Explainable machine learning/artificial intelligence model for estimating postoperative survival in renal cell carcinoma brain metastasis patients
Neuro-Oncology Advances, cilt.8, sa.1, 2026 (ESCI, Scopus)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 8 Sayı: 1
- Basım Tarihi: 2026
- Doi Numarası: 10.1093/noajnl/vdag140
- Dergi Adı: Neuro-Oncology Advances
- Derginin Tarandığı İndeksler: Emerging Sources Citation Index (ESCI), Scopus, EMBASE, Directory of Open Access Journals
- Anahtar Kelimeler: AI, brain metastasis, craniotomy, machine learning, RCC
- Süleyman Demirel Üniversitesi Adresli: Hayır
Özet
Background: Brain metastases (BM) from renal cell carcinoma (RCC) are associated with poor prognosis and limited survival. Prognostic tools specific to patients with RCC undergoing surgical resection of BM are lacking, and current models do not incorporate advanced machine learning (ML) approaches. This study aimed to develop and validate an ML-based model to predict overall survival (OS) after BM resection in RCC. Methods: We retrospectively analyzed 253 patients with histologically confirmed RCC and radiographically or pathologically confirmed BM who underwent neurosurgical resection at a tertiary referral center (1993-2021). Clinical and radiologic features were used to train and internally validate multiple ML models for OS prediction. Model performance was assessed using the concordance index (C-index) and time-dependent Area Under the Curve (AUC) at 1, 2, and 5 years. Feature importance and interpretability were evaluated using SHapley Additive exPlanations (SHAP). Results: The XGBoostCox + plsRcox model outperformed other algorithms, achieving a test C-Index of 0.59. AUCs at 1, 2, and 5 years were 0.61, 0.64, and 0.69 in the test cohort. SHAP analysis revealed extracranial disease status, number of BM, preoperative symptoms, and age at surgical resection as the most influential predictors. Kaplan-Meier analysis using an optimal cutoff based on the training cohort demonstrated significant survival differences between high- and low-risk groups in the test cohort (HR: 2.06 [1.26-3.35], P = .004). Conclusions and Relevance: An explainable XGBoostCox + plsRcox model accurately predicts OS after BM resection in RCC and enables personalized risk assessment via an online calculator (https://hasanovlab-rcc-bm-resect.share.connect.posit.cloud/).