Will artificial intelligence be the route to personalized medicine in psychiatry?

Artificial intelligence (AI) is entering and transforming multiple sectors of society and has potential benefits also in the field of psychiatry. Biotyping and phenotyping may help identify patients most likely to benefit from different agents, integrative AI models can predict therapeutic response, and ecological momentary assessment could one day help monitor symptoms and guide psychiatrists in patient care.

Artificial intelligence (AI) is entering and transforming multiple sectors of society and has potential benefits also in the field of psychiatry. Biotyping and phenotyping may help identify patients most likely to benefit from different agents, integrative AI models can predict therapeutic response, and ecological momentary assessment could one day help monitor symptoms and guide psychiatrists in patient care.

Drug development in psychiatry has been manifesting ‘innovation stasis’, remarked Professor Roger McIntyre, University of Toronto, Canada at AsCNP Virtual 2021. Response rates for anti-depressants have remained relatively static over the past 40 years he said. At the same time, there has been a change in the focus of illness, shifting away from absence of disease to more domain-based and patient-reported outcomes.1

 

Inflammatory biotypes may inform clinical phenotypes in MDD

The link between inflammatory disturbances and mental illness has become a hot topic in psychiatry. Studies show that alterations in levels of inflammatory cytokines are associated with pathological changes in patients with MDD and can alter response to conventional anti-psychotics.2-4

Biotyping will identify groups of patients that will most likely benefit from a particular drug

However, not all patients with MDD display alterations in cytokine levels and heterogeneity exists with respect to the underlying inflammatory biotypes that are emerging, so we need to try and refine the biophenotypes of different types of depression said Professor McIntyre.

 

Integrative models in machine learning most accurately predict therapeutic response

Artificial intelligence using dimension reduction techniques can identify patterns across multiple different cytokines. This technique has allowed patients to be separated into distinct biotypes of bipolar depression that can help predict which patients will respond best to a treatment.5

Meta-analysis of more than 20 studies conducted to evaluate the usefulness of machine learning algorithms in accurately predicting therapeutic outcome in adults with bipolar depression found that accuracy was significantly greater in integrative models informed by multiple data types compared with lower dimension data types. The data types included in the integrative model included phenomenological patient features, neuroimaging and peripheral gene expression data.6

AI will help us integrate interactions across multiple different biological systems

 

Tracking and monitoring MDD - a facilitator for case management

The ability to digitally fingerprint patients could also be a crucial aid in psychiatry. Using ecological momentary assessment and the mind.me app, symptoms in adults with clinically-relevant depression were tracked by collecting data such as voice recognition, texting and vocabulary use. This allowed prediction of depressive symptoms with an accuracy of 91%, a sensitivity of 98%, and a specificity of 93%.7

Artificial intelligence is not going to invent new drugs said Professor McIntyre, but it will help us to integrate interactions that take place within and across different biological systems that are impossible to measure without AI.

Our correspondent’s highlights from the symposium are meant as a fair representation of the scientific content presented. The views and opinions expressed on this page do not necessarily reflect those of Lundbeck.

References

  1. Manderscheid et al. Prev Chronic Dis 2010; 7:A19. Epub 2009.
  2. Hepgul N, et al. Neuropsychopharmacology 2016;41:2502-11.
  3. Osimo EF, et al. Brain Behav Immun 2020;87:901-909.
  4. Haroon E, et al. Psychoneuroendocrinology 2018;95:43-49.
  5. Lee Y, et al. Mol Psychiatry 2021:26: 3395-3406.
  6. Lee Y, et al. J Affect Disord 2018;241:519-532.
  7. McIntyre RS, et al. J Psychiat Res 2021;135:311-317.