Virtual Reality for Pulmonary Rehabilitation: Comprehensive Review

Introduction

Background

Pulmonary rehabilitation focuses on breathing exercises that can help people with chronic lung diseases improve lung function and reduce symptoms of chest tightness, chronic cough, and wheezing. Apart from lung diseases, pulmonary rehabilitation is commonly used for treatments related to hypertension, chronic pain, and cardiovascular disorders, such as coronary artery issues, arrhythmia, and myocardial infarction [1,2]. Breathing exercises may also offer effective and simple solutions for depressive and anxiety episodes, and other mental health–related disorders [3,4].

In recent years, virtual reality (VR) technology has been used in a wide variety of medical applications, including but not limited to areas involving the delivery of treatments for pulmonary diseases, such as asthma [5,6], chronic obstructive pulmonary disease (COPD), [7], and lung cancer [7], and mental health conditions, such as anxiety and stress-related disorders [5,8]. This is because exploration around physiological signals collected in VR can offer holistic breathing guidance options to users and provide them with breathing benefits [8-10]. Nowadays, individuals may explore and seek the assistance of professional coaches or use advanced devices that are specially designed for the purpose of breathing; however, these means often have a high cost and are time-consuming [5,7]. VR technology, on the other hand, is becoming one of the most accessible and low-cost solutions in the health care domain for breathing interventions [8,11]. Further, VR allows users to have full control over the environment they are exposed to. In combination with the use of biosensing technology, which provides acoustic, visual, and biofeedback guidance, VR users are offered the ability to consciously and self-effectively control and monitor their respiratory rate (RR) [2,5,12].

COVID-19 and Pulmonary Rehabilitation

The novel coronavirus SARS-CoV-2 was identified in late 2019. A couple of months later, the World Health Organization (WHO) declared COVID-19 as a pandemic, as it affected 412,351,279 people (5,821,004 deaths) worldwide (February 13, 2022) [13]. The clinical symptoms in patients with COVID-19 included high fever, sore throat, cough, exhaustion, and dyspnea [13,14].

During the COVID-19 pandemic, patients’ medical care, including admission to clinics and use of emergency services, was affected owing to the danger of contamination and the limitations of medical service resources [14]. In this situation, clinical visits, nonurgent treatments, and nonearnest clinical issues, especially among vulnerable populations like people with pulmonary diseases, were initially interrupted and later resumed with a diminished scope [15]. 

Doctors had to confront the quandary of who could be treated at clinical centers or at home, or who could be allocated to the set number of beds in intensive care units [16]. New technologies helped support vulnerable populations during the pandemic, and VR helped overcome a variety of clinical challenges. This technology is quickly changing clinical training, patient therapies, and rehabilitation [14,15,17]. The pandemic has changed the clinical framework and placed standard methods with virtual telemedicine and software systems to provide clinical benefits for alleviating the effects of COVID-19 [18]. A significant part of the clinical framework involves rehabilitation, and it is significant owing to the pandemic period [19]. This might be driven by telehealth stages, as with the use of VR.

The fundamental objective of pulmonary rehabilitation is to further develop the patient’s psychophysical state [19]. Regardless of restricted admittance to hospitals for rehabilitation owing to COVID-19, VR technology can be applied to this group of patients. It can provide extensive help in various areas, including patient management and clinical treatment, monitoring of patient progression in rehabilitation or assessment of changes, and evaluation and advancement of body function, exercise, and consecutive participation [16,19]. Pulmonary rehabilitation is also beneficial after COVID-19 infection, even in patients who are recovering, who need assisted ventilation or oxygen therapy [20]. Additionally, COVID-19 survivors experience stress, depression, and low quality of life, and the symptoms of dyspnea and fatigue can last more than 3 months after infection. Recent evidence has shown that pulmonary rehabilitation can alleviate these symptoms and can improve exercise performance, lung function, and quality of life in COVID-19 patients and survivors [21-23].

However, scientific studies examining and evaluating the opportunities and challenges of VR for breathing remain limited. In this literature review, we introduce state-of-the-art VR technologies relevant to pulmonary rehabilitation and breathing exercises by analyzing recent related articles on the subject. Owing to the lack of studies related to breathing gaming exercises, for this review, we only examined and evaluated 32 available studies from the last decade [5-8,10,24-50]. Evidence from related empirical and experimental studies that comprised several types of breathing exercises and patterns was systematically reviewed to address the following research questions:

• Is VR an effective solution for breathing exercises?
• Which are the most common VR contents used for breathing exercises in gaming?
• How feasible is gamified biofeedback breathing VR for real-world deployment?
• What are the current barriers to biofeedback VR technologies?
• What are the future directions of biofeedback VR technologies?

