Background: Predicting the molecular mechanisms and therapeutic effects of National Health Insurance herbal Prescriptions (NHPs) is challenging because of their multi-compound nature. We aimed to predict the therapeutic mechanisms of 56 NHPs using a Traditional Chinese Medicine (TCM) database and bioinformatic tools. Methods: We used a TCM database and bioinformatic techniques to construct networks of 56 NHPs. The network’s predicted potential therapeutic symptoms were compared with clinical study results and known indications to evaluate their validity. Finally, we identified the potential diseases and predicted the molecular mechanisms of 13 selected NHPs. Results: Of the initial 56 NHPs, 13 were selected for a detailed analysis. A five-layer network was constructed, which linked an average of 1,359 potential diseases per prescription. The concordance rate between network predictions and clinical papers was 45.6%, whereas that between network predictions and previously known major indications was 37.8%. Molecular mechanism analysis revealed that NHPs affect multiple pathways, including those related to metabolic diseases and nerve regulation. Conclusion: Our findings suggest that combining bioinformatics and clinical data can provide insights into the therapeutic effects of NHPs. This approach can guide future research by predicting potential therapeutic mechanisms and applications, providing insights into the therapeutic effects of NHPs.

Background: National Health Insurance Herbal prescriptions (NHPs) are widely used; however, their multi-component composition complicates mechanistic interpretation and impedes the development of evidence-based approaches in traditional medicine and healthcare policy. In this study, we applied a systems biology approach to link molecular mechanisms to clinical effects. Methods: From 56 NHPs, 13 with sufficient clinical evidence were selected. Multi-layer networks connecting herbs, ingredients, genes, and diseases were constructed using SymMap, with interactions filtered for oral bioavailability and statistical significance (false discovery rate < 0.05). Network-predicted diseases were validated against a clinically validated benchmark using permutation-based null model analysis, and gene set enrichment analysis (GSEA) was used to identify key molecular pathways. Results: Networks predicted an average of 1359 diseases per NHP, reflecting their polypharmacology. Importantly, the overall predicted disease sets for 10 of 13 NHPs showed statistically significant overlap with known clinical uses (p < 0.05, several with p < 0.001). GSEA indicated that NHPs commonly modulate three biological axes—hormone–metabolic regulation, neural signaling, and cell proliferation control. Conclusions: NHPs act as potential systemic homeostasis regulators. Our study introduces a computationally validated framework integrating network pharmacology with permutation-based statistical testing, providing a data-driven rationale for NHP use. These computational findings are exploratory and require future biological and clinical validation.