Table 2.

Technological advancements in neglect rehabilitation. The table briefly summarizes (a) sample size and description; (b) features of the VR device employed; (c) the conditions and tasks of the study; (d) main outcomes

VR AND TECHNOLOG* FOR NEGLECT REHABILITATION
References	Characteristics of sample	Characteristic of VR applications	Sessions	Main outcomes
Kim et al., 2015	Fourteen patients with hemispatial neglect (n = 14, mean age = 73.1 ± 5.8; nine males, five females)	An amiraglos SX® (Deocom) head-mounted display (HMD); a liquid crystal display (LCD) Screen. OKS and test management software were made with Visual C +  + (version 6.0) and Direct X (version 6.0)	Patients performed the line bisection task under several conditions: (i) screen—OKS: patients observed a stationary horizontal red line presented on the LCD screen; (ii) screen + OKS: the red line is presented with background blue OKS moving leftward; (iii) HMD—OKS; (iv) HMD + OKS. These latter two conditions employed the same paradigm as the former two, within an HMD	OKS projected onto a screen overcorrected the hemispatial neglect and outperformed HMD, whereas the OKS + HMD was more effective in decreasing patients' rightward deviation. Leftward HMD + OKS provided a better correction of patients' rightward deviation toward the midline and seemingly distracted patients to a lesser degree while performing the task
Faria et al., 2016	Eighteen USN patients randomly assigned to the experimental group (n = 9, aged 48 to 71 years) or the control group (n = 9, aged 50 to 65 years)	Reh@City, a VR-based simulation of a city, implemented using the Unity 3D game engine; desktop computer running Windows 7, and an arcade joystick (Topway’s Digiusb Joystick)	The experimental group completes increasingly difficult tasks within familiar places (i.e., a post office, a bank, a pharmacy, a supermarket). The cues were gradually removed and reintroduced as soon as they failed to respond correctly. The control group performed a traditional cognitive rehabilitation	The experimental group improved in terms of global functioning, attention, memory and visuospatial abilities, executive functions, social participation, emotion, and in the physical domain. The control group reported a significant worsening in verbal fluency and an improvement in attention and processing speed
Fordell et al., 2016	Clinical trial comparing pre-post-intervention measures of spatial attention in people with chronic neglect due to an ischemic right-sided infarction. 15 participants (mean age = 72.8 ± 5.7 years; 11 males, 4 females)	VR method RehAtt™. 27″ monitor and 3D Nvidia vision glasses. 3D pen and a haptic force feedback interface “Robotic pen” (Phantom omni haptic device) was used as a pointer by the paretic hand. Software developed with the open-source platform Colloseum 3D	5-weeks baseline followed by a 5-weeks training (3 h weekly, 15 h total) with RehAtt™ setup. Follow-up was collected within a week and after 25 weeks. Patients performed one neglect test battery (VR-DiSTRO™); (i) a mental rotation task; (ii) a visuomotor exploring task; (iii) a visuospatial and scanning task	At follow-up, neglect was significantly improved in tests and in spatial attention in activities of daily living. The effect lasted after a 6-month follow-up. The 5-week training with RehAtt™ improved the spatial attention of patients with chronic neglect, transferring the improvements in activities of daily living as well
Glize et al., 2017	Two groups were considered: right-brain-damaged patients with visual neglect (n = 7; mean age = 65.5 ± 4.1 years; six males, one female); healthy controls (n = 10; mean age = 63.3 ± 5.9 years; six males, four females)	Virtual supermarket VAP-S. Participants had to navigate within the supermarket and shop, as quickly as possible, a list of 8 items, 4 placed on the right side and 4 on the left side. The list was randomized on each session. Each session lasted a maximum of 45 min. The PA procedure exposed patients to a rightward optical shift of 10°, produced by the prismatic lenses, by means of glacier goggles (total visual field = 110°; monocular eye field = 80°). The PA session lasted a maximum of 10–20 min, each patient performing 10 sessions over a 2-week period, with at least 200 pointing movements each session. Patients completed the Subjective Straight-Ahead (SSA), pointing straight-ahead with the right arm while the body was aligned with the sagittal axis	Two days of VR training, followed by two days pre-test, patients could familiarize with the virtual environment. The degree of assistance during the performance decreased gradually. Generalization of PA effects to the visuospatial domain, spatial representation, and topographic memory was tested asking subjects to draw the VAP-S map from memory. During the prism exposure, patients had to point visual targets placed on the left or right side of the body midline	PA reduced the rightward attentional bias in a VR task, is possibly associated with an expansion of its effects to a supra-modal representation of space. PA enhanced both navigation and topographic memory, showing a positive effect on access to semantic information. These improvements persist after a 1-month follow-up