Methods

Design

This review was conducted according to Bargas-Avila and Hornbæk [51] and the Cochrane methodology [52,53], which involved 5 phases. The phases are described below.

Procedure

Phase 1: Detailed Assessment of Publications

Electronic libraries: The research was conducted with the use of 7 electronic libraries, which cover a balanced choice of multidisciplinary sources. The libraries were as follows: (1) ACM Digital Library (ACM), (2) Google Scholar, (3) IEEE Xplore (IEEE), (4) MEDLINE, (5) PubMed, (6) Sage, and (7) ScienceDirect (SD). The search was delimited to a timeframe of 13 years (2010 to 2023).

Search terms: The following 3 queries were used in all the libraries since the aim was to cover any type of VR technology for breathing:

1. Virtual Reality AND Breath
2. Virtual Reality AND Pulmonary Rehabilitation
3. Virtual Reality AND Breathing games

Search procedure: The search terms were used to examine the publication’s title, abstract, and keywords.

Search results: The search results in Phase 1 can be seen in Table 1.

Phase 2: Publications Retrieved for Detailed Evaluation

First exclusion: All search results from Phase 1 were imported into Mendeley electronic library. Possible entries with wrong years were excluded (625 wrong-year entries were removed). This elimination decreased the number of papers to 1694.

Second exclusion: Duplicate papers, either extracted from one or more libraries, which either produced or concluded the same outcome, were removed from this review. Moreover, duplicate papers, extracted from one or more similar terms, which either produced or concluded the same outcome, were also removed. A total of 375 duplicate publications were removed, leaving 1319 different papers.

Third exclusion: The entries were narrowed down to original full papers that were written in English. We excluded papers that did not have the full text available (ie, we did not have access to the full text) and papers that were not original full papers, such as workshops, posters, speeches, reviews, magazine articles, and generally grey literature without formal peer review. As a result, 312 papers were excluded, leaving 1007 papers.

Phase 3: Final Exclusion

Since this review was focused on VR technologies related to breath and breathing exercise gaming, we excluded papers that examined and used other types of technologies that were not related to VR or to breathing exercise gaming. Moreover, research that was only related to breathing without the use of VR was excluded. In this phase and based on these conditions, any irrelevant studies that appeared in Phase 1 and were not removed in Phase 2 were excluded. However, these studies may appear in our findings since relevant words were contained in our research but did not match the specific technical content. Based on these conditions, we removed 376 studies unrelated to VR, 215 studies unrelated to breathing, and 284 studies unrelated to breathing exercise gaming. We ended up with 32 relevant papers (24 journal articles and 8 conference papers; there were no book chapters). The flowchart is presented in Figure 1. All relevant studies were downloaded for examination.

‎

Phase 4: Data Gathering

All related information from the studies was extracted for examination. An Excel file was created, and the following data were extracted from each paper: sample size of the population studied, methodology, instruments, apparatus, VR content, VR interventions, types of biofeedback sensors, types of biosignals, VR feasibility, key findings, current VR limitations, and VR future directions. Moreover, we categorized each study based on the results as positive (+), negative (−), or neutral.

Phase 5: Data Analysis

The data gathered in Phase 4 were analyzed using descriptive statistics. Consequently, the literature was reviewed to support and enhance the additional knowledge provided by the study. Thematic analysis was used as an additional methodology to classify our findings based on themes. The following themes were considered: VR training, VR interaction, types of virtual environments (VEs), VR effectiveness, VR feasibility, VR design strategies, VR limitations, and VR future directions. Intercoder reliability was assessed between the researcher and research assistant. The Cohen kappa formula was used to calculate the similarity between the researcher and research assistant, and the similarity value was 0.89.

Results

Objectives, Study Design, and Interventions

The search identified 32 studies related to the effective use of VR for breathing exercises. The sample, study design, objective, and intervention of each study are presented in Table 2. Of the 32 studies, 24 involved healthy individuals [6,8,10,24,26,28-38,40,41,43-48], 1 involved individuals at high risk of developing anxiety disorders [39], 1 involved patients with lung cancer [7], 1 involved patients with pneumonia [49], 2 involved patients with COPD [27,42], 2 involved patients with COVID-19 [49,50], and 1 involved patients with asthma [5]. The breathing exercises were mostly related to breathing therapy (9/32) [5,6,10,25,27,33,48,49,54] and anxiety management (13/32) [8,28,29,32,34,36,37,39-41,43-45]. Most of the reviewed papers based their research on pilot (16/32) or control studies (11/32), while some of the papers (4/32) only described the design and development processes of their systems. Most of the systems (24/32) were designed based on breathing patterns and techniques related to the needs of the populations. For example, for participants who had lung cancer, the VR environments were created for them to perform normal breathing exercises (ie, 10-20 breaths per min) [7], as opposed to the environments for those who had pneumonia or COVID-19 [49], and the VR environments were created to navigate the participants’ breathing into a positive expiratory pressure technique (ie, involving breathing with an expiratory resistance that allows air to flow freely when inhaling but harder due to resistance when exhaling) [8]. A different example was provided in a study that included participants who had COPD, and the VR system offered them an avatar assistant, who presented educational content and physical exercises [27].

