MUMBAI, India, June 26 -- Intellectual Property India has published a patent application (202641071441 A) filed by Chaitanya Bharathi Institute Of Technology on June 09, 2026, for Shap-Guided Adaptive Deep Neural Architecture For Interpretable Medical Risk Prediction From Heterogeneous Healthcare Records.
Inventors include Dr. K. Mary Sudha Rani; Dr. Y. Rama Devi; Dr. Y. C. A. Padmanabha Reddy; Dr. T. Sridevi; Smt. K. Swathi; Dr. E. Padmalatha; Dr. T. Suvarna Kumari; Ms. G. Kavita; Sri. B. Srinivasa S. P. Kumar; and Dr. B. Sateesh.
The application for the patent was published on June 19, 2026, under issue no. 25/2026.
Abstract: A computer-implemented system and method for interpretable medical risk prediction from heterogeneous healthcare records are disclosed. The invention comprises a healthcare data acquisition module configured to collect data from electronic health records, laboratory systems, medical imaging repositories, physiological monitoring devices, wearable sensors, genomic databases, pharmacy systems, and clinical documentation platforms. A data harmonization engine standardizes and integrates the heterogeneous data into unified patient representations for processing by an adaptive deep neural architecture. A SHAP-based interpretability engine computes feature attribution values and provides explanatory information to an adaptive architecture controller that dynamically adjusts feature weights, attention mechanisms, neural pathways, and model parameters. The system generates patient-specific risk predictions together with clinically interpretable explanations, confidence metrics, and feature contribution analyses. The architecture supports continuous learning, adaptive optimization, multimodal data fusion, and federated deployment while improving predictive accuracy, transparency, clinical trust, and decision-support effectiveness in healthcare environments.
Disclaimer: Curated by HT Syndication.
