System and methods for processing neural signals

US10653330B2 · kind B2 · utility

Assignee

• Paradromics, Inc. · US

Inventors

• Matthew Angle · Austin, US
• Edmund Huber · San Francisco, US
• Robert Edgington · Austin, US

Key dates

• Filing date: Feb 13, 2019
• Grant date: May 19, 2020
• Expiry date: Feb 13, 2039

Classification

• Technology area (CPC G): Physics
• CPC primary: G16H50/70
• WIPO field: Medical technology
• WIPO sector: Instruments

Abstract

Systems and methods for processing neural signals are provided. A neural data analysis system may comprise a feature extraction module configured to extract a plurality of features from neural signal waveforms obtained by an implanted neural interface probe with a plurality of channels or electrodes, and transmit the extracted features as a plurality of discrete outputs. The neural data analysis system may also comprise a feature-event coalescence module configured to receive the plurality of discrete outputs from the feature extraction module, and construct a model-based inference of bioelectric activity based on feature event statistics, prior knowledge of bioelectric signals, and/or a behavioral model of the feature extraction module. The neural data analysis system may further comprise an approximator module configured to receive a plurality of coalesced events from the feature-event coalescence module, and apply a series of transformations to the coalesced event data to generate a higher entropy neural code.