The goal of this scoping review was to identify sensing modalities in recent studies that evaluate exergame efficacy. For this, 3 systematic searches were conducted in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement [21] and PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) [22]. A PRISMA-ScR checklist is provided in Multimedia Appendix 1. The database searches were conducted at different points in time by one of the researchers, each without a detailed review protocol.

Due to the interdisciplinary nature of the research field, 3 different databases were included in the review. To cover both computer science and clinical research, the ACM Digital Library and the subject-specific PubMed database were searched on September 14, 2022, and again on July 14, 2023. A final search on the engineering-focused IEEE Xplore database was conducted on January 15, 2024.

We defined 4 main requirements for studies to be included in our review. First, studies have to feature the evaluation of an exergame. Second, the evaluation has to be conducted quantitatively with sensors with regard to the game’s efficacy. Third, the games should focus on general fitness or sports to avoid studies focusing on activities of daily living. Finally, the studies should be recent, which we defined as being published between 2015 and 2023. Based on these requirements, relevant search terms were identified, combined, and generalized as follows:

(exergam* OR “fitness game”)

AND (efficacy OR evaluat* OR “heart rate” OR vo2 OR oxygen)

AND (fitness OR sport)

For ACM Digital Library, the search terms were searched for in the categories “Title,” “Abstract,” and “Author Keyword,” each category connected with an OR operator, whereas for PubMed and IEEE Xplore, the search terms were typed “as is” into the “Query Box” and “Command Search,” respectively. A detailed search strategy is given in Multimedia Appendix 2.

Afterward, the following eligibility criteria based on the previously stated requirements were defined. The results were screened and filtered accordingly:

The studies were split among the authors to be screened by abstract and assessed for eligibility. The results of this screening process were then presented and discussed conjointly.

Eligible studies were evaluated with regard to their general study design (including participant numbers and focus group), their evaluation methods (including evaluation metrics, vital signs, and sensor usage), and the exergames used (including exergame sensors as well as additional motion sensing). The results were again discussed, assessed for relevance for the review, and generalized conjointly.

The first search yielded 253 studies of which 30 were removed before the screening process as they were duplicates or not available in English or German. An additional 31 studies were identified through previous work of included authors and citation searching, resulting in a total of 258 studies included in the screening process. 208 studies were excluded in this screening process based on the eligibility criteria stated above, leaving a total of 49 eligible studies. The second search yielded an additional 40 studies, of which 32 were excluded from the screening process. The final search, conducted on the IEEE Xplore database, yielded another 50 studies. After screening by title, abstract, and eligibility criteria, 10 studies were deemed fitting, resulting in a total of 67 studies that were included in our review.

Table 1 gives a brief summary of the vital sign-sensing statistics. Multimedia Appendix 3 [23-89] and Multimedia Appendix 4 [23,24,26-58,60-90] feature additional tables that present details for all 67 studies included in the review. Multimedia Appendix 3 provides an overview of the study design, participants, and how efficacy was assessed. Participation numbers ranged from 6 [23] to 360 [24], with a median of 28. There was a large variety of different focus groups, with the biggest group being healthy adults (n=19 studies). Multimedia Appendix 4 details the exergames and corresponding sensors used in each study and gives information on additional motion sensing if there were any. Not all studies explicitly mention all the exergames evaluated; some feature the evaluation of a multitude of different games, and some tested games they developed themselves. Overall, the 67 studies included in our review feature the evaluation of approximately 49 different exergames. The overall review process with all screening steps is outlined in the PRISMA flow diagram in Figure 1.

As per our inclusion criterion, all 67 studies aim to assess efficacy. For that, 54 studies used sensors to measure vital signs, while 11 studies relied only on motion sensing, and 2 studies assessed efficacy without any sensors but with standardized tests only. The majority, that is, 52 of the 54, measured the player’s heart rate (HR), and 22 measured the oxygen uptake (VO₂) (more details in Table 1). Furthermore, 18 studies measured other vital signs such as blood pressure, blood lactate concentration, and carbon dioxide production.

