Overview

The goal of the research roundtables is to foster discussion between the participants and senior researchers in the field on several topics of high relevance in the ML4H community. Each roundtable will have a small group of senior researchers and practitioners who are experts in the selected topics and a few junior chairs who will lead the discussion on open problems in ML4H which were crowdsourced from the community.

How to participate

The research roundtables will take place at 13:40 ET on December 4th in Gather.Town. The session will be split into two parts, with a 5 minutes break in the middle where participants can move around to a different roundtable. Participants are encouraged to pick at most two roundtables that they would like to join and engage in the discussion points proposed by the junior chairs and ask further questions to the senior chairs. To avoid the disruption of conversations, the participants should only move between roundtables during the break.

Research Round-tables

Multimodal learning in healthcare and representation learning on clinical data

  1. Description: Patient’s electronic health records contain a diverse set of information such as clinical notes, lab test results, medical imaging, etc. Being able to obtain useful representations and integrate across these data modalities represents an important first step for building machine learning methods that can be applied in the clinical setting. This research roundtable will discuss the challenges in obtaining such representations and highlight some current state-of-the-art methods.
  2. Senior chairs: Marzyeh Ghassemi, Polina Golland
  3. Junior chairs: Sneha Jha, Rishab Khincha

Population health

  1. Description: Machine learning has been widely applied in population health, in which we are concerned with the health outcomes of a group of population and the distribution of outcomes within the group. In this roundtable, we discuss how ML-based approaches can change population health and whether reporting of machine learning predictive models aligns with established reporting guidelines. We also discuss opportunities, challenges, and solutions needed within ML to create the best possible outcomes.
  2. Senior chairs: Rumi Chunara, David Buckeridge
  3. Junior chairs: Hang Yuan, Serifat Folorunso and Favour Nerrise

Regulation in Health AI

  1. Description: Although AI offers unique opportunities to improve health care and patient outcomes, it seriously challenges the robustness and appropriateness of current healthcare regulatory models. Several countries and organizations have proposed regulations addressing the use of AI in health care, but there are still no concrete laws or robust regulations have been adopted. In this roundtable, we will discuss key legal and ethical issues that have arisen as a result of the adoption of AI in healthcare.
  2. Senior chairs: Regina Geierhofer, Thomas J. Fuchs, Junaid Nabi.
  3. Junior chairs: Payal Chandak, Rutwik Shah

Causality and inductive bias for stability, robustness, and generalization & Detecting failure modes of machine learning systems

  1. Description: As machine learning models start to be more widely used in clinical practice, their robustness and reliability becomes critical. However, it is widely known that such models can be brittle and significantly underperform when deployed in environments with distribution shifts from the training datasets. This research roundtable will explore ways of determining when models may fail to generalize in practice and highlight how principles from causality can be used to improve the reliability of such machine learning models in the healthcare setting.
  2. Senior chairs: Adarsh Subbaswamy, Wojciech Samek, Federico Cabitza, Lukas Ruff
  3. Junior chairs: Kaushik Manjunatha, Golam Rasul, Utkarshani Jaimini

NLP for Medical Conversations

  1. Description: Summaries generated from medical conversations can improve recall and understanding of care plans for patients and reduce the documentation burden for doctors. Recent advancements in automatic speech recognition (ASR) and natural language understanding (NLU) offer potential solutions to generate such summaries and unlock additional value from medical conversations automatically. In this roundtable, we aim to present an overview of the space and discuss opportunities and challenges. In addition, we also hope to share our learnings from building and shipping medical conversation understanding technologies at Abridge (https://abridge.com).
  2. Senior Chairs: Sandeep Konam, Zack Lipton

What are important considerations to build tools that solve clinical problems? A brainstorming session with clinicians.

  1. Description: Machine learning for health research is often criticized for missing the mark when it comes to clinically meaningful endpoints. Critics argue that research on novelty tasks or datasets does little to improve patient outcomes. In this roundtable we aim to discern what parameters and considerations should go into better selecting important clinical problems.
  2. Senior Chairs: Naomi Lee, Mark Sendak, Vidur Mahajan, Sergio Uribe, Jesse Ehrenfeld
  3. Junior Chairs: Huiqi Yvonne Lu, Kaivalya Deshpande

Fairness and ethical AI in healthcare

  1. Description: In the high octane space that is ML4H, where hyped technology meets an application domain with formalized ethics standards, alignment negotiations ensue. These take place in a complex ecosystem comprising technological, business, regulatory, medical and social activist concerns, among others. Yet, as recent work demonstrates, dealing with concrete evidence of ethics concerns in ML4H applications can be challenging on all these levels. In this roundtable we aim to discern how broad ethics chartas, the technological needs for precise forensics and political action relate to each other.
  2. Senior Chairs: Judy Gichoya, Rohit Malpani, Jeff Lockhart, Roxana Daneshjou
  3. Junior Chairs: Girmaw Abebe, Elora Schörverth, Jerry Fadugba