Exploring customised virtual environments in patients with cognitive decline and responsive behaviours: protocol for a proof-of-concept and feasibility study in a long-term care facility (iEMBRACE)

Abstract

Introduction

The global rise in the population aged over 65 has led to a corresponding increase in cognitive impairment diagnoses, with dementia as a predominant condition characterised by diverse aetiopathogenetic profiles. Behavioural and psychological symptoms of dementia (BPSD) encompass a range of psychiatric, behavioural and cognitive symptoms associated with cognitive impairment. BPSD significantly affects patients, caregivers and healthcare providers, often necessitating interventions like sedatives or physical restraints that may worsen patient outcomes. Emerging evidence underscores the need for innovative, non-pharmacological interventions to manage BPSD effectively.

The current study intends to investigate the feasibility and acceptability of customised, immersive virtual reality environments (iVRe) to reduce responsive behaviours among individuals with dementia. Building on prior findings demonstrating virtual reality (VR) potential in reducing anxiety and fostering positive emotional states, this pilot study assesses the feasibility, safety and user engagement of customised iVRe interventions.

Methods and analysis

A longitudinal, mixed-methods design will be employed, enrolling 20 elderly participants with varying levels of cognitive impairment from the APSP ‘Margherita Grazioli’ long-term care facility in Trento. Participants undergo three VR exposure sessions featuring tailored environments adjusted in real-time for visual and auditory preferences. Data collection integrates standardised self-report questionnaires, observational tools and clinical records. Measures include the Functional Assessment Staging Tool, Neuropsychiatric Inventory and Cohen-Mansfield Agitation Inventory, as well as tools assessing pain, anxiety and emotional states before, during and after VR sessions.

Ethics and dissemination

The study protocol has been approved by the Comitato Etico Territoriale della Provincia Autonoma di Trento per le Sperimentazioni Cliniche, Azienda Provinciale per i Servizi Sanitari—Trento, Italy (Rep. Int. 12090, 15 May 2025). All participants or their legal representatives will provide written informed consent prior to enrolment. Deidentified data will be securely stored on institutional servers at the Fondazione Bruno Kessler and the APSP ‘Margherita Grazioli’, curated in compliance with the General Data Protection Regulation, and retained for 3 years after study completion. Any data shared externally will be provided in fully anonymised form, and only for scientific purposes, subject to prior ethical and legal approval. Study findings will be disseminated through peer-reviewed publications, conference presentations and executive summaries shared with participating institutions and stakeholders.

Trial registration number

NCT06693193.

STRENGTHS AND LIMITATIONS OF THIS STUDY

• The study evaluates a personalised virtual reality intervention for individuals with cognitive impairment and responsive behaviours in a real-world long-term care setting.

• The longitudinal mixed-methods design allows integration of quantitative and qualitative measures to capture user experience comprehensively.

• Data will be collected through validated self-report and observational tools, complemented by caregiver input, to enhance methodological rigour.

• Recruiting participants from a single long-term care facility may limit generalisability to other populations and settings.

• The reliance on proxy reporting for some outcomes may introduce subjective bias in interpreting emotional and behavioural responses.

Introduction

In response to the increasing demographic of individuals aged over 65, the incidence of cognitive impairment diagnoses, for which ageing represents the principal risk factor, is on the rise.¹ Within the diagnostic context of dementia—a condition associated with diverse and distinguishable aetiopathogenetic profiles—the International Psychogeriatrics Association (IPA)² has introduced a category encompassing behavioural and cognitive symptoms closely linked to cognitive impairment diagnoses. This category is labelled behavioural and psychological symptoms of dementia (BPSD).

