Differential Treatment Benefit Prediction for Treatment Selection in Depression: A Deep Learning Analysis of STAR*D and CO-MED Data
Depression affects one in nine people, but treatment response rates remain low. There is significant potential in the use of computational modeling techniques to predict individual patient responses and thus provide more personalized treatment. Deep learning is a promising computational technique th...
| Main Authors: | , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Ubiquity Press
2020-01-01
|
| Series: | Computational Psychiatry |
| Subjects: | |
| Online Access: | https://cpsyjournal.org/articles/26 |
| _version_ | 1828420197447368704 |
|---|---|
| author | Joseph Mehltretter Robert Fratila David A. Benrimoh Adam Kapelner Kelly Perlman Emily Snook Sonia Israel Caitrin Armstrong Marc Miresco Gustavo Turecki |
| author_facet | Joseph Mehltretter Robert Fratila David A. Benrimoh Adam Kapelner Kelly Perlman Emily Snook Sonia Israel Caitrin Armstrong Marc Miresco Gustavo Turecki |
| author_sort | Joseph Mehltretter |
| collection | DOAJ |
| description | Depression affects one in nine people, but treatment response rates remain low. There is significant potential in the use of computational modeling techniques to predict individual patient responses and thus provide more personalized treatment. Deep learning is a promising computational technique that can be used for differential treatment selection based on predicted remission probability. Using Sequenced Treatment Alternatives to Relieve Depression (STAR*D) and Combining Medications to Enhance Depression Outcomes (CO-MED) trial data, we employed deep neural networks to predict remission after feature selection. Treatments included were citalopram, escitalopram, bupropion SR plus escitalopram, and venlafaxine plus mirtazapine. Differential treatment benefit was estimated in terms of improvement of population remission rates after application of the model for treatment selection using two approaches: (1) using predictions generated directly from the model (the predicted improvement approach) and (2) using bootstrapping for sample generation and then estimating population remission rate for patients who actually received the drug predicted by the model compared to the general population (the actual improvement approach). Our deep learning model predicted remission in a pooled CO-MED/STAR*D dataset (including four treatments) with an area under the curve of 0.69 using 17 input features. Our actual improvement analysis showed a statistically significant 2.48% absolute improvement (corresponding to a 7.2% relative improvement) in population remission rate ('p' = 0.01, CI 2.48% ± 0.5%). Our model serves as proof-of-concept that deep learning approaches, with further refinement and work to address concerns about differences between studies when multiple datasets are used for training, may have utility in differential prediction of antidepressant response when selecting from a number of treatment options. |
| first_indexed | 2024-12-10T15:07:38Z |
| format | Article |
| id | doaj.art-5f288e65ef4b4f29bc101b52c847c38b |
| institution | Directory Open Access Journal |
| issn | 2379-6227 |
| language | English |
| last_indexed | 2024-12-10T15:07:38Z |
| publishDate | 2020-01-01 |
| publisher | Ubiquity Press |
| record_format | Article |
| series | Computational Psychiatry |
| spelling | doaj.art-5f288e65ef4b4f29bc101b52c847c38b2022-12-22T01:44:00ZengUbiquity PressComputational Psychiatry2379-62272020-01-014617510.1162/cpsy_a_0002924Differential Treatment Benefit Prediction for Treatment Selection in Depression: A Deep Learning Analysis of STAR*D and CO-MED DataJoseph Mehltretter0Robert Fratila1David A. Benrimoh2Adam Kapelner3Kelly Perlman4Emily Snook5Sonia Israel6Caitrin Armstrong7Marc Miresco8Gustavo Turecki9Department of Computer Science, University of Southern California, Los Angeles, California, US; Aifred Health, Montreal, QuebecAifred Health, Montreal, QuebecAifred Health, Montreal, Quebec; Department of Psychiatry, McGill University, Montreal, QuebecDepartment of Mathematics, Queen’s College, New York City, NYAifred Health, Montreal, Quebec; Department of Psychiatry, McGill University, Montreal, QuebecAifred Health, Montreal, Quebec; Department of Psychiatry, McGill University, Montreal, QuebecAifred Health, Montreal, Quebec; Department of Psychiatry, McGill University, Montreal, QuebecAifred Health, Montreal, QuebecDepartment of Psychiatry, McGill University, Montreal, QuebecDepartment of Psychiatry, McGill University, Montreal, QuebecDepression affects one in nine people, but treatment response rates remain low. There is significant potential in the use of computational modeling