Delgado-García, GuillermoEngbers, Jordan D. T.Wiebe, SamuelMouches, PaulineAmador, KimberlyForkert, Nils D.White, JamesSajobi, TolulopeKlein, Karl MartinJosephson, Colin B.Calgary Comprehensive Epilepsy Program Collaborators2024-08-092024-08-092023-07-08Delgado-García, G., Engbers, J. D. T., Wiebe, S., Mouches, P., Amador, K., Forkert, N. D., White, J., Sajobi, T., Klein, K. M., Josephson, C. B., & Calgary Comprehensive Epilepsy Program Collaborators. (2023). Machine learning using multimodal clinical, electroencephalographic, and magnetic resonance imaging data can predict incident depression in adults with epilepsy: A pilot study. Epilepsia, 64(10), 2781–2791. https://doi.org/10.1111/epi.17710https://hdl.handle.net/1880/119379Objective: To develop a multi-modal machine learning (ML) approach for predicting incident depression in adults with epilepsy. Methods: We randomly selected 200 patients from the Calgary Comprehensive Epilepsy Program registry and linked their registry-based clinical data to their first-available clinical electroencephalogram (EEG) and magnetic resonance imaging (MRI) study. We excluded patients with a clinical or Neurological Disorders Depression Inventory for Epilepsy (NDDI-E)-based diagnosis of major depression at baseline. The NDDI-E was used to detect incident depression over a median of 2.4 years follow-up (IQR 1.5-3.3 years). A ReliefF algorithm was applied to clinical as well as quantitative EEG and MRI parameters for feature selection. Six ML algorithms were trained and tested using stratified three-fold cross-validation. Multiple metrics were used to assess model performances. Results: Of 200 patients, 150 had EEG and MRI data of sufficient quality for ML, of which 59 were excluded due to prevalent depression. Therefore, 91 patients (41 women) were included with a median age of 29 (IQR 22-44) years. A total of 42 features were selected by ReliefF, none of which were quantitative MRI or EEG variables. All models had a sensitivity >80% and 5 of 6 had an F1 score ≥0.72. Multilayer perceptron model had the highest F1 score (median 0.74; interquartile range [IQR] 0.71-0.78) and sensitivity (84.3%). Median area under the receiver operating characteristic curve and normalized Matthews correlation coefficient were 0.70 (IQR 0.64-0.78) and 0.57 (IQR 0.50-0.65), respectively. Significance: Multimodal machine learning using baseline features can predict incident depression in this population. Our pilot models demonstrated high accuracy for depression prediction. However, overall performance and calibration can be improved. This model has promise for identifying those at risk for incident depression during follow-up, though efforts to refine it in larger populations along with external validation are required.enAttribution-NonCommercial-NoDerivatives 4.0 Internationalhttp://creativecommons.org/licenses/by-nc-nd/4.0/Articlehttps://doi.org/10.1111/epi.17710