ecancer shared on LinkedIn:
“Prospective validation supports the potential of machine learning to improve risk stratification in newly diagnosed, asymptomatic chronic lymphocytic leukaemia (CLL).
In this ecancer interview from EHA 2026, Carsten Utoft Niemann (Rigshospitalet) discusses the first prospective, multicentre validation of the Chronic Lymphocytic Leukemia Treatment-Infection Model (CLL-TIM).
The study, conducted across nine international sites within the PreVent-ACaLL trial, evaluated whether CLL-TIM could reliably predict the risk of severe infection or treatment requirement within two years.
Patients were stratified according to predicted risk and model confidence.
The high-risk/high-confidence group experienced substantially worse outcomes, with composite infection – and treatment-free survival of 54%, compared with up to 95.9% in lower-risk groups.
Importantly, model confidence also showed reliable calibration, with high-risk/high-confidence patients experiencing poorer outcomes than those classified as high-risk but with lower prediction confidence.
The high-risk/high-confidence group also had lower overall survival, despite survival not being an outcome used to train the model.
These findings support the potential clinical utility of CLL-TIM as a decision-support tool to identify patients who may benefit from closer monitoring, earlier infection prophylaxis strategies and more intensive clinical management.
Watch the full interview with Prof Carsten Niemann here.”
Other articles about ecancer on OncoDaily.