Prediction of renal cell carcinoma: Development and validation of machine learning model

Comprehensive Summary

Zheng et al. developed a machine learning model to predict renal cell carcinoma (RCC) risk based on clinical and laboratory data. The study analyzed data from 1,057 individuals treated at a single hospital in China, including 376 RCC patients and 681 healthy controls. Seven machine learning algorithms were tested, and XGBoost showed the best performance. After further selection using recursive feature elimination, the final model incorporated 21 clinical variables, such as age, albumin, creatinine, glucose, inflammatory markers, and electrolyte ratios. In the validation cohort, the model achieved an AUC of 0.955 with strong accuracy and calibration, which depicted agreement between predicted and observed risk. SHAP analysis was also used to show how each variable contributed to predictions, with age and several inflammatory and metabolic markers having the greatest influence. Overall, the model demonstrated strong potential for identifying individuals at higher risk of RCC using accessible and noninvasive data.

Outcomes and Implications

Renal cell carcinoma is often diagnosed incidentally and at later stages, when symptoms appear and treatment options become more limited. This study is important because it shows how machine learning could help identify individuals at higher risk using routine blood tests and basic clinical data. They also relied on commonly available clinical variables, which allowed their approach to be relatively low-cost and easier to scale compared to imaging-based screening. If validated more broadly, models like this could be used in primary care settings to identify patients who may benefit from closer monitoring. However, because the study was retrospective and based on data from a single center in China, broader validation in more diverse populations will be necessary. With further testing, this type of model could become a useful tool for identifying individuals who may be at risk for RCC and positively affect RCC patient outcomes.