Google has recently developed an algorithm capable of predicting the length of hospital stays of patients and the risks of readmission or even death, according to Développez.com.
Tech giant Google has recently opened several research centres dedicated to artificial intelligence including ones in France and China, and has made incredible progress in this field. The latest technical advancement is an algorithm which has proven to be able to calculate the risks of fatal recurrences in a patient suffering from the advanced stages of breast cancer, according to Bloomberg. Using 175,639 data points, the AI announced a risk of death of 19.9%. The patent passed away a few days later in hospital.
Ultra-accurate predictions in record time
The calculations carried out by the Google algorithm proved to be more accurate than those of the hospital’s computers which announced a mortality risk of 9.3%. The AI analyses all of a patient’s data in great depth and in record time, including notes in PDF and even hand-written form, and was even capable of providing invaluable predictions regarding the length of hospital stays and risks of readmission.
The tool can calculate and also provide documental summaries very rapidly, and according to Nigam Shah, an associate professor at the University of Stanford and co-author of the Google research document published in the "Nature" review, would be a godsend for health-care professionals, who he says spend up to 80% of their time inputting and formatting patient data.
A revolutionary new tool for clinics?
"The next stage would be to deploy this new tool in clinics," said Jeff Dean, head of AI at Google and who is also in charge of directing research into other tools that can predict with great accuracy the appearance of symptoms and illnesses. This would be a potential field of application for AI that would be highly useful for medicine and also one that makes great commercial sense for Google.
According to Vik Bajaj, former head of Alphabet Healthcare Verily, Google’s approach goes beyond the software currently used in medical centres which are primarily coded by hand. Thanks to this IA, machines can learn how to analyse the data themselves. "They understand which problems are worth resolving," Bajaj explains.
Furthermore, engineers in Dean’s team are already working on a system that can guide doctors in diagnostics and choice of treatment.
Contact Allianz Partners
Mar 15, 2018
DeepMind – the Google subsidiary dedicated to artificial intelligence –and the Moorfields Eye Hospital in London have developed a form of IA which can detect anomalies on retinal sc [...]