MUMBAI, India, Oct. 26 -- Intellectual Property India has published a patent application (202611107261 A) filed by Manipal University Jaipur on September 07, 2026, for A Systematic Evaluation Of Multimodal Longitudinal Wound-Infection Risk Prediction Using Temporal Biomarkers And Explainable Machine Learning.
Inventors include Dr. Shyam Sunder Sharma; Dr. Ashu Yadav; Kartikay Mathur; Sabnam Das; and Bithika Bibhabari.
The application for the patent was published on October 23, 2026, under issue no. 43/2026.
Abstract: The present invention relates to a multi modal longitudinal wound-infection early-warning system for biomedical sensing and explainable machine learning. The system acquires pH, temperature, lactate, protease, glucose, and oxygen measurements over multiple time points and derives latest value, mean, velocity, and variability for each biomarker, generating a 24-feature temporal representation. The features are processed using a patient-level leakage-controlled machine-learning framework evaluating Logistic Regression, Random Forest, and XGBoost, with Random Forest selected as the final model. In simulation-based independent evaluation, the system achieved an ROC-AUC of 0.7662, with a 95% bootstrap confidence interval of 0.6982 - 0.8274 and PR-AUC of 0.7084. The best tested early observation window was 0–3, with an ROC-AUC of 0.7776. The system further incorporates SHAP-based explainability, probability calibration, and simulated sensor-noise robustness analysis. The model was independently verified and integrated into a prototype providing biomarker input, infection-risk prediction, risk stratification, temporal visualization, and AI-based explanation.
Disclaimer: Curated by HT Syndication.