In the context of exergames, efficacy refers to the effectiveness or ability of exergames to achieve their intended goals or their desired effect under optimal or controlled conditions. Efficacy is distinguished from effectiveness, with efficacy focusing on how well an approach works in ideal conditions, while effectiveness considers its performance in real-world or practical situations [91]. However, this distinction is not always clearly made by the researchers [25]. Therefore, in the context of this review, the 2 terms are used interchangeably. Typical goals for exergames can be defined as an increase in the participants’ performance, physical activity, or motivation [10,92]. Exergame efficacy is usually evaluated by analyzing these metrics over time or in comparison with conventional physical exercises. Often, they are estimated by measuring vital signs, doing standardized tests, or filling out questionnaires. Efficacy is a highly individual metric and will vary for different activities, different target demographics, and from person to person [10].

When looking at the studies’ approaches to assessing exergame efficacy, several different study designs were identified. First of all, a distinction can be made by whether the studies evaluate a single group or compare 2 or more groups. The former is referred to as a within-subject or cross-over study. The latter category is referred to as controlled studies, which again can be split into 2 subcategories. Randomized controlled studies compare randomly split groups, while cross-section studies compare nonrandomized groups based on preexisting conditions. The studies can also be categorized by how many measurements are done. Single-measurement studies use measured metrics and questionnaires to compare the participants’ performance, motivation, and general physical activity with nongaming-related exercises. Alternatively, studies with the repeated-measures design, focusing on the long-term effects of exergaming, do the same measurements more than once to observe longitudinal developments. The latter approach was used by 22 studies, while the other 45 studies focused on the first approach. Details on the study design of every included study can be found in Multimedia Appendix 3.

A total of 53 out of the 67 (79%) studies focus on efficacy evaluation by measuring vital signs in their studies. Out of all analyzed studies (more details in Table 1), 35 reported mean HR, and 31 reported peak HR or proportionate maximum HR (%max HR), with the maximum HR commonly being age-predicted. Ventilation-based evaluation took place in 21 analyzed studies, reporting peak VO₂ or proportionate maximum VO₂ (%max VO₂).

Based on the recorded vital signs, 15 studies processed their participants’ data to indicate the recorded energy expenditure (EE) for the exergame activity. In total, 16 studies calculated and reported the metabolic equivalent of task (MET) values for each activity. A total of 8 studies overlapped with studies reporting EE. Furthermore, noteworthy vital-sign-based evaluation metrics are systolic and diastolic blood pressure [26-29], heart rate reserve [30-33], pulse wave velocity with total peripheral resistance and stroke volume [34], respiratory exchange rate [35,36], lactate concentration [37,38], maximum power output [39], muscle activity [40], concentration through EEG (electroencephalography) [41]. In 3 studies, other derived indicators are used: play intensity and hemodynamic reactivity are calculated based on players’ heart rate [37], and the so-called activity counts are based on accelerometer measurements [42,43]. Many studies used subjective questionnaire scales for assessment, which we do not evaluate as they are unrelated to vital sign sensing.

Although many analyzed studies (35 out of 67) comprised more than 1 session of playing exergames, only a minority evaluated changes over time (22 out of 67), and thus qualifies as a longitudinal study in the context of this paper. The evaluation time for longitudinal studies ranged from 4 weeks [44-47] to 6 months [48].

As detailed in Multimedia Appendix 3, most researchers analyzed players’ performance. The performance evaluation was predominantly done objectively by tracking changes in anaerobic [46,49,50] or muscular fitness [51,52], functional strength [50], peak VO₂ [28,48], postural control [45], reaction time [24], or cognitive function [53]. The performance was sometimes evaluated subjectively using exercise self-efficacy score [54].

Other evaluation metrics analyzed over time include changes in intensity, usually based on HR or EE [29,45,55,56], whereas further research publications evaluated playtime for different physical activity levels [28,48,51]. In addition to performance and intensity, 6 research publications analyzed changes in motivation, such as play frequency and duration [56] as well as flow [45], player engagement [55] or enjoyment [54], and happiness [57].

