Artificial intelligence (AI) is expected to play a key role in diagnosing and treating depression more effectively. The newfangled technology could help make more accurate diagnoses and recommend better treatments for depression.
According to data shared by JAMA Psychiatry, nearly 20 per cent of people will have depression at least once in their lifetimes. The Lancet reported that 300 million people globally are currently experiencing depression.
Unsurprisingly, the World Health Organization (WHO) describes depression as the single biggest contributor to ill health around the world. So, how exactly could AI help?
Spotting depression can be an arduous task
Depression can be hard to spot and difficult to diagnose. In fact, general practitioners can only accurately detect depression in less than half of cases, The Lancet reported.
This doesn't come as a surprise given that there is no specific test for depression. To make a diagnosis, doctors rely on various factors including clinical observations, questionnaires and self-reported symptoms.
However, it is worth noting that symptoms of depression vary from person to person. While some people feel miserable without knowing why, many people with depression have symptoms that are severe enough to cause problems in day-to-day activities, as per a Mayo Clinic report.
People who are accurately diagnosed with depression can choose between a wide range of treatment options such as lifestyle change, medications and talk therapy. However, each person responds to treatment differently and it is hard to predict which treatment will work ahead of time.
Machine learning is one the branches of AI and it aims at training computers to learn, find patterns in data and make data-informed predictions without human intervention. In recent years, there has been a significant surge in research surrounding AI's application to illnesses like depression.
#AI can now detect #depression by hearing a person talk, or even reading #writing style and choice of words - accuracy is close to 80%.
— Bryan Druzin (@BryanDruzin) September 15, 2018
Consider the implications for online communication and #surveillance.#MachineLearning #DeepLearning #ArtificialIntelligence #innovation pic.twitter.com/cE1NSvCRQH
With AI, computers achieve human-like thinking ability and three crucial behaviours including self-correction, reasoning and learning. For instance, a new tool called Life2vec can accurately predict an individual's potential lifespan.
Similarly, Google is reportedly working on a program codenamed Project Ellmann, which will give users a "bird's-eye" view of their lives using the company's newly launched Gemini AI.
ChatGPT vs. doctors
According to some scientists, ChatGPT diagnoses and medical recommendations are as good as those of real-life doctors. Research, which was published in BMJ Journals, pits ChatGPT against primary care physicians and it showed surprising results.
ChatGPT mostly recommended talk therapy when it was given information about fictional patients with different levels of depression severity, sex and socioeconomic status. Doctors, on the other hand, recommended antidepressants.
It is worth noting that US, British and Australian guidelines also recommend talk therapy before resorting to medication. This is a major sign that ChatGPT is more likely to follow clinical guidelines, while doctors tend to overprescribe antidepressants.
Moreover, ChatGPT is less likely to be influenced by sex and socioeconomic biases. However, a research article published on Plos One implies doctors have a tendency to prescribe antidepressants to men, particularly those in blue-collar jobs.
The currently available diagnostic tools
AI-assisted MRI continues to show signs of future success. In the meantime, the process of detecting depression is getting simpler and easier, thanks to some diagnostic tools you might already have.
Powered by wearable AI, wearable devices such as smartwatches are expected to predict depression in the future. Smartwatches can turn out to be helpful since they can collect data such as social interaction, sleep data, metabolic rate, step counts and heart rates.
A recent review published in Nature shows wearables correctly predicted depression 70 to 80 per cent of the time. In fact, the research suggests wearable devices might offer unique data since they are normally worn all day and all night.
On the downside, smartwatches aren't easy on the pocket and may be inaccessible to many. Researchers have also turned to social media to detect depression. Scientists have used AI to predict the presence and severity of depression, according to a post on Sage Journals.
The emojis we use have also been used to successfully detect depression, as per a study published on ScienceDirect.
