MUMBAI, India, Nov. 2 -- Intellectual Property India has published a patent application (202641126808 A) filed by Vellore Institute Of Technology on October 24, 2026, for Machine Learning Method For Rare Disease Severity Prediction With Cohort-Size Validity Flagging.
Inventor includes Sanjay Kumar Mohanty.
The application for the patent was published on October 30, 2026, under issue no. 44/2026.
Abstract: ABSTRACT MACHINE LEARNING METHOD FOR RARE DISEASE SEVERITY PREDICTION WITH COHORT-SIZE VALIDITY FLAGGING A method of predicting rare-disease severity includes receiving, by a processor (102), sparse patient records (112) of a patient cohort having values for clinical fields, some values being missing. A missingness-preserving feature engineering stage (120) substitutes an imputed value for each missing value. A multi-family model selection stage (140) trains a classifier on features derived from the sparse patient records (112), the classifier generating a predicted severity and a confidence score. A confidence calibration and validity-flagging stage (160) estimates the confidence score from leave-one-out cross-validation results on the patient cohort without a held-out test set, compares a number of the sparse patient records (112) to a minimum sample-size threshold, and automatically attaches a statistical-validity caveat to the confidence score when the number is below the threshold. The predicted severity, the confidence score, and any caveat are output together to a clinician-readable output interface (170). (Fig. 1)
Disclaimer: Curated by HT Syndication.
