Computer Methods and Programs in Biomedicine Update (Jan 2024)

An evaluation of the commonly used portable medical sensors performance in comparison to clinical test results for telehealth systems

  • Rafiqul Islam Maruf,
  • Saori Tou,
  • Rieko Izukura,
  • Yoko Sato,
  • Mariko Nishikitani,
  • Kimiyo Kikuchi,
  • Fumihiko Yokota,
  • Subaru Ikeda,
  • Rakibul Islam,
  • Ashir Ahmed,
  • Masashi Miyazaki,
  • Naoki Nakashima

Journal volume & issue
Vol. 5
p. 100147

Abstract

Read online

Background and Objective: One of the primary challenges faced by telehealth systems is the accurate transmission of patient information to remote doctors. In this context, portable medical sensors deployed at the remote patients' end play a crucial role in measuring vital information. There are many sensors available in the market. However, the accuracy of the sensors has been always a concern. The objective of this study is to verify different sensors and create awareness for using accurate sensors to avoid misdiagnosis for the patients’ safety. Methods: This study considered the test result of a Japanese clinical pathology laboratory as the reference gold standard. The clinical pathology laboratory uses 1) Hexokinase UV method for blood glucose, 2) Enzymatic Determination method for cholesterol, 3) Automatic Analyzer (EDTA-2 K) of Hemoglobin, and 4) Uricase POD method for uric acid. To assess the performance of a medical sensor, its test results were compared to the gold standard test results obtained from the laboratory using the same sample. A Normalized Root Mean Square Error (NRMSE) threshold of less than 0.2 was established as the criterion for determining whether the medical sensor's performance fell within an acceptable range. Results: Among the eight most commonly used blood glucose devices in the Asian market, only one device was deemed acceptable with NRMSE less than 0.2. However, all four devices found in the Japanese market showed their acceptability. In the case of cholesterol, hemoglobin, and uric acid devices, only a limited number of items were available in Asian markets. Some of the hemoglobin and uric acid devices were found to be somewhat acceptable, while all the cholesterol sensors were found erroneous. Conclusions: This study has clearly shown the issues with the portable medical sensors and recommends the device approval authority of each country to approve sales of the quality sensors only for patients’ safety.

Keywords