Prediction of treatment response to antipsychotic drugs for precision medicine approach to schizophrenia: randomized trials and multiomics analysis

Table 2 Classification and regression performance of the optimal prediction model (C + PGM model) for treatment response

Cohort		AUC	MAE^*	MAPE^*	RMSE^*	R²
Discovery cohort (CAPOC, n = 2307)
APD	All	0.874	0.209	1.332	0.272	0.478
	Aripiprazole	0.882	0.220	1.203	0.290	0.481
	Haloperidol	0.850	0.179	2.243	0.236	0.501
	Olanzapine	0.878	0.180	1.644	0.239	0.510
	Perphenazine	0.901	0.211	1.185	0.267	0.517
	Quetiapine	0.856	0.241	1.081	0.305	0.392
	Risperidone	0.892	0.181	1.089	0.239	0.521
	Ziprasidone	0.865	0.235	1.291	0.299	0.414
Sex	Male	0.875	0.196	1.316	0.256	0.536
Sex	Female	0.880	0.221	1.347	0.287	0.418
Validation cohort (CAPEC, n = 1379)
APD	All	0.851	0.154	1.470	0.205	0.507
	Aripiprazole	0.882	0.151	1.114	0.195	0.564
	Olanzapine	0.840	0.155	1.628	0.208	0.468
	Risperidone	0.831	0.155	1.671	0.212	0.484
Sex	Male	0.843	0.151	1.926	0.205	0.500
Sex	Female	0.860	0.156	1.043	0.205	0.513

AUC area under the curve; MAE mean absolute error; MAPE mean absolute percentage error; RMSE root mean squared error; R² coefficient of determination; APD antipsychotic drug; CAPOC Chinese Antipsychotics Pharmacogenomics Consortium; CAPEC Chinese Antipsychotics Pharmacogenetics Consortium;
^*Predicted value and record positive and negative syndrome scales (PANSS) reduction rate were scaled into the range from − 1 to 1

ISSN: 2054-9369