MUMBAI, India, Aug. 17 -- Intellectual Property India has published a patent application (202641096181 A) filed by Mr. Shyam Kondeti; Mr. Ajit Ratnakar Pradnyavant; S. Balakrishnan; Kandula Rojarani; Yashika Gaidhani; Dr. Sunil Babu Melingi; Nameet Kumar Sethy; Mrs. Ramya Nellutla; Mrs. Yerram Sneha; Dr. D. Sheefa Ruby Grace; Dr. Amit K Patel; and R. Augasthega on August 09, 2026, for Explainable Ai Framework For Transparent Decision-Making And Model Accountability.
Inventors include Mr. Shyam Kondeti; Mr. Ajit Ratnakar Pradnyavant; S. Balakrishnan; Kandula Rojarani; Yashika Gaidhani; Dr. Sunil Babu Melingi; Nameet Kumar Sethy; Mrs. Ramya Nellutla; Mrs. Yerram Sneha; Dr. D. Sheefa Ruby Grace; Dr. Amit K Patel; and R. Augasthega.
The application for the patent was published on August 14, 2026, under issue no. 33/2026.
Abstract: The present invention discloses an Explainable Artificial Intelligence (XAI) framework for transparent decision-making and model accountability that enhances the interpretability, reliability, fairness, and traceability of artificial intelligence systems. The proposed framework comprises a Data Acquisition Module, a Data Pre- processing Module, an Adaptive AI Prediction Engine, an Explainability Engine, a Confidence Estimation Module, a Fairness and Bias Evaluation Module, an Accountability Management Module, a Continuous Monitoring and Adaptive Learning Module, and a Visualization Interface. The framework acquires and pre- processes data from heterogeneous sources before generating intelligent predictions using one or more machine learning or deep learning models. The Explainability Engine provides both local and global explanations by identifying influential features, decision pathways, and model behaviour for each prediction. The Confidence Estimation Module computes prediction certainty, while the Fairness and Bias Evaluation Module detects and mitigates demographic, statistical, and algorithmic bias to support ethical decision-making. The Accountability Management Module maintains comprehensive audit logs, including model versions, prediction records, explanation reports, confidence scores, and operational metadata, thereby ensuring complete traceability and regulatory compliance. The Continuous Monitoring and Adaptive Learning Module identifies concept drift, performance degradation, and explanation inconsistencies, enabling adaptive model refinement and long-term operational reliability. The framework further presents prediction outcomes together with interpretable explanations, confidence indicators, fairness assessments, and accountability records through an interactive visualization interface. The proposed invention is scalable and suitable for deployment in healthcare, finance, cybersecurity, manufacturing, transportation, education, legal analytics, smart governance, and other domains requiring transparent, trustworthy, and accountable artificial intelligence systems.
Disclaimer: Curated by HT Syndication.
