We are pleased to announce that our article "Explainable SHAP-space fall risk profiling from electronic health records for inpatient prevention planning" has now been published in npj Health Systems.

Our latest paper addresses the development of risk profiles for individualized intervention planning for patients at risk of falling. Based on routine data from the electronic health records of a large university hospital, comprising approximately 932,000 inpatients from 2016 to 2022, of whom 1.12% had a documented fall, we adapted clustering of SHAP values for fall prevention. In the held-out test set, the ensemble model achieves an AUROC of 0.95 and an AUPRC of 0.36, compared to 0.82 and 0.05, respectively, for a reference model based solely on guideline-based risk factors. The nine identified SHAP profiles separate observed fall incidence among alerted patients from 12.5% to 45.0% and were mapped onto guideline-consistent prevention domains of the World Falls Guidelines.

The full article is available as Open Access here:
https://doi.org/10.1038/s44401-026-00138-4

Many thanks and congratulations to everyone involved.