AI in Healthcare: Mitigating Disparities, Biases & Misinformation

Actions and Detail Panel


Event Information

Share this event

Date and time



Online event

Event description
A virtual symposium hosted by the Hariri Institute's Leveraging AI to Examine Disparities and Bias in Health Care Focused Research Program.

About this event

Artificial Intelligence (AI) and social media are finding useful applications in health care, yet, their use may perpetuate or even accentuate inequities, disparities, and the critical role of social determinants of health, or even facilitate the spread of health-related misinformation. The symposium will convene AI experts, medical researchers and practitioners, and computer and social scientists. The objective is to seek consensus and synthesize ideas coming from different vantage points on how to (1) approach algorithmic or data biases and develop methods to mitigate them, (2) develop a framework to leverage AI as part of a learning, continually improving health systems, and (3) grapple with the increasing influence of social media in the public perception of health issues, especially their role in spreading misinformation during the COVID-19 pandemic.

Co-sponsored by the BU Center for Information & Systems Engineering, the BU College of Communication Division of Emerging Media Studies, and the BU School of Public Health.

10:00am - 10:10am Welcome and Opening Remarks

10:10am - 10:50am Panel Discussion: The Challenges, Risks, and Opportunities of AI in Healthcare

  • Dr. Sandro Galea, Dean and Robert A. Knox Professor, BU School of Public Health
  • Dr. David Bates, Chief of the Division of General Internal Medicine, Brigham & Women's Hospital; Professor, Medicine, Harvard Medical School; Professor, Health Policy and Management, Harvard T.H. Chan School of Public Health
  • John Bryden, Executive Director and Senior Research Scientist, Observatory on Social Media at Indiana University
  • Derry Wijaya, Assistant Professor, Computer Science, Boston University

Panel moderated by Eric Kolaczyk, Director, Hariri Institute, & Professor, Mathematics & Statistics, Boston University

10:50am - 12:15pm AI in Healthcare, Algorithmic & Data Bias, and Mitigation Strategies: The speakers will explore new AI methods for health applications and outline how to mitigate algorithmic biases or biases inherent in the data used for training AI models.

Emma Pierson, Assistant Professor, Computer Science, Jacobs Technion-Cornell Institute at Cornell Tech

  • Title: "Using Machine Learning to Increase Equality in Healthcare"
  • Abstract: Our society remains profoundly unequal. Worse, there is abundant evidence that algorithms can, improperly applied, exacerbate inequality in healthcare and other domains. This talk pursues a more optimistic counterpoint -- that data science and machine learning can also be used to illuminate and reduce inequality in healthcare and public health -- by presenting three vignettes about pain, women's health, and COVID-19.

Rich Caruana, Senior Principal Researcher, Microsoft Research

  • Title: "Friends Don’t Let Friends Deploy Black-Box Models: The Importance of Interpretability and Privacy in Machine Learning Used for HealthCare"
  • Abstract: In domains such as healthcare it is critical that our models are correct, safe to deploy and protect privacy. For these reasons, whenever possible it is important to use glass-box learning methods that make it possible vet, correct and insure privacy before clinical deployment. In the talk I’ll present pneumonia, pregnancy, cancer and COVID-19 case studies where interpretable learning methods uncover surprising --- and risky --- patterns in the data that are learned by all ML methods, and I’ll show how glass-box methods can be used to discover and correct these problems while also guaranteeing differential privacy with little or no loss in accuracy.

Jenna Wiens, Associate Professor, Computer Science and Engineering and Associate Director, Artificial Intelligence Lab, University of Michigan

  • Title: "AI and Medical Imaging: Exploiting and Preventing Shortcuts"
  • Abstract: While deep learning has shown promise in improving the automated diagnosis of disease based on chest radiographs, deep networks may exhibit undesirable behavior related to shortcuts. In this talk, I will explore the problem of using chest x-rays to learn to diagnose the etiology of acute hypoxemic respiratory failure and will describe how biased training data can lead to models that exploit shortcuts and ultimately fail to generalize. In particular, I will demonstrate how deep nets can easily exploit shortcuts related to demographic attributes (e.g., sex and age) and will present a simple transfer learning approach to mitigate such behavior.

Panel moderated by Yannis Paschalidis, Professor, Electrical & Computer Engineering, Systems Engineering, Biomedical Engineering & Computing & Data Sciences, Boston University

12:15pm-1:15pm Lunch and Poster Session, GatherTown

1:15pm-2:40pm Learning Health Systems that Leverage AI: Learning Health Systems implement knowledge acquisition processes and embed them in their daily practice. In this evidence-based setting of continuous improvement, AI methods have an increasing role to play.

