Gastric cancer survival prediction using artificial intelligence models based on electronic health records: a systematic review and meta-analysis - Report - MDSpire
Advertisement
Gastric cancer survival prediction using artificial intelligence models based on electronic health records: a systematic review and meta-analysis
Clinical Report: Predicting Survival in Gastric Cancer Patients Using AI Models
Overview
This systematic review and meta-analysis evaluates the effectiveness of AI models using electronic health record data to predict 5-year overall survival in gastric cancer patients post-surgery.
Background
Gastric cancer remains a significant cause of cancer-related mortality, with survival rates after treatment being limited. Current prognostic models, including the TNM staging system, often lack the precision needed for personalized treatment strategies. The integration of artificial intelligence in survival prediction could enhance prognostic accuracy using readily available clinical data.
Data Highlights
Model Type
Pooled Mean AUC Improvement
p-value
Machine Learning vs Traditional
0.04
0.001
Boosting vs Bagging
0.02
0.04
Key Findings
Machine learning models showed a statistically significant improvement in predictive efficacy for 5-year overall survival (5-OS) compared to traditional methods.
The pooled mean AUC improvement was 0.04 (95% CI 0.02–0.07; p = 0.001).
Boosting techniques outperformed bagging strategies with an AUC improvement of 0.02 (p = 0.04).
Key prognostic factors included age, T stage, tumor size, serum albumin or prealbumin levels, and the metastatic-to-examined lymph node ratio.
The nature of clinical input data, particularly blood-derived biomarkers, influenced algorithm performance.
AI models utilizing routinely available clinical information can enhance survival predictions for gastric cancer patients.
Clinical Implications
The findings indicate that the characteristics of input data should be considered when selecting AI algorithms to optimize predictive accuracy.
Conclusion
AI-driven prognostic models utilizing electronic health record data can enhance the accuracy of survival predictions in gastric cancer patients.
