Deep Learning Approach for Imputation of Missing Values in Actigraphy Data: Algorithm Development Study

BackgroundData collected by an actigraphy device worn on the wrist or waist can provide objective measurements for studies related to physical activity; however, some data may contain intervals where values are missing. In previous studies, statistical methods have been appli...

Full description

Bibliographic Details
Main Authors: Jang, Jong-Hwan, Choi, Junggu, Roh, Hyun Woong, Son, Sang Joon, Hong, Chang Hyung, Kim, Eun Young, Kim, Tae Young, Yoon, Dukyong
Format: Article
Language:English
Published: JMIR Publications 2020-07-01
Series:JMIR mHealth and uHealth
Online Access:http://mhealth.jmir.org/2020/7/e16113/
id doaj-c7ff85c6fb74417188dcd1b07857c436
record_format Article
spelling doaj-c7ff85c6fb74417188dcd1b07857c4362021-05-02T19:28:34ZengJMIR PublicationsJMIR mHealth and uHealth2291-52222020-07-0187e1611310.2196/16113Deep Learning Approach for Imputation of Missing Values in Actigraphy Data: Algorithm Development StudyJang, Jong-HwanChoi, JungguRoh, Hyun WoongSon, Sang JoonHong, Chang HyungKim, Eun YoungKim, Tae YoungYoon, Dukyong BackgroundData collected by an actigraphy device worn on the wrist or waist can provide objective measurements for studies related to physical activity; however, some data may contain intervals where values are missing. In previous studies, statistical methods have been applied to impute missing values on the basis of statistical assumptions. Deep learning algorithms, however, can learn features from the data without any such assumptions and may outperform previous approaches in imputation tasks. ObjectiveThe aim of this study was to impute missing values in data using a deep learning approach. MethodsTo develop an imputation model for missing values in accelerometer-based actigraphy data, a denoising convolutional autoencoder was adopted. We trained and tested our deep learning–based imputation model with the National Health and Nutrition Examination Survey data set and validated it with the external Korea National Health and Nutrition Examination Survey and the Korean Chronic Cerebrovascular Disease Oriented Biobank data sets which consist of daily records measuring activity counts. The partial root mean square error and partial mean absolute error of the imputed intervals (partial RMSE and partial MAE, respectively) were calculated using our deep learning–based imputation model (zero-inflated denoising convolutional autoencoder) as well as using other approaches (mean imputation, zero-inflated Poisson regression, and Bayesian regression). ResultsThe zero-inflated denoising convolutional autoencoder exhibited a partial RMSE of 839.3 counts and partial MAE of 431.1 counts, whereas mean imputation achieved a partial RMSE of 1053.2 counts and partial MAE of 545.4 counts, the zero-inflated Poisson regression model achieved a partial RMSE of 1255.6 counts and partial MAE of 508.6 counts, and Bayesian regression achieved a partial RMSE of 924.5 counts and partial MAE of 605.8 counts. ConclusionsOur deep learning–based imputation model performed better than the other methods when imputing missing values in actigraphy data.http://mhealth.jmir.org/2020/7/e16113/
collection DOAJ
language English
format Article
sources DOAJ
author Jang, Jong-Hwan
Choi, Junggu
Roh, Hyun Woong
Son, Sang Joon
Hong, Chang Hyung
Kim, Eun Young
Kim, Tae Young
Yoon, Dukyong
spellingShingle Jang, Jong-Hwan
Choi, Junggu
Roh, Hyun Woong
Son, Sang Joon
Hong, Chang Hyung
Kim, Eun Young
Kim, Tae Young
Yoon, Dukyong
Deep Learning Approach for Imputation of Missing Values in Actigraphy Data: Algorithm Development Study
JMIR mHealth and uHealth
author_facet Jang, Jong-Hwan
Choi, Junggu
Roh, Hyun Woong
Son, Sang Joon
Hong, Chang Hyung
Kim, Eun Young
Kim, Tae Young
Yoon, Dukyong
author_sort Jang, Jong-Hwan
title Deep Learning Approach for Imputation of Missing Values in Actigraphy Data: Algorithm Development Study
title_short Deep Learning Approach for Imputation of Missing Values in Actigraphy Data: Algorithm Development Study
title_full Deep Learning Approach for Imputation of Missing Values in Actigraphy Data: Algorithm Development Study
title_fullStr Deep Learning Approach for Imputation of Missing Values in Actigraphy Data: Algorithm Development Study
title_full_unstemmed Deep Learning Approach for Imputation of Missing Values in Actigraphy Data: Algorithm Development Study
title_sort deep learning approach for imputation of missing values in actigraphy data: algorithm development study
publisher JMIR Publications
series JMIR mHealth and uHealth
issn 2291-5222
publishDate 2020-07-01
description BackgroundData collected by an actigraphy device worn on the wrist or waist can provide objective measurements for studies related to physical activity; however, some data may contain intervals where values are missing. In previous studies, statistical methods have been applied to impute missing values on the basis of statistical assumptions. Deep learning algorithms, however, can learn features from the data without any such assumptions and may outperform previous approaches in imputation tasks. ObjectiveThe aim of this study was to impute missing values in data using a deep learning approach. MethodsTo develop an imputation model for missing values in accelerometer-based actigraphy data, a denoising convolutional autoencoder was adopted. We trained and tested our deep learning–based imputation model with the National Health and Nutrition Examination Survey data set and validated it with the external Korea National Health and Nutrition Examination Survey and the Korean Chronic Cerebrovascular Disease Oriented Biobank data sets which consist of daily records measuring activity counts. The partial root mean square error and partial mean absolute error of the imputed intervals (partial RMSE and partial MAE, respectively) were calculated using our deep learning–based imputation model (zero-inflated denoising convolutional autoencoder) as well as using other approaches (mean imputation, zero-inflated Poisson regression, and Bayesian regression). ResultsThe zero-inflated denoising convolutional autoencoder exhibited a partial RMSE of 839.3 counts and partial MAE of 431.1 counts, whereas mean imputation achieved a partial RMSE of 1053.2 counts and partial MAE of 545.4 counts, the zero-inflated Poisson regression model achieved a partial RMSE of 1255.6 counts and partial MAE of 508.6 counts, and Bayesian regression achieved a partial RMSE of 924.5 counts and partial MAE of 605.8 counts. ConclusionsOur deep learning–based imputation model performed better than the other methods when imputing missing values in actigraphy data.
url http://mhealth.jmir.org/2020/7/e16113/
work_keys_str_mv AT jangjonghwan deeplearningapproachforimputationofmissingvaluesinactigraphydataalgorithmdevelopmentstudy
AT choijunggu deeplearningapproachforimputationofmissingvaluesinactigraphydataalgorithmdevelopmentstudy
AT rohhyunwoong deeplearningapproachforimputationofmissingvaluesinactigraphydataalgorithmdevelopmentstudy
AT sonsangjoon deeplearningapproachforimputationofmissingvaluesinactigraphydataalgorithmdevelopmentstudy
AT hongchanghyung deeplearningapproachforimputationofmissingvaluesinactigraphydataalgorithmdevelopmentstudy
AT kimeunyoung deeplearningapproachforimputationofmissingvaluesinactigraphydataalgorithmdevelopmentstudy
AT kimtaeyoung deeplearningapproachforimputationofmissingvaluesinactigraphydataalgorithmdevelopmentstudy
AT yoondukyong deeplearningapproachforimputationofmissingvaluesinactigraphydataalgorithmdevelopmentstudy
_version_ 1721488240952737792