Dr. Irene Dankwa-Mullan, Deputy Chief Health Officer, IBM Watson Health

  • Title: "Developing a Framework to Integrate Health Equity and Social Justice into Design and Development of AI Tools"
  • Abstract: This presentation will provide an overview of the translational aspect of AI and machine learning research and ways in which bias could be introduced along the design, development, and implementation phases- and integrated into the learning continuum. The range of ethical issues, health equity and social justice principles to be considered when leveraging AI technologies will be discussed.

Dr. David Bates, Chief of the Division of General Internal Medicine, Brigham & Women's Hospital; Professor, Medicine, Harvard Medical School; Professor, Health Policy and Management, Harvard T.H. Chan School of Public Health

  • Title: "The Digital Divide and Implications for Artificial Intelligence"
  • Abstract: Digital equity is a major concern today. While artificial intelligence offers great potential benefit in healthcare and is likely to be increasingly widely used, it could also worsen equity and the digital divide if not actively managed. This talk will review some of the evidence around digital equity, and will discuss patient portals, use of mobile apps, and telemedicine, and the implications around equity and artificial intelligence.

Dr. Leo Anthony Celi, Clinical Research Director, Laboratory of Computational Physiology, Massachusetts Institute of Technology; Co-Director, MIT Sana; Staff Physician, Division of Pulmonary, Critical Care and Sleep Medicine, Beth Israel Deaconess Medical Center

  • Title: "Better than Humans: Removing Prejudice and Bias from Artificial Intelligence"
  • Abstract: Data routinely collected in the process of care are heavily influenced by long-standing social, cultural, and institutional biases. Unless the underlying inequities in our communities are addressed, algorithms will perpetuate, if not magnify, existing health disparities. We need to build artificial intelligence that is better than humans: less prejudiced, less biased, more fair.

Panel moderated by Rebecca Mishuris, Assistant Professor, General Internal Medicine, Boston University School of Medicine

2:40pm-2:50pm Break

2:50pm-4:15pm COVID-19 Misinformation: The speakers will discuss the threat that misinformation poses to public health, with a focus on COVID false information spread on social media. Several perspectives will be explored, from computer science to communication studies. The panel will also discuss what can be done to mitigate health misinformation.

Angela Xiao Wu, Professor, Department of Media, Culture, and Communication, New York University

  • Title: "What Do Platform Trace Data Obscure?"
  • Abstract: Platform trace data can be impressive due to their unprecedented granularity and volume, as well as the fact that they are seemingly “unobtrusive” recordings of our activities when no one is watching. These apparent strengths of data for social research are outweighed by weaknesses accompanying their “measurement conditions”—that is, platform data are platforms’ records of their own behavioral experimentation. The case of Chinese platform trace data, where these weaknesses are particularly salient due to a combination of sociocultural and political circumstances, is used to illustrate what’s at stake.

John Bryden, Executive Director and Senior Research Scientist, Observatory on Social Media at Indiana University

  • Title: "Online Misinformation is Linked to COVID-19 Vaccination Hesitancy and Refusal"
  • Abstract: We study the impact of online misinformation on health decisions, investigating the extent to which COVID-19 vaccination rates and vaccine hesitancy are associated with levels of online misinformation about vaccines. We present findings that demonstrate these association are present across U.S. geographical regions, even after accounting for many political, demographic and socioeconomic factors. We also find evidence for a directional relationship from online misinformation to vaccine hesitancy. Our results support a need for interventions that address online misinformation at a geographical level.

Fernando Bermejo, Faculty Associate, Berkman Klein Center for Internet and Society at Harvard University; Research Director, Media Cloud; profesor asociado, IE University in Madrid

  • Title: "Challenges in Health Misinformation Research"
  • Abstract: Following the ABC model (actors, behavior, content), this talk will provide an overview of the limits and possibilities of current research on health misinformation.

Panel moderated by Elaine Nsoesie, Assistant Professor, Global Health, Boston University School of Public Health

4:15pm - 4:30pm Closing Remarks

To learn more about this series or view our other upcoming events, click HERE.

Don't miss out: Follow us on Twitter and LinkedIn! Interact with us the day of the event on Twitter using the hashtag #AIandHealthcareBU

Share with friends

Date and time


Online event

Organizer Hariri Institute for Computing, Boston University

Organizer of AI in Healthcare: Mitigating Disparities, Biases & Misinformation

Save This Event

Event Saved