Sepsis is a potentially life-threatening complication of a bacterial infection—it kills one in five of the 30 million people who contract it worldwide every year. But a new study in Nature suggests a system built using reinforcement learning could significantly reduce that number.
What is reinforcement learning? It’s a machine-learning technique inspired by the way animals learn through positive feedback. It’s the same technique DeepMind used to create a program that taught itself to play Go.
The research: This method was used by a team from Imperial College London who fed a system data on the way 96,156 sepsis patients admitted to 133 separate intensive care units in the US had been treated. The data included information on medication doses, intravenous fluids, and vasopressors (medicines that constrict blood vessels) given over the first 72 hours after admission to hospital. The system’s end goal was patient survival after 90 days.
Better outcomes: It’s yet to be tested in hospitals, but when compared with an independent validation sample, the program was significantly better at recommending sepsis treatment than human doctors, the team found.
AI assistant: The software is yet another in a growing cadre of AI-powered systems aimed at helping doctors treat tough diseases, especially when it comes to diagnosis. For example, Google’s deep-learning tool recently proved better than human pathologists at spotting metastatic tumors. A number of new apps also use AI techniques to help triage patients.
This artist is dominating AI-generated art. And he’s not happy about it.
Greg Rutkowski is a more popular prompt than Picasso.
What does GPT-3 “know” about me?
Large language models are trained on troves of personal data hoovered from the internet. So I wanted to know: What does it have on me?
An AI that can design new proteins could help unlock new cures and materials
The machine-learning tool could help researchers discover entirely new proteins not yet known to science.
DeepMind’s new chatbot uses Google searches plus humans to give better answers
The lab trained a chatbot to learn from human feedback and search the internet for information to support its claims.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.