Developing a Serious Video Game to Engage the Upper Limb Post-Stroke Rehabilitation

Abstract

Stroke often leads to severe motor impairment, especially in the upper limbs, greatly reducing a patient’s ability to perform daily tasks. Effective rehabilitation is essential to restore function and improve quality of life. Traditional therapies, while useful, may lack engagement, leading to low motivation and poor adherence. Gamification—using game-like elements in non-game contexts—offers a promising way to make rehabilitation more engaging. The authors explore a gamified rehabilitation system designed in Unity 3D using a Kinect V2 camera. The game includes key features such as adjustable difficulty, real-time and predominantly positive feedback, user friendliness, and data tracking for progress. The evaluations were conducted with 18 healthy participants, most of whom had prior virtual reality experience. About 77% found the application highly motivating. While the gameplay was well received, the visual design was noted as lacking engagement. Importantly, all users agreed that the game offers a broad range of difficulty levels, making it accessible to various users. The results suggest that the system has strong potential to improve rehabilitation outcomes and encourage long-term use through enhanced motivation and interactivity.

1. Introduction

Stroke represents a highly prevalent and life-threatening neurovascular emergency that remains one of the leading causes of mortality worldwide. According to 2019 data, it was identified as the fifth leading cause of death globally, with implications for long-term disability [1]. The burden of stroke commonly results in physical and cognitive impairments that affect patients’ quality of life and show the need for effective rehabilitation [2]. Approximately 101 million individuals worldwide have experienced a stroke [3], nearly doubling over the last three decades. Numerous studies underscore the prevalence of post-stroke cognitive impairment (PSCI), affecting 25% to 80% of stroke survivors [4]. It is estimated that approximately 50–80% of stroke survivors exhibit some degree of upper limb impairment in the early phase following a stroke [5]. When a stroke occurs, the sudden interruption of blood flow to the brain, caused by either a blood clot blocking a brain vessel (ischemic stroke) or the rupture or leakage of a blood vessel in the brain (hemorrhagic stroke), leads to the immediate death of brain cells. Prompt intervention is crucial. The impact of a stroke depends on the affected area of the brain, influencing speech, comprehension, emotions, sensations, motor functions, and vital functions such as heart rate, swallowing, and breathing. The consequences of a stroke may manifest as weakness, paralysis (hemiplegia), coordination difficulties (apraxia), changes in muscle tone (hypertonia or hypotonia), subluxation, contracture, swelling (edema), and pain [6]. Stroke affects not only motor functions but also cognitive functions, which can impact various domains, such as memory, focus, attention, control, and executive functions. These cognitive impairments influence rehabilitation outcomes and the way different techniques and technologies are applied [7].

Stroke rehabilitation is broadly characterized as any aspect associated with stroke treatment, with the overarching objective of mitigating disability and fostering engagement in everyday activities. Its objectives encompass the prevention of deterioration, improving patient functionality, and achieving the utmost level of independence—physically, psychologically, socially, and financially—within the enduring limitations imposed by the stroke [1]. In stroke rehabilitation, various innovative approaches are being explored to improve patient outcomes, including traditional therapies, technology-assisted solutions, and gamification, which introduces interactive, rewarding activities to enhance engagement and motivation during recovery.

Gamification can be characterized as a method of augmenting a particular activity through the incorporation of motivational elements. Huotari and Hamari [8] define gamification as the creation of a psychological experience that has the same positive impact as games. The term serious game is often closely associated with gamification and refers to a game designed with education or rehabilitation as its primary objective. Such games integrate elements of entertainment, attentional engagement, and problem solving to challenge functionality and performance. For the creation of these games, virtual reality (VR) can be implemented as a form of human–computer interaction (HCI), wherein individuals engage with a three-dimensional (3D) interface within a simulated environment comprising digital objects. The games mentioned in the articles by AlMousa et al. [9], Postolache et al. [10], Jayasree-Krishnan et al. [11], and John et al. [12] are some examples of the use of VR in game creation for upper limb stroke rehabilitation that employ various hardware devices, including Leap Motion [9,10,11,13], smart gloves (equipped with embedded sensors) with headbands [10], or an instrumented fork and knife with a 3D printed pressure pad, which measure and communicate information [11].

