TCMPR: TCM Prescription Recommendation Based on Subnetwork Term Mapping and Deep Learning

Figure 2

Symptom distribution, herb distribution, and symptom-herb correlation of clinical samples. (a) The distributions of symptom number before and after screening are similar to the Poisson distribution. (b) The distributions of the herb number are also similar to the Poisson distribution. (c) No matter which symptom segmentation algorithm is adopted, the average similarity of symptoms increases with the increase in prescription similarity, and our SSTM method can achieve the best symptom-herb correlation.

(a)

(b)

(c)