Deep Neural Network-Based Survival Analysis for Skin Cancer Prediction in Heart Transplant Recipients
Summary
Heart transplant recipients face high skin cancer risks. Deep learning models accurately predict Squamous Cell Carcinoma (SCC) and Basal Cell Carcinoma (BCC) incidence, outperforming traditional methods.
Area of Science:
- Oncology
- Transplant Surgery
- Medical Informatics
Background:
- Heart transplant recipients have an elevated risk of developing skin cancers, specifically Squamous Cell Carcinoma (SCC) and Basal Cell Carcinoma (BCC).
- Early detection and risk stratification are crucial for managing post-transplant complications.
Purpose of the Study:
- To evaluate the performance of deep learning (DL) models in predicting the incidence of SCC and BCC in heart transplant recipients.
- To compare the accuracy of DL models against traditional Cox regression for survival analysis in this patient population.
Main Methods:
- Utilized the United Network for Organ Sharing (UNOS) database to identify heart transplant recipients.
- Applied Cox proportional hazards model and two deep neural network-based survival models.
- Employed Lasso regression, Chi-square test, and Wilcoxon signed-rank test for risk factor identification.
Main Results:
- Deep learning-based survival models demonstrated superior accuracy in predicting skin cancer incidence compared to the standard Cox regression model.
- DL models effectively assessed skin cancer incidence rates across various time intervals post-transplantation.
Conclusions:
- Deep learning approaches offer significant advantages for survival analysis and risk prediction of post-transplant skin cancer.
- DL models show promise for improving clinical management and patient outcomes in heart transplant recipients at risk for skin cancer.
