Development and validation of a recurrence risk prediction model for elderly schizophrenia patients
Abstract Objective To construct a recurrence prediction model for Elderly Schizophrenia Patients (ESCZP) and validate the model’s spatial external applicability. Methods The modeling cohort consisted of 365 ESCZP cases from the Seventh People’s Hospital of Dalian, admitted between May 2022 and April...
Saved in:
| Main Authors: | , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
BMC
2025-01-01
|
| Series: | BMC Psychiatry |
| Subjects: | |
| Online Access: | https://doi.org/10.1186/s12888-025-06514-y |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract Objective To construct a recurrence prediction model for Elderly Schizophrenia Patients (ESCZP) and validate the model’s spatial external applicability. Methods The modeling cohort consisted of 365 ESCZP cases from the Seventh People’s Hospital of Dalian, admitted between May 2022 and April 2024. Variables were selected using Lasso-Logistic regression to construct the recurrence prediction model, with a nomogram plotted using the “RMS” package in R 4.3.3 software. Model validation was performed using 1,000 bootstrap resamples. Spatial external validation was conducted using 172 cases ESCZP from the Fourth Affiliated Hospital of Qiqihar Medical College during the same period. The model’s discrimination, accuracy, and clinical utility were assessed using Receiver Operating Characteristic (ROC) curves, Area Under the Curve (AUC), calibration curves, and Decision Curve Analysis (DCA). The Hosmer-Lemeshow test was used to evaluate model fit. Results A total of 537 cases ESCZP were included based on inclusion and exclusion criteria, with 150 recurrences within two years and 387 non-recurrences. Lasso-Logistic regression analysis identified Medication status, Premorbid personality, Exercise frequency, Drug adverse reactions, Family care, Social support, and Life events as predictors of ESCZP recurrence. The AUC for the modeling cohort was 0.877 (95% CI: 0.837–0.917). For the external validation cohort, the AUC was 0.838 (95% CI: 0.776–0.899). Calibration curves indicated that the fit was close to the reference line, demonstrating high model stability. DCA results showed good net benefit at a threshold probability of 80%. Conclusion The nomogram prediction model developed based on Lasso-Logistic regression shows potential in identifying the risk of recurrence in ESCZP. However, further validation and refinement are needed before it can be applied in routine clinical practice. |
|---|---|
| ISSN: | 1471-244X |