Table 2.

Overview of the selected studies

References	Year	Description	Main findings
Friedman et al. [53]	2011	These studies focus on the development of MusicGlove [48–51, 53]. This device targets functional movements such as pincer grip, key-pinch grip, and finger-thumb opposition while the user is playing a video game. The mechanism involves electrical leads on fingertips registering connections during specific movements. The video game consists of a modified version of the open-source game FOF. Five distinct musical notes scroll from the top to the bottom of the screen, creating the sensation of playing the song through the user’s movements. The main objective is to hit as many notes as possible. The difficulty level of each song is tailored to accommodate various degrees of hand impairment, aligning with the individual’s skill level	Effectively measures hand dexterity (% of correctly hit notes), showing a strong correlation with the BB score Incorporation of music into training sessions leads to significant improvements in both objective measures of hand motor performance and motivation
Friedman et al. [50]	2014		Participants improved hand function, particularly in grasping small objects, compared to conventional methods (BB score and 9 Hole Peg test) Strong correlations were found with the BB score Reported as more motivating than the conventional therapy
Sanders et al. [48]	2020		Amount of practice was not correlated with the average level of success experienced, but it was correlated with the amount of parameter exploration (hours of use) and completed grips were comparable to individuals in the chronic phase of stroke in a previous study [53] Feasible for autonomous home use and caused no adverse effects
Sanders et al. [49]	2022		Participants demonstrated higher compliance levels (hours of use and completed more grips) compared to individuals in previous stroke studies utilizing the same device Notable improvements in prehension ability and performance (Graded and Redefined Assessment of Strength, Sensibility, and Prehension subtests) Increased performance on the BB test compared to the conventional group
Zondervan et al. [51]	2016		Participants exhibited significantly greater enhancements in Motor Activity Log quality of movement and amount of use MusicGlove and conventional exercise groups significantly improved Box and Blocks test scores with no notable difference
Adamovich et al. [54]	2009	This study introduces a robotic/virtual environment dedicated to improving hand and arm coordination by simulating a piano with realistic visual, auditory, and tactile feedback. Enabling users to train both arms and hands concurrently, an algorithm dynamically adjusts task difficulty based on individual performance. The system supports CyberGloves for precise hand tracking and a CyberGrasp for haptic effects. The virtual piano trainer features a complete keyboard, associating each key with a corresponding sound file. Users follow configurable key sequences for songs, guided by visual cues highlighting the current key and corresponding finger. The CyberGrasp is utilized to resist flexion in inactive fingers, contributing to a comprehensive and tailored rehabilitation approach	Subjects showed improvements in both performance time and key press accuracy Two subjects improved aggregate time on the Jebsen Test of Hand Function Three out of the four subjects showed improvement in Wolf Motor Function Test aggregate time
Tingzhang et al. [56]	2014	This study comprises a sensor-based glove and computer software, creating an interactive interface with a piano for training in a home environment. The glove has five bending sensors on the fingers and integrated sensors such as a gyroscope, magnetometer, and accelerometer for precise data collection on detailed finger movements and hand position parameters that are transmitted signals via WiFi and are processed by a classifier. Correct performance triggers audio feedback, producing piano sounds corresponding to the user’s movements, and visual feedback includes an image of a marked piano with comment messages	The system is minimally intrusive, entertaining users. It operates in real time, and the classifier accurately identifies the played keys
Sun et al. [61]	2017	This study presents a wearable hand movement rehabilitation system for stroke patients, using a data glove and a keyboard game supported with hand gesture recognition. A 3D hand animation model enables patients to observe their performance during rehabilitation. The system incorporates five bend sensors, orientation sensors with 9 degrees of freedom, and gyroscopes, accelerometers, and magnetometers to capture hand movements. During the keyboard game, a metronome plays 60 beats per second, prompting subjects to alternate between different keyboard gestures every 2 s	Experimental results show high precision in recognizing simple gestures and moderate precision in complex key press gestures. The system offers user-friendly operation, avoids invasive methods, and processes data quickly
English et al. [55]	2017	This study presents an adaptive therapy gaming system designed to monitor patients’ frequency, duration, and physical motions during at-home therapy sessions. It explores the potential acceleration of motor skill learning through prior knowledge of musical cues within a robotic wrist rehabilitation system. The system utilizes an exoskeleton with a potentiometer to capture the patient’s wrist’s full range of motion, transmitting data via Bluetooth to an interactive therapy game. The game (RoboRockNRoll) encourages accurate completion of wrist therapeutic motions through strategically designed musical cues, and it involves users controlling a pick to catch music notes corresponding to popular songs during the gaming session	Participants exhibit enhanced precision in their movements when music is introduced, and this heightened precision is sustained even after the removal of musical cues. The study highlights the potential of leveraging existing knowledge of a song to facilitate users in anticipating motions, thereby accelerating the learning of a motor task Participants can acquire timing tasks more rapidly with audio-visual cues than solely on visual cues The drawback of multisensory stimuli lies in the potential impracticality of audio-visual cues in specific contexts
