The Holy Grail of public mental health? …AI?
Friday, December 15, 2023. Holiday Inn Fredericksburg. Leaving Virginia today Home by 8:30pm, I hope.
I’m a Marriage and Family Therapist. As far as mental health terminal degrees go, we’re the new kids on the block.
Marriage and Family Therapy is a distinct focus of psychotherapy practice, and didn’t emerge until the early 1960s. Only about 12% of mental health professionals have an MFT degree…
However, the essential conceptual and clinical forces that informed the development of Marriage and Family Therapy can be traced to a much earlier period.
Family therapy, as a treatment modality, was originally focused on the daunting task of understanding the etiology of schizophrenia, and treating it with talk therapy in a family setting.
In the late 1940s and early 1950s Marriage & Family Therapy thought leaders were enthralled by developments in cybernetics, systemic thinking, and communication theory…
These theories, informed by concepts from multiple disciplines including sociology, anthropology, and biology, provided powerful theoretical frameworks for a more in-depth understanding of the complexities of all family interactions, but the early thought leaders of Marriage and Family Therapy always returned to the intractable problem of helping families and their loved ones struggling with schizophrenia.
These theories provided the conceptual framework for much of the early family-centered research on schizophrenia.
Unfortunately, despite a heroic amount of effort, many of their ideas were heartbreakingly wrong. Some of them, problematically so.
What can AI now, do that the best minds in social science in the 1960’s could not?
We are apparently at an inflection point where Artificial Intelligence can diagnose schizophrenia in seconds from speech patterns produced in response to a few simple questions.
Humans with schizophrenia, one of the most compelling and grave mental disorders, typically have a loosening of semantic connections.
That is why they jump between totally unrelated subjects, which is what AI can detect in their speech.
AI’s diagnostic power is only getting started…
The diagnosis of other mental health conditions by AI could also also be expected to follow. I anticipate AI to become a widespread and diagnostically relevant tool by 2030.
This would represent a huge shift in mental health, which currently relies mostly on talking with patients and family for diagnosis, with only a minimal role for brain scans and blood tests.
Dr. Matthew Nour, the study’s first author, said:
“Until very recently, the automatic analysis of language has been out of reach of doctors and scientists. However, with the advent of artificial intelligence (AI) language models such as ChatGPT, this situation is changing.
This work shows the potential of applying AI language models to psychiatry—a medical field intimately related to language and meaning.”
How the study was conducted
The research included 26 people all diagnosed with schizophrenia who were compared with a control group of the same size.
All participants were given verbal fluency tasks, such as:
In 5 minutes, name as many animals as you can beginning with the letter ‘A’.
The answers were then fed into an AI language model program for analysis. This demonstrated that the AI could predict the answers of humans in the control group more accurately.
Those with schizophrenia gave more ‘unusual’ answers — and, the more severe their symptoms, the more unusual their answers.
Cognitive maps
The explanation for this difference between those with schizophrenia and without could be down to the ‘cognitive maps’ that the brain uses to store information and link concepts together.
This means that when a person recalls that ‘ant’ is an animal that begins with the letter ‘a’, they are also more likely to recall that ‘antelope’ is another animal starting with the same exact letter.
In schizophrenia, though, these linkages are not as strong, which produces more disconnected thoughts.
There is obviously a different but similar application here for AI and neurodiversity, and the study of a constellation of cognitive linkages between a couple. AI will design therpeutic talk-therapy interventions… that’s inevitable.
Indeed, brain scans in this study backed up the idea, pointing to a disruption in associative memory in those with schizophrenia.
This intersection of applied neuroscience and AI is incredible, profoundly significant, and a sign of potentially wonderous things to come for psychiatry…
Dr. Nour summed it all up:
“We are entering a very exciting time in neuroscience and mental health research.
By combining state-of-the-art AI language models and brain scanning technology, we are beginning to uncover how meaning is constructed in the brain, and how this might go awry in psychiatric disorders.
There is enormous interest in using AI language models in medicine.
If these tools prove safe and robust, I expect they will begin to be deployed in the clinic within the next decade.”
Final thoughts…
Exciting times, indeed. We might be facing a paradox of sorts. As AI rolls out in clinic settings, the power of AI will reshape the hierarchy of expertise, and further subordinate the diagnostic function of psychiatry to a powerful machine intelligence.
All while enjoyed an uptick in the quality of services. More done… with less.
Once the AI revolution is monetized, it will massage the greed is good ethos, Clinics, of course, will be cutting costs and chasing profits, but unable to avoid doing massive social good in the process.
Be well, stay kind, and Godspeed.
RESEARCH:
Dr. Nour’s Amazing study was published in the journal Proceedings of the National Academy of Sciences (Nour et al., 2023).
Data, Materials, and Software Availability
Analysis code and data to reproduce the results in the paper will be made available at github.com/matthewnour/verbal_fluency_trajectories and github.com/YunzheLiu/TDLM. The manuscript additionally relates behavioral measures to MEG data originally presented in a previous study: (11).
Acknowledgments
Dr. Nour thanks Dr. Atheeshaan Arumuham (King’s College London) for help with patient recruitment, and Professor Zeb Kurth-Nelson (DeepMind) for discussions on task design and analysis methods.
Author contributions
M.M.N. and R.J.D. designed research; M.M.N. performed research; M.M.N., D.C.M., and Y.L. contributed new reagents/analytic tools; M.M.N. analyzed data; and M.M.N. and R.J.D. wrote the paper.
Competing interests
The authors declare no competing interest.
Musa, A., Khan, S., Mujahid, M. et al. The shallow cognitive map hypothesis: A hippocampal framework for thought disorder in schizophrenia. Schizophr 8, 34 (2022). https://doi.org/10.1038/s41537-022-00247-7
