Advances in AI has allowed for computers to help doctors in diagnosing disease and help monitor patients’ vital signs from any location.
A team of researchers from the University of Colorado Boulder are working to apply machine learning artificial intelligence (AI) in psychiatry, with a speech-based mobile app that can categorise a patient’s mental health status as well as, or better than, a human can.
The university research paper has been published in Schizophrenia Bulletin, and lays out the promise and potential pitfalls of AI in psychiatry.
Peter Foltz, a research professor at the Institute of Cognitive Science and co-author of the paper, said: “We are not in any way trying to replace clinicians, but we do believe we can create tools that will allow them to better monitor their patients.”
Accessing mental health care
In Europe, the WHO estimated that 44.3 million people suffer with depression and 37.3 million suffer with anxiety.
Diagnosis of mental health disorders are based on an age-old method that can be subjective and unreliable, notes paper co-author Brita Elvevåg, a cognitive neuroscientist at the University of Tromsø, Norway.
Elvevåg said: “Humans are not perfect. They can get distracted and sometimes miss out on subtle speech cues and warning signs.
“Unfortunately, there is no objective blood test for mental health.”
Elvevåg and Foltz teamed up to develop machine learning technology that is able to more precisely detect day-to-day changes in speech that hint at mental health decline.
For instance, sentences that don’t follow a logical pattern can be a critical symptom in schizophrenia. Shifts in tone or pace can hint at mania or depression, and memory loss can be a sign of both cognitive and mental health problems.
“Language is a critical pathway to detecting patient mental states,” says Foltz. “Using mobile devices and AI, we are able to track patients daily and monitor these subtle changes.”
AI in psychiatry
The new mobile app asks patients to answer a five to 10 minute series of questions by talking into their phone. Among various other tasks, they’re asked about their emotional state, asked to tell a short story, listen to a story and repeat it and given a series of touch-and-swipe motor skills tests.
The team developed an AI system that assesses the speech samples, compares them to previous samples by the same patient and the broader population, and then rates the patient’s mental state.
The team asked human clinicians to listen to and assess speech samples of 225 participants – half with severe psychiatric issues; half healthy volunteers – in rural Louisiana and Northern Norway. They then compared those results to those of the machine learning system.
If the app detected a worrisome change, it could notify the patient’s doctor to check in.
Foltz said: “We found that the computer’s AI models can be at least as accurate as clinicians.
“Patients often need to be monitored with frequent clinical interviews by trained professionals to avoid costly emergency care and unfortunate events, but there are simply not enough clinicians for that.”
In the paper, the researchers lay out a call for larger studies to prove efficacy and earn public trust before AI technology could be broadly brought into clinical practice for psychiatry.
The paper states: ‘The mystery around AI does not nurture trustworthiness, which is critical when applying medical technology.
‘Rather than looking for machine learning models to become the ultimate decision-maker in medicine, we should leverage the things that machines do well that are distinct from what humans do well.’
The idea of AI transforming self-care has been exploited to a great extent. However, AI-powered tools can’t replace therapy, they can only serve as a helpful addition to check the symptoms, keep track of the mood etc.: https://madappgang.com/blog/mental-health-app-development-how-to-build-an-app-backed-by-science/
For a mental health service, it’s more important to be credible and have practitioners onboard than integrating innovative technologies like AI.
What we do in Denmark, is to assign the diagnosis to an under graduate. They ask the patient standard questions from a single page questionnaire. 15 minutes later you have a diagnosis. A psychiatrist will then create the prescription without having to meet with you in person. For the best healthcare system in the world, which we believe we have, you don’t need an AI. An app with a few questions is enough. The App can send a prescription to the pharmacy and you can start using anti-psychotics immediately, for example as sleeping pills. Tax payers money ensures this to be free to access for everyone. The side effects can of course include death, brain damage, disability and so forth, so it’s really up to yourself whether or not you undergo treatment. We also use social workers to diagnose people with mental illness. A psychiatrist is not necessary anymore. The Danish system has been perfected. Every level of the welfare system has become fit to diagnose anyone with any thing.
Eventually I chose to change my lifestyle and believe systems instead because the side effects were too harsh. In a matter of a few weeks I was cured and haven’t had any symptoms or issues since.
A doctor is not always your best choice when your health fails. The internet can sometimes be a much better choice.
Hi Hordur, my name is Giovani Missio. I’m a Psychiatrist in Brazil and I’m launching a startup to make a better treatment for depression. We use an app with some questions do diagnose and to assess patients do find the better treatment. Could you help me telling me more about your experience with this aplication in Denmark that you use?
Giovani
gmissio@gmail.com
The AI can’t do anything unless it’s programmed to recognize language indicators better than typical vague questionaires in use in most failed western systems, and that requires much better categorizations of diseases.
Step #0 in the coming neo-psychiatry revolution is strictly drawing the very obvious distinction between emotional/mood disorder and contexual-cognitive disorders. The idea that it’s 2020 and psychiatry isn’t adequately categorizing obvious symptomatic differences in practice is both astounding and shameful. The superstitious psychologization of neurological damage has set back brain health for over a century. I cringe whenever I even encounter the word ‘mental’ due to its abuse and contribution to the neglect of millions. I see many clueless colleagues who exclusively track mood in patients with clear and severe untreated cognitive disfunction all the time. It’s criminal malpractice as far as I’m concerned, and there is no effectual oversight or public policy by which they can be challenged. Depression, severe though it can be, isn’t even in the same universe as the diseases of the Schizophrenia syndrome, and it should be very easy diagnostically to distinguish between a person with insight who reports symptoms of depression vs someone without insight presenting suffering from extreme paranoia and disorganized or noncontextual speech. I find it necessary to make this point because reliance on AI analysis essentially depends on brain disease victims cooperating and openly sharing accurate information about mental states with an app daily, which already bespeaks total diagnostic failure on the part of clinicians — no paranoid person (i.e. the severe cognitive cases) would ever cooperate with such an analytical system. Unstable patients probably already think AI’s or similar are persecuting them specifically. So, excluding force, right out of the gate the results of the experiment will be — as usual with today’s chronically failed methods– skewed to overrepresent less severe conditions, while doing nothing to help or improve popular awareness of those with the most severe brain disabilities imaginable. Equality for all is an ideal that still matters and will always matter as the basis of good medicine. The new psychiatry must respect the obvious primacy of the brain and empirical results, it must prioritize new physical treatments, and it must strictly prevent categorical conflation in the field that causes no end of unenlightened public policy and wasted funding. The status quo is simply unacceptable, and everybody knows it.
I am really worried about this new technology, I saw that big company are using similar tools today to allow them to scale up their interviewing processes…
Great analysis, thank you. Despite the fact that now mental health apps and software are on a rise, there is still much to be done. Enhancing healthcare software with AI might have a great potential for future welfare.
Physicians diagnosis is very inconsistent in psychiatry. With speech recognition and machine intelligence it should be feasible to have a computer making good diagnoses consistently.
Eventually I can see that the treatment step should be good as well, perhaps with genetic mapping for person specific best.