Neural Navigator: Decoding Psychiatric Disorder Signatures from Patient-Derived Cerebral Organoid Network Dynamics

About This Grant

Neuropsychiatric disorders such as schizophrenia (SCZ) and bipolar disorder (BPD) remain a major clinical and translational challenge. These conditions affect over 6% of the U.S. population and account for substantial long-term disability and healthcare burden. Despite extensive genetic and imaging studies, most cases remain idiopathic with no identifiable molecular etiology, limiting our ability to stratify patients or design mechanism-based treatments. A critical barrier is the lack of tools to quantify circuit-level dysfunction in human brain tissue at sufficient resolution to resolve disease- specific features. Traditional approaches—animal models, neuroimaging, and postmortem tissue— either lack species relevance or cannot capture dynamic circuit activity with cellular precision. This project introduces Neural Navigator, a standardized platform for quantifying circuit pathophysiology in patient-derived cerebral organoids using multi-electrode array (MEA) recordings and machine learning–based modeling. Preliminary data show that core electrophysiological features—spike rate, inter-spike interval, burst duration—differ reproducibly across SCZ, BPD, and control lines. Using dynamic modeling, Neural Navigator achieves >90% classification accuracy following electrical stimulation. Aim 1 will generate a longitudinal dataset of spontaneous and evoked MEA activity across 75 cerebral organoids (25 per condition) from days 90 to 270, establishing reproducible profiles of circuit development and disease-associated phenotypes. Aim 2 will validate Neural Navigator as a quantitative pipeline that integrates stimulus-response network modeling (SRDNM), a sink–source index of information flow, MRMR-based feature selection, and interpretable classification models. This approach will produce the first reproducible framework for stratifying idiopathic psychiatric disease using human brain tissue. All models, code, and datasets will be released in FAIR-compliant formats. The resulting platform will support mechanistic disease quantification and provide quantitative endpoints for future therapeutic discovery and drug screening applications.