Closely associated with serious games are two critical factors: feedback and management of difficulty. Feedback represents the manner in which the game reacts to alterations or decisions executed by the user, and it is indispensable for the creation and maintenance of meaningful gameplay [14]. In the realm of rehabilitation, heightened attention is placed on positive feedback, given its pivotal role in fostering patient motivation and engagement. Negative feedback should be used very cautiously, or not at all, as it may provoke feelings of demotivation and frustration, particularly among post-stroke individuals grappling with motor impairments and potentially experiencing depression. Also, a decrease in tolerance or motivation often leads to intentional or unintentional cheating or, in the worst-case scenario, avoiding performing the rehabilitation exercises altogether [14]. Typically, feedback manifests in three modalities: visual, auditory, and haptic. Visual feedback mechanisms encompass score displays [9,10,11,13] or graphs [9]. Auditory feedback methods entail the emission of sound cues signifying task completion and fostering a sense of accomplishment [10,13,15], along with in-game instructions designed to enhance exercise execution [15]. In a haptic context, there exists the potential for vibration of the input device [12].

Within the domain of rehabilitation gaming, addressing failures and management of difficulty must be taken even more seriously, particularly due to the technical inexperience of most post-stroke patients, which increases the likelihood of encountering setbacks. Consequently, these setbacks necessitate careful management to foster patient encouragement and facilitate their training regimen. Rather than terminating the game following a failure or defeat, or ascribing negative connotations to such instances, such as loss of points, patients should instead be incentivized for their participation. This approach encourages renewed engagement with the games, enabling them to strive toward achieving rehabilitation objectives more effectively. It is imperative to maintain a balance in the management of difficulty, ensuring that tasks are not overly simplified, which could lead to user boredom. Similarly, tasks should not be excessively challenging, as this may result in user frustration and disillusionment. In both scenarios, demotivation ensues, ultimately impacting the effectiveness of rehabilitation. The implementation of diverse difficulty levels is commonly observed across all genres of games. Level progression should be contingent upon the patient’s performance status [16]. Upon completion of a level, patients advance to the subsequent level, motivating them to achieve success in the new stage. Adaptivity is a strategy grounded in an algorithm that dynamically adjusts the difficulty of the game in real-time, based on the performance and skills of the patient. This ensures an appropriate level of challenge that can result in the game becoming either easier or harder, depending on the player’s performance [15].

Prior to the development of this game, several existing applications were thoroughly analyzed to create an innovative solution capable of addressing challenges observed in previous works. Regarding hardware, certain examples demonstrated the use of head-mounted displays (HMDs) to enhance the immersive virtual reality experience in serious gaming applications [11,13]. For motion capture, various approaches were identified, such as the use of motion capture cameras like Kinect V2 for full-body tracking [17], Leap Motion for hand tracking [9,13], and data gloves [10]. At the software level, the majority of reviewed examples utilized Unity3D (2022.3.27f1) [18] for the development of games and 3D environments [9,10,11,13]. In terms of gameplay, the primary objective consistently centered on encouraging users to perform movements in engaging and varied ways to maintain their motivation. Notable examples include games that involve lifting weights [13], grabbing and throwing balls at designated targets [10], or reaching specific positions when certain areas are activated (e.g., turning green) [12]. Specifically concerning Kinect-based applications, some studies were identified; however, to the best of the authors’ knowledge, none of these studies involved games created to be focused on upper limb stroke rehabilitation using Kinect. Existing examples primarily address the use of Kinect V2 for lower limb stroke rehabilitation [17] or the use of Kinect Xbox 360 for upper body applications related to cerebral palsy [19]. Some studies explored the use of Xbox 360 Kinect for upper limb stroke rehabilitation, but they primarily utilized existing commercial games that are not specifically designed for this purpose. Ain et al. [20] incorporated games such as “Tennis Player,” “Joy Riding,” and “Rally Ball” from the Kinect Adventure Package and Kinect Sports Package, which are not health focused. Afsar et al. [21] involved games like “Mouse Mayhem,” “Traffic Control,” “Balloon Buster,” and “Mathercising” from Dr. Kawashima’s Body and Brain Exercises package. While these games are health related, they are not primarily intended for upper limb stroke rehabilitation, which might not consider certain critical aspects, such as the correctness of movements (if the movements involve reaching for objects that are beyond the patient’s reach, the user will inevitably need to engage their body, extending beyond the use of their arms alone, to reach them), the handling of errors or failures within the game, or the design of the interface to accommodate individuals facing greater challenges, whether in executing the movements or in utilizing the technologies.

The primary objective was to develop an upper limb stroke rehabilitation system that integrates key gamification features to enhance user motivation and adaptability through adjustable difficulty, real-time feedback, and interactive gameplay features. The game emphasizes simplicity and is specifically designed to support individuals with low/moderate stroke impairment (cognitive and physical) in improving their motor function recovery. The system was developed with the primary objective of supporting stroke survivors who exhibit mild to moderate impairments in the upper limbs and possess sufficient cognitive and physical abilities to interact with motion-based gamified environments. At this stage, individuals with severe motor deficits or those requiring full assistance were not included in the system’s target group. Healthy participants were recruited to assess usability and motivation in a controlled environment prior to proceeding with clinical trials involving stroke patients.

