ICT and Physical Activity

Abstract

Recently, physical inactivity has emerged as a problem worldwide and the effective application of digital health using information and communication technology (ICT) has been focused on as a potential solution to this problem. An overview of research using web-based physical activity assessment systems and wearable devices is presented. The following three topics will be discussed: (1) mobile health market and sensing technology, (2) evaluation of physical activity using ICT, and (3) a study of physical activity promotion through digital intervention. Wearable devices (a generic term for information terminals that are worn and carried around) have made remarkable progress in recent years, and physical activity promotion and weight loss interventions using PCs (personal computers) and mobile devices are considered to have advantages in terms of human and economic costs compared to traditional face-to-face interventions. A wide variety of data will be collected using various wearable devices, and artificial intelligence (AI) technologies such as machine learning and deep learning will be incorporated to develop applications that introduce future risk projections and other information. Rather than people catching up with technological advances, it is important to take a viewpoint of how technology can be used to enrich people’s lives.

1. Introduction

In recent years, lifestyle changes resulting from technological developments in the work and transportation environment have led to widespread physical inactivity in people worldwide [1]. As long sedentary behavior and physical inactivity are respectively associated independently with risk factors for noncommunicable diseases such as diabetes, coronary heart disease, and some cancers, in 2020, WHO issued guidelines for physical activity and sedentary behavior, recommending at least 150–300 minutes of moderate-intensity aerobic physical activity throughout the week for adults, at least 75–150 minutes of high-intensity aerobic physical activity, and at least 2 minutes of muscle-strengthening exercise and less sedentary behavior per week for adults [2].

Although the ecological approach [3] such as community intervention, sports promotion, health education, and environmental improvements such as maintenance of the parks and walkways are needed to promote physical activity among a large number of people, efforts to promote physical activity using ICT represented by smartphones, which have been remarkably developed and spread in recent years [4], have been the focus of public interest. However, there are a variety of physical activity intervention methods using online services, and the effectiveness of these methods is not clearly described. In other words, clinical research has not caught up with technological advances. This paper reviews the research to the present and summarizes the future needs of the society.

2. Mobile health market and sensing technology

Healthcare management, medical treatment, and medical support using smartphones and tablets for health-related information are known collectively as mobile health (m-Health). The global m-health market is growing rapidly, from $4.5 billion in 2013 to $23 billion in 2017 [5] and from $36.43 billion in 2021 to $113.25 billion in 2026 [6]. The use of the Internet and mobile devices, especially SMS, has been shown to be a cost-effective intervention to promote physical activity for a large population [7]. Systematic reviews have reported a positive effect mainly in Europe and North America [8, 9].

Recently, wearable devices (a general term for information terminals that are carried around with the wearer) have made remarkable progress, and some representative examples include wristwatch-type smart watches, glasses-type smart glasses, smart rings with sensors mounted on them, smart contact lenses, and even skin-attached sensors for pulse monitoring of pulse rate by skin-attached sensors [10] has also been reported as a wearable device. The main sensing methods used in wearable devices include acceleration, gyro, light, infrared, bioelectrical potential, and GPS. The three main categories of applications are (1) information on the body and mind (blood flow, heart rate, body temperature, EEG, eye movement), (2) information on position and speed (location information, movement information), and (3) input (operation) and motion support (body movement, muscle movement) [11].

Health-related data can be collected using sensors installed in the smartphone itself. The wearable devices mentioned previously can be used to collect a wide variety of data from multiple sources, capture it on a smartphone via Bluetooth, and manage the data using applications. By utilizing a web-based application, data can be stored and processed on a cloud server via a 4G (LTE) network. By using the cloud, service providers have the advantage of expanding the fields in which they can create new and various services, and users can check and manage visualized data such as physical activity, sleep status, stress checks, and dietary calories.

Active lifestyle advice based on observational data, including elements of behavioral science and gamification, feedback content to increase user adherence, and social networking service (SNS)-based community sites could work on human cognition, arouse motivation, trigger physical activity, and contribute to the prevention of lifestyle-related diseases caused by physical inactivity.

