Artificial intelligence is proving increasingly impressive in the fight against deadly diseases. The latest achievement is the development (in just 30 days) of an innovative treatment against hepatocellular carcinoma.
Thanks to the collaboration between researchers at the University of Toronto and Insilico Medicine, an AI platform called Pharma (based on Alphafold) led to the discovery of an unprecedented treatment pathway. Doctors have been able to attack this form of liver cancer by designing a revolutionary molecule capable of binding to the target.
I take care of you and tell you how long you will live
A second AI system turned out to be able to predict with over 80% accuracy patient survival at 6, 36 and 60 months. He did it by analyzing the unique characteristics of each through the notes of the oncologists.
This extraordinary result was achieved thanks to AI training on a large sample. Very large: 47.625 patients from six cancer centers located throughout British Columbia.
What prospects are there?
John-Jose Nunez, a psychiatrist and clinical researcher at the UBC Mood Disorders Center and involved in the study, explains it clearly: “AI essentially reads data as a human would read it. These records contain many details such as the patient's age, type of cancer, underlying health conditions, past substance use, and family histories.
The only difference? In the extreme ability of AI to combine this data together (which is more than has been used to date to predict cancer survival rates) to paint a complete picture of the results”.
ChatGPT for medications and survival estimation
Today, generative artificial intelligence makes texts and images available to us on request (warming up the engines to soon give us videos generated from a few lines of text). Tomorrow morning he will give us back medicines based on the diseases we will ask him to treat. Not only that: it will perfectly calculate our chances of survival.
The typology of these neural networks makes me think of a very close turning point in the diagnosis and treatment of diseases. It's their scalability. Their portability. These are very powerful systems that don't require huge structured data sets to work.
They can be trained quickly using local data from hospitals in a single city or county to help improve healthcare performance across the affected area.
An intelligent weapon. A tool that just three years ago no one would have thought could exist.
