MUMBAI, India, Aug. 31 -- Intellectual Property India has published a patent application (202641100490 A) filed by St. Peters Engineering College on August 19, 2026, for An Adaptive Multimodal Artificial Intelligence System For Real-Time Decision Making Using Heterogeneous Data Fusion.
Inventors include Mrs. P. Rajitha, Assistant Professor In Department Of Cseaiml, St Peters Engineering College, Opposite Ts Forest Academy, Kompally Road, Dullapally, Maisammaguda, Medchal; Ms. Gudimetta Sreelalitha, Assistant Professor In Department Of Cseaiml, St Peters Engineering College, Opposite Ts Forest Academy, Kompally Road, Dullapally, Maisammaguda, Medchal, Hyderabad, Telangana; Ms. Arukonda Srujana, Assistant Professor In Department Of Cseaiml, St Peters Engineering College, Opposite Ts Forest; Ms. Sampa Biswas, Assistant Professor In Department Of; Mr. R. Sunilgavasker, Assistant Professor In Department Of Cse; and Dr. M. Nagabhaskar, Associate Professor In Department Of Mba, St Peters Engineering College, Opposite Ts Forest Academy Kompally Road, Dullapally, Maisammaguda, Medchal, Hyderabad, Telangana.
The application for the patent was published on August 28, 2026, under issue no. 35/2026.
Abstract: The present invention relates to an Adaptive Multimodal Artificial Intelligence System for Real-Time Decision Making Using Heterogeneous Data Fusion, designed to address the challenges associated with processing diverse, dynamic, and heterogeneous information for intelligent decision support. The system acquires data from multiple sources and modalities, including structured numerical data, sensor measurements, text, images, audio, video, and time-series data. A modality-specific preprocessing mechanism performs data cleaning, normalization, noise reduction, missing- data handling, synchronization, and quality assessment. The processed information is subsequently transformed into representative features using suitable artificial intelligence and machine learning models. A principal feature of the invention is an adaptive heterogeneous data fusion mechanism that dynamically evaluates the relevance, reliability, quality, confidence, temporal characteristics, and contextual significance of individual modalities. Based on these parameters, the system determines adaptive fusion weights to emphasize reliable information and minimize the influence of noisy, incomplete, redundant, or conflicting data. The fused multimodal representation is provided to a real-time AI decision-making module for classification, prediction, anomaly detection, risk assessment, recommendation, or automated decision generation. The system further incorporates confidence and uncertainty assessment and a feedback-driven adaptive learning mechanism for continuously updating fusion parameters and decision strategies according to changing data patterns and operational conditions. The invention thereby provides improved robustness, adaptability, scalability, reliability, and low-latency decision making. The proposed system is applicable to smart healthcare, intelligent transportation, industrial monitoring, cybersecurity, smart agriculture, surveillance, disaster management, and other real-time applications requiring reliable decisions from heterogeneous multimodal information.
Disclaimer: Curated by HT Syndication.
