# Developing a model for predicting suicide risk among prostate cancer survivors

## TL;DR

This study creates a model to predict suicide risk in prostate cancer survivors using clinical data, helping identify high-risk patients for early intervention.

## Contribution

The study introduces a novel predictive model for suicide risk in prostate cancer survivors using seven clinical variables and validates its effectiveness.

## Key findings

- The model achieved C-indices of 0.702 in training and 0.688 in validation cohorts, showing good discriminative ability.
- High-risk prostate cancer survivors had a 3.5 times higher suicide risk compared to low-risk individuals.
- The model's ROC values at 3, 5, and 10 years were 0.727/0.644, 0.700/0.698, and 0.735/0.708, respectively.

## Abstract

Given the significantly higher suicide risk among cancer survivors compared to the general population, and considering that prostate cancer survivors make up the largest group of cancer survivors, it is imperative to develop a model for predicting suicide risk among prostate cancer survivors.

Clinical data of prostate cancer patients were extracted from the surveillance, epidemiology, and end results (SEER) database and randomly divided into a training cohort and a validation cohort in a 7:3 ratio. Initial variable selection was performed using univariate Cox regression, Best Subset Regression (BSR), and Least Absolute Shrinkage and Selection Operator (LASSO). Variables to be included in the final model were selected using backward stepwise Cox regression. Model performance was evaluated using the Concordance Index (C-index), Receiver Operating Characteristic (ROC) curves, and calibration curves.

Data from 238,534 prostate cancer patients were obtained from the SEER database, of which 370 (0.16%) died by suicide. Seven variables including age, race, marital status, household income, PSA levels, M stage, and surgical status were included in the final model. The model demonstrated good discriminative ability in both the training and validation cohorts, with C-indices of 0.702 and 0.688, respectively. ROC values at 3, 5, and 10 years were 0.727/0.644, 0.700/0.698, and 0.735/0.708, respectively. Calibration curves indicated a high degree of consistency between model predictions and actual outcomes. High-risk prostate cancer survivors had a 3.5 times higher risk of suicide than the low-risk group (0.007 vs. 0.002, P < 0.001), a finding supported by data from the validation cohort and the entire cohort.

A reliable predictive model for suicide risk among prostate cancer survivors was successfully established based on seven readily obtainable clinical predictors. This model can effectively aid healthcare professionals in quickly identifying high-risk prostate cancer survivors and timely implementation of preventive interventions.