Role of artificial intelligence in predicting the renal function after nephrectomy in renal cell carcinoma: a systematic review and meta-analysis.

Abstract

Purpose: To explore and assess the role of artificial intelligence (AI) in predicting the postoperative renal function in Renal Cell Carcinoma (RCC) patients undergoing nephrectomy.

Methods: A comprehensive literature search was conducted across multiple databases, including PubMed, Embase, Scopus, and Web of Science. PRISMA guidelines were followed throughout the systematic review and meta-analysis. The studies that used AI models to predict renal function after nephrectomy were included in our review. The details of different AI models, the input variables used to train and validate them, and the output generated from these models were recorded and analysed. The risk of bias was assessed using the Prediction Model Study Risk of Bias Assessment Tool (PROBAST).

Results: After the screening, a total of nine studies were included for the final analysis. The most common AI algorithms that were used to predict were based on machine learning models, namely Random Forest (RF), support vector machine (SVM) and XGBoost. Different performance metrics of various AI models were analysed. The pooled AUROC (area under the receiver operating curve) of the AI models was 0.79 (0.75-0.84), I² = 15.26%.

Conclusion: AI models exhibit significant potential for determining postoperative renal function in RCC patients. They integrate multimodal data to generate more accurate results. However, standardising the methodologies and reporting, utilising diverse datasets, and improving model interpretability can lead to widespread clinical adaptation.