Tobler-Ammann et al., 2017a	VSN patients (n = 7, aged 64–78, two females, five males)	A 21-inch computer screen; a height-adjustable chin rest (Novavision GmbH); a haptic Falcon Novint device (Novint Technologies) to control games, providing sensory feedback; 9 increasingly difficult exergames, simulating real-life tasks (e.g., cooking a recipe, completing a puzzle). For the follow-up measurement, the Eye Tracker Neglect Test (ETNT), Zürich Maxi Mental Status Inventory (ZüMAX), and Neglect Test (NET) were administered	Five 30- to 45-min sessions per week, over a 3-week period. The exergames intensity was individually increased by the supervising therapist and patients selected 3 to 4 exergames to perform in each session, following individual preference and were able to change the exergame selected whether they felt bored or preferred another one. A follow-up was conducted after a 4-week break in order to test exergames’ reversibility	The primary outcome showed that neglect exergames are a safe and feasible complementary intervention for VSN patients, with a lack of adverse events and attrition and a 95% median adherence rate. For the secondary outcome (the limited efficacy testing), patients showed a group trend improvement in both cognitive and spatial exploration skills, measured by ETNT, ZüMAX, and NET
Tobler-Ammann et al., 2017b	Two groups of users testing exergames’ usability: VSN patients (n = 7, mean age = 68.6 ± 8.9; five males, two females) and expert group of therapists (n = 12; mean age = 33.3 ± 5.7; gender unspecified)	A 21-inch computer screen; a height-adjustable chin rest (Novavision GmbH); a haptic Falcon Novint device (Novint Technologies) to control games, providing sensory feedback; 9 increasingly difficult exergames, simulating real-life tasks (e.g., cooking a recipe, completing a puzzle)	Both patients (as end-users) and therapists (as experts) completed the 3-week, 30-min exergames training sessions. The intervention included a total of 15 training sessions. After the training completion, patients and therapists completed a Technology Acceptance Model (TAM)-based questionnaire; patients were individually interviewed, and therapists attended two focus group interviews	Patients generally rated exergames as motivating and interesting, despite their initial positive attitude decreasing over time, and games were perceived as boring, childish, or exhausting, also due to the sitting position. Some patients could not understand the purpose of the exergame and preferred conventional therapy. Therapists generally provided lower rates to the use of exergames and more negative attitudes toward them. Therapists had reservations on the therapeutic potential of exergames and reported no intention to use them in the future
Yasuda et al. 2017	Ten USN patients after an ischemic or hemorrhagic stroke, aged between 45 and 85 years. No control group was used	Head-Mounted Displayed (HMD; Oculus Rift Development Kit 2); a motion-tracking device (Leap Motion) and a personal computer running a Unity 5 software. The VR room consisted of a desk and a virtual screen placed in front of the subject. Patients performed the task with their right hand, using a pen in the near-space condition and a pointer in the far space condition	A 30-min session included attentional assessments and two VR training conditions: far space (patients had to orally identify visual stimuli, flashed for 6 s from the right to the left side of the screen); near-space training (patients had to move a VR hand and touch three objects). Attentional assessments included four tasks of the Behavioral Inattention Test (BIT): line cancellation and bisection, letter and star cancellation tasks. The near and far space neglect was assessed as well: the near-space assessment required patients to perform a task using a pen to respond to stimuli presented on a A4 sheet placed at 40 cm. The far space assessment projected the stimuli on a wall and patients were asked to answer using a laser pointer	The far space evaluation showed that total BIT scores improved further after training; three of the four subscales (line cancellation, star cancellation and letter cancellation) improved significantly following the VR rehabilitation and after the session as well, whereas the scores of the line bisection task showed no statistically significant difference. For the near-space evaluation, instead, neither the differences in total BIT scores nor the four subscales were statistically significant. The VR-based rehabilitation appears promising for the rehabilitation of far space neglect in USN patients, but more evidence and a control group are needed for the near-space condition
