Anomaly Detection Framework for Wearables Data: A Perspective Review on Data Concepts, Data Analysis Algorithms and Prospects

Abstract

Wearable devices use sensors to evaluate physiological parameters, such as the heart rate, pulse rate, number of steps taken, body fat and diet. The continuous monitoring of physiological parameters offers a potential solution to assess personal healthcare. Identifying outliers or anomalies in heart rates and other features can help identify patterns that can play a significant role in understanding the underlying cause of disease states. Since anomalies are present within the vast amount of data generated by wearable device sensors, identifying anomalies requires accurate automated techniques. Given the clinical significance of anomalies and their impact on diagnosis and treatment, a wide range of detection methods have been proposed to detect anomalies. Much of what is reported herein is based on previously published literature. Clinical studies employing wearable devices are also increasing. In this article, we review the nature of the wearables-associated data and the downstream processing methods for detecting anomalies. In addition, we also review supervised and un-supervised techniques as well as semi-supervised methods that overcome the challenges of missing and un-annotated healthcare data.

Wearable activity trackers–advanced technology or advanced marketing?

“…Currently, although wearable devices can track changes over time, wearable devices cannot detect what is a meaningful change. Rather, various algorithms have been created and must be applied to find anomalies in the user’s generated data to predict future illness (Lown et al 2020 ; Alavi et al 2021 ; Sunny et al 2022 ). These algorithms require multiple steps to be useful, including first downloading the data to an application programming interface (API), processing it for uniformity, and filling in missing data points (Sunny et al 2022 ).…”

“…Rather, various algorithms have been created and must be applied to find anomalies in the user’s generated data to predict future illness (Lown et al 2020 ; Alavi et al 2021 ; Sunny et al 2022 ). These algorithms require multiple steps to be useful, including first downloading the data to an application programming interface (API), processing it for uniformity, and filling in missing data points (Sunny et al 2022 ). For healthcare providers to accurately predict disease risk based on deviations in user’s wearable technology data, current restrictions require them to employ the use of these relevant algorithms, which are external to the wearable device itself (Guk et al 2019 ; Sunny et al 2022 ).…”

Smart Consumer Wearables as Digital Diagnostic Tools: A Review

“…An analysis of wearable device data is primarily focused on identifying anomalous behaviors in the recorded data, and also predicting future events [ 98 ]. This is achieved by training the machine-learning model with the recorded data of known anomalous events and testing the model’s performance with previously unseen data.…”

Considerations for Analyzing and Interpreting Data from Biometric Monitoring Technologies in Clinical Trials

“…It is paramount to identify statistical approaches to deal with missing data to ensure that derived endpoints are valid, accurate, and reliable [53, 54]. The taxonomy for classifying missing data mechanisms is based on the likelihood of being missed: MCAR (missing completely at random), MAR (missing at random), and MNAR (missing not at random) [53]. Currently, derived values of digital clinical measures are often left missing when some underlying measurements are not available.…”

“…Currently, derived values of digital clinical measures are often left missing when some underlying measurements are not available. Emerging statistical approaches for addressing missing BioMeT data include within-patient imputations across standard periods, functional data analysis, deep learning methods, imputation approaches, and robust modeling [53].…”