The remainder of the article is structured as follows. Section 2 provides a comprehensive discussion of the methodologies employed, including detailed justifications for the choices made regarding hardware, software, and game logic. The game implementation section outlines the implementation processes required for the development of the application, encompassing the database structure, techniques utilized, and the final outcomes related to the application and gameplay. Section 3 presents the results obtained from various tests conducted on the application. Lastly, Section 4 concludes the article by summarizing the work undertaken and proposing future tasks necessary for further improvement.

2. Materials and Methods

This section provides a comprehensive overview of the technologies utilized, ranging from hardware to software, and describes the decision-making process through the use of mockups, questionnaires, and diagrams. It further presents the project’s development from a technical standpoint, illustrating all stages of the game’s design and implementation. The study adopted an experimental design, integrating qualitative input from healthcare professionals with testing conducted by healthy participants in a controlled environment. Throughout the design process, iterative feedback from an additional healthcare professional was incorporated to enhance and refine the system.

2.1. Hardware

At the hardware level, the Kinect V2 sensor was utilized. This device is designed to capture detailed information regarding the movements of an individual’s body and can also detect multiple individuals simultaneously. Its operational range is between 1.5 m and 4 m from the subject, with 1.5 m being the optimal distance. For effective performance, the camera should be mounted at a height of approximately 1 m above the ground. It is essential to avoid direct illumination on the camera, as this may compromise its motion detection capabilities. Additionally, the camera should be operated in well-lit environments to ensure a clear and unobstructed view. Furthermore, the use of a screen will be essential to provide visual feedback to the player, accompanied by speakers to deliver auditory feedback. The architecture diagram illustrated in Figure 1 visually represents the integration of the previously mentioned equipment for the application’s operation.

2.2. Software

At the software level, multiple technologies were employed. Unity 3D [18] was utilized to develop the application’s interface and underlying logic. Unity allows the creation of 3D game scenarios with ease, utilizing a highly user-friendly interface that simplifies all forms of testing and debugging in an efficient, rapid, and visual manner. The programming language selected for this project is C Sharp (C#) due to its readability and ease of implementation [22]. To acquire avatars for the application, the Mixamo [23] website was selected, as it offers pre-existing human models that are highly compatible with Unity 3D. To enable the functionality of Kinect V2 on a Windows-based computer, the Kinect V2 Software Development Kit (SDK) (KinectSDK-v2.0_1409) was downloaded and installed from the Microsoft website. For other operating systems, such as Ubuntu or macOS, supplementary tools are required to achieve compatibility. Kinect V2 was selected for this work due to its seamless integration with the Unity 3D development platform. At the database level, SQLite was chosen for this initial version since data storage will be managed locally, as there is no need for communication with external servers to fully enjoy the game. SQLite uses a file-based data storage system; it is atomic, consistent, isolated, and durable (ACID) even in the event of system crashes and power failures. Additionally, it is lightweight, straightforward to use, does not require external dependencies, and supports cross-platform environments, such as Android, iOS, Unix, macOS, Solaris, and Windows [24]. It was utilized to store all necessary and relevant information for the game.

2.3. Statistical Analysis

This section presents information gathered through questionnaires administered to therapists, aimed at gaining deeper insights into how to best approach the development of the game to effectively achieve its intended objectives.

2.3.1. Therapist Statistical Questionnaire

For the statistical analysis, the focus was placed solely on the responses provided by five therapists (four females aged between 40 and 59 years old and one male aged between 18 and 39 years old). The questionnaire consisted of 15 questions (

Table 1. Therapist statistical questionnaire.

No.	Question
1	How would you assess your level of familiarity with using computers and video games?
2	To what extent are you open to integrating technology-based solutions, such as games and applications, into stroke rehabilitation therapy?
3	How interested would you be in utilizing a stroke rehabilitation application that incorporates virtual reality technology?
4	How interested would you be in the use of personalized exercise plans tailored to the individual needs and progress of each patient?
5	How interested would you be in the incorporation of gamification elements (e.g., points, rewards, and levels) to enhance patient engagement and motivation?
6	How interested would you be in the provision of feedback (e.g., success messages or sounds) based on the patient’s performance and progress?
7	How interested would you be in offering real-time feedback during exercises to guide proper form and technique?
8	How interested would you be in ensuring ease of use (user friendly) of the application for both patients and healthcare professionals?
9	How interested would you be in incorporating social interaction features, such as sharing progress with other participants and connecting with the broader stroke survivor community?
10	How interested would you be in accounting for the cultural and linguistic diversity of users in terms of content and interface design?
11	How interested would you be in including relaxation techniques to address patient stress and anxiety?
12	How interested would you be in providing progress reports and data tracking capabilities for both patients and healthcare professionals?
13	Do you believe that stroke rehabilitation applications should offer customizable settings to adjust difficulty levels and various abilities?
14	Is it possible for a patient who has experienced impairment in both arms to undergo post-stroke rehabilitation for both limbs?
15	What are the key features that the application or game should include to offer optimal support in upper limb stroke rehabilitation?

