Multimodal, Foundation and Agent AI for Real-World Clinical Decision Support: Patient-Level Diagnosis, Prognosis, and Treatment Selection

Precision medicine depends on the capacity to synthesize complex, heterogeneous patient data (from genomic profiles and histopathological images to clinical records and real-time sensor outputs) to individualize diagnosis and treatment. The emergence of foundation models and multimodal AI architectures offers transformative potential for this task, enabling the large-scale integration of imaging, omics, and clinical data streams that precision medicine pipelines require.

Yet despite rapid advances in AI capability, significant challenges persist in translating these tools into clinical settings. Issues of model interpretability, data interoperability, and generalization across diverse patient populations continue to limit uptake. Closing this gap requires a focused research effort at the intersection of AI methodology and precision medicine practice.

This Research Topic aims to advance understanding of how multimodal, foundation, and agent AI systems can strengthen precision diagnosis and individualized treatment. It invites research that applies AI-driven approaches to optimize patient care through enhanced biomarker discovery, genotype-phenotype correlation, treatment response prediction, and patient-level prognosis to guide individualized treatment selection.

Contributions that bridge AI innovation and real-world clinical deployment, in alignment with SDG 3: Good Health and Well-being, are particularly encouraged.

This Research Topic focuses on patient-level clinical applications: diagnosis, prognosis, treatment selection, and real-world implementation. Purely technical algorithmic work, and upstream drug discovery or pharmacological inhibitor development without direct clinical decision-making relevance, fall outside its scope.

To gather further insights into this evolving field, we welcome articles addressing, but not limited to, the following themes:

• Multimodal AI for integrating omics data (including gene expression profiling and next-generation sequencing) with imaging and clinical records
• Foundation models applied to molecular diagnostics, histology, and immunohistochemistry
• Agent AI for individualized clinical decision support and treatment selection
• AI-enhanced analysis of varied imaging modalities, including fluorescence in situ hybridization (FISH)
• AI for patient-level prognosis, risk stratification, and treatment selection at the point of care
• Explainable and trustworthy AI for precision oncology and immunotherapy response prediction
• Federated and privacy-preserving learning across multi-institutional precision medicine datasets
• Real-world validation and ethical deployment of multimodal AI in precision medicine settings

Professor Junhan Zhao reports an advisory role and receipt of consulting fees and equity from Encapsulate Inc, equity in Alvus Health Inc, and serves on the research advisory board of Shriners Hospitals for Children.