Dynamic Prediction of Rectal Cancer Relapse and Mortality Using a Landmarking-Based Machine Learning Model: A Multicenter Retrospective Study from the Italian Society of Surgical Oncology—Colorectal Cancer Network Collaborative Group DOI Open Access
Rossella Reddavid, Ugo Elmore, Jacopo Moro

et al.

Cancers, Journal Year: 2025, Volume and Issue: 17(8), P. 1294 - 1294

Published: April 11, 2025

Background: Almost 30% of patients with rectal cancer (RC) who submit to comprehensive treatment experience relapse. Surveillance plays a leading role in early detection. The landmark approach provides more flexible and dynamic framework for survival prediction. Objective: This large retrospective study aims develop machine learning algorithm profile the patient prognosis, especially risk onset RC relapse after curative resection. Methods: A cohort 2450 were analyzed using analysis. Model applied classical cause-specific Cox landmarking approach, while B implemented landmarking-based RSF (random forest) competing algorithm. two models compared terms predictive interpretative ability. bootstrapped validation strategy was employed validate model’s performance prevent overfitting. best-performing hyperparameters selected systematically, ensuring robustness within approach. assessed these factors’ importance interactions accuracy that model. Results: outperformed (mean C-index 0.95 vs. 0.78), capturing complex providing dynamic, individualized predictions. Clinical factors influencing outcomes identified across time allowing accurate timely Conclusions: offers an improvement over traditional methods By accommodating time-dependent variables evolving nature data, this precise tool profiling survival, thereby supporting informed clinical decision-making.

Language: Английский

Dynamic Prediction of Rectal Cancer Relapse and Mortality Using a Landmarking-Based Machine Learning Model: A Multicenter Retrospective Study from the Italian Society of Surgical Oncology—Colorectal Cancer Network Collaborative Group DOI Open Access
Rossella Reddavid, Ugo Elmore, Jacopo Moro

et al.

Cancers, Journal Year: 2025, Volume and Issue: 17(8), P. 1294 - 1294

Published: April 11, 2025

Background: Almost 30% of patients with rectal cancer (RC) who submit to comprehensive treatment experience relapse. Surveillance plays a leading role in early detection. The landmark approach provides more flexible and dynamic framework for survival prediction. Objective: This large retrospective study aims develop machine learning algorithm profile the patient prognosis, especially risk onset RC relapse after curative resection. Methods: A cohort 2450 were analyzed using analysis. Model applied classical cause-specific Cox landmarking approach, while B implemented landmarking-based RSF (random forest) competing algorithm. two models compared terms predictive interpretative ability. bootstrapped validation strategy was employed validate model’s performance prevent overfitting. best-performing hyperparameters selected systematically, ensuring robustness within approach. assessed these factors’ importance interactions accuracy that model. Results: outperformed (mean C-index 0.95 vs. 0.78), capturing complex providing dynamic, individualized predictions. Clinical factors influencing outcomes identified across time allowing accurate timely Conclusions: offers an improvement over traditional methods By accommodating time-dependent variables evolving nature data, this precise tool profiling survival, thereby supporting informed clinical decision-making.

Language: Английский

Citations

0