Amid the endless stream of selfies and food photos, some more substantial mental health data might be buried in an Instagram feed that only artificial intelligence can decipher.
Using machine learning to sift through pictures posted to the social media app by 166 participants, researchers at Harvard and the University of Vermont identified users with depression by analyzing metadata like hue, color saturation and brightness. For example, depressed individuals tended to post photos that were bluer, darker and grayer, according to the study published in EPJ Data Science.
Perhaps even more notable, the algorithmic approach outperformed general practitioners in both accuracy and speed. While more than half of the diagnoses made by physicians were false positives, the majority of those made by a computer were accurate. In some cases, artificial intelligence was able to detect those signals earlier than the date of first diagnosis.
Although Instagram might not be the ultimate diagnostic tool, the researchers told the New York Times that the study speaks more broadly to the potential of machine learning to supplement mental health screenings.
“Paired with a commensurate focus on upholding data privacy and ethical analytics, the present work may serve as a blueprint for effective mental health screening in an increasingly digitalized society,” they wrote. “More generally, these findings support the notion that major changes in individual psychology are transmitted in social media use, and can be identified via computational methods.”
Mental health has become a breeding ground for potential innovation ranging from telemedicine to digital apps that supplement internet-based cognitive therapy in place of in-person treatment. At the same time, researchers have argued that digital biomarkers collected through mental health apps are creating a new “digital divide.”
Meanwhile, image-based applications of AI are particularly promising, with recent studies indicating machine learning can diagnose skin cancer as accurately as dermatologists and read thousands of X-rays and CT scans in a matter of minutes.