Illusory generalizability of clinical prediction models.

Autor: Chekroud AM; Spring Health, New York City, NY 10010, USA.; Department of Psychiatry, Yale University School of Medicine, New Haven, CT 06520, USA., Hawrilenko M; Spring Health, New York City, NY 10010, USA., Loho H; Department of Psychiatry, Yale University School of Medicine, New Haven, CT 06520, USA., Bondar J; Spring Health, New York City, NY 10010, USA., Gueorguieva R; Department of Biostatistics, Yale University, New Haven, CT 06520, USA., Hasan A; Department of Psychiatry, Psychotherapy and Psychosomatics, University Augsburg, 86159 Augsburg, Germany., Kambeitz J; Department of Psychiatry and Psychotherapy, University of Cologne, Faculty of Medicine and University Hospital of Cologne, Cologne, Germany., Corlett PR; Department of Psychiatry, Yale University School of Medicine, New Haven, CT 06520, USA., Koutsouleris N; Department of Psychiatry and Psychotherapy, Ludwig-Maximilians-University, Munich, Germany., Krumholz HM; Center for Outcomes Research and Evaluation, Yale New Haven Hospital, New Haven, CT 06520, USA., Krystal JH; Department of Psychiatry, Yale University School of Medicine, New Haven, CT 06520, USA., Paulus M; Laureate Institute for Brain Research, Tulsa, OK 74136, USA.
Jazyk: angličtina
Zdroj: Science (New York, N.Y.) [Science] 2024 Jan 12; Vol. 383 (6679), pp. 164-167. Date of Electronic Publication: 2024 Jan 11.
DOI: 10.1126/science.adg8538
Abstrakt: It is widely hoped that statistical models can improve decision-making related to medical treatments. Because of the cost and scarcity of medical outcomes data, this hope is typically based on investigators observing a model's success in one or two datasets or clinical contexts. We scrutinized this optimism by examining how well a machine learning model performed across several independent clinical trials of antipsychotic medication for schizophrenia. Models predicted patient outcomes with high accuracy within the trial in which the model was developed but performed no better than chance when applied out-of-sample. Pooling data across trials to predict outcomes in the trial left out did not improve predictions. These results suggest that models predicting treatment outcomes in schizophrenia are highly context-dependent and may have limited generalizability.
Databáze: MEDLINE
Externí odkaz:
Nepřihlášeným uživatelům se plný text nezobrazuje