AI in Medicine: Inclusion and Equity

(AiMIE) 2018

Agenda

August 22, 2018

CEMEX Auditorium 

Stanford Graduate School of Business

655 Knight Way, Stanford, CA


Morning Sessions


8:00am

Registration and Breakfast

9:00am

9:05am

Artificial Intelligence (AI) and Machine Learning (ML) in Medicine 101

Jonathan Chen

So-called AI technologies can automatically label every cat video on the internet and fill your social media feeds with the perfect clickbait to keep you distracted. Medicine is ripe for the application of AI, given enormous volumes of real world data and ballooning healthcare costs.

This talk will provide a brief introduction to demystify buzzwords from AI to big data to deep learning and machine learning, by drawing analogies to well-established medical tools like risk prediction scores and observational research, and contrasting why mastering Go and operating self-driving cars has not been the same as solving the unique challenges of medicine.

9:20am

Stanford Center on Poverty & Inequality Panel: A Big Data Approach to Reducing Poverty and Increasing Social Mobility in the United States

Moderator David Grusky

Panelists Geoffrey Cohen and Emma Brunskill

The U.S. has had extremely high rates of poverty for decades. The government spends almost $1 trillion per year, nearly one-fourth of the total federal budget, on means-tested programs to reduce economic hardship and improve social welfare. It is altogether unclear that we are securing anything approaching an optimal return on this investment. To the contrary, the U.S. compares unfavorably to other well-off countries on such key outcomes as poverty, economic mobility, and educational outcomes.

What accounts for this state of affairs? The premise of this work is that substantial headway can be made once we have built credible predictive models of poverty that tell us who is at risk and who will benefit from interventions. In conventional social science models of poverty, a shockingly small amount of the variability is explained, yet no one believes that poverty is a truly random affair. The field lurches from one flavor-of-the-day account to another, and fails to deliver the comprehensive model we need. Meanwhile, scholars study the problem within their independent disciplinary silos and seldom build synergistic collaborations. We propose to harness the power of machine learning to analyze new troves of big data at multiple disciplinary levels of analysis and develop the first powerful predictive models of poverty.

9:45am

Future of AI in Health and Equity: A Perspective from the National Academy of Medicine

Victor Dzau

Victor Dzau will discuss the current state of health disparities in the United States as well as the impact of social determinants on health outcomes. Furthermore, he will discuss how AI may transform health and medicine and highlight the potential for AI to address or exacerbate existing health disparities.

10:05am

A Stanford Perspective: Big Data and AI: Will They Reduce Disparities?

Mark Cullen

Despite the promise, the legacy of new technological advances in medicine has not typically been associated with uniform diffusion into practice. As a result, most of the initial benefit of these technologies is conferred to those who already enjoy the most advanced forms of care, widening rather than narrowing health disparities in the population. In this brief talk, Dr. Cullen will present a Stanford perspective on reasons for concern as well as optimism as the benefits of AI are realized by the medical and public health communities.

10:25am

Break

10:45am

Session: Levers and Possibilities for Inclusive AI in Healthcare

Introductions and hosted by Margaret Levi

10:50am

Automating Austerity - High-Tech Challenges to Wellness and Care

Virginia Eubanks

In her new book Automating Inequality: How High-Tech Tools Profile, Police, and Punish the Poor, Virginia Eubanks reports on three new automated decision-making systems having profound impacts on poor and working communities — and frontline care workers — across the United States. In this session, she will suggest lessons that automated eligibility in welfare, algorithmic resource matching in homeless services, and predictive models in child welfare offer to the sustainable and just practice of medicine.

11:10am

An Inclusive Moral Political Economy for Healthcare

Margaret Levi

In the inclusive moral political economy the value of equity and inclusion is foremost. How do we ensure that a majority of the population has access to the benefits of AI in healthcare, and do not bear a burdensome cost? How do we potentially achieve this?

