Exploring Compatibility Laws in Traditional Chinese Medicine Prescriptions Through Data Mining.

To investigate the regulations governing medicines in Traditional Chinese Medicine (TCM) prescriptions using data mining techniques, this study aims to extract frequent item sets with high lift degrees from a substantial dataset, thereby providing theoretical references for exploring new TCM prescriptions. Using a crawler to collect a good deal of TCM prescriptions through the Pandas package in Python to read and process the data to get normalized data of medicines, and then perform frequency statistics, rule analysis, and association rules network analysis. After the completion of data processing, the study yielded a total of 34,518 prescriptions and identified 8,398 unique flavors of Traditional Chinese Medicine (TCM). Frequency statistics revealed a distribution pattern consistent with Zipf's law, emphasizing the prominence of specific medicines. Subsequently, the compatibility laws of medicines were analyzed through association rules, and they were ranked in descending order based on the degree of lifting. This analysis resulted in the identification of 18 frequent 2-item sets and 125 frequent 3-item sets. By scrutinizing the rules governing medicines across a vast array of prescriptions, this study extracts embedded mathematical patterns. The predominant therapeutic effects within the frequent item sets revolve around activating blood circulation and relieving pain, tonifying Qi and strengthening the spleen, dispelling wind and removing dampness, as well as clearing heat and eliminating toxins. These findings offer theoretical references for Chinese medicine practitioners, aiding them in the nuanced application of medicines and enhancing the formulation of effective treatment plans.