Reinforcement learning
Machine learning that learns the best actions by trial and error in its environment, guided by rewards. A robot vacuum learns the layout of a home this way, memorising every misstep to improve the next run.
A supplier pitches an ICU tool that recommends fluid and vasopressor doses for septic patients, trained by reinforcement learning on years of logged ICU records. The model never got to try a dose and see what happened, so wherever it recommends something the clinicians in those records rarely did, it is estimating outcomes it has barely observed. Ask how the policy was evaluated offline and how often its advice departs from what was actually given, then plan a silent run alongside the team before anyone acts on a recommendation.
Terms like this come up in real clinical scenarios across the HelloAI courses: bite-sized modules with verifiable certificates. An account takes one minute, no password needed.
Sign in →