The 67 reviewed studies assess various exergames that use different motion sensing technologies, among which the Kinect for Xbox 360 (Microsoft Corp) [93] is the most commonly used (26 out of 67). The Kinect combines a regular RGB (red-green-blue) camera with a depth sensor and can thus be considered a camera-based system. In addition, 1 study each included an exergame using the EyeToy (London Studio) for PlayStation 2 [54] and PS Move (Sony Interactive Entertainment) for PlayStation 3 [44], resulting in 28 out of 67 studies featuring at least one camera-based system. Ergometers, dance mats, balance boards, and other sensing technologies specialized for specific motions are used in 21 out of 67 studies and, thus, make up the second-most common type of motion sensing technology.

VR systems and body-worn inertial measurement units (IMUs) are featured less commonly. Only 5 out of 67 studies feature body-worn IMUs for motion sensing, and only one of them uses IMUs not included in a smartphone or game controller. Furthermore, 7 studies evaluate exergames that use the Nintendo Wii remote controllers held in the player’s hands. These controllers are not body-worn and combine an IMU and an infrared sensor with an external emitter for tracking. Together with VR systems, which were used in 16 out of 67 studies as either standalone or within the so-called ExerCube (Sphery AG) [94], they make up the category of hybrid sensing (overall n=22). Finally, one study [58] features a game that does not use any form of motion sensing.

Out of the 67 studies, 21 did additional motion sensing during the exergaming for evaluation purposes. Common motion sensors used for this task are body-worn accelerometers and IMUs (used in 12 studies) and force plates (used in 4 studies). Furthermore, 2 studies used hybrid motion-capture systems, and 2 used camera-based systems. A specialized system to measure reaction times, an exercise bike, surface electromyography, and a handgrip dynamometer were used in 1 study each.

Motion quality, as a term, is not clearly defined and is often visually evaluated by a professional physiotherapist or sports scientist. One of the most comprehensive approaches to defining the term is given by Skjaerven et al [20], where they investigated the lived experiences of a group of expert physiotherapists in search of essential characteristics and features of the term. They found that motion quality has many different aspects, including the biomechanical “characteristics of path and form in [motion]” as well as the physiological and temporal “characteristic of flow, elasticity, vitality, and rhythm in [motion].” Therefore, to provide specific feedback on the full motion quality, it would be necessary to assess the whole motion by tracking and evaluating the motion of all relevant body parts at every point in time, independent from the gameplay.

An essential part of exergaming is the game’s ability to track and analyze the player’s motions to judge the player’s gameplay and their exercise quality. Often, exergames focus on the gameplay aspect, only evaluating motions implicitly by requiring players to interact with the virtual environment, for example, slashing, collecting, or avoiding virtual objects [3,39,40,95,96]. Alternatively, some games combine both evaluations by having players fit a predefined shape [59,97,98]. While such an approach might be suitable to verify if players are performing a specific motion at all, it only enables detecting static poses or certain joints without analyzing the holistic motion execution.

None of the 49 games featured in the studies explicitly assess the quality of exercises performed. However, 7 studies use additional motion sensing to do some form of quality assessment during their evaluation. In addition, 4 of these 7 studies focus on specific motion aspects, such as assessing the angular displacement or range of motion of certain joints [41,55], quantifying postural sway [45], and evaluating shoulder flexibility [26]. Only 3 studies [44,51,60] assess the players’ motion quality as a whole, using a multitude of sensors and either a standardized test or a professional physiotherapist. However, for 2 of the studies, these assessments do not happen during the exergame but are used to identify general motion quality at times when the subjects are not playing. The final study [44] employs an expert to do an analysis of gross upper body biomechanics based on video recordings of the subjects during the exergames. Therefore, this is the only study in our review that fulfills the condition of holistically evaluating all relevant body parts at every point in time during the exercise, which we consider necessary for motion quality assessment, according to Skjaerven et al [20]. However, this is not an automated process and instead requires manual input from an expert.