Yasuda et al., 2018	One male patient (age = 76 years) with left SN after a right middle cerebral artery infarct	Head-Mounted Displayed (HMD; Oculus Rift Development Kit 2); a motion-tracking device (Leap Motion) and a personal computer. Head and finger movements are tracked by both HMD and tracking devices	One 30-min session, once a day, five times a week, over a 6-week training period. Baseline data were obtained 6 weeks pre-intervention and symptoms were assessed pre- and post-intervention. The far space training (10 min) includes a visual search task in the VR space, orally identifying the flashing objects. The near-space training (10 min) required to touch each object presented in the VR space from right to left, followed by a moving slit	The patient showed a major improvement in omission rates for both the line cancellation task in far and near space. An improvement was found in shifting the midpoint in near and far space. No changes were observed in CBS scores, referred to daily life activities
Ekman et al., 2018	Twelve patients with chronic neglect due to right-sided infarction (mean age = 72.7 ± 6.0 years; four females, eight males)	VR method RehAtt™. 27″ monitor and 3D Nvidia vision glasses. 3D pen and a haptic force feedback interface “Robotic pen” (Phantom omni haptic device) was used as a pointer by the paretic hand. Software developed with the open-source platform Colloseum 3D. fMRI data were collected on a 3-T Discovery MR750 Scanner. Patients responded at the Posner cueing task using a response grip (NordicNeuroLab). The behavioral performance at Posner cueing task was recorded with an in-house program in Python (Python Software Foundation)	Two 30-min sessions with a 15-min break, including 5 min of listening to music before audio-spatial training. Patients performed 3 training sessions each week, 5 weeks total. Patients also underwent fMRI scanning 1 week before and 1 week after the VR training and performed a Posner cueing task during the scanning. Patients were presented with two empty boxes at the left and right sides of a red fixation cross, which turned green when the task changed. Patients then had to follow a central arrow cue shifting attention from left to right side of the visual screen (top-down processing) and, later, press a button whenever they targeted a target flashed either on the right or left side of their visual field (bottom-up processing)	RehAtt® training and the Posner cueing task combine an intensive top-down and bottom-up multisensory stimulation implemented within a VR environment. They also show an increased BOLD signal in cortical regions beyond the ventral (VAN) and dorsal (DAN) attentional networks, such as DLPFC, ACC, and bilateral temporal cortex. RehAtt® training is thus capable of inducing neuronal changes in patients with chronic spatial neglect
De Luca et al., 2019	A 57-year-old woman affected by USN following a subarachnoidal hemorrhage in the right frontal–temporal-parietal region	BTS NIRVANA is a movement-based VR semi-immersive system. The standard BTS NIRVANA equipment includes one or two marker-less infrared sensors, a touch-screen workstation, the camera supports, the BTS NIRVANA software, and Webcam. A projector is also connected to a big screen displaying a series of interacting exercises	BTS NIRVANA required the patient to complete two rehabilitation training: standard cognitive training (SCT) was either combined with a semi-immersive virtual training with the patient’s shadow (S-IVT_s) or without the patient’s shadow (S-IVT). Each training combination (SCT + S-IVT_s and SCT + S-IVT) lasted for one month, with twenty 45-min rehabilitation sessions, fine times a week. The motor performance was also of interest and evaluated with the Trunk Control Test. The SCT + S-IVT_s condition included both ideo-motor and attentional tasks while watching her shadow performing them. In the SCT + S-IVT condition, the patient completed similar tasks without seeing her shadow performing them	Following VR-based rehabilitation with the BTS NIRVANA system, the patient displayed improved motor performance, controlling trunk movements, and better cognitive performances. Integrating the S-IVT treatment to the SCT allowed an improvement in attention, scanning, visual search, and spatial cognition
