Hello, I would like to propose a Google Summer of Code 2026 project idea based on the open-source project *penuX*.
*Project Title:* Reproducible and Explainable Pathogen Prediction from Vitals (penuX) *Project Repository:* https://github.com/NETANELSTERN/penuX *Project Overview:* penuX is a research-oriented Python project that predicts likely pathogens from patient vital signs (such as temperature, WBC, and SpO₂), producing ordered probability outputs rather than a single hard classification. The project already includes prediction scripts, tests, and CI, but the workflow is not yet fully standardized, reproducible, or explainable for broader research and educational use. This GSoC project aims to evolve penuX into a reproducible, interpretable, and well-documented clinical ML pipeline while preserving its research focus. *Proposed Work and Expected Outcomes:* - A config-driven, reproducible inference pipeline (YAML/JSON based) - A single CLI entrypoint for preprocessing, inference, and evaluation - Explainability for predictions (e.g., permutation-based feature importance per sample) - Confidence calibration evaluation (reliability diagrams, ECE/MCE metrics) - Expanded deterministic tests and improved CI coverage - Clear documentation and a Quickstart guide for new contributors *Benefits to the Community:* - Promotes reproducible and auditable clinical ML research - Encourages responsible use of probabilistic predictions - Provides a concrete example of explainable ML in healthcare - Lowers the barrier for students and researchers to engage with clinical data science *Difficulty:* Medium *Duration:* 12 weeks *Mentor:* Netanel Stern I believe this project aligns well with GSoC’s goals of education, open collaboration, and meaningful open-source contributions. I would be happy to refine the scope further or adapt it to the expectations of a specific mentoring organization. Best regards, Netanel Stern בתאריך יום ג׳, 6 בינו׳ 2026 ב-21:30 מאת Karsten Hilbert via Python-list <[email protected]>: > Am Tue, Jan 06, 2026 at 06:52:10PM +0100 schrieb Karsten Hilbert via > Python-list: > > > One worthwhile project might be to improve the GNUmed > > package > > https://www.gnumed.de/documentation that is. > > Karsten > -- > GPG 40BE 5B0E C98E 1713 AFA6 5BC0 3BEA AC80 7D4F C89B > -- > https://mail.python.org/mailman3//lists/python-list.python.org > -- <https://netanel.ml> -- https://mail.python.org/mailman3//lists/python-list.python.org
