2023 Breakout Session

Breakout Session GoalsThe goal of these breakouts are to discern and deliberate on how to move forward in each of these topical areas, in the near term and in the long-term.

Charge to Breakout Participants: See links to each breakout session below.  Please refer to the 2023 MSM Mission statement in forming your remarks

Instructions to Notetakers:  Slides for this session that include the 'flash talk' slide before breakouts, a notes template for Round 1 & 2, and a template for report back is: Click here for all Breakout Session Notes!

Breakout Sessions (notes in links below)

1. Digital Twins: whole person, mental health (Click on the link for pre-Breakout Background Info)

MSM lead: Gary An; IMAG lead: Liz Ginexi

  • How would the MSM Consortium play a role?
  • How to integrate mechanism-based multi-scale modeling with machine learning/artificial intelligence to develop medical digital twins?
  • How to foster collaborations that integrate underlying digital twin model development with developers of sensor/assay technologies needed to provide the data feedback to refine the virtual twin?
  • Propose strategies for being able to capture inter-individual heterogeneity in a way that allows personalization of digital twins.
  • Propose strategies for dealing with uncertainty and parameterization when integrating models/modules that cross multiple scales.
  • Discuss how the concept of Multi-scale Mechanism-based Digital Twins can find synergies with the topics in the other Breakouts (Melding Mechanistic-ML models, Sociabehavioral and Social Determinants of Health, Quantum Computing and Translations/Incentivization)

 

2. New Mechanistic-ML Methods: PINN, transformer, XAI, Large Language Models

MSM lead: George Karniadakis; IMAG lead: Mauricio Rangel-Gomez, Ilana Goldberg, Julia Berzhanskaya

  • What are the similarities and differences between current methods; e.g. PINN, transformer, XAI, Large Language models?
  • What are the opportunities to improve these methods?
  • How can these methods be applied to blood diagnostics an diseases as it pertains to the IMAG interst group AI and Machine Learning for Blood Diagnostics and Diseases | Interagency Modeling and Analysis Group (nih.gov)
  • How can these methods be applied to neuroscience?
  • How can these methods be used in the IMAG initiatives discussed in Session 2.1
  • How do we quantify uncertainty in neural networks and specifically in scientific machine learning?
  • Graph neural networks and causal inference.

 

3. Sociobehavioral and Social Determinants of Health (SDoH) Models: network, probabilistic, stochastic models

MSM lead: Elsje Pienaar, Kyoko Yoshida, Bruce Y. Lee; IMAG lead: Julia Berzhanskaya, Asif Rizwan

  • What is the state of current modeling efforts - post-pandemic modeling?
  • How can SDoH be used to develop models to address health inequities?

 

4. Quantum Computing and Other Technologies for Modeling: quantum sensors, neuromorphic chips. emerging compute capabilities

MSM lead: Suvranu De; IMAG lead: Orlando Lopez, Raj Gupta

  • How can these technologies be used in the IMAG initiatives discussed in Session 2.1?

 

5. Translation and Incentivization: state of practice and future outlook

MSM lead: Feilim Mac Gabhann; IMAG lead: Elena Sizikova

  • What are new and emerging MSM tools in your field, who develops them and how are these tools translated from research prototypes into practice?
  • What are the domain-specific opportunities and hurdles in translation of MSM tools in your research domain? 
  • How do we make credible and trustworthy MSM tools? Do different organizations (e.g. academia/industry/regulatory) have different approaches to or considerations about credibility?
  • What technologies exist and what technologies are missing/needed for timely and effective tool dissemination?