Researchers from UCLA used an interactive voice application with artificial intelligence (AI) to monitor the well-being of patients who are being treated for mental illness. The app performed well at detecting a change in mental health.
The team followed 47 patients for up to 14 months using the MyCoachConnect app. All patients were being treated for serious mental illness, including bipolar disorder, schizophrenia and major depressive disorder.
Participants were asked to call a toll-free number one or two times a week and answer three open-ended questions that were prompted by a computer-generated voice. They made calls on either a mobile phone, landline or payphone. The questions were: how have you been over the past few days, what has been troubling you over the past few days and what has been particularly positive over the past few days? Each patient spoke for two to three minutes per question. This project was kept low tech on the patient’s end and high tech on the backend so participants of all incomes could be included.
The MyCoachConnect app was designed to collect personalized patient responses. The app's AI algorithm was trained to analyze individual words in response and offer a personalized analysis for each patient. The app focuses on the patient’s choice of words and how their responses change over time. It also places a small emphasis on audio features, like tone of voice.
After the study, some participants reported finding the system easy and enjoyable to use. They also said that speaking to a computer-generated voice allowed them to speak more freely and they felt less lonely.
The team found that the app’s data analysis was just as accurate at monitoring a patient’s mental states as a physician. In patients who saw a change in their mental health, the app reached the same conclusion as a physician. These results highlighted why language use and voice patterns are important to indicate a change in mental health.
The team hopes that AI can analyze data collected from healthcare apps to create more proactive and personalized care. The app could help improve treatment by enabling doctors to intervene early when patients are experiencing more symptoms.
The study was published in PLOS ONE.