US veterans diagnosed by deep-learning

30 November 2016
The deep learning tool shifts through huge amounts of data to find useful information. Flow Health focuses on building a knowledge graph, a database containing information about people’s genomes and phenotypes, to identify disease risks and recommend treatments.

“Our mission is to advance healthcare by applying the latest artificial intelligence techniques to improve the detection, diagnosis, treatment and management of diseases,” said Alex Meshkin, CEO of Flow Health. “The VA supports millions of Americans who have served our nation and deserve our honour, respect and the best care our country has to offer. Through our partnership with the VA, Flow Health is working to unleash the power of AI to benefit our nation’s veterans.”

Data from 22 million veterans

AI needs access to heaps of data to be able to make useful insights. Fortunately medicine provides huge amounts of this. There are an estimated 30 petabytes of clinical data locked inside the VA records from 22 million veterans over 20 years. Flow Health will be building the “world’s largest” knowledge graph on medicine and genomics.

AI is a very interesting subject for healthcare. DeepMind, Google’s British AI firm, has a team dedicated to health. And coincidently they have also just signed a five-year partnership with the Royal Free London NHS Foundation Trust to develop an app called Streams, as published last week.

With access to the UK’s National Health Service data, DeepMind will use Streams to alert clinicians if their patients show high risk of developing acute kidney injury, among other things. AI won’t be used only to analyse patterns in health care data, but eventually may even start dishing out health advice.

Gartner analysts have predicted that up to 50% of the population will rely on virtual personal health care assistants by 2025. Companies such as Baidu Research, the research arm of a Chinese web services company, have developed Melody, an “AI-powered conversational bot” to provide relevant information for doctors and patients.