Towards interpretable, medically grounded, EMR-based risk prediction models

Abstract Machine-learning based risk prediction models have the potential to improve patient outcomes by assessing risk more accurately than clinicians. Significant additional value lies in these models providing feedback about the factors that amplify an individual patient’s risk. Identification of...

Full description

Bibliographic Details
Main Authors: Isabell Twick, Guy Zahavi, Haggai Benvenisti, Ronya Rubinstein, Michael S. Woods, Haim Berkenstadt, Aviram Nissan, Enes Hosgor, Dan Assaf
Format: Article
Language:English
Published: Nature Portfolio 2022-06-01
Series:Scientific Reports
Online Access:https://doi.org/10.1038/s41598-022-13504-7
_version_ 1818239781482528768
author Isabell Twick
Guy Zahavi
Haggai Benvenisti
Ronya Rubinstein
Michael S. Woods
Haim Berkenstadt
Aviram Nissan
Enes Hosgor
Dan Assaf
author_facet Isabell Twick
Guy Zahavi
Haggai Benvenisti
Ronya Rubinstein
Michael S. Woods
Haim Berkenstadt
Aviram Nissan
Enes Hosgor
Dan Assaf
author_sort Isabell Twick
collection DOAJ
description Abstract Machine-learning based risk prediction models have the potential to improve patient outcomes by assessing risk more accurately than clinicians. Significant additional value lies in these models providing feedback about the factors that amplify an individual patient’s risk. Identification of risk factors enables more informed decisions on interventions to mitigate or ameliorate modifiable factors. For these reasons, risk prediction models must be explainable and grounded on medical knowledge. Current machine learning-based risk prediction models are frequently ‘black-box’ models whose inner workings cannot be understood easily, making it difficult to define risk drivers. Since machine learning models follow patterns in the data rather than looking for medically relevant relationships, possible risk factors identified by these models do not necessarily translate into actionable insights for clinicians. Here, we use the example of risk assessment for postoperative complications to demonstrate how explainable and medically grounded risk prediction models can be developed. Pre- and postoperative risk prediction models are trained based on clinically relevant inputs extracted from electronic medical record data. We show that these models have similar predictive performance as models that incorporate a wider range of inputs and explain the models’ decision-making process by visualizing how different model inputs and their values affect the models’ predictions.
first_indexed 2024-12-12T13:03:00Z
format Article
id doaj.art-45509ecd51824359a1fcef360ad7fb81
institution Directory Open Access Journal
issn 2045-2322
language English
last_indexed 2024-12-12T13:03:00Z
publishDate 2022-06-01
publisher Nature Portfolio
record_format Article
series Scientific Reports
spelling doaj.art-45509ecd51824359a1fcef360ad7fb812022-12-22T00:23:45ZengNature PortfolioScientific Reports2045-23222022-06-0112111210.1038/s41598-022-13504-7Towards interpretable, medically grounded, EMR-based risk prediction modelsIsabell Twick0Guy Zahavi1Haggai Benvenisti2Ronya Rubinstein3Michael S. Woods4Haim Berkenstadt5Aviram Nissan6Enes Hosgor7Dan Assaf8Caresyntax GmbHDepartment of Anesthesiology, The Chaim Sheba Medical CenterDepartment of General and Oncological Surgery – Surgery C, The Chaim Sheba Medical CenterCaresyntax GmbHCaresyntax GmbHDepartment of Anesthesiology, The Chaim Sheba Medical CenterDepartment of General and Oncological Surgery – Surgery C, The Chaim Sheba Medical CenterCaresyntax GmbHDepartment of General and Oncological Surgery – Surgery C, The Chaim Sheba Medical CenterAbstract Machine-learning based risk prediction models have the potential to improve patient outcomes by assessing risk more accurately than clinicians. Significant additional value lies in these models providing feedback about the factors that amplify an individual patient’s risk. Identification of risk factors enables more informed decisions on interventions to mitigate or ameliorate modifiable factors. For these reasons, risk prediction models must be explainable and grounded on medical knowledge. Current machine learning-based risk prediction models are frequently ‘black-box’ models whose inner workings cannot be understood easily, making it difficult to define risk drivers. Since machine learning models follow patterns in the data rather than looking for medically relevant relationships, possible risk factors identified by these models do not necessarily translate into actionable insights for clinicians. Here, we use the example of risk assessment for postoperative complications to demonstrate how explainable and medically grounded risk prediction models can be developed. Pre- and postoperative risk prediction models are trained based on clinically relevant inputs extracted from electronic medical record data. We show that these models have similar predictive performance as models that incorporate a wider range of inputs and explain the models’ decision-making process by visualizing how different model inputs and their values affect the models’ predictions.https://doi.org/10.1038/s41598-022-13504-7
spellingShingle Isabell Twick
Guy Zahavi
Haggai Benvenisti
Ronya Rubinstein
Michael S. Woods
Haim Berkenstadt
Aviram Nissan
Enes Hosgor
Dan Assaf
Towards interpretable, medically grounded, EMR-based risk prediction models
Scientific Reports
title Towards interpretable, medically grounded, EMR-based risk prediction models
title_full Towards interpretable, medically grounded, EMR-based risk prediction models
title_fullStr Towards interpretable, medically grounded, EMR-based risk prediction models
title_full_unstemmed Towards interpretable, medically grounded, EMR-based risk prediction models
title_short Towards interpretable, medically grounded, EMR-based risk prediction models
title_sort towards interpretable medically grounded emr based risk prediction models
url https://doi.org/10.1038/s41598-022-13504-7
work_keys_str_mv AT isabelltwick towardsinterpretablemedicallygroundedemrbasedriskpredictionmodels
AT guyzahavi towardsinterpretablemedicallygroundedemrbasedriskpredictionmodels
AT haggaibenvenisti towardsinterpretablemedicallygroundedemrbasedriskpredictionmodels
AT ronyarubinstein towardsinterpretablemedicallygroundedemrbasedriskpredictionmodels
AT michaelswoods towardsinterpretablemedicallygroundedemrbasedriskpredictionmodels
AT haimberkenstadt towardsinterpretablemedicallygroundedemrbasedriskpredictionmodels
AT aviramnissan towardsinterpretablemedicallygroundedemrbasedriskpredictionmodels
AT eneshosgor towardsinterpretablemedicallygroundedemrbasedriskpredictionmodels
AT danassaf towardsinterpretablemedicallygroundedemrbasedriskpredictionmodels