| Objective1、To study the relationship among teicoplanin trough concentration(Cmin),efficacy and safety in adult patients.To explore the best safe and effective range of teicoplanin.To explore the factors that affect the trough concentration and clinical efficacy of teicoplanin,in order to provide references for clinical rational use of drugs.2、Establishing a neural network model to predict the teicoplanin serum concentration of patients.Investigating the effect and characteristics of artificial neural network which was applied in predicting teicoplanin serum concentration to provide references for individualized medication design.Methods1、Prospectively collect information on adult patients who used teicoplanin intravenously in a tertiary hospital from January 2019 to October 2020.We used the high performance liquid chromatography(HPLC)to determine teicoplanin trough concentration.The relationship among teicoplanin trough concentration,efficacy and safety was analyzed by non-parametric tests.We drawed the receiver operating characteristic(ROC)curve to explore the reference concentration of effective,hepatotoxicity and nephrotoxicity.The best effective and safe treatment window was analysised by group.Multiple linear regression and logistic regression were used to explore the influencing factors of teicoplanin trough concentration and clinical efficacy.2、The patients’ serum concentration of teicoplanin from January 2019 to June 2020 was used as the modeling data set.The patients’ teicoplanin concentration from July 2020 to October 2020 was used as the verification data set.The artificial neural network method(ANN)was used to predict teicoplanin trough concentrations,simultaneously the feasibility of the method was verified by verification data set.Results1、The information of total of 247 patients were collected,the clinical effective rate and the bacteriological effective rate were both 65.18%.The serum concentration of teicoplanin was significantly correlated with clinical efficacy,bacteriological efficacy,hepatotoxicity and nephrotoxicity.Group analysis showed that when the concentration of teicoplanin was in the range of 15～25mg/L,the patient had good effect and less incidence of hepatotoxicity and nephrotoxicity.The mean initial dose over the first 3days(MID),estimated glomerular filtration rate(e GFR),gender and combined use of loop diuretics had effect on teicoplanin trough concentration(P<0.05).Trough concentration,serum albumin(ALB),MID,duration of medication,and combined sepsis shock were factors that affect clinical efficacy(P<0.05).2、The teicoplanin trough concentration of 137 patients were collected as training samples to build a artificial neural network(ANN)model.The performance of the ANN model was evaluated by mean square error(MSE)and pearson correlation coefficient.The pearson correlation coefficient between the predicted concentration and the measured concentration was 0.943,and the MSE was 0.0249.Through Analyzing 31 cases of verification data we found that the absolute error which less than 20% was 93.55%,mean prediction error(MPE)was-0.59%,and mean absolute prediction error(MAPE)was 0.74%,which showed the result that the established ANN model had good ability of predicting.Conclusion1、This study found that teicoplanin trough concentration was correlated with clinical efficacy,bacteriological efficacy,hepatotoxicity and nephrotoxicity.The optimal therapeutic window was found to be 15～25mg/L.MID,e GFR,gender and combined loop diuretics were factors that affect teicoplanin trough concentration.Trough concentration,ALB,MID,medication time and combined sepsis shock were factors that affect the clinical efficacy of teicoplanin.2、The ANN model for predictiing teicoplanin trough concentration was explored for the first time,and it was confirmed that the ANN method can be used to predict the concentration of teicoplanin with high accuracy and good predictive ability,which can provide a reference for the customization of teicoplanin’s individualized medication regimen. |
PDF Full Text Request