Generally, BPSD includes psychiatric, behavioural and cognitive symptoms that may emerge at any stage of cognitive impairment. Identifying BPSD is crucial, as these manifestations contribute to the decline in cognitive and functional abilities, often leading to pharmacological interventions (eg, neuroleptic or sedative administration) or physical restraints (eg, alarms, locks, Buxton chairs), potentially worsening patient prognosis. BPSD also significantly impacts the psychological well-being and functionality of caregivers and healthcare providers assisting these individuals.³,⁷

BPSD encompasses ‘reactive’ behaviours, which may include physical actions (eg, throwing objects, biting, pushing) or verbal expressions (eg, swearing, shouting).^{8 9} These behaviours follow stable patterns throughout the day in people with dementia (PwD), often manifesting at specific times (eg, sundowning syndrome) or being triggered by particular treatments or activities (eg, wound care, hygiene procedures). Given the limitations and potential side effects of traditional interventions for managing BPSD, there is a need for innovative approaches that may engage and distract PwD during BPSD episodes and reduce reliance on pharmacological treatments.^{10 11}

Some studies suggest exploring virtual reality (VR) as a management tool for activating behaviours, alongside its use for recreational and therapeutic purposes across multiple sessions.¹²,¹⁹ Immersive virtual reality environments (iVRe) have demonstrated the potential to reduce isolation and enhance engagement in recreational activities aimed at well-being across various populations. VR devices expose users to diverse sensory stimulation and can promote multiple levels of immersion within VR contexts.²⁰ Immersive VR environments enable interactive experiences, minimising interference from stimuli external to the virtual context.²¹

To date, few studies have examined the feasibility and efficacy of iVRe in managing anxiety among elderly individuals with physical and cognitive impairments.^{22 23} Research indicates that VR headsets may aid in alleviating depressive and anxious symptoms in older adults, with recent studies investigating the feasibility of relaxing VRe in promoting positive emotions, engagement and overall well-being while managing negative emotions.^{22 23} Findings suggest that exposure to realistic, natural, immersive scenarios via VR headsets is safe for individuals with cognitive and physical impairments and can promote relaxation and positive emotions. In this regard, Walden and Feliciano²⁴ examined the use of virtual nature scenes to reduce agitation in individuals with dementia through a single-case design. Although their intervention did not employ fully customisable VR, participants were exposed to calming nature environments that promoted relaxation and behavioural stability. This study provides preliminary evidence supporting the potential of immersive VR for managing responsive behaviours in dementia and strengthens the rationale for exploring personalised VR approaches in the present study. Future studies are recommended to explore the effects of customising VR content based on individual preferences within the virtual environment. Personalisation in VR has gained attention in other research contexts,^{25 26} where exposure to relaxing, customisable virtual environments, including personal or meaningful material, has been shown to facilitate relaxation, positively influence mood regulation and manage activation and anxiety.

A personalised approach allows users to experience environments that align with their preferences, enhancing presence and safety within the VR context. Specifically, a customisable VR environment entails identifying conditions likely to be pleasant, relaxing and suitable for the individual’s needs. A recent study by Appel et al³ aimed to assess VR’s efficacy in managing responsive behaviours characterised as verbal or non-verbal reactions to aversive, frustrating or incomprehensible stimuli or events encountered by individuals with dementia. While some responsive behaviours may be adaptive strategies to communicate discomfort or unmet needs, the current study focuses specifically on maladaptive behaviours that manifest as signs of distress or agitation, such as repetitive vocalisations, yelling, restlessness or physical aggression. Clarifying this distinction is essential to target behaviours that are disruptive to the individual’s well-being and to the care environment, while recognising that these reactions can also reflect an underlying attempt to cope with stress or confusion. This study incorporated two experimental procedures: one focused on studying the impact of VR exposure on ‘responsive’ behaviours, while the other involved VR sessions within recreational activities for individuals with dementia. The findings highlighted positive outcomes regarding acceptability, feasibility and potential reduction in reactive behaviours, alongside increased relaxation and enjoyment during the activity. Staff organisational and planning challenges for sessions were noted. However, these did not hinder the positive effects of VR in terms of relaxation and reactive behaviour management, mainly when VR was explicitly applied to reduce these reactions. Reminiscence episodes were more common during multiple sessions conducted as recreational activities. To address these challenges, Appel et al³ recommend thorough planning with healthcare staff and incorporating training as a preliminary step before experimentation. Another recommendation involves evaluating VR exposure at multiple time points outside VR sessions.