), 14 of which were multiple choice. Of these, 12 questions employed a Likert scale, with 1 representing the most negative response and 5 representing the most positive. Two additional questions were yes/no responses, and all 14 multiple choice questions included the option “No opinion.” The final question was open-ended to gather therapists’ views on the most important characteristics of the game.

2.3.2. Therapist Open-Ended Questionnaire

To obtain additional insights from therapists, a series of open-ended questions was developed, focusing on key issues pertinent to the project’s development. These questions were specifically tailored for therapists to provide relevant information. A response was received from a female therapist, offering diverse perspectives on various topics, which served to enhance the foundational understanding necessary for advancing the application’s implementation. Furthermore, before the application’s development, a series of mockups was created to provide a visual representation of the application’s overall functionality (the mockup images are shown in Appendix A Figure A1).

2.4. Game Implementation

This section provides a detailed explanation of the application’s development process and a comprehensive overview of all the features included in the game. Figure 2 contains a class diagram that offers a visual representation of the application’s overall structure and logic.

2.4.1. Database Tables

The SQLite implementation utilized was sourced from the GitHub repository SQLite4Unity3d, developed by Roberto Huertas [25]. The application database was organized into three tables: healthcare professional information, patient information, and game statistics:

• Healthcare Professional Table—Regarding personal information, the system stores the first and last names of the respective professional. Additionally, a username and password are retained to allow authentication. As a security measure for safety against dictionary or brute force attacks, the system also tracks the number of consecutive failed authentication attempts.
• Patient Table—Regarding personal information, the system stores first and last names, age, and gender. For game-related attributes, it encompasses the player’s position, preferred playing arm, difficulty level, time spent in the game, and username. Furthermore, specific variables are designated to record the lateral elevation limits for each patient and the date and time of their last login.
• Score Table—This table is connected to the patient table and stores the history and the highest score achieved at different levels.

2.4.2. Kinect V2 Information

Kinect V2 SDK includes pre-built code that, using its camera, detects human bodies by creating a representation in which joints are depicted as cubes, with connections made between them using green renderer lines to facilitate the recognition of a human body, as shown in Figure 3. This structure includes 25 points, spanning from the head to the feet. The process began by modifying the base code. First, the feature that allows the detection of multiple individuals simultaneously was disabled, ensuring that if the patient is detected, any other person, such as a therapist or someone passing behind the patient, will not be recognized.

Specific conditions were established to track and manage determined joints to perform key tasks:

• SpineBase—SpineBase data were used to determine the avatar’s position relative to the game environment. To maximize the game’s potential and ensure more accurate position tracking, the user should always remain centered with the camera and with a straight back.
• WristRight/WristLeft—These points provide the position of the user’s wrists at a given frame, which will inform the future positioning of the avatar’s arm.
• ElbowRight/ElbowLeft—These points give the position of the user’s elbows at a given frame, which is used to understand how the elbows should bend to ensure natural, humanly possible movements of the avatar’s arm.

2.4.3. Avatar

For this work, two avatars were selected (as shown in Figure 4), one female and one male, to replicate the patient’s movements while using the application. The avatars were selected from Mixamo due to their seamless integration with Unity and their suitability for accurately replicating movements.

Following the completion of the initial configurations to integrate Mixamo avatars with Unity 3D, a method was required to transfer the points captured by the Kinect V2 sensor to the avatars, enabling them to replicate the movements of the user. The approach employed to accomplish this was Inverse Kinematics (IK). IK in Unity refers to the computational technique used to determine the necessary joint rotations and positions within a hierarchical skeletal structure to achieve a specific target position for an end effector, such as a hand or foot. Unlike forward kinematics, where joint angles are directly manipulated to compute the position of the end effector, IK works in reverse, calculating the required joint configurations to reach a predetermined position. In Unity, IK is utilized primarily in character animation to ensure that parts of a character’s body, like limbs, can dynamically and naturally interact with the environment [26]. The Two Bone IK constraint enables the inversion of control within a simple hierarchy of two GameObjects, allowing the tip of a limb to reach a designated target position. An additional Hint GameObject can be used to define the desired orientation of the limb when it bends [27]. With all configurations applied, the avatar in the development scene environment will appear as shown in Figure 5.