3. Evaluation of physical activity using ICT

Up to the present, various methods have been developed to assess physical activity, such as questionnaires, accelerometers, and more recently Global Positioning System (GPS), depending on the purpose of use. However, there is little evidence for the promotion of physical activity for a large population of several tens of thousands of residents. This is because accelerometers are difficult to widely distribute to a large number of people due to their cost, and self-reports have low validity and are not suitable for longitudinal evaluation of the effects of interventions. Conventional physical activity measurement methods based on questionnaires have the advantage of measuring physical activity in a large number of subjects, but they have limitations in measurement accuracy. The correlation coefficient is about 0.3–0.6 when compared to Doubly-Labeled Water (DLW), which is the gold standard method for measuring total energy expenditure (TEE) [12].

The validity of self-reported physical activity measurement methods using the web and cell phones has been reported in several studies. A web-based questionnaire called Active-Q was developed to select the intensity of each activity from four categories (work, transportation, leisure, and sports), and its validity was verified using the DLW method on 37 general adult subjects, and a significant association of r = 0.52 (p < 0.001) was reported for TEE [13]. Thirty men and women with cardiovascular diseases were asked to self-report their activity levels during the day and at night using a smartphone-based system and validated with an accelerometer, and an association of r = 0.45 (p < 0.05) was reported in physical activity levels [14]. These self-reported physical activity measurements using the web or cell phones are superior in that they can simultaneously assess the physical activity of many people at the same time, but their measurement accuracy is similar to that of traditional questionnaires, and the content of the questions is not well designed. On the other hand, a study has been published in which a high correlation of 0.88 was found between TEE using the DLW method and the physical activity measurement method using the paper-based 24-hour recall method [15]. The validity of this measurement method for assessing TEE exceeds the accuracy of accelerometers when the DLW method is used as the standard. The paper-based method is difficult to use for a large number of people due to the lack of interactive responses and difficulty in memory recall.

A physical activity system using IT devices based on the 24-hour look-back method has been developed, and its validity has been shown to correlate with TEE by the DLW method with r = 0.874 (p < 0.001) and with activity energy expenditure (AEE) with r = 0.679 (p < 0.001) [16]. It is reported that the measurement accuracy is equivalent to that of a 3-axis accelerometer in estimating TEE. A feature of this system is the method of dragging and dropping illustrations on the web screen to fill in the timeline (Figure 1). Using such a system, it is possible to assess the physical activity of a large number of people simultaneously and at the same time with high accuracy at low cost [17] and to assess each activity intensity (Figure 2) [18]. It is also possible to evaluate the physical activity of a large population with COVID-19 online, and the effects of COVID-19 on physical activity have been clarified [19, 20]. The 24-hour activity recording method suggests the possibility of physical activity assessment by voice input (Figures 3 and 4) by introducing a voice recognition application programming interface (API) [21].

Figure 1.

Screenshot of the web-based simplified physical activity record system [17].

Figure 2.

Activity time per activity intensity [18].

Figure 3.

A behavior-recording system with a voice input app [21].

Figure 4.

Correlation of average measured METs with voice recognition app [21].

4. A study of physical activity promotion through digital intervention

Literature studies on e-health and m-health regarding physical activity, sedentary behavior, and diet [22] reported an average annual increase of 26% in the number of articles between 2000 and 2016, with a rapid increase since 2014. A review [23] that conditioned digital interventions on healthy lifestyles cited 107 articles from 2015 to 2020 and found that methods for successful digital interventions include monitoring, motivation, goal setting, personalized feedback, participant engagement, psychological empowerment, persuasion, digital literacy, self-efficacy, and authenticity.

Forty review articles published up to March 2021 on digital interventions to promote physical activity were analyzed [4]. Physical activity was assessed by objective methods (wearables and smartphone active trackers) in 30%, objective methods plus self-report in 60%, and evaluation based on behavioral change theory, including goal setting, self-monitoring, and feedback as a theoretical framework, in 55% of the cases. In order to clarify the effectiveness of digital interventions for physical activity promotion, the need to introduce an evaluation framework at the system development stage is noted.