Future studies should explore the optimal number of VR sessions to determine intervention efficacy. Additionally, assessing VR’s impact on reactive behaviours caused by various factors (eg, physical pain, emotional discomfort) is essential. A varied selection of VRe reflecting users’ interests may enhance VR’s effectiveness in reducing reactive behaviours. Initial feasibility studies conducted at the APSP ‘Margherita Grazioli’ in Povo (Trento) produced promising findings in terms of feasibility and acceptability, consistent with previous research.^{27 28} Overall, VR exposure was well-tolerated by participants, who reported neither moderate nor severe side effects related to headset use (eg, nausea, dizziness, confusion). In line with previous studies,²⁹ participants verbally expressed or displayed a general sense of interest, awareness, engagement and pleasure during VR exposure, appreciating the experience. Feedback indicated that exposure to realistic, natural and personalised VR environments—tailored to users’ preferences and needs—elicited reactions such as reminiscence of nostalgic memories, aesthetic appreciation of the virtual scenery and a sense of security within an enjoyable setting. Participants expressed feelings of self-realisation and a desire to be helpful by providing feedback and suggestions for improving the virtual scenarios. The observed increases in relaxation and decreases in post-VR anxiety and stress align with previous findings,^{27 28} supporting the hypothesis that VR could be a valuable stress management tool.

The current study suggests important considerations for future research. Conducting multiple VR sessions could mitigate anxiety related to the novelty of VR devices and allow further exploration of VR’s potential impact on participants’ emotional states. An ecological experimental setting, where familiar healthcare providers conduct the experimental phase, is recommended to reduce activation and anxiety when using VR devices with external researchers, reproducing the experimental phase closer to a real-world setting. Indeed, the current research broadens the scope of VR applications by focusing on managing responsive behaviours in dementia patients, mainly targeting anxiety and agitation. At the same time, our previous study²³ emphasises engagement and relaxation. Based on a multisession, longitudinal design, the current study intended to explore sustained VR effects compared with a single-session approach.²³ It introduces advanced customisation with real-time adjustments via a smartphone interface and emphasises ecological validity through tailored VR experiences for individual comfort over multiple sessions. By outlining the methodological approach, participant recruitment and planned assessments, the protocol aims to ensure transparency and reproducibility of the study design. The findings obtained from this feasibility phase will serve to refine procedures, identify potential barriers and inform the design of future large-scale clinical trials investigating the clinical effectiveness of VR interventions in dementia care. Overall, the current study’s primary scope is to significantly advance the research by extending VR’s therapeutic potential, addressing responsive behaviours, enhancing customisation and leveraging a longitudinal and user-centred approach to improve outcomes in dementia care.

Objectives of the study

Based on the above, the current research project aims to engage a sample of elderly residents with cognitive impairment and explore the feasibility of using realistic, customisable VR environments in response to the onset of Responsive Behaviours. Specifically, the study will assess (1) feasibility through acceptance metrics (including recruitment rate, adherence rate, attrition rate and tolerance to the intervention); (2) safety (monitoring for adverse effects related to VR headset use) and (3) user engagement, enjoyment and overall experience. This is to determine whether the current experimental procedure warrants further investigation through rigorous randomised controlled trials assessing clinical and statistical efficacy. The study also seeks preliminary insights into whether exposure to customisable, relaxing VR environments might positively impact responsive behaviours observed throughout the day, relaxation levels, state anxiety and general emotional state. A longitudinal design with multiple sessions will promote user familiarity with the technology. Data will be collected through standardised self-report measures (administered in an interview format) and observational assessments by the long-term care facility staff and researchers. Notably, this information will not influence the clinical evaluation of the patient.

Methods

Study design

The study design is based on a longitudinal, mixed-methods approach inspired by the Obesity-Related Behavioural Intervention Trials framework for the design (phase Ib) of digital interventions and their preliminary testing (phase IIa³⁰).

