| Objective:Through the large sample and retrospective data of real clinical diagnosis and treatment,the effect of Qinggan Jianpi method on the titer of thyroid peroxidase antibody(TPOAb)and thyroglobulin antibody(TgAb)in patients with Hashimoto’s thyroiditis(HT)was studied,and the characteristics of the dominant population with decreased antibody titer were analyzed,and then the prediction model of the dominant population was constructed.In order to optimize the clinical diagnosis and treatment of HT treated with traditional Chinese medicine,to find the advantages of treatment from the HT population treated with traditional Chinese medicine,to predict the possibility of a significant decrease in TPOAb and TgAb titers,and to provide realworld evidence-based medical evidence for the clinical practice of traditional Chinese medicine in the treatment of HT.Method:1 Effect of Qinggan Jianpi method on TPOAb and TgAb titers in HT patientsA total of 1649 patients with HT who were treated in the Dongzhimen hospital Thyroidology department from June 2018 to September 2021 were included.According to the time of taking traditional Chinese medicine,the patients were divided into low exposure group(≤30 days),medium exposure group(30-90 days)and high exposure group(90-180 days).The effects of different time of taking traditional Chinese medicine on the decrease percentage,decrease difference and treatment response of TPOAb and TgAb titers were observed.2 The dominant population characteristics of TPOAb and TgAb titer decline in HT patientsThe patients were divided into dominant population and non-dominant population according to whether TPOAb and TgAb titers decreased by more than 30%.Logistics regression,least absolute shrinkage and selection operator(LASSO)regression,combined with clinical practice,were used to compare the dominant population and non-dominant population,and to study the characteristics of dominant population of HT patients with significantly decreased antibody titers after treatment with traditional Chinese medicine.3 Establishment and verification of prediction model for TPOAb and TgAb titer decline dominant population based on machine learningThe data set is randomly divided into training set and test set according to 7:3.The training set is used to establish the model,and the test set is used to verify the model.The characteristics of the dominant population were introduced as predictors,and whether it was the dominant population was used as a clinical outcome.The discrimination,calibration and clinical utility of the model were evaluated by combining machine learning algorithms of logistic regression,random forest and artificial neural network.The model was displayed in the form of nomogram,and the optimal model was selected to construct the clinical prediction model of TPOAb and TgAb titer decline dominant population.Results:1 Effect of Qinggan Jianpi method on TPOAb and TgAb titers in HT patients1.1 Effect on TPOAb titerIn terms of the decrease of TPOAb titer,the medians of the drop percentage in the low exposure group,the medium exposure group and the high exposure group were 3.0%,8.0%and 21.0%,respectively,and the medians of the drop difference were 3.61 IU/mL,14.2 IU/mL and 28.9 IU/mL,respectively.There were 39 patients(13.9%),131 patients(21.2%)and 140 patients(37.9%)with a drop percentage≥30%.There were statistically significant differences in the percentage of decrease in TPOAb titer,decrease difference and treatment response among the three groups(P<0.05).Pairwise comparison,in terms of the percentage and difference of TPOAb titer decline,the differences between the low exposure group and the medium exposure group,the low exposure group and the high exposure group,the medium exposure group and the high exposure group were statistically significant(P<0.05).There was no significant difference in the decrease percentage≥30%between the low exposure group and the medium exposure group(P>0.05),while there were significant differences between the low exposure group and the high exposure group,the medium exposure group and the high exposure group(P<0.05).1.2 Effect on TgAb titerIn terms of the decrease of TgAb titer,the medians of the drop percentage in the low exposure group,the medium exposure group and the high exposure group were 1.0%,10.0%and 23.0%,respectively,and the medians of the drop difference were 3.15 IU/mL,39.72 IU/mL and 94.70 IU/mL,respectively.There were 44 patients(15.5%),163 patients(25.9%)and 165 patients(42.4%)with a drop percentage ≥30%.There were significant differences in the percentage of TgAb titer decrease,decrease difference and treatment response among the three groups(P<0.05).Pairwise comparison,low exposure group and medium exposure group,low exposure group and high exposure group,medium exposure group and high exposure group,the differences were statistically significant(P<0.05).Subgroup analysis and sensitivity analysis were performed on the percentage of TPOAb and TgAb titers.The results showed that the antibody titers in the high-exposure group of traditional Chinese medicine decreased more significantly,which was consistent with the above research results.2 The