About

Foundation models are quickly emerging as powerful tools to solve a variety of biomedical challenges, such as clinical text generation and summarization, radiograph analysis, disease prediction, etc. [1, 2, 3, 4, 5, 6]. Foundation models are characterized by their ability to solve multiple prediction tasks across diverse domains. These models are first pre-trained on vast quantities of unlabeled data, and then fine-tuned using task and domain specific examples. This stands in contrast to the mechanistic approach to machine learning in healthcare, which explicitly incorporates prior knowledge of physics/biology/chemistry into the models that are then refined by training from data. While both these approaches have the potential to streamline healthcare administration, reduce costs, enhance accessibility, and ultimately improve the quality of patient care, many questions remain unanswered:

  • What are the key differences between the two modeling paradigms?
  • What qualifies as a clinical foundation model?
  • Which healthcare challenges can be effectively addressed by each of these approaches, and which remain beyond their respective scope?
  • What are the challenges associated with training and applying these models in a healthcare context?
  • What can these models learn and from what sources of information?
  • How can such models be best integrated with the routines of healthcare professionals?
  • How about challenges limiting their prospective adoption (usability, maintainability, trustworthiness, robustness, fairness, etc.)?
  • How should these technologies be regulated?

References

  1. Zhang, Kai, et al. "BiomedGPT: A Unified and Generalist Biomedical Generative Pre-trained Transformer for Vision, Language, and Multimodal Tasks." arXiv preprint arXiv:2305.17100 (2023).
  2. Park, Sangjoon, et al. "AI can evolve without labels: self-evolving vision transformer for chest X-ray diagnosis through knowledge distillation" Nature communications 13.1 (2022): 3848.
  3. Li, Chunyuan, et al. "LLaVA-Med: Training a Large Language-and-Vision Assistant for Biomedicine in One Day." arXiv preprint arXiv:2306.00890 (2023).
  4. Zhang, Sheng, et al. "Large-Scale Domain-Specific Pretraining for Biomedical Vision-Language Processing." arXiv preprint arXiv:2303.00915 (2023).
  5. Moor, Michael, et al. "Foundation models for generalist medical artificial intelligence." Nature 616.7956 (2023): 259-265.
  6. Steinberg, Ethan, et al. "Language models are an effective representation learning technique for electronic health record data." Journal of biomedical informatics 113 (2021): 103637.

Relevant Topics

Relevant topics for the symposium include, but are not limited to:

Submission Tracks

We invite submissions for two non-archival tracks: Traditional and Non-traditional.

Traditional Track

In this track, we seek outstanding papers which explore novel approaches or offer fresh insights into the core questions targeted by this symposium. Full papers can be up to 4 pages in length with unlimited pages for references and appendices. All submissions will undergo a double-blind review process, and authors are required to ensure proper anonymization of their submissions. Examples of similar tracks include the AAAI Special Track on AI for Social Impact, MLHC Research Track, and ML4H Proceedings track.

Dual-submission policy: Papers submitted to this track should not have been previously published or be currently under review at any other archival venue. Nevertheless, given the non-archival nature of this symposium, authors are free to subsequently publish their work in alternative venues.

Non-traditional Track

In this track, we seek all sorts of non-traditional research artifacts. We welcome submissions such as papers that describe software tools, datasets and benchmarks, clinical abstracts demonstrating the application of foundational models and machine learning to specific clinical challenges, or previously published research findings that can stimulate insightful discussions among the symposium attendees. Additionally, we invite position papers that address policy considerations for contemporary machine learning models within the healthcare context. We highly encourage submissions from student first authors. Papers in this track should be concise, with the maximum length of 2 pages and unlimited pages for references and appendices. The review process will be single-blind, i.e. the author names and affiliations can be included in the submissions. Examples of similar tracks include the AAAI student abstract track, MLHC Clinical Abstract Track, and ML4H findings track.

Dual-submission policy: Papers submitted to this track can be already published or under-review at any other archival venue. Furthermore, accepted papers can also be later published in another archival venue.

Submission site

Openreview Submission Site

Format: Authors should follow the formatting guidelines outlined in the AAAI-24 Author Kit. Depending on the track, please utilize the appropriate templates: Authors submitting to the traditional track should employ the anonymized templates (AnonymousSubmission), while those submitting to the non-traditional track should make use of the camera-ready templates (CameraReady).

Supplementary Sections and Appendices

Supplementary Material and Appendices: Unlimited pages of supplementary materials such as appendices, proofs, and derivations may be attached to the paper. However, reviewers will not be obligated to review these materials, as part of the paper evaluation process.

We strongly encourage authors to include two additional sections into their submissions, when applicable. These two optional sections would not count towards the page limit, but should be limited to a maximum of 1 page each.

Ethical Considerations and Reproducibility Statement: We urge authors to include a dedicated paragraph that addresses the ethical implications of their work. Furthermore, please provide a paragraph outlining the measures taken to ensure the reproducibility of your research. This section can also include links to anonymized code and data submissions. We recommend that authors familiarize themselves with the Ethics Statement and Reproducibility sections in the ICLR 2024 Author Guide.

Ethics Board Approval: If your research involves datasets that necessitate Institutional Review Board (IRB) approval or its equivalent, please ensure that you mention these details in the camera-ready version for traditional track papers. For non-traditional track submissions, this information should be included at the time of initial submission. In the case of traditional track papers, at the time of submission, it is sufficient to include a statement indicating that relevant ethics approval information will be provided if the paper is accepted.

Important Dates

Event Date
Submission deadline Friday, January 5, 2024 Friday, January 26, 2024
Rolling review Saturday, January 27, 2024 – Friday, February 9, 2024
Notification of acceptance Friday, February 9, 2024 Friday, February 23, 2024
Final versions of papers due Friday, Mar 8, 2024 Friday, March 22, 2024
Symposium Monday, March 25, 2024 – Wednesday, March 27, 2024
All deadlines are 11:59pm UTC-12:00 (anywhere on Earth).
We will conduct rolling reviews from January 27 to February 6,
with papers reviewed in the order of submission during this period.

Presentation and Attendance

Authors of accepted papers will be invited to present a spotlight and/or a poster on their work at the symposium. At least one author of each accepted paper must register and be available in person to present their work.



We welcome participation by a broad range of participants: students, academics, clinicians, industrial researchers, exhibitors, with or without accepted papers! More information about registration can be found in the AAAI-24 Spring Symposium website .

Program

We will have 5 keynote talks, 3 panel discussions, 2 poster sessions, and several contributed talks (presentations of accepted papers) and technical demonstrations or tutorials throughout the 2.5-day symposium.

Confirmed Keynotes and Panelists

Confirmed Tutorials and Demos

Organizers

Student Members

Faculty Members

  • (Program Chair) Dr. Artur Dubrawski, Alumni Research Professor of Computer Science, Carnegie Mellon University
  • Dr. Su-In Lee, Paul G. Allen Professor of Computer Science & Engineering, University of Washington
  • Dr. Frederic Sala, Assistant Professor, University of Wisconsin-Madison
  • Dr. Jimeng Sun, Health Innovation Professor, Computer Science Department and Carle's Illinois College of Medicine, University of Illinois Urbana-Champaign
  • Dr. Gilles Clermont, Professor of Critical Care Medicine, Mathematics, Clinical and Translational Science, and Industrial Engineering, Department of Critical Care Medicine, University of Pittsburgh

Industry Members