Autoplan

Autoplan automates the most time-consuming step in radiotherapy: the design of the patient-specific treatment plan. It turns roughly three hours of expert manual work into about twenty-five minutes of automated computation, at a plan quality that matches or exceeds what even experienced clinics deliver consistently.

Demand for radiotherapy will roughly double to 20 million patients per year by 2040, while the world heads into a structural shortage of more than 80,000 medical physicists by 2050. The bottleneck is not the treatment machines but the specialists who plan the treatments.

Autoplan runs inside Varian Eclipse, the planning system used in more than 3,000 centres worldwide. It is deterministic rather than AI-learned, and therefore reproducible, explainable, and free of any need to train on local patient data before go-live. It covers the ten most common cancer types, is configurable per clinic, and requires no new staff to start.

Autoplan has been in daily clinical use at Amsterdam UMC, the Netherlands’ largest radiotherapy centre, for eight years, used in the treatment of more than 9,000 patients and serving as the default planning tool there, trusted by the physicists who plan with it and the clinicians who approve the resulting plans. An academic partner clinic abroad provides independent validation, with clinical go-live scheduled for August 2026.