Healthcare Monitoring Using Low-Cost Sensors to Supplement and Replace Human Sensation: Does It Have Potential to Increase Independent Living and Prevent Disease?

Abstract

Continuous monitoring of health status has the potential to enhance the quality of life and life expectancy of people suffering from chronic illness and of the elderly. However, such systems can only come into widespread use if the cost of manufacturing is low. Advancements in material science and engineering technology have led to a significant decrease in the expense of developing healthcare monitoring devices. This review aims to investigate the progress of the use of low-cost sensors in healthcare monitoring and discusses the challenges faced when accomplishing continuous and real-time monitoring tasks. The major findings include (1) only a small number of publications (N = 50) have addressed the issue of healthcare monitoring applications using low-cost sensors over the past two decades; (2) the top three algorithms used to process sensor data include SA (Statistical Analysis, 30%), SVM (Support Vector Machine, 18%), and KNN (K-Nearest Neighbour, 12%); and (3) wireless communication techniques (Zigbee, Bluetooth, Wi-Fi, and RF) serve as the major data transmission tools (77%) followed by cable connection (13%) and SD card data storage (10%). Due to the small fraction (N = 50) of low-cost sensor-based studies among thousands of published articles about healthcare monitoring, this review not only summarises the progress of related research but calls for researchers to devote more effort to the consideration of cost reduction as well as the size of these components.

Keywords: body area network; healthcare; human skin; low-cost sensor; physiological parameter.

Figure 1

Flow chart of literature selection through the review phases. The contributions of IEEE, EI, and WOS were similar: 1963, 2356, and 1863, respectively, while PubMed contributed more than half of all publications (6798) in the initial search.

Figure 2

Summary of typical algorithms used for processing data acquired by low-cost sensors including LR [20], DT [21], FFT [22], SVM [22], NB [23], SA [29], WT [34], LDA [40], ANN [42], HMM [56] and KNN [65].

Figure 3

Comparison of different techniques used for sensor data transmission and storage.

Figure 4

Number of publications between 2002 and 2021 using low-cost sensors for healthcare monitoring.

Figure 5

Word cloud of keywords and titles: (a) keywords and (b) title. The larger the font is, the more frequently the word appears. (a) Word clouds of the keywords from the selected publications and (b) word clouds of the article titles from the selected publications.