Various studies involving the treatment of anxiety disorders decided to choose healthy participants for the needs of the experimentations. In particular, among 15 studies, healthy individuals tried several VEs to perform slow (ie, 4-10 breaths per min) [8,10,25,26,41,44,45], normal [7,8,31,32,48], and diaphragmatic breathing exercises (ie, inhale through the nose moving the air toward the lower belly and exhale through the mouth) [39,40,46]. Moreover, healthy individuals participated in studies where VEs were created to guide them on breathing patterns like respiratory sinus arrhythmia (ie, synchronization between heart rate variability [HRV] and RR) [6] and pursed-lip breathing (ie, inhale through the nose with the mouth closed and exhale through tightly pressed lips) [38]. Finally, few studies examined the effectiveness of VR in children for treating asthma by practicing breathing exercises with the use of spirometry [5] and anxiety by practicing diaphragmatic breathing [40].

Instruments

Self-Reported Data

As mentioned above, most of the studies enhanced conventional breathing training through VR interventions. To evaluate the effectiveness of the VR system, studies used several measures, which are presented in Table 3. All the reviewed studies (32/32) collected participants’ demographic information (eg, age, sex, ethnicity, educational level, etc). Half of the reviewed studies (14/32) examined the effect of VR breathing training on mental health (ie, anxiety) [6,25,26,28-30,32,34,36,37,39-41].

To assess the level of anxiety, the following protocols were used: State-Trait Anxiety Inventory (STAI; 3/32 studies) [25,40,44], Generalized Anxiety Disorder-7 (GAD-7; 2/32 studies) [27,49], Hospital Anxiety and Depression Scale (HADS; 2/32 studies) [42,50], Perceived Stress Scale (PSS; 2/32 studies) [28,39], and Edmonton symptom rating scale (1/32 studies) [50]. The STAI is a 4-point Likert scale form, which contains 40 questions measuring pressure, worry, and anxiety [55,56]. GAD-7 is a 7-question screener that assesses participants’ psychological well-being status [27,57]. The HADS is a 14-item scale that scores 0 to 3 for each item. The first 7 items relate to anxiety, and the remaining 7 relate to depression. A higher score is associated with greater anxiety and symptoms of depression [42]. The PSS is a 14-item scale with 7 positive and 7 negative items rated on a 5-point Likert scale [58]. The Edmonton symptom rating scale is a questionnaire used to rate the intensity of 9 common symptoms experienced by cancer patients, including pain, tiredness, nausea, depression, anxiety, drowsiness, appetite, well-being, and shortness of breath [59]. Moreover, the Cognitive Interference Questionnaire (CIQ) was used to measure the performance, task-oriented worries, and off-task thoughts of participants through a 22-item questionnaire based on a 5-point scale [5,60]. The Mini-Mental State Examination (MMSE; 2/32 studies) is a set of 11 questions to check if participants have cognitive impairments such as problems with thinking, communication, understanding, and memory [61]. Lastly, the visual analog scale (VAS; 2/32 studies) [5,25] was used to evaluate participants’ calmness and relaxation self-efficacy. The VAS is usually used for measuring pain on a 10-point Likert scale, but since the scale is easy to reflect, studies also use it to measure other emotional reflections as well [62].

Some studies (9/32) examined the VR experience [6,10,31,33,38,42,44-46] through self-reported questionnaires. The questionnaires included questions related to the participants’ positive and negative emotions, game flow, engagement, ability, capacity, pressure that the said user was experiencing, and challenges that the user perceived [62,63]. Two studies (2/32) also evaluated the motivation of the participants, using a multidimensional self-reported 7-point scale with a Likert-type format called the Intrinsic Motivation Inventory (IMI) [40,44]. The IMI consists of 7 subscales that measure participants’ experiences related to a target activity, such as interest/enjoyment, perceived competence, effort, value/usefulness, pressure/tension, perceived choice, and relatedness [64]. The Just Noticeable Difference (JND) was used in a study (1/32) to stimulate the perception level of the user on system latency [46]. Another study (1/32) used the Igroup Presence Questionnaire (IPQ) to assess the sense of presence in different VEs [48]. Lastly, an instrument named the State Mindfulness Scale (SMS; 1/32) [25] was designed for mindfulness assessment with 21 questions self-reported on a 7-point Likert scale. In particular, the SMS counts the level of present-moment attention to and awareness of activity (mindful or mental) [65].

