Construction of a therapeutic effect prediction model for bronchial asthma.

Objective We aimed to investigate the feasibility and method of constructing a traditional Chinese medicine (TCM) curative effect prediction model based on the data of Kechuanting acupoint plastering therapy in the treatment of bronchial asthma (asthma). Methods Data from the Chronic Disease Management Research Platform of 303 patients with asthma who were treated with Kechuanting acupoint plastering therapy for 6 weeks in the Department of Acupuncture and Rehabilitation of Jiangsu Hospital of Traditional Chinese Medicine from June to August 2018 to 2021 were selected. We used Phyton 3.10 for statistical analysis. After data preprocessing, the influencing factors were used to build models by Logistic regression, support vector machine, K-means clustering algorithm, Bayes algorithm, random forest method and Light gradient boosting machine (LightGBM) respectively, with the improvement of asthma control test score (ACT), forced expiratory volume in one second (FEV1) and exhaled nitric oxide (FeNO) as the outcome indicators. Then, the models were compared and analyzed. Subsequently, the superior model was used to establish the efficacy prediction model and verify its stability to obtain the accuracy rate and eliminate the relatively important factors. Results The accuracy rate of the Kechuanting acupoint plastering therapy curative effect prediction model established by the LightGBM model was more than 70%. Five important factors were selected, including allergic history, tabacco and alcohol abuse, plastering duration, ACT before treatment, and FeNO before treatment. According to the classification analysis and the relationship between the important factors and the outcome indicators, Kechuanting acupoint plastering therapy significantly improved the ACT of patients with no history of allergy, no tabacco and alcohol abuse, and poor ACT:5-15 points (P < 0.05). Furthermore, Kechuanting acupoint plastering therapy improved FeNO more significantly in patients with more than 3 years of treatment than those with no more than 3 years (P < 0.05). However, Kechuanting acupoint plastering therapy only improved FeNO in a few patients with poor asthma control levels (P < 0.05) and severe airway inflammation (FeNO>50x10-9) (P>0.05). Conclusion Acupoint plastering application has a significant effect on improving the control level of asthma, but its effect on improving airway inflammation is limited. It is feasible to use data from the chronic disease management research platform to construct the prediction model. After optimization and testing, the predictive model established based on the data of this study may provide an effective evaluation tool for targeted clinical treatment. [ABSTRACT FROM AUTHOR]