AI Algorithm’s Ability to Predict Antidepressant Effectiveness Called into Question by Expert

An article in NRC (8/2) stated that an AI algorithm can predict whether the antidepressant sertraline will work or not. An AI model was used to determine which clinical and brain variables correlated with sertraline treatment outcomes. But this does not mean that the algorithm has predictive value. True predictive power can only be demonstrated if the algorithm were applied to a completely new group of patients. Only then is the AI ​​model tested on data that it has not yet ‘seen’ and can it be determined whether it really predicts something. Basic treatment facts were forgotten. The number of patients who need to be treated with sertraline to see more improvement in one patient than would have been the case with placebo is four. This means that in three out of four patients who improve on sertraline, this is actually due to the placebo effect. So if the AI ​​algorithm predicts anything, it is the placebo response, even in the sertraline group.

Biological systems with relevance to psychiatry are characterized by three important properties: non-linearity, regulation and interaction. This complicates making predictions based on single variables, as in this study, since the effects of a single factor can be amplified, reduced, or modified by the presence or activity of other factors. AI models, no matter how advanced, are based on mathematical principles and have the same fundamental limitations as other mathematical models when dealing with the complex systems of the brain.

Jim van Os

Head of the Brain Division UMC Utrecht

Share Email the editor
2024-02-11 21:49:33
#Algorithm