AI is Learning to Diagnose Schizophrenia from a Smartphone Video

September 23, 2019


Speaking into your smartphone for 2 minutes could reveal whether you have a mental health condition. That is according to the developers of an app that analyses facial expressions and speech to diagnose schizophrenia.


The company behind the app, AICure, hopes that it could be used to better support and monitor people with schizophrenia, and eventually those who have other mental health conditions. The current version of the app was developed to measure symptoms of schizophrenia like low mood and difficulty thinking, says Isaac Galatzer-Levy at AI Cure. Conventionally, these have been harder to measure than symptoms like hallucinations and delusions, he says.


To do this, the app tracks facial movements, as well as the content, tone and pitch of a person’s speech. Some people with schizophrenia move more slowly, and show less emotion on their faces, says Galatzer-Levy. It can then send a score to the person’s doctor, rating these symptoms. However, the app isn’t designed to spot all symptoms associated with the condition, such as hallucinations or delusions.


The team tested the app with 21 people who have schizophrenia and nine people who don’t. The participants made weekly recordings over 12 weeks. Each person was also evaluated by a clinician at Mount Sinai Hospital in New York at the start and end of the study.


The results of this trial suggest that the app’s ratings “are highly correlated” with those of a clinician, says Galatzer-Levy, who presented the work at the European College of Neuropsychopharmacology annual meeting in Copenhagen, Denmark.


The app has also been trialled by pharmaceutical company Takeda, which plans to use it to assess how well schizophrenia drugs work in clinical trials.


However, AICure doesn’t yet have enough data to prove its app works, says Saeed Farooq at Keele University, UK. “The sample size is very small,” he says.


“We see these results as proof of concept more then as a complete diagnostic model,” says Galatzer-Levy. “We are working to capture more data,”


Others are worried about how the app could be used. “These kinds of technology can be used outside of the clinical care setting,” says Effy Vayena at the Swiss Federal Institute of Technology in Zurich. She can imagine versions of the app being used by employers, for example. “That is something that I’d be worried about.”


Galatzer-Levy says that although his app is regulated as a healthcare technology, it wouldn’t be difficult for another firm to use the same technology in a similar way. He hopes that the app will help clinicians take better care of their patients, by allowing them to assess people more often.


Pritzker Group Home Page

Pritzker Group Venture Capital

Pritzker Group Private Capital

Pritzker Group Asset Management