Biomarkers Derived from Neurophysiological Computational Modeling

Assessing neurological conditions often requires sophisticated, invasive methods or complex imaging studies, leading to a more time-consuming and expensive process. Given the importance of early detection and the increasing prevalence of neurological conditions in society, a non-invasive, affordable, and efficient method of assessment will fill a significant unmet need. Current approaches struggle to provide real-time, dynamic assessment of neurophysiological conditions. Traditional methods rely heavily on static symptom observation or medical imaging, which require several repeat evaluations and don't provide continuous monitoring. Furthermore, these techniques require professional interpretation, which could house subjective errors. This creates a need for a method that can promptly detect neurological conditions using a more accessible and standardized tool.

Technology Description

The proposed system and method analyze the condition of a subject based on specific features extracted from the speech signal. This process incorporates a neurophysiological computational model, which interprets the speech data to derive control parameters. These parameters act as biomarkers or indicators of the subject's neurophysiological condition. To do so, the model compares speech-associated features with its predicted patterns and utilizes the derived error signal to update its own parameters. This innovative approach stands out because of its ability to constantly update the control parameters within the model based on the error signal it encounters in the comparison process. As each update better aligns the model predictions and actual features, it promotes more accurate biomarker extraction. Furthermore, by comparing the updated parameters with disorder-associated parameters stored in a library, it provides valuable insights into the potential presence of neurological disorders.