MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202641089241 A) filed by Ms. Puppala Sireesha on July 22, 2026, for A System For Privacy-Preserving Clinical Speech Acquisition And Biomarker Reporting For Neuropsychiatric Assessment.
Inventors include Ms. Puppala Sireesha; Ms. Nallari Ranga Sai Shivani; Ms. Shipra Shukla; and Mr. Gadagoju Soumith.
The application for the patent was published on July 31, 2026, under issue no. 31/2026.
Abstract: Disclosed is a privacy-preserving clinical speech acquisition system and method for collecting and processing speech data for neuropsychiatric assessment, schizophrenia and depression included. A speech task controller administers a plurality of standardized speech tasks and captures an audio recording for each. A recording quality evaluation engine scores each recording by a task-dependent weighted combination of normalized acoustic quality metrics. Where a recording falls short of a quality threshold, a recording guidance module diagnoses a probable physical cause and issues corrective guidance, and a segment-level re-recording and merging module re-records only the defective temporal segment and merges it with the previously accepted segments to rebuild a complete recording. A speech processing engine derives a transcript, a spectrogram, and a set of acoustic and linguistic biomarkers, each carrying an algorithmically computed confidence value and, where warranted, a reason code. The resulting data is stored against a session identifier in a privacy-preserving storage engine, patient-identifying metadata being joined to the derived representations only through a one-way, controlled-access reference. A longitudinal clinical analytics engine normalizes and confidence-weights a patient's biomarkers across sessions to yield a progression score, a structured reporting engine compiles a clinician-readable report, and a structured export module exports the data in standardized formats, including FHIR, JSON, and CSV.
Disclaimer: Curated by HT Syndication.
