinson's today.
Robert Wood Johnson Foundation (RWJF) is the nation’s largest philanthropy dedicated solely to health.
RWJD supports research and programs targeting some of America’s most pressing health issues—from
substance abuse to improving access to quality health care.
Patient Advocacy
National Patient Advocate Foundation (NPAF), the advocacy affiliate of the Patient Advocate Foundation,
represents the patient voice, both the powerful stories of individuals and the collective needs of the
community. The NPAF’s primary objective is to prioritize the patient voice in health system delivery reform
to achieve person-centered care.
Patient-Centered Outcomes Research Institute (PCORI} helps people make informed healthcare decisions,
and improves healthcare delivery and outcomes, by producing and promoting high-integrity,
evidence-based information that comes from research guided by patients, caregivers, and the broader
healthcare community.
Patient Privacy Rights’ purpose is to honor and empower the individual’s right to privacy through personal
control of health information wherever such information is collected and used. They educate, collaborate
and partner with people to ensure privacy in law, policy, technology, and maximize the benefits from the use
of personal health information with consent.
Private Sector
Allscripts is a leader in healthcare information technology solutions that advance clinical, financial and
operational results. Its innovative solutions connect people, places and data across an Open, Connected
Community of Health.
Amazon Web Services is a subsidiary of Amazon that provides on-demand cloud computing platforms to
individuals, companies and governments, on a metered pay-as-you-go basis.
Apple is an American multinational technology company headquartered in Cupertino, California, that
designs, develops, and sells consumer electronics, computer software, and online services.
Blackfynn helps the neuroscience and neurology communities make optimal use of data by powering an
innovative platform that integrates and puts complex data in context.
Booz Allen Hamilton provides management and technology consulting and engineering services to leading
Fortune 500 corporations, governments, and not-for-profits across the globe. Booz Allen partners with
public and private sector clients to solve their most difficult challenges through a combination of consulting,
analytics, mission operations, technology, systems delivery, cybersecurity, engineering, and innovation
expertise.
Epic hires smart and motivated people from all academic majors to code, test, and implement healthcare
software that hundreds of millions of patients and doctors rely on to improve care and ultimately save lives
around the globe.
Flatlron Health is a healthcare technology and services company focused on accelerating cancer research
and improving patient care. Their mission is to improve lives by learning from the experience of every
cancer patient.
Geisinger Healthcare is a coordinated intersection of services and providers - primary care and specialists,
hospitals and trauma centers, insurance, medical education and research. Geisinger has expanded and
evolved to meet regional needs and developed innovative, national programs in the process.
Google LLC is an American multinational technology company that specializes in Internet-related services
and products, which include online advertising technologies, search engine, cloud computing, software, and
hardware.
Google Al conducts research that advances the state-of-the-art in the field, applying Al to products and to
new domains, and developing tools to ensure that everyone can access Al.
Health Catalyst is dedicated to enabling health care organizations to fundamentally improve care by
building the most comprehensive and fully integrated suite of healthcare data warehousing and process
improvement solutions available.
IBM Research is a community of thinkers dedicated to addressing some of the world’s most complex
problems and challenging opportunities for the benefit of all. They are one of the world’s largest and most
influential corporate research labs, with more than 3,000 researchers in 12 labs located across six
continents.
Intel Corporation invents at the boundaries of technology to make amazing experiences possible for
business and society for everyone. Leading on policy, diversity, inclusion, education and sustainability, we
create value for our stockholders, customers, and society.
KB Stack Consulting helps government and non-profit organizations develop creative strategies for using
data, evidence, and innovation to improve the impact of government social programs.
Mathematica Policy Research is dedicated to improving public well-being by bringing the highest standards
of quality, objectivity, and excellence to bear on public policy. It advances its mission through objective,
evidence-based standards, superior data, and collaboration.
Microsoft is an American multinational technology company with headquarters in Redmond, Washington. It
develops, manufactures, licenses, supports and sells computer software, consumer electronics, personal
computers, and related services.
Mpirica Health is a digital health company that uses machine learning, backed by a robust methodology, to
scores hospitals and surgeons based on objective clinical outcomes. Our cloud-based platform and API helps
patients and payers, especially self-insured employers, avoid surgery risks and costs.
Oncology Analytics, Inc. provides health plans with an evidence-based, technologically driven approach to
utilization management, which is purpose-built for oncology. They provide a technology enabled service to
ensure cancer patients get the right treatment at the right time, and at an affordable price.
Pfizer’s purpose is grounded in their commitment to fund programs that provide public benefit, advance
medical care and improve patient outcomes. Their belief is that all people deserve to live healthy lives. This
drives their desire to provide access to medicines that are safe, effective, and affordable.
TrialX brings researchers and patients together, accelerating clinical research to find cures for millions.
They put the patient at the center and build solutions that work together seamlessly to meet the patient
where they are, helping them find the trials they are looking for and connect with investigators close by.
Verily Life Science’s mission is to bring together technology and life sciences to uncover new truths about
health and disease. As an independent company, they are focused on using technology to better understand
health, as well as prevent, detect, and manage disease.
Viz.ai’s world class team of technologists, doctors, executives, and advocates has developed an Al-driven
approach to care that is fast, effective, collaborative, and sustainable.
11 Health and Technologies have a complete care management platform that is the new gold standard in
healthcare for both patients and clinicians by combining Smart Technology with the world's first one to one
Patient Coach Program that provides both technical and emotional support to patients.
Roundtable on Sharing and Utilizing Health Data for Al Applications
U.S. Department of Health and Human Services | April 16, 2019
Purpose: Identify high-priority health applications of Al and key issues for an HHS Al strategy to address.
Registration and Light Breakfast
Welcome
Mona Siddiqui, Chief Data Officer, U.S. Department of Health and Human Services (HHS)
Opening Remarks
Ed Simcox, Chief Technology Officer, HHS
Structure of the Day
Joel Gurin, President, Center for Open Data Enterprise (CODE)
Lightning Talks: Emerging Applications of Al
Sohini Chowdhury, Deputy Chief Executive Officer, Michael J. Fox Foundation
Jason Jones, Chief Data Scientist, Health Catalyst
Vivian Lee, President of Health Platforms, Verily Life Sciences
James Wiggins, Senior Solutions Architect for Academic Medical Centers, Amazon Web Services
為什麼會看到廣告
3會員
60內容數
留言0
查看全部
你可能也想看
Google News 追蹤
*合作聲明與警語：
本文係由國泰世華銀行邀稿。
證券服務係由國泰世華銀行辦理共同行銷證券經紀開戶業務，定期定額(股)服務由國泰綜合證券提供。
剛出社會的時候，很常在各種 Podcast 或 YouTube 甚至是在朋友間聊天，都會聽到各種市場動態、理財話題，像是：聯準會降息或是近期哪些科
*合作聲明與警語：
本文係由國泰世華銀行邀稿。
證券服務係由國泰世華銀行辦理共同行銷證券經紀開戶業務，定期定額(股)服務由國泰綜合證券提供。
剛出社會的時候，很常在各種 Podcast 或 YouTube 甚至是在朋友間聊天，都會聽到各種市場動態、理財話題，像是：聯準會降息或是近期哪些科