We are excited to announce the first Medical Decision Extraction, Analysis, and Classification Task (MedExACT) shared task, co-located with the BioNLP workshop at ACL 2026.
- Website (Details): https://sites.google.com/view/medexact-acl2026
- Google Group (Registration & Results): https://groups.google.com/g/medexact-acl2026
- Organizers Email: medexact-acl2026+owner at googlegroups dot com
=== Overview ===
MedExACT focuses on detecting and labeling medical decisions in ICU discharge summaries. Given a full discharge summary, systems must detect contiguous text spans that express medical decisions and assign each span one of ten categories defined by the the Decision Identification and Classification Taxonomy for Use in Medicine (DICTUM) (Ofstad et al., 2016) (e.g., Defining Problem, Therapeutic Procedure, Drug, Evaluation).
=== Data ===
Participants will use the MedDec dataset (https://aclanthology.org/2024.findings-acl.975.pdf) (Elgaar et al., 2024), which contains over 56k expert-annotated decision spans drawn from de-identified ICU discharge summaries in MIMIC-III. The data is supplemented with patient demographics to facilitate robustness analysis.
=== Evaluation & Prize ===
- Robustness: Systems will be evaluated for accuracy and robustness across patient subgroups (Sex, Race, and Language Proficiency) at both span and token levels. The final ranking metric is the average of the Base Score and the Worst-Group F1 across patient subgroups.
- Prize: We will cover the full ACL 2026 registration fee for the top three teams (one member per team).
=== Important Dates ===
- Train/Validation data release: January 1, 2026
- Test set release: March 23, 2026
- Prediction submission deadline: April 5, 2026
- Results notification: April 24, 2026
- System papers due: May 5, 2026
- Workshop day (tentative): July 7, 2026
=== Organizers ===
- Mohamed Elgaar (UMass Lowell)
- Jiali Cheng (UMass Lowell)
- Nidhi Vakil (UMass Lowell)
- Mitra Mohtarami (Anselm College)
- Adrian Wong (BIDMC)
- Hadi Amiri (UMass Lowell)
- Leo A. Celi (MIT, BIDMC)
We warmly invite the community to participate in this task to advance clinical decision extraction.
Best regards,
MedExACT Organizers