Task Force: Kickoff Meeting
- June Burke
Key Takeaways
Tuesday, January 23rd, 2024
Gregory Kirk, Vice Dean of Research, and Nilanjan Chatterjee, Professor of Biostatistics, co-lead the initiative with school-wide representation to address the public health impact of AI. PHAISE objectives broadly include initiating in-depth discussion within our school community on AI in public health, identifying areas of strategic interest and need at BSPH, and enhancing the School’s understanding and appropriate adoption and development of AI tools.
Introductions
Thomas Hartung – Chair for Evidence-based Toxicology, EHE. Founding editor, Frontiers in AI, journal that has published over 900 articles in past 9 years. Leads Medicine and Public Health section. AI is used commonly for data retrieval or predictive toxicology. Led interest group in AI in EHE Department.
Shared examples of journal articles in chat:
§ Artificial Intelligence (AI) – It’s the end of the tox as we know it (and I feel fine)
§ Artificial intelligence as the new frontier in chemical risk assessment
Allison West – Assistant Professor, Population, Family and Reproductive Health, Child and Family Researcher. Communities and organizations she works with strong interest in AI. Directs a national research network that focuses on home visiting services.
Johannes Thrul – Associate Professor, MH. Psychologist by training. Digital/mobile health and addiction and substance use.
Benjamin Hunyh -- Assistant Professor, EHE. Practices data science / AI for different applications in health for 10 years. Focus on data science for planetary health.
Alyssa Columbus – PhD candidate, biostatistics. Research interest and experience in information security and data science.
Carl Latkin – Vice Chair, HBS. Looking for more collaboration across departments.
Ahmed Hassoon – Assistant Research Professor, EPI and Neurology at School of Medicine. Co teaches Data Science for Public Health class in BioStat. Works on AI and machine learning. Doctorate project was a clinical trial using a small language model for cancer patient coaching. Developed tech and data science tools for ED and primary care at JHU.
Kadija Ferryman – Assistant Professor, HPM. Medical anthropologist and work focuses on ethics, race and policy as they relate to health information technology. Experience with race in clinical algorithms. FDA policy and regulation of AI as a medical device.
Honkai Ji – Professor, BIOS. Develops statistical computational and machine learning approaches. Interested in AI to understand basic biology from longitudinal cohort data.
Rose Weeks – DrPh student in implementation science. Full time faculty member in IH. Research Associate in Center for Global Digital health Innovation, Center for Indigenous Health and IVAC. Developing an AI chatbo to address vaccine misinformation for STI testing.
Elham Ali – DrPh student working with local and state governments to apply human-centered design/ data science in solving local problems.
Abdul Bachani – Associate Professor, IH. AI to assess risk factors for injuries, using camera footage. AI algorithm to identify adolescents at risk of suicide in partner with Whiting.
Amy Wesoloski – Associate Professor, EPI. Modeling and statistical approaches, incorporating machine learning for research in infectious disease transmission.
Joseph Ali – Professor, IH. Background in law, ethics, philosophy and interest in ethics and global digital health, governance of digital health.
Brian Caffo – Director, Academic Programs for Data Science and AI Institute. Personal work in computational AI methods and AI development.
Ilinca Ciubotariu -- Background in infectious diseases with interest in AI & ML. Working with R3ISE program in MMI
Keri Althoff – Professor, EPI. Background in coordinating and analysis of large multi-cohort -datasets.
What does a task force do? What are the resources associated with a task force?
Our mandate is to guide development of school-wide strategy for AI.
Our recommendations will inform the need and level for school-wide investments.
Task-force designations reflects the time-limited nature where within the next year, we’d like to elevate the conversation and define the School level positioning and investments.
We’d like to get broader input on developing a comprehensive AI strategy tailored to the skills, unique needs and expertise of colleagues. What are the most favorable outcomes from this group?
Host workshop/conference types of activities. For example: GIS Day. Could be by department or cohosted between departments.
Bias/representativeness of data area involves many BSPH groups. School is well-positioned to have representative datasets.
Master’s or PhD students at Whiting might conduct theses jointly with BSPH. This could consist of a “dating” system to see where there are potential mutually beneficial connections.
Develop areas for collaboration with education mission. Pilot AI as TAs.
Broadening data access at JHM data warehouse. Primary appointed with BSPH grants only 499 records per request -- makes it extremely difficult to train an AI model.
A means of sharing datasets or sources between PIs.
Translating BSPH knowledge into a publicly accessible generative model.
To take PubMed Central and generate the chatGPT embeddings. Create the embedding language database for PubMed with labels.
Commercialization of AI tools – there is a current gap between what market wants and what the BSPH researches.
Create organizational structure to promote smaller models.
Engage with the public to determine their interests, values, and priorities for AI.
Help determine role AI should have in health and society.
AI Research Day with lunch and in-person speed-dating tables.
Help organize public data that can be used to build AI models.
Use of intellectual property rights and content with cultural heritage rights.
To focus and hone on how BSPH and JHU are stewards of good data, and we want our data to be used although we are protective of it and the people included in it as an important public health principle.
Market involvement with AI Institute at the University level.
Advocate for career path in AI + public health.
Market support services in AI around the School as offered by Whiting, etc.
Clarifying computational purchasing.
Work with Natalia Trayanova, Director of Research at AI-X Foundry.
Generate more collaboration with Whiting.
How should the Task Force operate?
Develop topics by specific problem areas.
Targeting students will create action and solidify the relationship to a certain extent, recognizing that lack of salary competitiveness with tech makes trainee difficult long-term.
Are there specific areas that the Task Force should address, based off past work members with AI-X Foundry?
This task force should be the primary home for public health research in AI. Collaboration and advocacy is focus with University initiatives
Relevant External Efforts
NIH AI Focus Groups
Arizona State University and OpenAI – testing OpenAI collaboration for education. Integrating ChatGPT into education to create individualized TA’s for students.
Ahmed Hassoon and Brian Caffo to cohost panel discussion about AI policy on Washington, D.C. campus on February 9, 2024.
Task Force Members Before Next Meeting
Review list of potential speakers and send recommendations or feedback to Junie.
Engage with Department members for potential PHAISE topics.