Study sample

A convenience sample of 20 participants, stratified based on varying levels of cognitive functioning impairment,³¹ will be recruited from the Azienda Pubblica di Servizi alla Persona ‘Margherita Grazioli’ in Povo, Trento (APSP ‘Grazioli’; Public Agency for Personal Services, a residential, long-term, sociohealthcare facility for adults and elderly individuals). This sample size ensures a manageable and accessible group for exploratory analysis while capturing a range of issues that will enrich data diversity, essential for understanding varied impairment impacts in a small-scale pilot setting. As this is a feasibility and proof-of-concept study, we intentionally plan to include participants with different degrees of cognitive impairment to explore the acceptability and applicability of the intervention across a realistic clinical spectrum. To ensure methodological consistency, participants will be selected among those presenting a level of cognitive impairment that still allows basic comprehension, interaction and safe participation in the procedures. Individuals whose condition prevents understanding simple instructions, maintaining attention or providing observable behavioural responses will be excluded. This approach aims to balance ecological validity with feasibility, allowing the study to examine how the virtual-reality intervention can be applied across different degrees of cognitive decline while ensuring participant safety and comfort.

The data collection will also be conducted at the APSP ‘Grazioli’. To ensure uninterrupted administration, the experimental setting will be free from distracting external interferences. The study will occur in an ecological mode within APSP settings familiar to each participant, selecting environments where participants will likely experience activation states.

Inclusion criteria

Inclusion criteria comprise:

• Age ≥ 65 years.

• Presence of cognitive decline confirmed by the attending physician.

• Ability to tolerate short sessions of VR exposure.

• Provision of written informed consent by the participant or their legal representative.

Exclusion criteria

• Presence of any psychiatric diagnoses as recorded in medical records.

• A total score of ≥40 on the neuropsychiatric inventory (NPI) indicates neuropsychiatric symptoms.

• A partial NPI score >6 in the domains of agitation/aggression or delusions.

• Enrolment in palliative care services.

• Severe neurological damage, a positive diagnosis of epilepsy or having first-degree relatives diagnosed with epilepsy.

• Presence of a cardiac pacemaker or other metallic devices.

• Infectious or gastrointestinal disorders.

• Open facial wounds, motor or visual impairments, or neuromuscular pain that would prevent using the Meta Quest 2 headset.

• Refusal to consent by the individual or their legal representative for study participation.

• Any condition judged by the clinical team to make participation unsafe, overly distressing or not feasible given the individual’s cognitive and physical state.

Screening and monitoring procedures

Before each VR session, participants’ vital signs (blood pressure and heart rate) will be measured by the nursing staff to ensure clinical stability. If the attending physician or qualified nurse considers the values to be significantly divergent from the participant’s usual baseline or indicative of clinical instability, the session will be postponed for safety reasons. The participant will be reassessed later the same day or the following day, according to medical advice. All decisions regarding eligibility for VR exposure on a given day will be made exclusively by healthcare personnel, in accordance with institutional clinical procedures and the recommendations of the local ethics committee. The research team will not apply predefined numerical thresholds to ensure that safety assessments remain within the clinical domain. Any postponed sessions or temporary exclusions will be documented, and missing data resulting from these occurrences will be reported descriptively and included in the feasibility analysis. Throughout all sessions, trained healthcare staff and researchers will continuously monitor participants to ensure safety and comfort. A pain assessment scale (Pain Assessment in Advanced Dementia, PAINAD) is also administered to ensure the experience is not distressing for the patient. Designated personnel will closely monitor the participant’s state to ensure patient safety and comfort throughout the procedure.

FBK researchers will oversee the APSP ‘Grazioli’ staff throughout the experimental procedure to ensure adherence to the research protocol.

Weekly data collection reviews will be conducted between the research team and the healthcare staff involved in participant recruitment and assessment. These brief meetings will be used to verify data completeness, resolve inconsistencies in case report forms and discuss any procedural or logistical issues encountered during the week. The sessions will also include a review of potential technical issues related to the VR equipment, such as headset calibration, software performance or user comfort, to ensure smooth operation throughout the study. The purpose of these reviews is to maintain data accuracy and procedural consistency across all sessions and to promptly identify and address any barriers to implementation. Any issues raised, including technical adjustments or corrective actions, will be documented in a study log for quality assurance and feasibility reporting.

Measures

The following self-report questionnaires will be administered in an interview format by the health staff operator during different stages of the experimental procedures (preliminary assessment, presession evaluation, during-session evaluation and postsession evaluation). The participant’s current clinical condition may also be retrieved from clinical records.