Biosignals and Physiological Data

Biosignals are physical signals that describe the state of human living. A wide variety of biosignals are regularly used in hospitals and in home monitoring. The most well-known include electrocardiography (ECG), electroencephalography (EEG), and photoplethysmography (PPG). Alternatively, physiological data consist of heart rate (HR), blood pressure, RR, etc.

The studies analyzed the above signals and physiological data in detail. In particular, several studies extracted ECG signals to collect HR data to calculate the average beats per minute [10,25,28,31,37,38,49,50] and HRV to measure the specific changes in time between heart beats [6,29,31,44]. Moreover, a few studies preferred the use of the PPG signal to collect HR [27], blood pressure [50], and oxygen saturation [28,49,50]. Respiration signals were collected by most of the reviewed studies, and they assessed data like lung function [5,36,42], lung volume [8,28], expansion [6,40], and breathing force [48]. Further, respiration signals were used to calculate physiological data, such as RR [8,27,28,30,32,33,35,38-40,45,49,50], and exhalation and inhalation durations [8,39], while accelerometer data were used to identify participants’ chest movements [40]. Moreover, EEG data were collected by researchers to reflect participants’ brain activity [29,38]. Lastly, electrodermal signals were used by a few studies to record the electric characteristics of the skin and allow researchers to assess participants’ stress levels [37,38].

Apparatus

VR Technology

VR allows its users to have full control over the environment they are exposed to. VR technology can provide acoustic, visual, tactile, or olfactory interactions between the user and the system. Based on the above abilities, VR systems can be categorized as nonimmersive, semi-immersive, and fully immersive [66]. Our review found that 21 of the 32 evaluated studies [6,10,25-29,31-33,36,37,39,41,43,45-49] used fully immersive equipment, where the participant’s vision is fully enveloped with a head-mounted display (HMD) system and the interactions with the system are based on natural gesture recognition processes. However, 5 of the 32 studies used nonimmersive VR equipment, such as a smartphone [29,44], tablet [5], or laptop [8,34], and these 3D graphical systems allowed users to navigate VEs. Lastly, 6 of the 32 studies did not report the type of VR equipment used [7,8,24,33,39,40].

Biofeedback Equipment

All the reviewed studies used biosignal responses to assess the accuracy of the delivered solutions. In particular, the reviewed biosignals were measurements of the physiological changes in respiration, PPG, ECG, EEG, and electrodermal activity. Most of the studies (27/32) used already existing systems to record biosignals [5-8,10,24-34,36-41,44,45,47,49,50]. The search results for biofeedback equipment are presented in Table 4. Impressively, only 4 of the 32 studies developed their own systems [35,43,46,48]. Heng et al [43] developed a breathing input sensor consisting of Arduino Uno, a Rev C. Wind Sensor chip, and an ESP8266 ESP-01S Wi-Fi module. The Arduino Uno board acts as the main processor for the breathing input sensor and the power supplier for the whole system. The wind sensor chip picks up the wind signal from human breath and translates it into raw reading data. After that, the Arduino Uno board translates the raw reading data into wind speed. Finally, ESP-01S communicates with the system software through a WebSocket protocol. Moreover, Hu et al [35] used the same Arduino Uno board as mentioned above but with a different sensor. A gas pressure sensor (XGZP6857A) was applied for respiratory measurements. Tao et al [46] established their own system hardware based on an Adafruit Feather M0 Bluefruit LE board. This board has 2 main features: processing of audio signals with an ATSAMD21G18 ARM Cortex M0 processor and Bluetooth Low Energy communication with an nRF51822 chipset. A MEMS wired microphone was used for the board analog input to read the signal (Figure 2). The system could take breath as input for instrument playing in the VR game. The microphone detected exhalation, which was mapped to the instrument audio in the VR game.

‎

Tatzgern et al [48] created the AirRes mask to measure participants’ breathing according to VE interactions (Figure 3). The system depends on a Sensirion SFM3300 mass flow sensor 2. This sensor allows the assessment of a participant’s airflow with high-precision measurements from both directions (exhalation and inhalation). A protective gear 3M mask frame was used, and the custom-made equipment (electronic circuit) and breathing sensor were added on the frame. Moreover, a disk was applied to the mask to change the amount of air. A Servo Motor SG-90 device was used to rotate and control this disk. A custom circuit board was connected to the airflow sensor and servomotor. The above custom-made system controlled resistance wirelessly in the VR game through an ESP32 Bluetooth module.

