Enhancing Clinical Data Analysis by Explaining Interaction Effects between Covariates in Deep Neural Network Models
Deep neural network (DNN) is a powerful technology that is being utilized by a growing number and range of research projects, including disease risk prediction models. One of the key strengths of DNN is its ability to model non-linear relationships, which include covariate interactions. We developed...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-01-01
|
| Series: | Journal of Personalized Medicine |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2075-4426/13/2/217 |
| _version_ | 1797619945447620608 |
|---|---|
| author | Yijun Shao Ali Ahmed Edward Y. Zamrini Yan Cheng Joseph L. Goulet Qing Zeng-Treitler |
| author_facet | Yijun Shao Ali Ahmed Edward Y. Zamrini Yan Cheng Joseph L. Goulet Qing Zeng-Treitler |
| author_sort | Yijun Shao |
| collection | DOAJ |
| description | Deep neural network (DNN) is a powerful technology that is being utilized by a growing number and range of research projects, including disease risk prediction models. One of the key strengths of DNN is its ability to model non-linear relationships, which include covariate interactions. We developed a novel method called interaction scores for measuring the covariate interactions captured by DNN models. As the method is model-agnostic, it can also be applied to other types of machine learning models. It is designed to be a generalization of the coefficient of the interaction term in a logistic regression; hence, its values are easily interpretable. The interaction score can be calculated at both an individual level and population level. The individual-level score provides an individualized explanation for covariate interactions. We applied this method to two simulated datasets and a real-world clinical dataset on Alzheimer’s disease and related dementia (ADRD). We also applied two existing interaction measurement methods to those datasets for comparison. The results on the simulated datasets showed that the interaction score method can explain the underlying interaction effects, there are strong correlations between the population-level interaction scores and the ground truth values, and the individual-level interaction scores vary when the interaction was designed to be non-uniform. Another validation of our new method is that the interactions discovered from the ADRD data included both known and novel relationships. |
| first_indexed | 2024-03-11T08:34:14Z |
| format | Article |
| id | doaj.art-15aaec0a64344ea5896a3bcd99d12677 |
| institution | Directory Open Access Journal |
| issn | 2075-4426 |
| language | English |
| last_indexed | 2024-03-11T08:34:14Z |
| publishDate | 2023-01-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Journal of Personalized Medicine |
| spelling | doaj.art-15aaec0a64344ea5896a3bcd99d126772023-11-16T21:32:21ZengMDPI AGJournal of Personalized Medicine2075-44262023-01-0113221710.3390/jpm13020217Enhancing Clinical Data Analysis by Explaining Interaction Effects between Covariates in Deep Neural Network ModelsYijun Shao0Ali Ahmed1Edward Y. Zamrini2Yan Cheng3Joseph L. Goulet4Qing Zeng-Treitler5Department of Clinical Research and Leadership, School of Medicine and Health Sciences, George Washington University, Washington, DC 20037, USADepartment of Clinical Research and Leadership, School of Medicine and Health Sciences, George Washington University, Washington, DC 20037, USADepartment of Clinical Research and Leadership, School of Medicine and Health Sciences, George Washington University, Washington, DC 20037, USADepartment of Clinical Research and Leadership, School of Medicine and Health Sciences, George Washington University, Washington, DC 20037, USAVA Connecticut Healthcare System, New Haven, CT 06516, USADepartment of Clinical Research and Leadership, School of Medicine and Health Sciences, George Washington University, Washington, DC 20037, USADeep neural network (DNN) is a powerful technology that is being utilized by a growing number and range of research projects, including disease risk prediction models. One of the key strengths of DNN is its ability to model non-linear relationships, which include covariate interactions. We developed a novel method called interaction scores for measuring the covariate interactions captured by DNN models. As the method is model-agnostic, it can also be applied to other types of machine learning models. It is designed to be a generalization of the coefficient of the interaction term in a logistic regression; hence, its values are easily interpretable. The interaction score can be calculated at both an individual level and population level. The individual-level score provides an individualized explanation for covariate interactions. We applied this method to two simulated datasets and a real-world clinical dataset on Alzheimer’s disease and related dementia (ADRD). We also applied two existing interaction measurement methods to those datasets for comparison. The results on the simulated datasets showed that the interaction score method can explain the underlying interaction effects, there are strong correlations between the population-level interaction scores and the ground truth values, and the individual-level interaction scores vary when the interaction was designed to be non-uniform. Another validation of our new method is that the interactions discovered from the ADRD data included both known and novel relationships.https://www.mdpi.com/2075-4426/13/2/217deep learningrisk analysisAlzheimer’s disease and related dementia |
| spellingShingle | Yijun Shao Ali Ahmed Edward Y. Zamrini Yan Cheng Joseph L. Goulet Qing Zeng-Treitler Enhancing Clinical Data Analysis by Explaining Interaction Effects between Covariates in Deep Neural Network Models Journal of Personalized Medicine deep learning risk analysis Alzheimer’s disease and related dementia |
| title | Enhancing Clinical Data Analysis by Explaining Interaction Effects between Covariates in Deep Neural Network Models |
| title_full | Enhancing Clinical Data Analysis by Explaining Interaction Effects between Covariates in Deep Neural Network Models |
| title_fullStr | Enhancing Clinical Data Analysis by Explaining Interaction Effects between Covariates in Deep Neural Network Models |
| title_full_unstemmed | Enhancing Clinical Data Analysis by Explaining Interaction Effects between Covariates in Deep Neural Network Models |
| title_short | Enhancing Clinical Data Analysis by Explaining Interaction Effects between Covariates in Deep Neural Network Models |
| title_sort | enhancing clinical data analysis by explaining interaction effects between covariates in deep neural network models |
| topic | deep learning risk analysis Alzheimer’s disease and related dementia |
| url | https://www.mdpi.com/2075-4426/13/2/217 |
| work_keys_str_mv | AT yijunshao enhancingclinicaldataanalysisbyexplaininginteractioneffectsbetweencovariatesindeepneuralnetworkmodels AT aliahmed enhancingclinicaldataanalysisbyexplaininginteractioneffectsbetweencovariatesindeepneuralnetworkmodels AT edwardyzamrini enhancingclinicaldataanalysisbyexplaininginteractioneffectsbetweencovariatesindeepneuralnetworkmodels AT yancheng enhancingclinicaldataanalysisbyexplaininginteractioneffectsbetweencovariatesindeepneuralnetworkmodels AT josephlgoulet enhancingclinicaldataanalysisbyexplaininginteractioneffectsbetweencovariatesindeepneuralnetworkmodels AT qingzengtreitler enhancingclinicaldataanalysisbyexplaininginteractioneffectsbetweencovariatesindeepneuralnetworkmodels |