Background: Although higher-generation TKIs are associated with improved progression-free survival in advanced NSCLC patients with EGFR mutations, the optimal selection of TKI treatment remains uncertain. To address this gap, we developed a web application powered by a reinforcement learning (RL) algorithm to assist in guiding initial TKI treatment decisions. Methods: Clinical and mutational data from advanced NSCLC patients were retrospectively collected from 14 medical centers. Only patients with complete data and sufficient follow-up were included. Multiple supervised machine learning models were tested, with the Extra Trees Classifier (ETC) identified as the most effective for predicting progression-free survival. Feature importance scores were calculated by the ETC, and features were then integrated into a Deep Q-Network (DQN) RL algorithm. The RL model was designed to select optimal TKI generation and a treatment line for each patient and was embedded into an open-source web application for experimental clinical use. Results: In total, 318 cases of EGFR-mutant advanced NSCLC were analyzed, with a median patient age of 63. A total of 52.2% of patients were female, and 83.3% had ECOG scores of 0 or 1. The top three most influential features identified were neutrophil-to-lymphocyte ratio (log-transformed), age (log-transformed), and the treatment line of TKI administration, as tested by the ETC algorithm, with an area under curve (AUC) value of 0.73, whereas the DQN RL algorithm achieved a higher AUC value of 0.80, assigning distinct Q-values across four TKI treatment categories. This supports the decision-making process in the web-based ‘EGFR Mutant NSCLC Treatment Advisory System’, where clinicians can input patient-specific data to receive tailored recommendations. Conclusions: The RL-based web application shows promise in assisting TKI treatment selection for EGFR-mutant advanced NSCLC patients, underscoring the potential for reinforcement learning to enhance decision-making in oncology care.

Background: This study aimed to investigate the effect of cytoreductive nephrectomy (CN) on the survival outcomes of nivolumab used as a subsequent therapy after the failure of at least one anti-vascular endothelial growth factor (VEGF) agent in patients with metastatic clear-cell renal-cell carcinoma (ccRCC). Methods: We included 106 de novo metastatic ccRCC patients who received nivolumab after progression on at least one anti-VEGF agent. Multivariate Cox regression analysis was performed to investigate the factors affecting survival in patients receiving nivolumab. Results: Of the 106 de novo metastatic ccRCC patients, 83 (78.3%) underwent CN. There were no statistical differences between the two groups in terms of age, gender, Eastern Cooperative Oncology Group (ECOG) score, tumor size, International Metastatic RCC Database Consortium (IMDC) risk group, number of previous treatment lines, first-line anti-VEGF therapy, or metastasis sites (p = 0.137, p = 0.608, p = 0.100, p = 0.376, p = 0.185, p = 0.776, p = 0.350, and p = 0.608, respectively). The patients who received nivolumab with CN had a longer time to treatment discontinuation (TTD) [14.5 months, 95% confidence interval (CI): 8.6–20.3] than did those without CN 6.7 months (95% CI: 3.9–9.5) (p = 0.001). The median overall survival (OS) was 22.7 months (95% CI: 16.1–29.4). The patients with CN had a median OS of 22.9 months (95% CI: 16.3–29.4), while those without CN had a median OS of 8.1 months (95% CI: 5.6–10.5) (p = 0.104). In the multivariate analysis, CN [hazard ratio (HR): 0.521; 95% CI: 0.297–0.916; p = 0.024] and the IMDC risk score (p = 0.011) were statistically significant factors affecting TTD; however, the IMDC risk score (p = 0.006) was the only significant factor for overall survival. Conclusions: Our study showed that the TTD of nivolumab was longer in metastatic ccRCC patients who underwent cytoreductive nephrectomy.

Background: We developed clinical–laboratory models for early progression or death within 90 days (EPD90) during PD-1/PD-L1 blockade in advanced non-small cell lung cancer. EPD90 is a pragmatic early-risk endpoint, not a diagnosis of hyperprogressive disease. Methods: This retrospective single-centre cohort included patients treated between January 2018 and March 2026. The four-item baseline score, EPS-4, combined bone metastasis, lactate dehydrogenase (LDH), albumin and C-reactive protein. Among patients alive and progression-free at day 42, D1 added six-week changes in neutrophil-to-lymphocyte ratio (NLR), and D2 additionally included LDH changes. Internal validation used bootstrap resampling and repeated cross-validation. Results: Among 282 patients, 83 (29.4%) had EPD90. In the common baseline complete-case set (n = 252; 76 events), EPS-4 had an apparent area under the curve (AUC) of 0.729 (95% confidence interval [CI] 0.665–0.793). In 238 landmark-eligible complete cases (56 subsequent events), adding NLR change increased AUC from 0.718 to 0.801. Adding LDH yielded an AUC of 0.816, a modest, statistically non-significant increment (delta AUC 0.016; p = 0.094). Bootstrap-corrected calibration slopes were 0.980 for D1 and 0.957 for D2. Conclusions: NLR provided the principal dynamic contribution. D1 is emphasized as the main dynamic model; D2 remains an exploratory extension with uncertain incremental value. Unavailable laboratory collection dates and lack of external validation limit interpretation. The models and proposed thresholds require independent validation before clinical use.

Background: Tumour-growth kinetics required to diagnose hyperprogressive disease are often unavailable in routine care. We developed baseline and dynamic clinical–laboratory models for progression or death within 90 days (EPD90), a pragmatic endpoint for early clinical deterioration, and assessed whether six-week laboratory change improved risk estimation. Methods: This retrospective single-centre cohort included consecutive adults with advanced or unresectable non-small cell lung cancer treated with PD-1/PD-L1 blockade between January 2018 and March 2026. EPS-4 combined bone metastasis, lactate dehydrogenase (LDH) >240 U/L, albumin < 3.5 g/dL and C-reactive protein >10 mg/L. Day-42 landmark models added log2 ratios of neutrophil-to-lymphocyte ratio (NLR) and LDH. Performance was assessed with 2000 bootstrap resamples and repeated cross-validation. Results: Among 282 assessable patients, 83 (29.4%) had EPD90. In common complete cases (n = 252; 76 events), EPS-4 had an AUC of 0.729 (95% CI 0.665–0.793). Among 238 patients alive and progression-free at day 42 (56 subsequent events), AUC increased from 0.718 for EPS-4 alone to 0.801 after adding NLR change and 0.816 after adding LDH change. The final increment from LDH was modest (ΔAUC 0.016; P = 0.094). Conclusions: Baseline EPS-4 and early NLR change identified higher-risk patients, but the models remain development-stage and require independent prospective validation.