dominant population characteristics of TPOAb and TgAb titer decline in HT patients2.1 Characteristics of dominant population with decreased TPOAb titerLogistics regression results showed that the clinical features that were conducive to a significant decrease in TPOAb titers included female(odds ratio(OR)=2.430,95%confidence interval(CI):1.403 to 4.463,P<0.05),positive thyrotropin receptor antibody(TRAb)(OR=1.676,95%CI:1.056 to 2.643,P<0.05),medium exposure to traditional Chinese medicine(OR=1.795,95%CI:1.205 to 2.726,P<0.05)and high exposure to traditional Chinese medicine(OR=3.839,95%CI:2.534 to 5.926,P<0.05).The clinical features that are not conducive to the significant decrease of TPOAb titer include age 30-50 years old(OR=0.648,95%CI:0.470 to 0.896,P<0.05),age≥50 years old(OR=0.343,95%CI:0.224 to 0.520,P<0.05).The results of LASSO regression showed that the clinical characteristics related to the outcome and the corresponding regression coefficients were gender(0.022),age(-0.051),FT3(0.018),TSH(0.001),TRAb(0.016),thyroid function(0.031)and the time of taking traditional Chinese medicine(0.079).2.2 Characteristics of dominant population with decreased TgAb titerLogistics regression results showed that the clinical features that were beneficial to the significant decrease of TgAb titer included TRAb positive(OR=1.721,95%CI:1.062 to 2.770,P<0.05),medium exposure to traditional Chinese medicine(OR=2.149,95%CI:1.471 to 3.198,P<0.05)and high exposure to traditional Chinese medicine(OR=4.263,95%CI:2.871 to 6.443,P<0.05).The clinical features that were not conducive to the significant decrease of TgAb titer included age 30-50 years old(OR=0.718,95%CI:0.526 to 0.981,P<0.05),age ≥50 years old(OR=0.533,95%CI:0.364 to 0.777,P<0.05)and TgAb high antibody(OR=0.582,95%CI:0.448 to 0.754,P<0.05).The results of LASSO regression showed that the clinical characteristics related to the outcome and the corresponding regression coefficients were age(-0.015),FT3(0.036),FT4(0.005),TgAb(-0.028),TRAb(0.028),thyroid function(0.040)and the time of taking traditional Chinese medicine(0.089).Logistic regression,LASSO regression and clinical practice were combined to screen the population characteristics.The clinical characteristics related to the dominant population with decreased TPOAb titer included gender,age,TRAb,thyroid function status and time of taking traditional Chinese medicine;the clinical features related to the dominant population with decreased TgAb titer include age,TgAb,TRAb,thyroid function status and time of taking traditional Chinese medicine.3 Establishment and verification of prediction model for TPOAb and TgAb titer decline dominant population based on machine learning3.1 Prediction model of dominant population with decreased TPOAb titerLogistic regression model in the training set,AUC=0.720,95%CI:0.681 to 0.759,Brier score was 0.159;in the test set,AUC=0.655,95%CI:0.593 to 0.717,Brier score was 0.187.The random forest model was in the training set,AUC=0.696,95%CI:0.655 to 0.737,Brier score was 0.187;in the test set,AUC=0.631,95%CI:0.568 to 0.693,Brier score was 0.228.Artificial neural network model in the training set,AUC=0.756,95%CI:0.719 to 0.792,Brier score was 0.151;in the test set,AUC=0.620,95%CI:0.553 to 0.687,Brier score was 0.196.3.2 Prediction model of TgAb titer decline in dominant populationLogistic regression model in the training set,AUC=0.702,95%CI:0.665 to 0.739,Brier score was 0.178;in the test set,AUC=0.644,95%CI:0.584 to 0.704,Brier score was 0.200.The random forest model was in the training set,AUC=0.700,95%CI:0.661 to 0.739,Brier score was 0.202;in the test set,AUC=0.632,95%CI:0.573 to 0.690,Brier score was 0.256.Artificial neural network model in the training set,AUC=0.749,95%CI:0.715 to 0.784,Brier score 0.162;in the test set,AUC=0.623,95%CI:0.563 to 0.683,Brier score was 0.217.In the training set and the test set,the logistic regression model performs better in terms of discrimination,calibration,and interpretability.Therefore,the logistic regression model was selected as the optimal model to construct the prediction model of TPOAb titer decline dominant population and the prediction model of TgAb titer decline dominant population.Conlusions:1 The longer time of taking traditional Chinese medicine(90-180 days)is more conducive to reducing the titer of TPOAb and TgAb than the medium time(30-90 days)and the shorter time(≤30 days).2 The common dominant population characteristics of TPOAb and TgAb titers in HT patients were TRAb positive,thyroid function status(hypothyroidism,hyperthyroidism)and Chinese medicine taking time(30-90 days,90-180 days).The decreasing trend of TPOAb titer in female patients was more obvious.3 The TPOAb titer decline advantage population prediction model and the TgAb titer decline advantage population prediction model have good prediction performance,which can objectively predict the probability of antibody titer decline significantly.At the same time,they can be used as an effective tool for predicting antibody titer decline in HT patients treated with traditional Chinese medicine,and provide individualized reference for clinical practice. |
PDF Full Text Request