AI/ML Subgroup Meeting

Date: January 23, 2024 7 pm EST

Meeting Chairs: Ma’n Zawati & Rosalyn Ryan

Zoom meeting ID: 850 2874 4504

https://us02web.zoom.us/j/85028744504?pwd=Sm5KYllYdUdwM3dwajlKNGxNK2JUQT09
Passcode: 811299

Zoom recording:

https://us02web.zoom.us/rec/share/gwKxiGer65TND9H4-V_ju9Vvnu5hz62a6NXG5SmL5rozFkAVZMXtGlbP2g2sb88.PtgvsR6FScfk4kIC  

Passcode: 2%L6x0?F

Attendees:

1.  Ma’n Zawati
2. Rosalyn Ryan
3. Tina Hernandez-Boussard
4. Diya Uberoi
5. Jessica Seegobin

Minutes:

1.  Introductions:

JS:  Thanked everyone for coming and joining at such a short notice.

Everyone then went around the room and introduced themselves. MZ is an associate professor at McGill University and the research director of the Centre of Genomics and Policy. RR, is the CEO of Health Fox US, a company that implements programs in hospitals. THB, is the associate dean of research at Stanford medical school, where her work centers on the application of AI in healthcare. All three have AI-related projects or are/been part of AI initiatives.

2. Recap of last meeting:

MZ provided a brief summary of the first meeting between the co-chairs. He highlighted the way AI has proliferated over the years and how central conversational AI is to data sharing processes. Given the focus of GA4GH on data sharing standards, he clarified the need to streamline data sharing processes and the role AI could play in facilitating such outcomes.  

Ga4gh has implemented a range of policies, but questions arise as to how we ensure transparency, security and privacy. A tool that can address these issues, but is also concrete and that is usable by the broader GA4GH  community would be important.

Regardless of the approach adopted, the group will have to consider what is to be included in the scope. The scope will also change depending on whether the focus is on health care v. research.

RR clarified the need for a narrative that addresses the hope of AI and how it facilitates medical care  and makes it easier for people to do their jobs. Need to look at how it’s used, its validation and how it will help solve the problems we have, especially in genomics where data sets are larger.

In summary, the co-leads suggested that since their last meeting, they are still working on coming up with a clear mandate for the group and that the next step would be to then consider the input of colleagues who are in the group.

3. TINA’s INTEREST

Tina was asked to be involved as a member of the group. She commented upon the importance of the initiative and her willingness to provide comments on any documents or summaries produced. She agreed to be a member of the group.  Her work in the field was discussed and how important it is to ensure transparency in the process given the way these tools can go wrong and the harm they can cause. Data sets are often biased and issues arise when it is then transferred to other settings. The need for standards was highlighted and how important it is that policies/ tools are broad to address the various harms that can arise from use of AI. Furthermore, the importance of the stakeholders and their expectations was discussed. RR and MZ mentioned their interest in engaging Tina for additional feedback, etc., where needed.

4. Key Takeaways:

• AI is a growing and important field. Going forward however, it will be important to clearly define a focus and agree upon key outputs or deliverables that the group will work on.
• Tools/ standards to be developed will differ depending on the players and the context decided upon
• MZ and RR expressed interest in wanting some time to concretize their thoughts, re-convene and then meet with the larger group. Aim is to have this done before Connect takes place in Switzerland.

5. Next Steps:

• MZ & RR to concretize the group’s aims and the re-convene with each other