The avatar’s movements are utilized at three distinct stages within the application. During gameplay, motion capture data from Kinect V2 are applied for generating brief animations to demonstrate specific movements, such as illustrating the proper technique for performing a lateral arm raise, as depicted in Figure 6, and for the movement of specific body parts to determine the type of gameplay (for instance, changing the leg position to simulate a standing or sitting position).

The data used by the code to modify the avatar’s movements are provided in the form of three-dimensional spatial coordinates (x, y, z). This design enables the system to accept motion data from cameras other than Kinect V2, facilitating the game’s operation. For integrating alternative input sources, it was enough to develop code that outputted three-dimensional spatial coordinates and supplied them as input to the avatar.

2.4.4. Unity Project

This section provides a concise overview of the functionality of all scenarios included in the application. Initially, an infinite sky image environment was created as the background to function as an initial scene, as depicted in Appendix A Figure A2.

Following the initial scenario, a page appears requesting the healthcare professional’s login credentials, as shown in Appendix A Figure A3. If the therapist enters an incorrect password three times for the same username, the account will be locked. This security measure was designed to protect the account from dictionary or brute force attacks. To be unlocked, it is necessary to communicate with the person responsible for managing the database. If the therapist does not have an account, a button is provided to navigate to the account creation process. In the scene illustrated in Appendix A Figure A4, specific fields, including first and last name, username, and password, must be accurately completed. Upon clicking the save changes button and completing all required fields, a confirmation panel will appear, allowing therapists to verify their decision to create the account. Once the account is successfully created, a confirmation message will be displayed indicating that the data have been saved successfully.

After the therapist has successfully logged in, a new panel appears to verify whether the patient already has an account in the application or if it is necessary to proceed with the account creation process, as illustrated in Appendix A Figure A5.

Upon selecting the option indicating that an account has not yet been created, the user is directed to an account creation menu presented in Appendix A Figure A6, with several required fields, such as first and last name, age, gender, injured arm, playing position (sitting or standing), username, difficulty level, game time and notes. All fields, except for the notes, are marked with a red asterisk to indicate that they are mandatory. Upon clicking “Save Changes,” the system validates the input fields for errors. If all fields are correctly completed, a confirmation message will appear, asking the user to verify whether they wish to save the data to the database.

The system then transitions to a scenario designed to assess the patient’s maximum lateral reach. In this scenario, an animation of an avatar performing a lateral raise is displayed on the screen, accompanied by a brief explanation of the expected movements and a “Ready” button, which prompts the patient to replicate the movements. The system updates the recorded position to the best value achieved by the patient. This value is then stored in the database and later used as input for the game, ensuring that certain objects are not placed in positions where the patient is unlikely to reach. For instance, if the patient has significant difficulty raising their arm, game objects will not be positioned at great heights, as they would be unreachable. As the patient progresses through subsequent sessions, the objects will gradually appear at more challenging, yet attainable, distances fostering realistic recovery and progression.

Upon selecting the option indicating that the user already has an account, the system retrieves all patients with existing accounts in the application. The resulting list of patients will be organized by the most recent login, ensuring that the most recently accessed records appear at the top, as can be seen in Appendix A Figure A7.

Upon selecting the patient, the system navigates to the main menu, as illustrated in Appendix A Figure A8, which presents the following options:

• Play: Initiates a level-based game session.
• Train: Opens a game session where the therapist configures the game’s characteristics.
• Statistics: Displays the user’s statistics based on his/her progress across levels.
• Profile: Provides access to edit all fields within the user’s profile.
• Delete Account: Deletes the selected patient’s account.
• Exit: Returns the user to the patient account selection screen.

This application currently features a single game, which involves the patient interacting with touch-sensitive objects. These objects are equipped with a capsule collider to enable the patient’s selected hand to collide with the object. Upon creation, the object is blue, and once the patient’s hand detects the touch, the object changes to green and then disappears. To prevent actions, such as arm crossing, that are not recommended for rehabilitation purposes, the capsule will award a point and disappear only when touched by the designated hand, chosen before the game, regardless of any attempts to use the opposite hand. A new capsule, identical in characteristics, subsequently appears in a different location. Depending on the level, the capsule may be subject to a time limit; after a predetermined number of seconds without the patient making contact, the object begins to blink, alternating between white and red, as shown in Figure 7, before disappearing and reappearing elsewhere. This feature is designed to enhance the level of difficulty and challenge, ensuring that the patient remains engaged and does not become disinterested due to the game being too easy. Upon successfully interacting with all of the objects for a given level, the patient receives a congratulatory message, and the system returns to the main menu. The game may also conclude if the healthcare professional selects the “End!” button displayed on the screen. Following each game, the system verifies whether the score achieved is the highest for the respective level. If it is, the database is updated to record the new best score. Furthermore, regardless of the outcome, the score and date of each game are stored in the database for future use in statistical analysis.