‎

VEs and Interventions

As mentioned above, one of the most important advantages of VR is its ability to provide the feeling of being immersed in a simulated environment [66]. Even though the participants of VR technology acknowledge and recognize that the environment provided by VR is not real, they act as they would in a real environment. The reviewed studies focused on specific types of VEs for mental and anxiety rehabilitation and pulmonary rehabilitation. The study results for VEs are presented in Table 5.

Types of VEs for Mental and Anxiety Rehabilitation

Numerous studies (12/32) used nature scenes, such as scenes of beaches, forests, oceans, and mountains [10,25,26,28,29,33,34,36-40]. Below, we present 4 of the most impressive natural VEs. The first environment was a beach scenery at sunset with several dynamic parameters like lights and clouds, which shifted according to breathing to provide feedback to the participant (Figure 4) [25]. If the participant’s breathing was below the threshold, which was preset by the system, the breathing pace was considered to be correct. As a result of correct breathing, the sky turned clear, and the participant could enjoy a star-spangled sky. A campfire was also included in the said scenario as a dynamic object to make the participant feel more relaxed and to provide the participant with an indication of whether breathing was at the appropriate pace.

‎

The second environment was a dynamic scenario with a tree submerged in the middle of water (Figure 5) [34]. The goal of this intervention was to help the participant to practice pursed-lip breathing. The participant was told to wear an unidentified HMD and sit in a comfortable position with the legs crossed. Before the start of the intervention, the participant was advised to go through a breathing exercise introduction. After that, the participant had to follow rhythmic breathing by inhaling and exhaling until the tree started to expand on inhalation and contract on exhalation. If the participant continued to breathe rhythmically, the tree started to bloom. If the breathing of the participant was nonrhythmic, the view of the participant started to blur in the monitor until the right rhythm was found. Leaves were also used to provide feedback to the participant on the breathing rhythm. More colorful leaves indicated that the participant was following the correct breathing rhythm.

‎

The third environment was the ZenVR environment, which was a dedicated VE that included an open room in a mountain environment with plants, where the participant could train on different meditation techniques (Figure 6) [47]. The VE contained several objects related to meditation like meditation cushions and candles. Several dynamic parameters, such as the weather and a bonsai tree, changed according to the level of the breath training. During the duration of the training, the bonsai tree grew to indicate the participant’s progress. An hourglass was present as a marker of time for the participant and to indicate that the program continued during the silent meditation. Additionally, numbers appeared with different sizes to illustrate the breathing exercises.

‎

The last environment was an underwater experience for diaphragmatic breathing practiced with children with anxiety issues (Figure 7) [40]. The participant was instantaneously informed of the state of breathing by a dynamic circle in the VE that expanded according to breathing. The system applied gravity to the participant’s avatar if the lung capacity was more than half. The participant’s breathing pace was able to determine the direction and magnitude of the force. When the participant inhaled, an upward force was applied, and when the participant exhaled, an extra forward force was applied so that the participant was able to dive into the deep ocean. The combination of slow and deep breathing allowed the participant to swim better and have more control in the game. The type of VR equipment was not reported in the study.

‎

Types of VEs for Pulmonary Rehabilitation

Several studies (10/32) developed different VEs for breathing exercises to improve respiration [5,6,8,29,31,32,34,41,47,49]. In particular, Prpa et al [26] generated a 3D element of a body of water (an ocean) (Figure 8). This minimal environment displayed the ocean and the sky through a variety of grayscale shades. A continuous breathing pattern allowed the participant to control the ocean environment. It started with a light grey sky and a stationary participant position above the ocean surface. When the participant found the breathing flow, the game continued with movements of ocean waves, which were based on the participant’s breathing pace. The right breathing pattern was mapped to wave movement and musical rhythm. The sky changed from grey to black until the ocean was stationary again, as was seen by the participant at the beginning of the procedure, with the only difference being the color of the sky.

Another study developed a VR intervention for alleviating dyspnea in patients recovering from COVID-19 pneumonia (Figure 9) [49]. Specifically, it created a room with a matched gender body lying on a couch. The goal was to illuminate the virtual body synchronously or asynchronously according to the patient’s chest movements.

