Objective To evaluate the consistency between the intelligent interpretation of antidepressant therapeutic drug monitoring （TDM） results using blood drug concentration data processing software and reference standards， and assess its clinical utility in assisting junior psychiatrists in identifying abnormal TDM results. Methods From March 1 to 31， 2024， 226 patients with depressive disorder hospitalized at Beijing Anding Hospital affiliated with Capital Medical University were selected as research subjects. Firstly， the consistency between the TDM results interpreted by the software and the reference standards was verified. Subsequently， 10 junior physicians who passed the assessment and had good consistency were recruited from three clinical centers for TDM result interpretation. All TDM results were randomly and evenly distributed， and each physician interpreted the TDM results of the same case twice in sequence， once by an independent physician and once by software assisted physician （software + physician）. Paired χ2 test was used to analyze the differences between two interpretation methods and reference standards， as well as the differences in sensitivity， specificity， and accuracy. The area under the receiver operating characteristic （ROC） curve （AUC） was used to compare the efficiency of two interpretation methods. Results Among the TDM results of which of 226 patients， 109 （48.23%） were abnormal. The consistency rate between the abnormal TDM results identified by the software and the reference standard was 100%. The accuracy of individual physician interpretation was 65.93% （95%CI=59.35%-72.08%）， lower than the assistance of "software"， of which the accuracy was 99.12% （95%CI=96.84%-99.89%）， and the difference was statistically significant （χ2 =75.00， P ＜ 0.001）. Paired χ2 test showed that the sensitivity （χ2 =64.00， P ＜ 0.001） and specificity （χ2 =6.23， P＜ 0.05） of the software + physician combined interpretation of abnormal TDM results were superior to those of independent physician interpretation， and the differences were statistically significant. The comparison of AUC between the two interpretation methods showed that the accuracy of software + physician combined interpretation was better than that of independent physician interpretation （0.991 vs 0.647）， and the difference was statistically significant （Z=12.51， P＜ 0.001）. Conclusions The serum concentration data processing software shows high consistency between the TDM results and the reference standards， which helps to improve the identification and cause analysis of abnormal TDM results by junior psychiatrists.