In training mode, the same game is used, but the therapist has the flexibility to select both the number of objects the patient must interact with and the time limit for each capsule’s disappearance. It can also select options such as game duration and preferred hand, as these settings are already available in standard gameplay mode.

In the statistical menu option, the values of points and their corresponding dates, stored within the score database, are utilized to generate a set of graphs (as exemplified in Figure 8). There are two types of graphs: level-specific graphs displaying the score, level, and date for each individual level; and general graphs presenting the points achieved along with their respective dates, regardless of the levels.

An illustrative video, submitted in the Supplementary Material, was created to showcase all the application’s features to facilitate a comprehensive understanding of the system.

3. Results

This section will commence by presenting a concise visual representation of motion capture, as observed through Kinect V2, within the game environment and in real-world scenarios. Subsequently, it will proceed to discuss the results of the tests conducted using this application.

3.1. Visual Representation

Figure 9 presents a series of images to illustrate the results of keypoint detection by Kinect V2, which are subsequently transformed into a skeleton model, and the avatar during gameplay and the same image in a real-world scenario.

3.2. Interface Test Results

For the interface test, the primary focus was on the person’s interactions with the application, specifically in relation to data entry, error validation, data monitoring, and overall ease of use. To assess these aspects, a brief questionnaire was developed and administered after the person tested the application, consisting of the following questions:

• Is the application user-friendly and intuitive to use?
• Did any errors occur while using the application?
• What recommendations do you have for improving the application in the future?

For this test, four healthy individuals (not therapists) participated: three women and one man, with ages ranging from 19 to 55 years. All participants (100%) found the application to be easy and intuitive to use, and no errors were encountered while interacting with the game interface. One participant recommended enhancing the connection between the avatar and the user for a more immersive experience by modifying the visual attributes of the avatar to more accurately reflect the unique characteristics of each user.

3.3. Therapist Statistical Questionnaire Results

This section presents an analysis of the results derived from the statistical questionnaires completed by the therapists, as outlined in the methodology section. Figure 10 presents the data from the 12 questions using the five-point scale. As observed, the majority of responses were rated 4 or higher, indicating that the topics addressed by these questions were perceived as important, with the exception of question 1, which assessed familiarity with computer games. Here, older respondents (ages 40–59) reported lower familiarity, which was to be expected. Question 9 yielded a mean score of 3.8, indicating some uncertainty about the importance of social interaction between participants. For the yes/no questions (questions 13 and 14), all responses were affirmative, except for one case where the respondent expressed no opinion. From the open-ended responses to Question 15, several key insights emerged. These included the importance of real-time feedback, the need for session result storage, the inclusion of varying difficulty levels, and the significance of user friendliness, particularly given the generally older age of the patient population.

3.4. Therapist Open-Ended Questions Results

This section provides an overview of the questions presented to the therapist and her corresponding responses, as outlined in the Section 2.

• What exercises and activities are recommended for upper limb rehabilitation based on the patient’s pathology? The intervention should be individualized and rooted in the clinical reasoning process tailored to each specific case. Physiotherapy procedures must be selected according to the identified compromised functions, with a focus on enhancing motor relearning through the incorporation of functional activities meaningful to the patient. In upper limb rehabilitation, reaching movements are central to most functional activities.
• How is the correct and safe execution of exercises ensured? During one-on-one sessions, the physiotherapist, as an expert in human movement analysis, identifies compensatory movements and provides feedback to the patient, primarily through somatosensory and proprioceptive cues, to ensure proper performance.
• What functionalities or tools are essential for an upper limb rehabilitation game to effectively support recovery? The game should facilitate motor relearning by allowing patients to safely practice movement patterns, minimizing compensatory strategies while providing appropriate challenges to promote progression.
• How can the balance between challenge and progression be maintained while minimizing the risk of frustration in a rehabilitation game? Offering a broad range of difficulty levels allows the challenge to be adjusted according to the patient’s abilities and progress, ensuring a positive experience.
• What strategies can ensure sustained engagement and interest in an upper limb rehabilitation game over an extended period? The game should balance the need for challenge and progression with the risk of frustration, providing a motivating yet manageable experience for the patient.