‎

‎

A particularly interesting study used multi-user VR to simultaneously immerse two or more participants into an underwater world with jellyfish and a growing glass sponge (Figure 10) [33]. The aim was to synchronize the breathing between the participants, and enhance the breathing awareness, breathing pace, and relation between the participants. In this environment, each participant’s breath was represented by a jellyfish, which moved and glowed in such a way as to provide clear breathing feedback. As the participants synchronized their breathing, the glass sponge was structured to begin to grow and emit light.

‎

Interestingly, 2 of the reviewed studies used existing predeveloped systems for pulmonary rehabilitation [42,43]. Rutkowski et al [42] used the virtual therapeutic garden game, released by the European Association of Psychotherapy (Figure 11). Initially, the garden appears grey (untidy and unkept), and the watering pot is on its side. Diaphragmatic exercises with resistance, prolonged exhalation exercises, and chest percussion activate the watering pot to water the garden. With each rehabilitation session and the correct conduct of fitness exercises, the garden becomes increasingly colorful and alive, symbolizing the process of gaining health through the rehabilitation sessions.

‎

Bubble Tower [43] investigated the development of mapping breathing techniques in VR gameplay mechanics (Figure 12). The whole platform offered a set of 8 mini-games, where Bubble Tower was designed to train the participant in the fundamentals of breathing techniques. The mini-games teach the participant basic breathing skills like long and strong breathing, and breathing strength control. These mini-games include stages where the participant is required to pop the bubble with a strong breath, blow candles in a room with a long breath, build up the momentum for a windmill to spin fast with a long breath, etc.

‎

An interesting study presented a VE based on the escape room game philosophy (Figure 13) [48]. In the escape room scenario, the participant used different breathing pattern interactions to answer a sequence of tasks to escape from the room. The room had all the objects that the participant needed to escape. During the training, the participant interacted with different individual objects. The first task was to blow out candles on a cake, and then, the participant had to blow bullets through a blow tube toward a target. The third task included the movement of ships based on breathing force. The participant continued with inflating balloons and sorting them by resistance. After that, the participant had to reveal numbers on a mirror with the breath. The last task involved shooting with a toy gun by holding the breath.

‎

Discussion

Principal Findings

Based on all the studies that were reviewed (29 studies had a positive label and 3 had a neutral label), it is suggested that VR can be an effective solution for pulmonary rehabilitation among patients with lung cancer, patients with COPD, patients with asthma, and individuals and children who are dealing with mental health–related disorders such as anxiety. Overall, the results indicated that VR can enhance the functional outcomes of pulmonary rehabilitation, increase breathing body awareness, and improve relaxation techniques. In the COVID-19 crisis, evidence showed the need for VR technology adoption. VR pulmonary rehabilitation presents an opportunity for the safe and effective recovery of COVID-19 patients and survivors at home. This technology could be adopted on a large scale to further develop the well-being of individuals during pandemics like COVID-19 and could similarly advance autonomous medical care. In the reviewed studies, most of the VR systems included features of natural environments, like a beach, sky, and forest. Further, it is highly recommended for future studies to incorporate water elements (eg, bubbles) and undersea sceneries (eg, seabed, jellyfish, and sea plants) to enhance relaxation. Some systems involved the use of music and vibrant colors. As technology continues to advance and progress, it is expected that biofeedback in VR systems will have a vital role in the practice of breathing in both the medical setting and the real world. The reviewed studies proved the feasibility of implementing a VR system for pulmonary rehabilitation enhanced with biofeedback. Such a system can be a reliable solution to enhance participant training. An assortment of minimal-cost sensors and biofeedback frameworks can incorporate VR and provide exact and significant information from tasks in gamified biofeedback interventions for breathing. Regarding the adequacy of pulmonary rehabilitation, biofeedback VR innovations must bridge explicit obstacles like equipment assembly, participant population, experiment length, and breathing patterns. The boundless breathing direction in VR frameworks can cause issues in the legitimate decisions of breath-detection equipment. Accordingly, many specialists battle with equipment assembly, equipment incorporation, and its compatibility with VR systems. There have been hindrances in adopting accepted procedures of biofeedback VR in various populations requiring breathing training for different purposes, including overcoming mental health issues and stress pressure, and achieving overall health benefits. The restricted length of examinations can prevent the assessment of the longer-term impacts of pulmonary rehabilitation.

