MUMBAI, India, Feb. 6 -- Intellectual Property India has published a patent application (202641007343 A) filed by Vellore Institute Of Technology, Vellore, Tamil Nadu, on Jan. 25, for 'system and method for bias-aware and fairness-optimized learning on healthcare-related datasets for identifying rare medical events.'
Inventor(s) include R Sujatha; Amal Sumesh; and Arun Eshwar Senthil Kumar.
The application for the patent was published on Feb. 6, under issue no. 06/2026.
According to the abstract released by the Intellectual Property India: "A system for bias-aware and fairness-optimized learning on healthcare-related datasets for identifying rare medical events is provided. The system 100 includes an input data source 102 that obtains datasets and a training server that receives datasets from input data source 102; analyzes the datasets to determine statistical indicators and compute empirical group distribution; computes, (i) a demographic correction term based on ratio of target group distribution to empirical group distribution, (ii) a rarity term based on empirical probability of associated outcome label, and (iii) an uncertainty term derived from prediction probability entropy for each dataset; combines the demographic correction term, rarity term, and uncertainty term using weighted coefficients to generate bias-aware weight for each dataset; normalizes each bias-aware weight across datasets to generate normalized bias-aware weight; and applies the normalized bias-aware weights during training of machine-learning model 106A to balance learning across demographic groups and rare medical events."
Disclaimer: Curated by HT Syndication.
