A Noninvasive, Economical, and Instant-Result Method to Diagnose and Monitor Type 2 Diabetes Using Pulse Wave: Case-Control Study

Abstract

Background: We should pay more attention to the long-term monitoring and early warning of type 2 diabetes and its complications. The traditional blood glucose tests are traumatic and cannot effectively monitor the development of diabetic complications. The development of mobile health is changing rapidly. Therefore, we are interested in developing a new noninvasive, economical, and instant-result method to accurately diagnose and monitor type 2 diabetes and its complications.

Objective: We aimed to determine whether type 2 diabetes and its complications, including hypertension and hyperlipidemia, could be diagnosed and monitored by using pulse wave.

Methods: We collected the pulse wave parameters from 50 healthy people, 139 diabetic patients without hypertension and hyperlipidemia, 133 diabetic patients with hypertension, 70 diabetic patients with hyperlipidemia, and 75 diabetic patients with hypertension and hyperlipidemia. The pulse wave parameters showing significant differences among these groups were identified. Various machine learning models such as linear discriminant analysis, support vector machines (SVMs), and random forests were applied to classify the control group, diabetic patients, and diabetic patients with complications.

Results: There were significant differences in several pulse wave parameters among the 5 groups. The parameters height of tidal wave (h₃), time distance between the start point of pulse wave and dominant wave (t₁), and width of percussion wave in its one-third height position (W) increase and the height of dicrotic wave (h₅) decreases when people develop diabetes. The parameters height of dominant wave (h₁), h₃, and height of dicrotic notch (h₄) are found to be higher in diabetic patients with hypertension, whereas h₅ is lower in diabetic patients with hyperlipidemia. For detecting diabetes, the method with the highest out-of-sample prediction accuracy is SVM with polynomial kernel. The algorithm can detect diabetes with 96.35% accuracy. However, all the algorithms have a low accuracy when predicting diabetic patients with hypertension and hyperlipidemia (below 70%).

Conclusions: The results demonstrated that the noninvasive and convenient pulse-taking diagnosis described in this paper has the potential to become a low-cost and accurate method to monitor the development of diabetes. We are collecting more data to improve the accuracy for detecting hypertension and hyperlipidemia among diabetic patients. Mobile devices such as sport bands, smart watches, and other diagnostic tools are being developed based on the pulse wave method to improve the diagnosis and monitoring of diabetes, hypertension, and hyperlipidemia.

Keywords: diagnosis; hyperlipidemia; hypertension; pulse wave analysis; type 2 diabetes.

Figure 1

Traditional Chinese medicine pulse life informatics analysis system.

Figure 2

The wristband pulse wave information acquisition terminal.

Figure 3

The pulse wave image collected.

Figure 4

A typical single pulse wave. h: height; h₁: height of dominant wave; h₃: height of tidal wave; h₄: height of dicrotic notch; h₅: height of dicrotic wave; mm: millimeter; t: time; s: second; t₁: time distance between the start point of pulse wave and dominant wave; t₄: time distance between the start point of pulse wave and dicrotic notch; t₅: time distance between dicrotic notch and the end point of pulse wave; W: width of percussion wave in its one-third height position.