Independent Factors Associated with Short-Term Treatment Response in Schizophrenia: A Nomogram-Based Predictive Model

ObjectiveThis study sought to identify independent factors influencing short-term treatment responses in schizophrenia, covering both positive and negative symptoms, and developed a validated nomogram prediction model. Method: The study examined clinical data from 446 schizophrenia patients diagnosed between 2019 and 2023, focusing on health status, admission vital signs, and initial lab results. Participants were split into training and validation groups (6:4 ratio) and classified as improvers or non-improvers based on a ≥ 30% symptom score reduction. LASSO and multivariable logistic regression analyzed the data, with a nomogram illustrating the prediction model. The Bootstrap method generated the ROC curve, and the AUC was calculated. Calibration and clinical decision curves evaluated the model's discrimination, calibration, and clinical utility. Results: The study involved 446 schizophrenia patients, divided into 267 for training and 179 for validation. Of these, 122 improved in psychiatric symptoms, while 324 did not. Key factors linked to improvement- gender, age, smoking, LDL, homocysteine, and folic acid, were identified using LASSO and logistic regression. These factors formed a predictive model, visualized with nomograms. The model's performance, tested with 1000 Bootstrap iterations, had a C-index of 0.7817. ROC analysis showed an AUC of 0.814 for the training group and 0.7863 for the validation group. Conclusion: Gender, age, smoking habits, LDL, homocysteine, and folic acid independently influence the improvement of psychiatric symptoms in schizophrenia. Predictive models using these factors are effective and clinically valuable, with the resulting nomogram serving as a practical tool for evaluating potential symptom improvement in patients.