Federated learning

A way to train one AI model across several hospitals or sites without moving any patient data out of where it lives. Instead of pooling everyone's records in one place, the model is sent to each site, learns locally, and only the model updates are shared back and combined. It lets AI learn from large, diverse data while the records stay put, which is why it is attractive for privacy-sensitive healthcare data.

Brain imaging is the clearest case. No single hospital has enough scans of a rare tumor to train a good model alone, and the scans can't legally leave the building. With federated learning, each site trains on its own images and sends back only the math, so a model learns from many hospitals' worth of cases while every patient's MRI stays on its own servers. The Ontario Brain Institute built exactly this with Brain-CODE and its NeuroFL platform, training across neurological data from more than 20,000 people without pooling it.

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.