Dr. David Mohr runs the program, called "Mobilyze!, which has completed an initial pilot. The app is now being tweaked and prepared for a second pilot program.
"We're trying to use context sensing," Mohr said. "We're using the in-dwelling sensors in the smart phone to identify specific patient states that may be relevant to depression and the treatment of it."
Mobile phones today have myriad sensors, such as GPS, gyroscopes, accelerometers, light sensors and microphones.
Mobilyze!, under development in Feinberg's newly opened Center for Behavioral Intervention Technologies, harnesses cell phone sensor data on a near continuous basis, and uses computer algorithms to detect specific patient states.
"The states we're trying to detect are location, activity, social context and mood," Mohr said. "We're using a machine-learning paradigm that initially requires some user input."
For example, the application can use sensor data to determine a person's location, activity level, and level of social interaction.
If a patient inputs that one of their positive activities may be calling a friend or meeting someone for coffee, the phone can determine if the patient has performed that activity of a period of time, and if they've haven't, it will remind them.
"When people are depressed they tend not to engage in activities that are rewarding or enjoyable. If you increase those activities, it will improve your mood," Mohr said. "If you're at home on Saturday for four hours or more, for example, then you're at greater risk for depression."
But don't expect smartphone consultations anytime soon. JSA Health Telepsychiatry's Fishkind said size still matters.
"We believe, and not everyone does, that very high quality equipment on the other end makes patients more comfortable faster," he said. "If they were looking at a small picture on a cell phone, you can imagine there would be a very different sense of connectedness to the psychiatrist then if they're doing a 1080P transmission on a 42-inch television."
Sign up for Computerworld eNewsletters.