Most of the studies (14/32) examined the effectiveness of breathing through VR based on physiological data [6-8,24,25,29,35,36,43-45,47-49]. Overall, the review suggests that VR is a reliable and feasible solution for pulmonary rehabilitation. Specifically, one of the reviewed studies [8] revealed that VR rehabilitation can be a reliable solution to treat pneumonia. Moreover, it was found that participants who performed pulmonary rehabilitation through VR felt more confident compared to those who performed the training with face-to-face supervision from health care professionals [25]. It was explained that embodied interactions through VR and biofeedback responses made the participants more aware of their inhalation and exhalation rhythm [28]. Additionally, the reviewed studies about pulmonary rehabilitation in COVID-19 patients reported significant improvements in tiredness, shortness and comfort of breath, and vital signs, such as HR, RR, blood pressure, and SpO2 [48,49]. A study comparing patients with COPD undergoing VR pulmonary rehabilitation and those not undergoing this rehabilitation found that VR rehabilitation was associated with high stress release and a sharp reduction in depressive episodes [41]. The positive impact of VR on the enhancement of rehabilitation with different breathing patterns in emotional well-being has been documented by several studies (13/32), with most of these studies being focused on anxiety and stress monitoring [5,8,10,27,28,31,33,34,36,37,39-41,43,49]. The level of anxiety of participants was found to decrease within few minutes after using VR with biofeedback for controlling breathing. In particular, a study comparing induced anxiety between participants who used VR and those who did not use VR reported a reduction in the level of anxiety in those who used VR [39]. In addition, a study documented that apart from the positive effect VR has on anxiety, it can also increase concentration and positively motivate participants [54]. In particular, it was found that participants who performed pulmonary rehabilitation via VR were less reluctant to participate in the training activities. Anxiety levels have also been studied in young populations [39] and COVID-19 patients [49]. VR was suggested to be an effective solution for children at risk for anxiety disorders [39] and was reported to be effective at improving anxiety and increasing the feeling of well-being in patients with COVID-19 [49].

A study reported that participants were able to perform paced breathing techniques without distraction [27]. It suggested the use of vibrant, rich, and multi-dimensional VEs to deliver an effective and enjoyable VR experience. Another study mentioned that water manifestation is a key element to decrease anxiety. More specifically, it was reported that having water features can enhance stress management techniques and significantly expand HRV based on paced breathing [27,54]. Two studies suggested that participants were able to more accurately control the pace of their breathing on adding biofeedback [28,30]. Zafar et al [30] noticed that participants who were exposed to biofeedback systems were able to control their breathing more precisely as opposed to the traditional type of training. Correspondingly, participants of VR biofeedback systems scored higher in their subsequent stress test compared with the pretest. Furthermore, a study that examined the effectiveness of respiratory biofeedback during VR meditation by measuring EEG and ECG signals in a respiratory biofeedback state, control feedback stress state, and control no feedback state, proved that VR meditation is effective for relaxation and breathing exercises. The study findings suggest that if VR is used for meditation, no biofeedback equipment is needed to reduce arousal, providing a more affordable and less intrusive option to apply VR to relaxation exercises [28]. In summary, biofeedback is recommended for the effective deployment of VR systems for pulmonary rehabilitation. The feasibility and findings of the studies are presented in Table 6.

Limitations and Future Work

Even though the effectiveness of VR for pulmonary rehabilitation is well documented, several limitations have been reported in the reviewed studies. First, some of the reviewed studies measured the relaxing effect of VR through the use of psychophysiological responses, such as HR, and self-reported questionnaires. However, some of the studies suggested the need for additional instruments that explicitly assess different aspects of affect and mood [10,41]. For example, a study suggested the collection of 2 different biosignals (PPG and ECG) for HR accuracy [10]. Moreover, the study mentioned that a participant’s respiration has limited validation and recommended the use of an additional belt sensor [10]. It was suggested to measure stress levels using not only traditional reports, such as self-reports and the Stroop task, but also cortisol levels [41]. Future studies should consider triangulating the physiological data with interviews and other qualitative data to provide a holistic assessment of the impact of biofeedback systems [29,44]. Future studies can also investigate the factors that can enhance VR pulmonary rehabilitation to support existing relaxation and destress techniques, and this should be compared to traditional practices [27].

Second, the reviewed studies highlighted the limit of breathing guidance in VR systems, which can provoke complications and barriers in the correct choice of breathing sensing hardware [45,47]. As a result, many researchers struggled with aspects of the hardware apparatus, like the weight of the hardware on the user’s head in addition to the VR HMD [47], as well as with hardware integration and its compatibility with VR systems [45]. In addition, most hardware equipment involved high-end solutions, and the cost for the equipment in most of the studies was between €300 and €1200 (US $322 and US $1289, respectively) [5,6,10,27,29-33,36-39,45]. Three studies [34,46,47] developed their own affordable equipment, for which the cost was approximately €70 (US $75). Future studies should provide clear guidelines for the effective apparatus as well as the cost of the system. As mentioned previously [66], moving to low-cost and accessible solutions will decrease the need for technical support. This suggests that participants will be able to have their own personalized devices, which could lead to an increased quality of life.

