A novel model for exploring the correlation between patterns and prescriptions in clinical practice of traditional Chinese medicine for systemic lupus erythematosus

Objectives: Traditional Chinese medicines in treating systemic lupus erythematosus are not uncommon. However, logic of pattern diagnosis and consensus on treating this disease is lacking. This study aimed to explore the correlation between the pattern diagnosis of and its corresponding prescription in systemic lupus erythematosus.

Design: Clinical data including patterns and prescriptions from the electronic medical records of lupus patients in a medical center were collected. Using a specific coding system (B-code) to encode the patterns and prescriptions provided a platform for data processing and statistics, and nonlinear canonical correlation analysis was employed to examine the correlation between them.

Results: In the 261 valid visits collected, a total of 46 patterns and 193 prescriptions were encoded into 57 B-codes. In our database, "Yin, Vacuity, Heat, and Dampness" were the most common B-codes (more than 90%) in both patterns and prescriptions. "Anemarrhena, Phellodendron, and Rehmannia Pill (Zhi-Bai-Di-Huang-Wan)" and "Miltiorrhizae Radix (Dan-Shen)" were the most frequently used formula (52%) and herb (48%), respectively. Water-rheum, kidney, and stasis were among the three most effective variable sets for correlating the patterns and prescriptions.

Conclusion: By using B-code with nonlinear canonical correlation analysis, the clinical dataset can be examined to shed light on the logic of pattern differentiation and principle of treatment for a specific disease.