Methods for assessing many psychiatric and neurological conditions fail to objectively quantify nuanced behaviors that are powerful indicators of condition severity and early treatment response. Our novel machine learning models analyze a constellation of behavioral and physiological biomarkers across multiple modalities (audio, video, natural language processing) to accurately detect and monitor symptoms in ways that exceed human perception. With our more objective and sensitive measures of baseline severity and treatment response, we reduce variability across clinical trial sites and increase study power, patient selection, and enrichment. Our disease-agnostic platform works across psychiatric and neurological conditions, such as Schizophrenia, Major Depression, and Parkinson’s disease, and builds on decades of academic work and seminal papers by our team. Our model for assessing Major Depression recently won the Distinguished Poster Award at the International Society for CNS Clinical Trials Methodology (ISCTM)'s annual scientific meeting.