בתאריך יום א׳, 1 במרץ 2026 ב-19:41 מאת נתי שטרן <[email protected]>:
> Here is a brief explanation in English, including an invitation for > development: > > *Project Summary: penuX (MedExACT / BioNLP ACL 2026)* > > PenuX is a hybrid Deep Neural Network (DNN) pipeline developed for the > MedExACT shared task at BioNLP/ACL 2026. It focuses on medical decision > extraction and ICU outcome prediction using MIMIC-III/IV data. The model > predicts *in-hospital mortality (primary)* and *length of stay > (secondary)* based on patient vitals, labs, and medications from the > first 48 hours of ICU admission. The project is open-source and we actively > seek contributions, including *code development*, clinical advisory input > on DICTUM taxonomy, clinical feedback on subgroup robustness methodology, > and collaboration on cross-institutional validation. > > - > > > ---------- Forwarded message --------- > מאת: נתי שטרן <[email protected]> > Date: יום א׳, 1 במרץ 2026 ב-18:32 > Subject: Collaboration Inquiry: Open-Source ICU DNN (MIMIC-III/IV) — > MedExACT / BioNLP ACL 2026 > To: <[email protected]>, <[email protected]>, < > [email protected]> > > > Dear Mayo Clinic AI & Informatics Team, > > I am reaching out from the MedExACT shared task at BioNLP / ACL 2026, > where we are building open-source tools for medical decision extraction and > ICU outcome prediction on MIMIC-III/IV data. > > Given Mayo Clinic's two decades of leadership in clinical NLP — including > the NLPaaS framework and your work on EHR-based decision support — I > believe there may be meaningful alignment between our work and your > research agenda. > > Our project, penuX, is a hybrid DNN pipeline (no pandas, PyTorch) that > predicts: > • In-hospital mortality (primary) > • Length of stay (secondary) > > Using vitals, labs, and medications from the first 48h of ICU admission, > with calibration-first evaluation and subgroup fairness analysis. > > We are specifically looking for: > 1. Clinical advisory input on DICTUM taxonomy and ICU decision > documentation > 2. Feedback on subgroup robustness methodology from a clinical > perspective > 3. Potential collaboration on cross-institutional validation > > Would any member of your team be open to a brief conversation or async > exchange? > > 🔗 Repository: https://github.com/netanelcyber/penuX > 💬 Discussion: https://github.com/netanelcyber/penuX/discussions/11 > > Thank you for your time and for the foundational work your group has > contributed to clinical NLP. > > Best regards, > Netanel Stern > MedExACT Organizer, BioNLP / ACL 2026 > [email protected] > > > -- > <https://netanel.ml> > -- <https://netanel.ml> -- https://mail.python.org/mailman3//lists/python-list.python.org