3.5. Gameplay Test Results

In this testing phase, it is important to consider the conditions under which the tests were conducted. All tests took place in a well-lit environment, with the person either seated or standing 1.5 m away from the Kinect V2 camera(Microsoft Corporation Porto, Portugal). The camera was positioned approximately 0.80 m above the ground at the edge of a table, ensuring that no objects obstructed its view, as shown in Figure 11.

This section primarily focuses on the gameplay experience and gathers comprehensive feedback from the participants. After completing the session, each individual was asked to answer certain questions represented in

Table 2. Questionnaire after game session.

No.	Question
1	On a scale from 0 (no familiarity) to 5 (a high level of familiarity), how familiar are you with using computer games?
2	Have you ever used a virtual reality (VR) game?
3	On a scale from 0 (not motivating) to 5 (very motivating), how motivating did you find the game?
4	Do you think the difficulty settings were appropriate for any person?
5	Did you encounter any errors/limitations while playing the game?
6	What recommendations do you have for improving the gameplay in the future?

.

A test was conducted with 18 healthy participants (not therapists), 15 males and 3 females, 13 of whom with ages between 20 and 26 and 5 of whom with ages between 27 and 35. The participants were seated 1.5 m from the Kinect V2 camera, which was positioned at a height of 0.80 m. Among the participants, the majority reported a high level of familiarity with computer games. On a scale of 0 to 5, twelve participants rated their familiarity as 5, two rated it as 4, one as 3, one as 2, and two as 1 (represented in Figure 12 by the blue values). Additionally, two-thirds of the participants had prior experience with virtual reality games (represented in Figure 13 by the green values). Approximately 77% of the participants rated the application as highly motivating, giving it a score of 4 or 5 (represented in Figure 12 by the orange values). The most critical feedback was related to the visual environment, which was perceived as lacking in appeal, rather than the gameplay. All participants agreed that the application offers a versatile range of difficulty levels, suitable for a wide variety of users (represented in Figure 13 by the blue values). In terms of issues, four participants noted that the capsule’s position was too far back, making it difficult to see clearly on the screen. Another reported limitation was that, on occasion, the object spawned in the hand’s zone, automatically awarding a point.

Several participants offered suggestions for improvement, including increasing the number of games available, as most enjoyed the current game but desired additional options; introducing a friendly competition feature or a mode for two users to play simultaneously; displaying the level below the date in the general statistics section; and beautifying the overall game design.

3.6. Limitations

The project presents some limitations. First, the requirement for the user to maintain a straight back poses a challenge, particularly for older individuals who may have spinal issues and are unable to keep their back completely straight. This limitation can negatively affect the avatar’s interaction with the capsules. Additionally, due to the avatar’s fixed back position, arm movements may appear unnatural when the arms are extended behind the body.

From a visual standpoint, the project lacks some aesthetic appeal, making it somewhat difficult to perceive the depth of the capsules, which hinders the ability to determine whether they are near or far. A more defined visual environment, incorporating shadows and varied camera angles, could alleviate this issue.

Another limitation is the lack of a variety of games, which could impact the overall effectiveness of the application. Offering multiple game options would cater to different user preferences and prevent the experience from becoming monotonous. This variety could also extend the duration of rehabilitation, as patients may feel more motivated by having the ability to select from different games and engage in diverse activities. The randomness in the creation of capsules presents a potential issue. Capsules may occasionally appear within the patient’s scoring area, resulting in automatic points being awarded without any movement. This undermines the reliability of the scoring system and statistics.

From a database perspective, the use of SQLite is perfectly suitable for this initial version. However, in the event of a larger data volume and the migration to a server database system, a minor transition to PostgreSQL or MySQL would be necessary to ensure continued efficient and seamless data handling. The project and its code are designed to facilitate a simple migration process, requiring only minimal changes.

Additionally, it would be essential to increase both the number of tests and participants to enhance the reliability of the findings.

4. Discussion

This project successfully integrated essential features into a rehabilitation game specifically designed for upper limb stroke recovery, showcasing its potential to enhance patient motivation. Increased motivation is a key factor in improving patient engagement in therapeutic exercises, which may significantly contribute to a more complete and efficient recovery process.

The game utilizes a humanoid avatar capable of replicating the user’s movements, detected solely through a camera, within an immersive and interactive environment. A points-based reward system encourages patient participation—by simply moving their arm to interact with on-screen objects, users can earn points, making the experience more engaging. The use of camera systems, such as Kinect V2, offers a cost-effective and practical alternative to VR headsets, particularly benefiting older patients who may find VR technology challenging. This approach still delivers an experience that closely resembles virtual reality without the complexity or cost associated with wearable devices.

