Beware the blind spots in using AI to treat mental health issues

Artificial intelligence is enabling a whole new world of receiving and administering health services. But when it comes to mental healthcare, code should not necessarily replace the human factor. By Florence de Vries

- Florence de Vries

Recently, I found myself trying to keep up with the choreography of a setting in modern healthcare. I was sitting on a bed with a friendly human nurse rushing between me and a virtual doctor a screen delivering instructions and asking medical questions with the warmth of a vending machine - to administer a vitamin B injection.

I answered nearly 30 questions, including why I thought I needed the vitamin B injection in the first place. The nurse, clearly doing her best within the confines of this relatively modern protocol, assured me that it would be fine: “This is how we do it now.”

It was only later, hours after I had left, that I realised no one (neither physical nor virtual) had asked whether I'm prone to anaemia. And I had forgotten to mention that I am and take supplements for it.

I wondered if this information would have changed the decision to administer the vitamin B injection. I don’t know. Perhaps not. But what struck me wasn’t the oversight, it was the system. It was a reminder that as artificial intelligence (AI) becomes more embedded in medicine, we have to pause and ask what gets lost when care becomes code. And, more specifically, how do these shortcomings play out in the fields of psychiatry and psychology when working with people living with mental illnesses?

Leaps in healthcare settings

According to the World Health Organization, nearly one billion people live with a mental disorder. In the past three years, it has warned that the rising tide of mental health disorders worldwide is working against one grim constant: access to mental healthcare.