It was further suggested for future studies to build systems that include machine learning algorithms to empower participant rehabilitation. These kinds of systems can offer the advantage of higher relaxation levels. For this, it is recommended for model algorithms to automatically classify breathing states, compared with other sensors, and analyze different visual cues in VEs. Moreover, it was stated that machine learning methods should adapt logic modules to provide automatic adaptations in VEs [31]. It is worth mentioning that Fast Fourier Transformation implementation presents a limitation compared with other methods of low-frequency detection, although it is fully functional. For future studies, it is therefore suggested to implement a wavelet transform method, which can provide higher resolution at low frequencies without requiring a larger window size [33].

To improve the systems even further, future studies should examine the effectiveness of biofeedback compared with proper control conditions in different groups of participants under different types of circumstances to determine exactly when and why biofeedback might not be preferable. For example, the effects of biofeedback in children, as examined in the study by van Rooij et al [40], could differ from the effects of biofeedback in adults, as examined by Tinga et al [29]. In the study by Tinga et al [29], the reduction in arousal (on all outcome measures combined and HR specifically) was the largest in the control feedback placebo condition, indicating that respiratory biofeedback had no additional value in reducing arousal and was even less effective than the control feedback placebo. The above finding indicates no preference for respiratory biofeedback compared with control feedback placebo in lowering pain levels in participants with chronic back pain.

Third, a study suggested the extension of VR exposure time, since it was found that this might allow participants to develop their own strategies for producing respiration patterns [31]. Expectedly, most studies suggested that a large sample is required to verify the trend of average HRV and other bioindicators [30,43], as well as to address difficulties in VR design [8,45]. An enhanced sample size could have a positive impact on VR design since a wider set of participants can express their interests [48,49]. Finally, it has been suggested for future studies to extend experiments to multi-session investigations to examine the longer-term effects of pulmonary rehabilitation among participants [25,30,33,49].

Conclusion

The future directions of biofeedback VR technologies hold huge potential for significant advancements in the pulmonary field, ushering in a new era of personalized and adaptive experiences, enhanced sensor technologies, integration with artificial intelligence and machine learning, gamification and immersive exercises with integration into telehealth, and remote monitoring.

One of the most exciting prospects of VR is the ability to deliver personalized and adaptive experiences through biofeedback VR technologies. VR applications can dynamically tailor experiences to patients based on their specific needs. Real-time integration of user responses and physiological data enables these applications to optimize the effectiveness of biofeedback interventions, ensuring that patients receive the most relevant and impactful feedback. Advancements in sensor technologies are another crucial area of development. Wearable sensors, such as biometric devices, provide real-time data on BR, HR, skin conductance, muscle tension, etc. Continued improvements in sensor miniaturization, wireless connectivity, and comfort will enhance the usability and reliability of biofeedback VR technologies, making them more accessible and user friendly.

The integration of artificial intelligence and machine learning presents exciting possibilities for biofeedback VR technologies. These algorithms can analyze massive amounts of biofeedback data, identify patterns, and provide personalized recommendations for stress reduction, relaxation breathing techniques, or breath performance enhancement. Machine learning can also help in monitoring progress, evaluating outcomes, and optimizing the effectiveness of biofeedback interventions. The cooperation between artificial intelligence, machine learning, and biofeedback VR technologies has the potential to unlock new levels of personalized and evidence-based interventions. Gamification and immersive experiences play crucial roles in engaging users and maximizing the benefits of biofeedback VR technologies. By incorporating game elements and designing interactive VEs, biofeedback VR applications can provide engaging and motivating experiences.

The reviewed studies showed that VR technology can be applied in various areas in the health field, such as stress management, anxiety disorders, pain management, phobia treatment, and rehabilitation. It is necessary to establish evidence-based practices and guidelines for the use of biofeedback VR technologies in health care through collaboration among researchers, health care professionals, and developers. The integration of biofeedback VR technologies in clinical settings will revolutionize the way doctors and researchers approach treatment, improving accessibility, reducing health care costs, and enhancing patient engagement and outcomes.

In the future, innovations in biofeedback VR technologies may also include developments in neurofeedback and brain-computer interfaces with the brain activity of patients. As technology continues to grow and VR progress is investigated, these future directions are poised to shape the field, leading to transformative applications and advancements in wellness, health care, education, etc. The continued exploration and integration of VR technologies and biofeedback technologies have the potential to revolutionize how we understand and enhance human performance, well-being, and quality of life in the future.