Instagram Photos May Help Diagnose Depression in the Future
A team of researchers has developed an artificial intelligence (AI) system that seemingly can identify whether someone suffers from depression or not by looking at their Instagram photos.1 The project is part of a joint venture between Harvard University and the University of Vermont. The AI system is in its preliminary trials, but is showing promise. Currently, studies on its efficacy show that its algorithm can identify depressed individuals with 70 percent accuracy. By comparison, general practitioners, given no metrics or scales to use (in other words, using just an interview), have about a 42 percent accuracy rate at diagnosing depression on initial presentation.2 This comparison is not very rigorous, however is offered by the researchers via championing the use of AI systems for initial screening in the future. In a world where more and more things are turning digital, it is not out of the question to imagine a future where your social media is accessible as part of medical intake.
AI System uses a Sophisticated Algorithm
The AI system uses a sophisticated algorithm which takes into account an array of factors of Instagram photo posting, including how many “likes” are given, and received. It looks at color preference, and what types of filters are used. Previous research has shown that depressed individuals tend to use black and white toned filters, and non-depressed individuals tend to prefer to lighten images to make them appear brighter. It also has face recognition software to determine whether there are people in the photos. It also screens comments.
AI Concluded Diagnosis of Accuracy 70 Percent of the Time
The most recent trial included 166 individuals, roughly half of which had been diagnosed with clinical depression within the last three years. The AI system analyzed the data from each person’s Instagram account and was found to conclude a diagnosis of depression accurately 70 percent of the time.
The tool is not being developed as a diagnostic measure, but more of an initial screening. Something that could indicate whether further screening may be important.
- Mitchell AJ, Vaze A, Rao S (2009) Clinical diagnosis of depression in primary care: a meta-analysis. Lancet 374(9690):609-619.
Image Copyright: <a href=’https://www.123rf.com/profile_dolgachov’>dolgachov / 123RF Stock Photo</a>
Node Smith, associate editor for NDNR, is a fifth year naturopathic medical student at NUNM, where he has been instrumental in maintaining a firm connection to the philosophy and heritage of naturopathic medicine amongst the next generation of docs. He helped found the first multi-generational experiential retreat, which brings elders, alumni, and students together for a weekend campout where naturopathic medicine and medical philosophy are experienced in nature. Three years ago he helped found the non-profit, Association for Naturopathic ReVitalization (ANR), for which he serves as the board chairman. ANR has a mission to inspire health practitioners to embody the naturopathic principles through experiential education. Node also has a firm belief that the next era of naturopathic medicine will see a resurgence of in-patient facilities which use fasting, earthing, hydrotherapy and homeopathy to bring people back from chronic diseases of modern living; he is involved in numerous conversations and projects to bring about this vision.