Preliminary assessment (participant selection)

The Sociodemographic Scale will be administered to assess the inclusion and exclusion criteria. Moreover, sociodemographic data will include gender, age, marital status, education level, past profession, use of glasses, hearing difficulties, use of a wheelchair, limited or no ability to move the head or body, duration of stay in the long-term care facility, type of service used, prior experience with VR and the presence of responsive behaviours.

The Functional Assessment Staging Tool (FAST^{31 32}): The FAST is a clinically validated measure designed for Alzheimer’s patients to classify functional stages across seven primary levels (stage 1: normal adult functioning; stage 2: subjective forgetfulness; stage 3: early deficits in functional performance (eg, difficulty managing finances); stage 4: assistance required for complex tasks; stage 5: dependence on others for daily tasks (eg, dressing); stage 6: assistance required for basic tasks (eg, toileting) and stage 7: severe dementia with loss of motor and communication skills and 16 substages, with stage 7F describing advanced dementia (eg, inability to lift the head). The original version of the FAST scale demonstrated robust psychometric properties (High consistency among evaluators using the tool; Strong correlations between FAST stages and cognitive measures such as the Mini–Mental State Examination; A hierarchical structure that aligns closely with the progressive stages of Alzheimer’s pathology, particularly the correlation between FAST stages and neurofibrillary tangle deposition in the brain). The Italian-translated version of the original FAST, commonly used in clinical and research settings in Italy, was applied. This translation—routinely adopted at the APSP ‘Margherita Grazioli’ long-term care facility as part of standard clinical screening and functional assessment procedures— was used solely to support sample selection and describe participants’ functional status, ensuring consistency with real-world clinical practice, ensuring methodological continuity and comparability between research and real-world data. It is part of the centre’s standard assessment procedures for individuals with cognitive impairment and is also described in national reference texts on dementia care as a core tool for evaluating disease stage and functional autonomy.³³ To our knowledge, no formal psychometric validation of the Italian FAST has been published to date; nevertheless, the tool remains the reference standard for staging functional decline in dementia at national (and local) level.

The NPI^{34 35} is a standardised instrument designed to evaluate neuropsychiatric and behavioural symptoms in individuals with neurological disorders, particularly dementia (eg, delusions, hallucinations, agitation, depression, anxiety, euphoria, apathy, disinhibition, irritability, aberrant motor behaviour, sleep disturbances and appetite changes). For each domain, both frequency (1–4) and severity (1–3) are rated, yielding a total score ranging from 0 (no symptoms) to 144 (maximum severity and frequency across all 12 domains). The NPI is administered through a structured interview with caregivers, who report on the person with dementia’s behaviours over the previous month. The Italian version, commonly used in clinical and research practice, has been widely applied in studies on Italian patients with dementia, which have documented its reliability and clinical validity.³⁴

The Cohen-Mansfield Agitation Inventory (CMAI^{36 37}): This inventory assesses the frequency of agitation-related behavioural symptoms in elderly individuals. The test, administered to caregiving staff, evaluates behaviour observed over the past 2 weeks. It consists of 29 items scored on a frequency scale of 1 (never) to 7 (several times an hour) based on caregiver observations over the past 2 weeks. The behaviours are categorised into four groups: aggressive behaviours, verbal agitation, physically non-aggressive behaviours and other behaviours.

The PAINAD^{38 39}: The PAINAD is a multidimensional scale that evaluates pain in patients with cognitive impairments. It examines five parameters—breathing, vocalisation, facial expression, body language and consolability—on a scale of 0–2 points. The score provides insights into pain severity, ranging from 0 (no pain) to 10 (severe pain). The original and Italian versions of the PAINAD demonstrated strong interrater reliability (IRR), internal consistency and validity as a pain assessment tool for non-verbal populations.

The State-Trait Anxiety Inventory-Form Y (STAI-Y^{40 41}): The STAI-Y is a self-report questionnaire that assesses anxiety for psychodiagnostic purposes or to evaluate the effectiveness of interventions. It includes 40 items scored on a 4-point Likert scale, divided into state anxiety (eg, ‘I feel calm’, ‘I feel secure’) and trait anxiety (eg, ‘I feel tense’, ‘I am satisfied with myself’).