11:30am

Panel: Public Health Implications 

Moderator Nirav Shah

Panelists John AuerbachDavid Buckeridge, and Chad Heilig

Issues include: infrastructure, priorities, needed reconciliations for incentives, predictive medicine and potential problems for preventive medicine; opportunities for the most out-of-date public health infrastructure systems i.e. at highest risk of failure (e.g. water supply in Flint) that might benefit from AI-facilitated tech-enabled solutions; current AI used today for surveillance, prevention, emergency response; and other public health priorities.

12:00pm

Closing panel discussion: Levers and Possibilities for Inclusive AI in Healthcare

Moderator Margaret Levi

Panelists Virginia EubanksNirav ShahJohn Auerbach, Chad Heilig and David Buckeridge

12:30pm

Lunch


Afternoon Sessions


1:30pm

Marc Tessier-Lavigne President, Stanford University

1:35pm

Session:  Legal and Institutional Challenges of Implementing Equitable AI in Medicine

Introductions and hosted by Roberta Katz

1:40pm

Mapping the Social Dilemmas of Law and AI

Mariano-Florentino Cuéllar

In law and public policy, the issue of how we regulate emerging forms (and applications) of AI is of increasing importance. Some of the really interesting questions are just now going from medium-term dilemmas to completely present-tense problems. Many of the legal/governance questions are neither entirely new (though the context and some of the policy implications are different) nor likely to be easily resolved anytime soon. What distinctive ethical, policy, and legal issues arise in the use of AI  in the medical context? How do broader issues associated with AI  –– including, for example, the “hand-off” problem, explainability, and long-term erosion of organizational competence –– arise in the medical context? How do we handle issues of human-machine interaction, system safety, bias, and even "personhood" in connection with the use of AI in the medical context?

1:55pm

Irreproducible Research in the AI Era

John Ioannidis

This session will discuss the challenges of irreproducible science, how the AI era creates new challenges and opportunities for reproducibility and what might this mean for equity and inclusion.

2:15pm

Healthcare Big Data, Predictive Analytics, and Machine Learning: Legal and Ethical Issues

I. Glenn Cohen

This talk will present a quick tour of legal and ethical issues raised by the use of machine learning and predictive analytics in healthcare. Among the topics we may touch on are AI and equitable healthcare: data and equality; physician role disruption, training gaps and reasons given; risks of automation bias; liability; scarcity and the inevitability of distributive choices; privacy threats; transparency; and trade secrecy.

2:35pm

Closing Panel: Legal and Institutional Challenges of Implementing Equitable AI in Medicine 

Moderator Roberta Katz

Panelists Mariano-Florentino CuéllarJohn Ioannidis, and I. Glenn Cohen

2:55pm

Break

3:10pm

Machine Learning for Personalizing Medicine: Estimating Individual Effects of Treatment for a More Diverse Patient Base

Sanjay Basu

Machine learning methods offer a unique opportunity to communicate between the results of randomized trials and the questions we need to answer for our patients in clinical settings. This session will discuss the prospects and challenges of translating the results of our most important source of scientific data into personalized clinical practice.

3:30pm

Keeping the Human in the Loop for Equitable and Fair Use of ML in Healthcare

Nigam Shah

While there are multiple challenges in clinical use of ML models, imagine how a patient's experience would change if we could predict risks of specific events and take proactive action. In this talk, we will review Stanford Medicine's Program for Artificial Intelligence (AI) in healthcare, with the mission of bringing AI technologies to the clinic, safely, cost effectively and ethically. Using our experience in deploying a prediction model to improve access to palliative care services, we will discuss potential solutions to issues relating to model correctness, interpretability, fairness, and equity as well as issues such as autonomy of decision making and fiduciary responsibility. We will review how keeping the human in the loop can enable fair use of ML in healthcare and why doing so is a necessary form of “presence."

3:50pm

Program Wrap-up: Frameworks for an Inclusive Future of AI in Healthcare

Moderator Tina Hernandez- Boussard

Panelists Timnit GebruCarla PughAlison PughMary MaddenTess Posner, and Marinka Zitnik

Scholars from Medicine, Engineering and the Humanities will explore frameworks for an inclusive future of AI in healthcare.

4:50pm

Closing & Summary

Abraham Verghese

5:00pm

Reception—GSB Community Courtyard