The Cancer Organoid Digital Twin: A Critical Review of AI-Enabled Patient-Specific Modeling of Drug Response, Resistance, and Tumor Evolution
Richard Afriyie Osei
*
Department of Nanoscience, University of North Carolina at Greensboro, Greensboro, NC, USA.
Olaitan Ebenezer Oluwadare
Applied Mathematics Department, School of Physics, Engineering, Mathematics and Computer Science, Delaware State University, USA.
Olatunde Simeon Awolola
School of Health Information Management, University College Hospital Ibadan, Ibadan, Nigeria.
*Author to whom correspondence should be addressed.
Abstract
The integration of patient-derived cancer organoids with artificial intelligence (AI) offers a potential approach for developing more individualised models of tumour behaviour and therapeutic response. Conventional precision oncology relies heavily on static molecular measurements, which may not fully capture intratumoural heterogeneity, treatment-induced adaptation, therapeutic resistance, or tumour evolution. Cancer organoids provide a functional experimental representation of patient-derived tumour biology, while AI enables the analysis and integration of complex molecular, phenotypic, imaging, and clinical data. Together, these technologies could provide a foundation for patient-specific computational models capable of predicting treatment response and adapting as new biological and clinical information becomes available. This critical review examines the biological fidelity of cancer organoids, the application of AI to organoid phenotyping and drug-response prediction, and the potential integration of these systems within a cancer digital-twin framework. Particular attention is given to intratumoural heterogeneity, clonal selection, culture-induced changes, incomplete representation of the tumour microenvironment, multimodal data integration, longitudinal updating, uncertainty quantification, and clinical validation. We argue that combining organoids with AI does not, by itself, constitute a digital twin. A clinically credible cancer organoid digital twin requires patient specificity, biological fidelity, multimodal integration, longitudinal updating, predictive capability, uncertainty awareness, and clinical actionability. The field therefore remains at an emerging translational stage. Future progress will depend on standardised organoid workflows, larger longitudinal datasets, improved modelling of the tumour microenvironment, hybrid mechanistic–AI approaches, and prospective multicentre validation. If these challenges are addressed, cancer organoid digital twins could shift precision oncology from static tumour characterisation towards dynamic prediction of treatment response, resistance, and tumour evolution.
Keywords: Cancer organoids, patient-derived organoids, artificial intelligence, digital twins, drug-response prediction, therapeutic resistance, tumour evolution, precision oncology