A machine learning model for predicting blood concentration of quetiapine in patients with schizophrenia and depression based on real‐world data

• DOI: 10.1111/bcp.15734
• Publication Date: 2023-04-03T04:04:30Z
• Authors: Yupei Hao; Jinyuan Zhang; Lin Yang; Chunhua Zhou; Ze Yu; Fei Gao; Xin Hao; Xiaolu Pang; Jing Yu
• Keywords: Therapeutic Drug Monitoring; Univariate; Quetiapine Fumarate; Univariate analysis; Depression

This study aimed to establish a prediction model of quetiapine concentration in patients with schizophrenia and depression, based on real-world data via machine learning techniques assist clinical regimen decisions.A total 650 cases therapeutic drug monitoring (TDM) from 483 at the First Hospital Hebei Medical University 1 November 2019 31 August 2022 were included study. Univariate analysis sequential forward selection (SFS) implemented screen important variables influencing TDM. After 10-fold cross validation, algorithm optimal performance was selected for predicting TDM among nine models. SHapley Additive exPlanation applied interpretation.Four (daily dose quetiapine, type mental illness, sex CYP2D6 competitive substrates) through univariate (P < .05) SFS The CatBoost best predictive ability (mean [SD] R2 = 0.63 ± 0.02, RMSE 137.39 10.56, MAE 103.24 7.23) chosen mean (SD) accuracy predicted within ±30% actual 49.46 3.00%, that recommended range (200-750 ng mL-1 ) 73.54 8.3%. Compared PBPK previous study, shows slightly higher ±100% value.This work is first predict blood depression using artificial intelligent techniques, which significance value medication guidance.