One of the foundational principles of this project is its use of gamified rehabilitation to enhance patient motivation, as opposed to relying solely on traditional physical therapy methods [7,8,9,10]. The game leverages a Kinect camera, enabling patients to interact with the system using natural body movements without the need for physically attached devices that may cause discomfort, such as VR headsets [10,11,12,13] or body-worn sensors [15] or gloves [10]. This approach also reduces the burden on healthcare professionals, as it eliminates the need for cleaning shared equipment between patients [19]. The system features an automatic difficulty adjustment mechanism, supervised by healthcare professionals, to ensure that each patient engages with levels that are appropriately challenging. This prevents users from becoming unmotivated due to tasks being too easy or overly difficult. The involvement of clinicians helps ensure that the difficulty level does not spike unexpectedly, preserving the continuity of the rehabilitation process [13]. Furthermore, the game design emphasizes exclusively positive feedback, as negative feedback can be detrimental to user motivation, particularly in a rehabilitation context, such as sounds associated with unsuccessful attempts [13].

Despite promising results, the testing phase was limited to healthy individuals, with a small sample size providing initial feedback. Participants generally found the game motivating and well balanced in terms of challenge, although they noted that the aesthetic design could benefit from future improvements. Suggestions for enhancing user engagement include introducing a customizable avatar to reflect the patient’s physical characteristics, thereby increasing immersion. Incorporating personalized motion data (e.g., lateral arm lift, elbow flexion, and circular motions) upon account creation could allow therapists to better tailor exercises to individual needs.

The tests conducted with healthy participants, rather than directly involving post-stroke patients, played a crucial role in several respects. Firstly, they allowed for verification of usability, ensuring that the game is easy to understand and play, functions reliably from a technical standpoint, and is free of major issues. Secondly, these tests supported refinement of game mechanics, such as adjusting difficulty levels, identifying design flaws, and assessing whether the proposed tasks are meaningful and appropriate. Additionally, this stage served to validate the game’s concept before engaging patients, representing an important ethical and methodological step. It helped to prevent exposing individuals with impairments to a system that may not yet be fully suitable or prepared. It also demonstrated the game’s potential for real rehabilitation use. Early testing enabled the collection of initial feedback, such as identifying unclear elements, observing user motivation, and gauging engagement levels. Ultimately, this preliminary testing formed a foundation for more rigorous studies. It provided justification for subsequent research involving patients, where more solid hypotheses can be formulated for clinical trials. This approach helped to avoid exposing vulnerable patients to poorly designed systems and ensured the game can be effectively adapted to the motor and cognitive realities of stroke survivors. Moreover, it helped to confirm that the game maintains user motivation—an essential factor in successful rehabilitation.

Looking forward, transitioning from a static to a server-based database system could foster a sense of community among patients by enabling cross-institutional interaction, competition, and motivation. Furthermore, the development of additional games with more realistic or context-based scenarios—such as simulating picking apples—could further refine rehabilitation strategies by targeting specific muscle groups and movements.

5. Conclusions

Gamification represents a transformative approach to upper limb stroke rehabilitation by increasing motivation, improving engagement, and supporting a more enjoyable recovery process. The use of camera-based tracking instead of wearable devices provides a non-intrusive, cost-effective, and accessible alternative suited for a broad patient demographic, especially older individuals.

The system developed through this project demonstrates the potential of combining technology, motion capture, and interactive gameplay to replicate effective rehabilitation exercises in a virtual setting. The adaptability of the game to different camera systems and the inclusion of safety-focused design elements reflects its readiness for wider application. Nonetheless, ongoing development is required to refine the aesthetics, expand testing with actual stroke patients, and tailor therapeutic exercises based on more comprehensive motion data.

Preliminary testing with healthy participants, rather than post-stroke patients, served as a critical ethical and methodological step to assess usability, refine game mechanics, and validate the system before clinical application. This approach not only prevented exposing vulnerable individuals to an unrefined system but also provided essential feedback and a foundation for more robust, hypothesis-driven clinical trials. In the future, the proposed plan is to conduct testing with patients who have experienced a stroke.

As the technology matures, the integration of customizable avatars, competitive elements, multiplayer game modes, and additional games tailored to specific rehabilitation goals could significantly enhance both the effectiveness of the platform and social interaction among participants. The inclusion of a multidisciplinary rehabilitation team—comprising physicians, occupational therapists, and physiotherapists—could further improve the clinical evaluation of the system. Ultimately, this game has the potential to become a foundational tool in clinical rehabilitation programs, contributing to a future in which recovery is not only more effective but also more engaging and personalized for each patient.