Insole Systems for Disease Diagnosis and Rehabilitation: A Review

Abstract

Some chronic diseases, including Parkinson's disease (PD), diabetic foot, flat foot, stroke, elderly falling, and knee osteoarthritis (KOA), are related to orthopedic organs, nerves, and muscles. The interaction of these three parts will generate a comprehensive result: gait. Furthermore, the lesions in these regions can produce abnormal gait features. Therefore, monitoring the gait features can assist medical professionals in the diagnosis and analysis of these diseases. Nowadays, various insole systems based on different sensing techniques have been developed to monitor gait and aid in medical research. Hence, a detailed review of insole systems and their applications in disease management can greatly benefit researchers working in the field of medical engineering. This essay is composed of the following sections: the essay firstly provides an overview of the sensing mechanisms and parameters of typical insole systems based on different sensing techniques. Then this essay respectively discusses the three stages of gait parameters pre-processing, respectively: pressure reconstruction, feature extraction, and data normalization. Then, the relationship between gait features and pathogenic mechanisms is discussed, along with the introduction of insole systems that aid in medical research; Finally, the current challenges and future trends in the development of insole systems are discussed.

Keywords: chronic diseases; gait analysis; insole systems; plantar pressure.

Figure 1

Demonstration of diseases, sensing techniques and current products. Sensing techniques are the foundation of insole systems, while insole systems are used to assist the diagnosis and rehabilitation of diseases.

Figure 2

Illustration of prototypes of (a) piezoresistive, (b) resistive, (c) piezoelectric, (d) capacitive and (e) temperature-humidity insole sensing techniques. These figures are inspired by references [10,14,18,19,20,32] but are originally drawn by authors of this article.

Figure 3

Three steps of processing gait data for medical purposes: 1. Reconstruction, 2. Extraction, 3. normalization. Suitable techniques of each step are mentioned in Section 3, Section 4 and Section 5.

Figure 4

Illustrations of four predominant gait features, complementary features like gait velocity and pressure distribution can be obtained using the features above.

Figure 5

Desired sensing points on the foot, they are suitable for different diseases.

Figure 6

Demonstrations of six chronic diseases mentioned. Each disease has different pathogenesis and causes different gait abnormaility with correspondent jeopardies.

Figure 7

Summarization of challenges (including black-box issue, algorithms overlook gait data fluctuation, and limited accuracy due to hardware in Section 6) and outlook (including predict users’ performance under different scenarios and multi-sensing based digital twin model in Section 7).