techniques to predict individual patient responses and thus provide more personalized treatment. Deep learning is a promising computational technique that can be used for differential treatment selection based on predicted remission probability. Using Sequenced Treatment Alternatives to Relieve Depression (STAR*D) and Combining Medications to Enhance Depression Outcomes (CO-MED) trial data, we employed deep neural networks to predict remission after feature selection. Treatments included were citalopram, escitalopram, bupropion SR plus escitalopram, and venlafaxine plus mirtazapine. Differential treatment benefit was estimated in terms of improvement of population remission rates after application of the model for treatment selection using two approaches: (1) using predictions generated directly from the model (the predicted improvement approach) and (2) using bootstrapping for sample generation and then estimating population remission rate for patients who actually received the drug predicted by the model compared to the general population (the actual improvement approach). Our deep learning model predicted remission in a pooled CO-MED/STAR*D dataset (including four treatments) with an area under the curve of 0.69 using 17 input features. Our actual improvement analysis showed a statistically significant 2.48% absolute improvement (corresponding to a 7.2% relative improvement) in population remission rate ('p' = 0.01, CI 2.48% ± 0.5%). Our model serves as proof-of-concept that deep learning approaches, with further refinement and work to address concerns about differences between studies when multiple datasets are used for training, may have utility in differential prediction of antidepressant response when selecting from a number of treatment options.https://cpsyjournal.org/articles/26depressiontreatment selectionartificial intelligencemachine learningtreatment predictionmood disorders |
| spellingShingle | Joseph Mehltretter Robert Fratila David A. Benrimoh Adam Kapelner Kelly Perlman Emily Snook Sonia Israel Caitrin Armstrong Marc Miresco Gustavo Turecki Differential Treatment Benefit Prediction for Treatment Selection in Depression: A Deep Learning Analysis of STAR*D and CO-MED Data Computational Psychiatry depression treatment selection artificial intelligence machine learning treatment prediction mood disorders |
| title | Differential Treatment Benefit Prediction for Treatment Selection in Depression: A Deep Learning Analysis of STAR*D and CO-MED Data |
| title_full | Differential Treatment Benefit Prediction for Treatment Selection in Depression: A Deep Learning Analysis of STAR*D and CO-MED Data |
| title_fullStr | Differential Treatment Benefit Prediction for Treatment Selection in Depression: A Deep Learning Analysis of STAR*D and CO-MED Data |
| title_full_unstemmed | Differential Treatment Benefit Prediction for Treatment Selection in Depression: A Deep Learning Analysis of STAR*D and CO-MED Data |
| title_short | Differential Treatment Benefit Prediction for Treatment Selection in Depression: A Deep Learning Analysis of STAR*D and CO-MED Data |
| title_sort | differential treatment benefit prediction for treatment selection in depression a deep learning analysis of star d and co med data |
| topic | depression treatment selection artificial intelligence machine learning treatment prediction mood disorders |
| url | https://cpsyjournal.org/articles/26 |
| work_keys_str_mv | AT josephmehltretter differentialtreatmentbenefitpredictionfortreatmentselectionindepressionadeeplearninganalysisofstardandcomeddata AT robertfratila differentialtreatmentbenefitpredictionfortreatmentselectionindepressionadeeplearninganalysisofstardandcomeddata AT davidabenrimoh differentialtreatmentbenefitpredictionfortreatmentselectionindepressionadeeplearninganalysisofstardandcomeddata AT adamkapelner differentialtreatmentbenefitpredictionfortreatmentselectionindepressionadeeplearninganalysisofstardandcomeddata AT kellyperlman differentialtreatmentbenefitpredictionfortreatmentselectionindepressionadeeplearninganalysisofstardandcomeddata AT emilysnook differentialtreatmentbenefitpredictionfortreatmentselectionindepressionadeeplearninganalysisofstardandcomeddata AT soniaisrael differentialtreatmentbenefitpredictionfortreatmentselectionindepressionadeeplearninganalysisofstardandcomeddata AT caitrinarmstrong differentialtreatmentbenefitpredictionfortreatmentselectionindepressionadeeplearninganalysisofstardandcomeddata AT marcmiresco differentialtreatmentbenefitpredictionfortreatmentselectionindepressionadeeplearninganalysisofstardandcomeddata AT gustavoturecki differentialtreatmentbenefitpredictionfortreatmentselectionindepressionadeeplearninganalysisofstardandcomeddata |