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

Abstract

Background Data 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. Objective The aim of this study was to impute missing values in data using a deep learning approach. Methods To 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). Results The 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. Conclusions Our deep learning–based imputation model performed better than the other methods when imputing missing values in actigraphy data.

Modeling Brain Volume Using Deep Learning-Based Physical Activity Features in Patients With Dementia

“…The basic structure and initial hyperparameter values were based on our previous model ( Jang et al, 2020 ). The model development was conducted based on the NHANES activity dataset, which was randomly divided into portions at ratios of 8:1:1 for use as a training set, a validation set, and a test set, respectively.…”

“…An autoencoder model can capture such invisible and non-linear information effectively. An autoencoder is an unsupervised deep learning model composed of an encoder and a decoder ( Kramer, 1991 ; Jang et al, 2020 ). An encoder receives high-dimensional input data and encodes it into a low-dimensional latent vector.…”

The Dilemma of Analyzing Physical Activity and Sedentary Behavior with Wrist Accelerometer Data: Challenges and Opportunities

“…Similarly, Mathur et al [84] successfully performed deep data augmentation training to address heterogeneities in wearable sensing devices' software and hardware. Furthermore, Jiang et al [85] validated a deep learning model to accurately impute missing PA data-a common issue with free-living, device-determined PA data-and observed it to outperform other known methods of imputing missing activity data. Furthermore, van Kuppevelt et al [86] employed an unsupervised classification approach, which successfully learned different human PA behaviors, based on the accelerometer-derived data it observed.…”

Exploring the potential of neural networks to predict statistics of solar wind turbulence

“…Specifically, a simple feed-forward neural network was found to accurately reproduce simulated stochastic processes and fill gaps that matched the original power spectrum with up to 50% missing data (Comerford et al, 2015). Generative adversarial networks (Y. and convolutional neural networks (Jang et al, 2020) have also been used to impute missing intervals.…”