MUMBAI, India, Sept. 22 -- Intellectual Property India has published a patent application (202621091698 A) filed by National Institute Of Technology, Raipur on July 28, 2026, for A System And Method For Neural Signature-Based Deception Detection.
Inventors include Dr Raghavendra Prasad; Dr. Arindam Bit; and Dr Shashikanta Tarai.
The application for the patent was published on September 18, 2026, under issue no. 38/2026.
Abstract: A SYSTEM AND METHOD FOR NEURAL SIGNATURE-BASED DECEPTION DETECTION A system and method for neural-signature-based deception detection is disclosed. Sixteen-channel EEG is recorded while a subject views Probe stimuli (approximately 20 % of trials) and Irrelevant stimuli (approximately 80 % of trials). The signals are epoched and analysed across five ERP time windows and four frequency bands (Delta, Theta, Alpha, Beta). Statistical features (F-value, p-value, partial eta-squared) derived by repeated-measures ANOVA are processed by a Multi-Layer Perceptron regressor (Custom Weight Allocation Model) trained with an Adam solver and an 80/20 train-test split to assign each channel a signed neural-resource-contribution weight. A Custom Rank Allocation Model ranks the channels by their computed resource allocation to generate a Deception Channel Priority hierarchy. A Pseudo Neural Activity Index combining ERP-, frequency- and CWAM-derived sub-indices yields a scalar deception-probability score between 0 and 1; a score of 0.65 or above indicates a Deceptive response. Principal deception biomarker channels comprise T8, P4, F3 and Cz/Pz. An optional standardised low-resolution electromagnetic tomography module maps scalp findings to Brodmann Areas 10, 11, 25 and 47. FIG. 7
Disclaimer: Curated by HT Syndication.
