Role of mHealth and Wearables in Disease Diagnosis in Healthcare

By Vibhuti Gupta, Assistant Professor of Computer Science & Data Science, Meharry School of Applied Computational Sciences

With the emergence of mobile health technologies and the widespread routine use of wearables, person-generated health data has become a promising data source for biomedical research. The data generated through wearables are unique as it is collected passively, continuously, and objectively in free-living conditions, providing a personalized and comprehensive picture of people’s health. These smart devices not only help people pursue a healthy lifestyle but also provide a constant stream of healthcare data for early disease diagnosis and treatment by actively recording physiological parameters and tracking metabolic status. A multitude of disease conditions have been diagnosed recently using wearable devices, such as skin diseases and injuries, cardiovascular disease detection, diabetes, cancer, and other chronic conditions. 

Wearable technology has many advantages compared to traditional medical monitoring devices, such as continuous monitoring, early diagnosis, and ease of use, but it has some cons that need to be addressed.

Rapid advancements in machine learning have assisted in identifying the digital biomarkers from the data generated through these devices that can aid in disease diagnosis, prognosis and treatment effects. Digital biomarkers are digitally collected data, such as heart rate, sleep and steps from wearable devices, that are transformed through predictive models into indicators of health outcomes. Some digital biomarkers have outperformed traditional clinical biomarkers, such as arrhythmia detection, because of their ability to monitor patients outside of the clinic continuously. The most successful digital biomarkers have been developed based on supervised, unsupervised and semi-supervised machine learning models. These machine learning models have the capability to learn from the digital data, and then predict the disease states automatically which assists doctors for early alerts for emergency conditions.

For hospital in-patient care, continuous monitoring has been used for decades through multiparameter patient vital signs monitors (e.g., BP, blood oxygen saturation [SpO2], heart rate, respiration rate). However, continuous monitoring of non-bedridden patients is more challenging because patients must travel around the hospital with bulky monitoring devices. The wireless nature of wearable devices makes the process of continuous monitoring of physiological parameters easy. Intensive care units (ICUs) are already benefiting from the analytics of wearables data through algorithms. Continuous monitoring helps identify the dynamic changes to patient status and warning signs prior to adverse health events and may provide an opportunity for early interventions to prevent severe health events. Thus, overall wearables and mHealth technologies combined with machine learning algorithms have revolutionized healthcare.  

To implement these mHealth and wearable technologies in real world has full of challenges. Firstly, the data collected is in a free-living environment. Thus, it has data quality issues. Wearable sensors data are highly unstructured, complex, and messy since it is generated continuously and with high frequency (thousands of observations per second) leading to rich streams of time series data. Moreover, the data is high dimensional, with many different variables that are hard to process. Secondly, there are data integration issues as a large amount of data is collected from disparate data sources, which has to be combined effectively to produce reliable clinical decisions. Thirdly, the predictive models developed become obsolete frequently due to the dynamic evolution of disease states. Overall, these challenges have to be addressed for producing reliable healthcare decisions using these technologies.

Wearable technology has many advantages compared to traditional medical monitoring devices, such as continuous monitoring, early diagnosis, and ease of use, but it has some cons that need to be addressed. Wearables tend to have short battery life, due to which the data monitoring is affected. Many users comply with charging their devices; however, most of them do not. There are data compliance issues with the users, too. Many users don’t wear the device for a sufficient amount of time to consider their data for analysis, which leads to inaccurate conclusions. Some wearables have been reported to measure data inaccurately on occasion. This can be especially dangerous when measuring data like heart rates. For individuals with heart conditions, this false reading could lead to overexertion and further health issues.

Overall, it is important to consider the challenges and issues that arise due to these technologies, which must be addressed to generate accurate and reliable medical decisions.