Xiao et al. [52]	2018	This study introduces a wrist strap incorporating FMG technology, featuring an interface to play a virtual piano. The wrist strap comprises eight FSRs and one IMU. The FSRs extract pressure patterns applied by the musculotendinous complex against the strap, while the IMU monitors arm movement with a 3-axis accelerometer, 3-axis gyroscope, and 3-axis magnetometer. The system lets users play virtual piano keys by pressing a finger on a flat surface, generating desired sounds	An initial evaluation demonstrated the practicality of utilizing FMG for virtual musical instrument control, with the user achieving successful performance within a short timeframe and encountering minimal false predictions
Taheri et al. [57]	2014	These studies present the design and initial testing of FINGER [57–59]. This device utilizes stacked single degree-of-freedom 8-bar mechanisms, individually assisting the index and middle fingers through a natural grasping motion. The mechanism’s optimization was based on trajectory data collected from healthy subjects using color-based motion capture. The rehabilitation protocol involves engaging subjects in a Guitar Hero®-like a game, where they play along with a song by flexing their fingers to hit notes on a visual display, receiving performance-based assistance from the robot. The study aims to test the hypothesis that optimal engagement in rehabilitation therapy occurs when subjects operate at their challenge level, achieved by controlling success rates	FINGER exhibited its capability to enable individuals with various impairment levels to engage in the game successfully FINGER, coupled with a gain-adaptation algorithm, confirmed the hypothesis that subjects could be assisted as necessary to achieve predetermined success levels in the game
Taheri et al. [58]	2012		The study observed a decrease in effort for both high and low-level subjects as their success rates increased, aligning with previous observations of user slacking when robotic assistance is excessive [57] While the study did not identify an optimal level of effort, it acknowledged the potential relationship between effort measures and the optimal challenge point, a direction for future research
Rowe et al. [59]	2017		Participants exhibited substantial improvements in functional and impairment-based motor outcomes, depression scores, and self-efficacy of hand function Notably, higher assistance levels correlated with increased motivation and secondary motor outcome enhancements, especially among individuals with more pronounced finger motor deficits
Thielbar et al. [60]	2014	Researchers developed an actuated virtual keypad (AVK) system, combining a custom actuated glove called the PneuGlove with a virtual scene consisting of a hand and five keys. This system aims to promote independence in finger movements and allows adjustments in task difficulty based on user capabilities. The PneuGlove controls air pressure to extend or prevent flexion of specific digits using air chambers, while real-time hand posture updates occur according to measured joint angles. The AVK system offers two training modes: Key Combination, which helps participants practice discrete key combinations, and Song Mode, where participants similarly play songs from the Guitar Hero video game	The AVK treatment demonstrated superiority over the intensive occupational therapy treatment for measures of ARAT and JTHFT. Additionally, changes in ARAT scores for the AVK group approached the Minimal Clinically Important Difference (MCID) of 5.7
Merians et al. [62]	2002	The study introduces a PC-based rehabilitation system that incorporates virtual reality simulation exercises. It employs the CyberGlove for free hand movement during virtual reality exercises and the RMII force feedback glove for force-exertion exercises and finger strengthening. The RMII glove, an exoskeleton device, uses non-contact sensors such as Hall-Effect and infrared to measure finger positioning and flexion. The virtual exercises are displayed on a flat screen without special 3D head-mounted displays, utilizing shadows and perspective cues for depth. The simulations involve four exercises targeting aspects of hand movement such as range, speed, fractionation, and strength. The fractionation exercise includes a piano keyboard. Before the exercises, patients’ hand movement parameters are assessed to set an initial difficulty level	The Jebsen Test of Hand Function and the Fugl-Meyer Assessment assessed each patient’s hand function improvement. Two out of three patients showed progress on the Jebsen Test, and objective measurements indicated that all patients experienced improvement in most hand parameters during the training
Mawase et al. [63]	2020	This study presents a device that measures isometric forces from each finger using a hand-shaped keyboard with ten keys, each featuring FSRs at the fingertip positions. Participants, seated comfortably facing a monitor, rest their hands on the keyboard with wrists strapped and supported by foam armrests. Only finger muscles activate the isometric forces, as forearm, arm, or trunk movements are not detectable. The device assesses finger strength and individuation through tasks that involve pressing specific keys with instructed fingers to match target force levels while maintaining low forces in non-instructed fingers	The study results indicated that finger impairment decreased following training, as assessed by the individuation task. This improvement was associated with better clinical hand function, including precision pinch performance. The training enhanced the trained task and positively impacted overall finger dexterity and movement quality, as evidenced by Fugl-Meyer and Motor Activity Log measures

FOF Frets on Fire, BB Box and Block, FMG force myography, FSR force-sensing resistors, IMU inertial measurement unit FINGER Finger Individuating Grasp Exercise Robot, FSR force-sensing resistors, RMII Rutgers Master II ARAT Action Research Arm Test JTHFT Jebsen-Taylor Hand Function Test