Predicting the Hallucinogenic Potential of Molecules Using Artificial Intelligence
ACS Chemical Neuroscience August 2, 2024 Fabio Urbina, Thane Jones, Joshua S. Harris et al. 9 citations
New drugs for serious mental health disorders should avoid causing psychedelic experiences. Analogs of psychedelic drugs called psychoplastogens show promise for treating opioid use disorder by reducing drug dependence, with rare serious side effects. This effect is linked to increased neuritogenesis and neuroplasticity. Some psychoplastogens act through the 5HT 2A receptor, but others have different pharmacology, making prediction of hallucinogenic potential difficult. Researchers developed machine learning classification models to predict psychedelic effects using in vitro PsychLight data (support vector classification, AUC 0.74) and in vivo human data from Shulgin and Shulgin (SVC, AUC 0.72). The models predicted known 5HT 2A agonists' psychedelic potential with AUCs of 0.97 and 0.71, respectively, aiding design of non-hallucinogenic psychoplastogens.