Objective To explore the biological factors influencing the prolonged length of stay in first-episode schizophrenia patients， construct a predictive model， and validate it. Methods A total of 203 schizophrenia patients in Shanghai Baoshan Mental Health Center from January 2022 to December 2023 were retrospectively selected for the study. Variables such as demographic and biological indicators， primary medications during hospitalization， blood concentrations， and length of stay were collected from patients before treatment and within one week before discharge. Pearson or Spearman correlation was used to analyze the correlation between length of stay and 25 hydroxyvitamin D［ （25（OH）D）］ in patients on different medications. The patients included in the prediction model were divided into a training set and a test set. LASSO regression was used to screen variables for the multifactor Logistic regression model. A predictive model was constructed by applying a nomogram to the independent risk factors for prolonged length of stay. Receiver operator characteristic （ROC） curve， calibration curve， and decision curve analysis were used to evaluate the predictive model. Results A total of 155 cases of first-episode schizophrenia were included in prediction model analysis and randomly assigned to training set（ 108 cases） and test set（ 47 cases） in a 7∶3 ratio. Follow up on indicators of the patient before treatment and within one week before discharge showed that， the levels of glycated hemoglobin， triglycerides， lipoproteins， cystatin C， prolactin， thyroid stimulating hormone， and 25（OH）D increased， levels of fasting blood glucose， total cholesterol， high-density lipoprotein， low-density lipoprotein， blood calcium， total triiodothyronine， total thyroxine， free triiodothyronine， free thyroxine， and cortisol decreased， levels of procalcitonin decreased and levels of immunoglobulin A increased in immune and inflammatory indicators within one week before discharge compared with those before treatment， and the differences were statistically significant（ all P＜0.05）. Multifactorial Logistic regression analysis showed that age［ OR=1.04， 95%CI（ 1.01， 1.07），P=0.002］ and 25（OH）D level［ OR=1.09，95%CI（ 1.02，1.16），P=0.002］ were risk factors for prolonged length of stay， and the difference was statistically significant. The area under the ROC curve of the training set and test set of the nomogram prediction model for prolonged length of stay was 0.698 and 0.716， respectively. For patients who applied aripiprazole， aripiprazole blood concentrations at one month post-treatment were negatively correlated with pre-treatment 25（OH）D levels with a statistical difference（ r=-0.40，P=0.027）. Conclusions This study constructs and validates a prediction model for prolonged length of stay in firstepisode schizophrenia， and the nomogram prediction model performs well. Older， higher 25（OH）D levels may result in patients having longer hospital stays， and high 25（OH）D levels may prolong hospital stays by affecting blood concentrations.