For a young man suffering from bipolar disorder an app was able to predict an emerging manic episode — and help avoid a hospital stay — more than a week before his closest friends.
"This was very exciting, because in psychiatry we use mostly subjective measures," said Dr. Dror Dolfin, a senior psychiatrist at the GEHA mental health center in Israel. Dolfin has been running clinical trials of new technology created by Israeli start-up Lifegraph.
"A day-to-day plot of the course of this person's manic episodes — it was looking at an X-ray for the first time," he said. "We don't have that in psychiatry."
Lifegraph draws on data collected by a person's smartphone to detect changes in their behavior and send alerts. Psychiatrists in Israel have used the technology to identify early warning signs of distressful behavior in patients. The company's software can detect a mental health episode a full month before a person requires hospitalization, clinical trials Lifegraph conducted in Israel have shown.
Patients download the app and then leave it running in the background for continuous tracking. The app monitors things like the pitch of a person's voice on the phone, how many hours they slept, the number of text messages that a person sends or how far they travel any given day.
Machine learning algorithms get to know the patient and send doctors alerts when a person exhibits unusual behavior. Doctors and caregivers can track all the data in a dashboard.
One in five people in the U.S. suffer from a mental health illness, and mood disorders are the third most common cause of hospitalization in the U.S. for youth and adults aged 18 to 44, according to the National Alliance of Mental Illness. Serious mental illness costs America $193 billion in lost earnings every year, according to NAMI.
"In the clinical setting, there are hopeful trends towards bringing new data to bear," Stanford researchers noted in the One Hundred Year Study on Artificial Intelligence, which highlighted Lifegraph's technology.
Lifegraph's technology holds great potential to improve diagnoses and supplement what patients tell doctors explicitly, said Dolfin. This will make it easier to monitor how patients respond to medication and enable doctors to intervene before a patient's mental state requires hospitalization, for example.
Psychiatric hospitalization is very destructive for a person's life, said Dolfin. It is also extremely costly for insurers, employers and state funded programs. Behavioral-illness related hospital stays in the U.S. cost $45 billion annually, according to Lifegraph.
The company, which came out of Tel Aviv University, is now lining up partnerships with tele-care centers in the U.S. that are looking for ways to reduce costs and improve care for mental health patients, said Lifegraph CEO Keren Sela in an email. Pharmaceutical companies are also interested in using the technology in clinical trials, she said.
Lifegraph received funding from the office of the Israeli chief scientist and Tel Aviv University, and is talking to investors about financing to fund further growth, said Sela.
Thousands of health related apps, combined with a rise in specialized fitness tracking devices, like Fitbit, and the increasing connectivity between those devices and people's homes has created a "vibrant new sector of innovation," Stanford researchers noted.
Of course, the collection of so much highly personalized data raises privacy concerns.
"I can actually see that this data would be tremendously helpful and I can see why this would be appealing," said Corynne McSherry, Legal Director of the Electronic Frontier Foundation. "Patients would want to be absolutely confident that this information was going to be transmitted in encrypted ways and that the database would have extraordinary protection."
Database breaches over the past several years have shown just how difficult it is for companies to keep data safe, she said.
CEO Sela said that Lifegraph patient privacy is strictly maintained, but the company does keep some of a users raw metadata.
"When comparing to all other very intrusive means, an app for preventing the next episode is much less intrusive," she said.