Abstract
Background Ninjin'yoeito (NYT), a traditional Japanese Kampo medicine, has shown potential in treating frailty and overactive bladder (OAB) symptoms. However, its effects are multifaceted and vary among individuals. This pilot study explored the use of topological data analysis (TDA) and natural language processing (NLP) to evaluate the effect of NYT on frailty in patients with OAB. Methods Fifteen patients with frailty aged 75 or older underwent pelvic floor muscle training (PFMT) and one month of NYT administration. The eight standardized health questionnaires were simplified into a 28-item format using NLP. Persistent homology analysis via TDA revealed the complex, multidimensional effects of NYT, while network graph clustering using the Louvain method identified key health domains influenced by NYT. Results TDA revealed multiloop structures in the therapeutic effects of NYT, indicating multiple pathways of improvement across physical and mental health domains. Network graph clustering identified four distinct communities linking OAB symptoms with energy, physical function, mental stress, and sleep quality. No significant adverse effects were noted. Conclusions This pilot study demonstrated the feasibility of using TDA and NLP to analyze the complex effects of NYT on frailty in patients with OAB. These findings suggest that NYT exerts multifaceted therapeutic benefits and further large-scale studies are warranted to explore its long-term efficacy.