Insilico and THPharm Expand AI Drug Development Strategy

Seoul, South Korea, August 12, 2026

AI Strategy Connects Drug Effects With New Indications

Traditional drug development generally begins by selecting a disease and then identifying or developing a therapeutic candidate that can address the underlying biology. THPharm’s approach reverses this sequence by starting with pharmacological effects and phenotypes confirmed through non-clinical research. AI-based analysis is then used to connect those observations with disease mechanisms, molecular pathways, gene expression patterns, and biomarkers that could reveal additional indications with development potential. The strategy is designed to make greater use of existing biological and pharmacological information while potentially reducing some of the risks associated with identifying entirely new therapeutic applications. By analyzing evidence already generated during development, THPharm aims to prioritize opportunities where the underlying biology has stronger supporting evidence. The companies intend to focus resources on candidates that demonstrate a combination of biological rationale, biomarker reproducibility, validation potential, and commercial opportunity. Insilico Medicine’s PandaOmics platform forms the core of the analysis infrastructure supporting the collaboration. The platform is being used to integrate drug-target information, literature-derived knowledge graphs, public omics datasets, and post-treatment gene expression data. Through this integrated analysis, the companies are seeking to identify relationships between drug targets, downstream molecular signals, disease pathways, and potential biomarkers.

PandaOmics Analyzes THP-001 Data

The initial program is focused on THP-001, which THPharm describes as a core asset currently being evaluated in multinational Phase 3 clinical trials. During the initial data-integration stage, the companies have secured 16 public datasets containing 269 samples to establish a foundation for gene-expression analysis and meta-analysis. The ongoing analysis is intended to connect pharmacological effects observed in non-clinical studies with disease-specific molecular mechanisms and biomarker candidates. Researchers are examining gene networks and downstream signals associated with the drug’s targets to identify indications where the observed biological effects may have therapeutic relevance. Once the analysis framework for THP-001 is established, THPharm plans to apply the strategy to additional DDS-based metabolic disease candidates. These follow-up programs have already generated non-clinical efficacy and safety or toxicity data, providing a foundation for further AI-supported indication analysis. The company is considering major metabolic disease areas including heart failure, fatty liver disease, and obesity.

THPharm Plans Validation of AI-Derived Candidates

Rather than pursuing multiple indications simultaneously, THPharm plans to prioritize potential opportunities based on the results of AI analysis and additional scientific and commercial criteria. These include biomarker reproducibility, feasibility of non-clinical validation, disease biology, and marketability. Candidate indications and biomarkers identified through PandaOmics will undergo further testing in disease-specific cell and animal models. Only candidates demonstrating reproducible efficacy and supporting mechanisms are expected to advance toward early-stage clinical development. This staged approach is intended to combine AI-powered discovery with conventional experimental validation, helping distinguish promising computational findings from candidates with sufficient biological evidence for further development.

The collaboration reflects the growing use of AI in drug discovery and development, particularly for extracting additional value from existing biological datasets. By connecting pharmacological effects with disease mechanisms and biomarkers, THPharm aims to create potential new development pathways from existing assets rather than relying exclusively on new molecule discovery. For Insilico Medicine, the collaboration demonstrates the application of PandaOmics to indication expansion and translational drug development. For THPharm, the strategy could provide a capital-efficient approach for expanding its metabolic disease pipeline. If AI-derived hypotheses are successfully validated experimentally and clinically, the approach could enable a single therapeutic asset to support multiple development opportunities while maintaining evidence-based selection throughout the process.

Source: Insilico Medicine, THPharm press relese