Wåhlin et al., 2019	Patients with visuospatial neglect (n = 13; mean age = 73 ± 6 years; four females, nine males)	The hardware setup included a PC (EMS Shuttle P4), video and sound, a monitor, 3D vision glasses (Nvidia), a robotic pen (Phantom Omni), and a numeric keyboard. The software employed for RehAtt™ used the open-source platform Colloseum3D VRlab	Patients underwent two fMRI scanning sessions, one week before the intervention and one week after the end of the intervention. Within fMRI, patients performed a scanner-adapted Posner task. Symptoms were assessed three times at baseline and at follow-up after the training. The intervention included a 5-weeks training (3 h weekly, 15 h total) with RehAtt™ setup (Fordell et al., 2016)	The intense scanning VR training increased DAN inter-hemispheric functional connectivity in patients affected by chronic visuospatial neglect: this suggests that chronic conditions as well could benefit from training that showed their positive effects for recovery from acute states of neglect. Moreover, the training increased the integration of the frontal eye field (FEF), controlling the saccadic eye movements to the left side of the space, with the left posterior parietal cortex
Choi et al., 2021	Twenty-four post-stroke patients randomized into two groups: a Digital Practice group (n = 12, mean age = 63.00 ± 10.02); a control group (n = 12, mean age = 61.58 ± 9.99)	The virtual environment was created by means of the Oculus Rift Developer Kit 2, with the Oculus Rift 1.3.2 Software Development Kit (SDK) and Windows Runtime 0.8.0-beta. Patients wore Oculus Rift DK2 and used the Leap Motion controller	The experimental group underwent twelve 30-min sessions of digital practice, whereas the control group underwent twelve 30-min sessions of conventional USN rehabilitation	Preliminary results show that only the experimental group reported greater recovery of cognitive and visual perception, as well as self-awareness of neglect. Digital practice could help patients increase their degree of arousal and attention, rotating their heat to a greater degree, with positive effects of attention and arousal on the contralesional side
Cogné et al. 2020	Forty-eight individuals divided into three groups: with unilateral auditory and visual neglect post-stroke (n = 22; mean age = 65.8 ± 8.8 years; 17 males, 5 females); without visual/auditory neglect post-stroke (n = 14; mean age = 63.9 ± 15.5 years; 9 males, 5 females); healthy controls (n = 12; mean age = 67.6 ± 10.0 years; 7 males, 5 females)	A 3D virtual environment reproducing a medium-sized North American town, displayed on a laptop. Participants were able to navigate the town by means of a joystick and a Tobii Pro TX300 eye tracker that detected eye movement. An earphone provided auditory cues in the navigational task	Randomized exposure to either one of the three conditions: (i) “without auditory cues”; (ii) “with auditory cues”; (iii) “auditory cues after prism adaptation”. The prism adaptation procedure itself involved three steps: (i) pre-exposure baseline measurement of pointing; (ii) exposure to prismatic displacement; (iii) post-exposure after effect measurement. A secondary task required a free recall and recognition of landmarks among a pictorial list including distractors. A tertiary analysis considered eye saccades and eye-fixation duration	Primary outcome: auditory cues had a positive effect on spatial navigation abilities in patients with visual and auditory neglect and were even more helpful after a single prism adaptation exposure. Secondary outcome: auditory cues decreased spatial memory abilities, which were compensated following the prism adaptation procedure. Tertiary outcome: the eye-tracking device showed increased duration of the eye fixation, following prism adaptation procedure
Huygelier et al., 2020	Phase 2 and 3 of the study recruited seven stroke patients aged 44 to 69 years	The Oculus Rift CV1 headset with integrated headphones and infrared sensors. The Oculus Touch Controller was used to provide responses; the VR game was developed in Unity 3D	The game world consisted of three scenes (lake, garden, and forest) presented in three lighting conditions (day, evening, and night), for a total of 9 possible combinations. Half of the trials presented a cue that predicted the location of the target. Then, the target was presented for 3 s, and patients pressed the button that corresponded to the target they saw while receiving visual and auditory feedback	The rehabilitation game is a promising tool for detecting visual neglect and improving patients’ performance in orienting their attention to the neglected side