The Observed Emotion Rating Scale (OERS⁴²): The OERS is a tool to assess observable emotional states in individuals with cognitive impairments. More in detail, the scale permits investigating the presence and intensity of positive (eg, pleasure) and negative emotions (eg, fear, anxiety, anger, sadness). Each emotional state is rated on a 5-point frequency scale (0=never, 4=frequently or continuously). Psychometric properties showed good IRR and validity for capturing affective responses in dementia patients.

The Ad-Hoc Observation Grid: This grid captures verbal and non-verbal behaviours, assessing usability, engagement, enjoyment, satisfaction and relaxation during the session. Moreover, the tool is used to assess the type and frequency of drugs used during the perioperative VR exposure.

The Non-Pharmacological Therapy Experience Scale (NPT-ES⁴³): The NPT-ES scale is an observational measure that evaluates non-pharmacological interventions’ affective, emotional and social impact in dementia care. It consists of five items (eg, participation, enjoyment, social interaction) scored on a scale from 0 (never) to 3 (always). The total score ranges from 0 to 15, with higher scores indicating a more positive experience. The psychometric properties of the NPT-ES demonstrated good internal consistency and IRR.

The Virtual Reality Symptom Questionnaire (VRSQ⁴⁴,⁴⁶): The VRSQ is a self-report tool designed to assess symptoms associated with VR exposure, particularly focusing on cybersickness. More in detail, the VRSQ evaluates physical and ocular symptoms that could be experienced related to VR exposure. The questionnaire assesses eight general physical side effects—general discomfort, fatigue, boredom, drowsiness, headache, dizziness, concentration difficulties and nausea—and five visual effects—tired eyes, aching eyes, eyestrain, blurred vision and difficulties focusing—on a 7-point scale ranging from 0 (not present) to 6 (severe). The VRSQ has demonstrated good reliability and validity in capturing VR-induced symptoms.

To obtain information from health staff operators about their perspective on the helpfulness of using the hardware and software apparatus with PwD, a focus group and the System Usability Scale⁴⁷,⁴⁹ will be deployed at the end of the experimental procedure sessions. Table 1 reports the outcomes and endpoints focus of the current study.

Table 1. Description of the outcomes and endpoints.

Group type (PwD, health staff operators)	Outcomes	Endpoint(s)
Health staff operators	Collection of information on users’ preferences to create iVRe. Collection of personal audio-visual material to render within the iVRe.	Survey ad hoc
Health staff operators	Staff operators’ training for the administration of the experimental procedure.	Focus group Simulation training
Health staff operators	Feasibility investigation by obtaining information from staff operators.	Focus group SUS
PwD	Inclusion and exclusion criteria investigation	Sociodemographic scale NPI
PwD	Sociodemographic and clinical diagnosis assessment.	Sociodemographic scale Medical record FAST
PwD	Preliminary investigation of the type, frequency and timing of the onset of responsive behaviour in patients who will adhere to the study.	CMAI
PwD	Assessment of pain.	PAINAD
PwD	Assessment of usability, feasibility, engagement and satisfaction related to the exposure in the VR environment. Assessment of the type and frequency of drugs used during the perioperative VR exposure.	Ad hoc Observation Grid NPT-ES
PwD	Assessment of the observed emotional states.	OERS
PwD	Assessment of the state of anxiety participants will experience before, during and after the VR exposure.	STAI-Y1
PwD	Assessment of the cybersickness symptoms VR-exposure-related.	VRSQ

CMAI, Cohen-Mansfield Agitation Inventory; FAST, Functional Assessment Staging Tool; iVRe, immersive Virtual Reality Environments; NPI, Neuropsychiatric Inventory; NPT-ES, Non-Pharmacological Therapy Experience Scale; OERS, Observed Emotion Rating Scale; PAINAD, Pain Assessment in Advanced Dementia; PwD, Patients with Dementia; STAI-Y, State-Trait Anxiety Inventory-Form Y; SUS, System Usability Scale; VRSQ, Virtual Reality Symptom Questionnaire.