To the best of our knowledge, this review represents the first work analyzing sensor usage in studies that examine exergame efficacy. We found 67 eligible studies, which we analyzed with regard to sensor usage for both efficacy evaluation and motion analysis. Overall, HR was the most commonly used vital sign to evaluate efficacy (n=52), while the Microsoft Kinect was the most commonly used exergame sensor (n=26). The results of the analysis show that the sensors used in the exergames and the sensors used in the evaluation are, in most cases, mutually exclusive.

From the review, 2 main conclusions can be drawn: First, the majority of papers focus on measuring vital signs to infer a physical activity metric. Second, an overall lack of motion quality assessment can be stated among exergame studies. For the studies, this lack of quality assessment can lead to undetected and unwanted influences on the results. For example, it might be hard to differentiate if a lack of efficacy should be attributed to the exergame or to the participants’ execution, especially in longitudinal studies. Furthermore, it also poses a problem for the exergames themselves since the players are incentivized to optimize for score, which might lead to optimal gameplay that does not coincide with a correct exercise execution and might lead to injuries instead of increased health [9]. Therefore, in the following outlook, we outline how different types of sensor technologies could be used with existing motion analysis techniques to assess motion quality in exergames.

This scoping review is prone to the same search-related limitations as other reviews of this type. First, only articles published in international journals and full articles published in conference proceedings written in English or German are considered. Therefore, potentially relevant studies published in other languages may have been missed. Second, some articles were excluded from the analysis because the required information to assess their eligibility based on the inclusion criteria was not provided. Third, we assess a lack of reproducibility for the ACM Digital Library. In general, a very high volatility can be noted in the amount of records ACM’s search function returns when changing individual words or operators. Since the current search results in less records than the original search, which were all included in the review, we consider this to be a minor limitation. Finally, out of the 67 studies included in this scoping review, 8 studies were not found firsthand using the search terms but instead through previous work of included authors and citation searching. This may indicate that there would have been even more fitting search terms for the review question. However, due to the author’s experience with the topic and the unambiguity of the results, we are confident that we were able to mitigate any possible negative effects and that the included studies present a complete overview of the current state of research.

In addition to limitations related to the search, we can also note 2 limitations related to the scope of evaluation. They are a deliberately chosen result of our research focus, which means to analyze the studies’ methodology with regards to sensing instead of their results. First, the definitions for the terms “efficacy” and “motion quality” used in this paper focus on how these 2 metrics are evaluated and do not go into detail on what qualifies as “good” efficacy or motion quality. Instead, we evaluate how well the studies are able to assess efficacy and motion quality with the methodology and especially the sensors they use. Therefore, we do not define what a “high/low quality motion” may look like as this question is highly specific to the individual motion and therefore cannot be answered generally. For the same reason, we also do not go into detail on the cohorts’ age, sex, and focus group as they are predominantly relevant to the studies’ results.

In this paper, we conduct a scoping review of recent studies that examine exergame efficacy to determine which sensors are being used and how they are being used during gameplay and evaluation. Our results show that most studies evaluate exergame efficacy by measuring vital signs, the most common being heart rate (52 out of 67) and oxygen consumption (22 out of 67). However, motion quality is only assessed in 7 out of 67 studies despite being an important factor in an exergame’s effectiveness and risk of injury. Furthermore, out of the 49 exergames evaluated in the reviewed studies, none feature quality assessment during or after gameplay, and only 3 studies feature motion quality assessment beyond the exergame’s feedback.

Since exergames already use motion sensing technologies to track the player’s motions, they could also be used for external quality assessment. Therefore, we discuss recent advances in the field of motion analysis and potential use cases of different sensors commonly used in exergames. We come to the conclusion that many of the same sensing technologies typically used in exergames and exergame studies are well-suited for additional motion quality assessment to ensure consistent exergame effectiveness and reduce the likelihood of injury while exergaming.