5. Introducing advanced research using smart shoe sensors

Smart shoes are Internet of Things (IoT) shoes with built-in motion sensors and other sensors in the shoes and AI. These sensors, which include accelerometers, gyros, and GPS, can evaluate walking and running paces, landing times, pronation, stride, pitch, and landing impact force during movement. It is important to evaluate the quality of walking because walking is a main part of physical activity for people living a typical life. Good gait quality means that a person requires less energy to move the same distance. In other words, if the mechanical efficiency of walking (= total work/energy expenditure) can be evaluated using smart shoes, the quality of each individual’s gait could be assessed and improved [24].

A walking test was conducted on 35 middle-aged and elderly subjects to clarify the relationship between mechanical energy efficiency in walking and the level of physical activity. The subjects performed three types of walking, normal walking, fast walking, and slow walking, on an approximately 10-m walking path for five strides each, and the energy efficiency of one walking cycle was evaluated using a force plate. At the same time, smart shoes (with a built-in small motion sensor) and plantar pressure distribution were evaluated and assessed to clarify whether the energy efficiency of walking could be evaluated even without a force plate (Figure 5). Furthermore, we evaluated energy expenditure by DLW for 1 week and physical activity by the triaxial accelerometer, in order to clarify the relationship between these data and the mechanical efficiency of walking (= total workload/energy expenditure).

Figure 5.

Basic data collection experiment using smart shoes.

A schema of effective algorithms for promoting physical activity via AI is shown for data obtained from smart shoes (Figure 6). Deep learning is incorporated using an application programming interface (API) in order to link with the software that has been developed so far. Data obtained from the smart shoes are transformed through the conversion process (C1–C7) and then matched (M1) to generate a program that determines the validity of the data using deep learning technology. Accuracy management of wearable sensors is the key to evaluating walking behavior with wearable sensors (smart shoes) and generating tailor-made advice provided by AI. The key point is to evaluate mechanical energy with high accuracy. Recent measurement technology has advanced dramatically, and there is a possibility of evaluating the energy efficiency of walking with a certain level of accuracy using smart shoes, which are linked to a smartphone and sensors and are considered to be capable of inexpensive, high-precision motion analysis.

Figure 6.

AI-based feedback system using smart shoes.

The final goal is to develop an AI feedback system using smart shoes based on data obtained during walking. If it is possible to increase physical activity within the optimal range for each individual, it could lead to the prevention of lifestyle-related diseases such as diabetes and hypertension, and contribute to solving the various problems confronting an aging society.

6. Conclusions

The history of digital intervention research using online systems is short, and technology is constantly advancing, but it is not always up to date with people’s needs. JMIR Publications, a research journal dedicated to digital health, launched in 1999, has a vision: “We believe that people should be able to make effective, well-informed decisions through health-related research and technology, and that they should be able to make the best use of technology. We envision a world where health research and technology enable people to make effective, informed decisions, self-manage their health and well-being, and live happier, healthier lifestyles [25]. In the future, this field will continue to develop applications that use various wearable devices to collect a wide variety of data and incorporate artificial intelligence (AI) technologies such as machine learning and deep learning to introduce predictions of future risks. It is important to consider how technology can be utilized to enrich people’s lives, rather than how people can catch up with technological advances.

Finally, as proposals for increasing physical activity using the ICT environment, the following four approaches should be considered: (1) an approach to elements that can be changed based on an analysis of individual behavior, (2) efforts to prevent health problems (sleep disorders, stiff shoulders, etc.) associated with increased use of PCs, smartphones, and other devices, (3) a system in which each person can choose the frequency of ICT-based interventions, and a system in which the frequency of ICT-based interventions can be adjusted as much as possible to the frequency of face-to-face meetings with ICT users, and (4) automated messages need to be moderated to the extent that they are not unpleasant and do not become boring.

Acknowledgments

I would like to thank my collaborators for their active support of the study and the subjects who cooperated in the research. This work was supported by JSPS KAKENHI 17 K01800 and 21 K11478.

Conflict of interest

The authors declare no conflict of interest.