MUMBAI, India, July 30 -- Intellectual Property India has published a patent application (202641087676 A) filed by Dr. R Vimala; R Premalathasri; Dr. M Deepa; Dr. S Sarala; A Henry; P Sathya; R Usharani; K Gowthame; and R Jeevanathan on July 17, 2026, for Ai-Based Probabilistic Risk Prediction System.
Inventors include Dr. R Vimala; R Premalathasri; Dr. M Deepa; Dr. S Sarala; A Henry; P Sathya; R Usharani; K Gowthame; and R Jeevanathan.
The application for the patent was published on July 24, 2026, under issue no. 30/2026.
Abstract: The present invention relates to an AI-Based Probabilistic Risk Prediction System configured to predict the likelihood of potential adverse events by analysing multidimensional, historical, real-time, and contextual data. The proposed system integrates data acquisition, intelligent pre-processing, feature analysis, artificial intelligence-based prediction, probabilistic inference, uncertainty estimation, dynamic risk scoring, adaptive risk classification, and decision-support mechanisms. Input data collected from heterogeneous sources is validated, cleaned, normalised, encoded, and transformed into suitable feature representations. An artificial intelligence prediction engine identifies hidden and nonlinear relationships among risk-related attributes and generates an initial prediction corresponding to a target event. A probabilistic inference module estimates the probability of occurrence of the predicted event, while an uncertainty estimation module determines the confidence and reliability associated with the generated prediction. The predicted probability, uncertainty value, contextual severity, and influential risk factors are combined to generate a dynamic risk score. The generated score is classified into low, moderate, high, or critical risk categories using predefined or adaptively adjusted thresholds. When an identified risk exceeds an applicable threshold, an alert and decision-support module generate warnings, recommendations, or automated control instructions. An adaptive learning module continuously or periodically updates model parameters, probability calibration, feature importance values, and risk thresholds using newly available data and verified event outcomes. The invention provides an intelligent, scalable, uncertainty-aware, and adaptive framework for early risk identification, reliable probabilistic assessment, transparent decision support, and proactive risk mitigation across healthcare, finance, cybersecurity, industrial, transportation, and other risk-sensitive environments.
Disclaimer: Curated by HT Syndication.
