Ensuring generalizability and clinical utility in mental health care applications: Robust artificial intelligence-based treatment predictions in diverse psychosis populations.

The model performance was decreased compared to the full model but still highly significant ( r = 0.53, P < 0.001).

also in 44,862 other papers

There was a larger rate of false non‐remissions predictions in women for the FEP model, although this did not quite reach statistical significance: the differences in model performance may therefore also be due to increased response rates in women, which is consistent with current literature, 52 leading to an unexpected better outcome.

also in 147 other papers