Objective: we developed an organoid-based AI-driven platform to classify patient specific molecular subtypes and predict drug efficacy through an AI-powered validation pipeline. Patient-derived induced pluripotent stem cells (iPSCs) provide a promising approach for personalized disease modeling. By differentiating iPSCs into midbrain dopaminergic and GABAergic neurons and integrating them into a microfluidic system, we established a functional cortico-midbrain-thalamic circuit that better mimics disease progression. AI-driven analysis enables unbiased disease classification and targeted therapeutic testing, advancing precision drug development.

Background: Parkinson’s disease (PD) affects approximately 12 million people worldwide and leads to progressive motor and cognitive impairments. Despite extensive research, clinical trial failure rates remain high as traditional models fail to capture PD’s molecular and cellular heterogeneity.

Method: To achieve this, iPSC-derived brain organoids were generated from PD patients and healthy controls, incorporating midbrain and cortical neuronal subtypes to reflect disease-relevant neural circuits. Multi-modal data collection included holotomography for real-time structural and morphological analysis, multi-electrode array (MEA) recordings to assess neuronal activity and connectivity, and high-content imaging for quantitative phenotyping. For AI-driven subtyping, deep learning, self-supervised models, and graph-based analysis classified PD subtypes and identified molecular signatures of disease progression. A validation pipeline was developed to predict patient-specific therapeutic responses, supporting early-phase drug development. AI-driven analysis of patient-derived organoids successfully classified distinct PD subtypes, revealing novel disease mechanisms.

Results: Multi-modal AI integration improved subtyping accuracy and predictive validity, while graph-based AI models enhanced interpretability of patient-specific disease networks. The platform also predicted individualized drug responses, demonstrating strong translational potential.

Conclusion: This study presents an AI-powered organoid-based platform integrating deep learning with patient-specific disease modeling to advance precision medicine in PD. By bridging preclinical research with clinical translation, this approach may reduce clinical trial failure rates and accelerate therapeutic development.

« Back to 2025 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/organoid-based-ai-clinical-trial-platform-for-phase-0-studies-in-parkinsons-disease/