Data collection and data management

User data will be collected via LimeSurvey, a General Data Protection Regulation (GDPR)-compliant platform employed by FBK for data collection and storage in research. Access to LimeSurvey will be managed through a shared account and password, accessible by the APSP ‘Grazioli’ staff and FBK researchers, allowing data access while ensuring security. After collecting informed consent from staff and patients, APSP ‘Grazioli’ personnel will audio-video record the administration sessions to gather research-relevant information, which will be shared with FBK personnel for data analysis. This data will be shared using the FBK-integrated Microsoft SharePoint Drive cloud system, which is compliant with institutional guidelines. Given that audio recordings constitute personal data, all collected data will be encrypted with a password (zip) shared exclusively with active members of APSP ‘Grazioli’ and FBK involved in the research, listed in the final paragraphs of this document.

Hardware and software equipment

The VRe deployed is implemented and supported by hardware comprising a Meta Quest 2 headset (formerly Oculus Quest 2; Meta Platforms, Menlo Park, California, USA) and a OnePlus Nord 3 5G smartphone.

The following hardware specifications characterise the Meta Quest 2: Product Dimensions: 191.5×102 mm×142.5 mm (strap folded in), 191.5×102×295.5 mm (strap fully extended); Weight: 503 g; Tracking: Supports 6 df for head and hand tracking via integrated Oculus Insight technology; Storage: 256 GB; Display Panel: Fast-switch LCD; Display Resolution: 1832×1920 pixels per eye; Refresh Rate: 72 Hz at launch; supports 90 Hz with updates; SoC: Qualcomm Snapdragon XR2 Platform; Audio: Integrated speakers.

The OnePlus Nord 3 5G has the following features

CPU: MediaTek Dimensity 9000; GPU: Arm Mali G710 MC10; RAM: 8GB/16GB LPDDR5X; ROM: 128GB UFS 3.1 storage; Battery: 5000mAh (dual 2500mAh cells, non-removable); Wi-Fi Support: 2×2 MIMO, 2.4G/5.1G/5.8G compatibility; supports Wi-Fi 6 (802.11ax), Wi-Fi 5 (802.11ac) and legacy standards (802.11a/b/g/n). The iVR environments were developed using the Unity framework and programmed in C#. Assets were sourced from reputable repositories, including Unity Asset Store; SketchFab, TurboSquid, Polyhaven; HDRIhaven; Freesound.org; freemusicarchive.org/. The codebase was managed and versioned using GitLab.

The software requirements, including interface design, VR interactions and visual and auditory elements selectable for the virtual context, were collected through collaboration with domain experts and user testing feedback. This ensured an interactive, customisable virtual experience optimised for end-user engagement (figure 1).

Figure 1. Representation of the VR scenarios that could be chosen during the experimental procedure. Author created (Pardini et al²³). VR, virtual reality.

Intervention procedures

Preparatory meetings, training and protocol development

Meetings will be held between FBK researchers and APSP ‘Grazioli’ healthcare staff to define organisational aspects. This preparatory phase contributed not only to structuring the operational procedures but also to developing the contents of the study protocol submitted to the ethics committee for approval.

These preliminary discussions ensured that the planned procedures were feasible, ethically sound and consistent with the daily operations of the long-term care facility.

A training phase will instruct staff on the independent use of the devices. Each preparatory meeting and training session will last approximately 60–90 min and will include hands-on guidance on the use of the VR headset, safety procedures and observation tools.

Staff training will be conducted by members of the research team and will continue until participants demonstrate competence in operating the headset, managing participant comfort and completing observational assessments independently under supervision.

The preparatory process continued until the start of the experimental sessions, ensuring smooth coordination between researchers and staff and maintaining operational readiness.

This process ensures that staff are adequately prepared before the start of the experimental phase. A written guideline document outlining procedural steps for staff during administration will also be provided(figure 2).

Figure 2. Overview of the steps to achieve the outcomes and the related milestones for the research project. Author created (Pardini et al²³). FBK, Bruno Kessler Foundation; VR, virtual reality.

Project presentation and PwD family engagement

Before the start of the study, and following approval by the local Ethics Committee, the project was formally presented by researchers from the Bruno Kessler Foundation to the families of individuals potentially eligible for participation at APSP ‘Margherita Grazioli’.

The main aim was to ensure transparent communication about the project’s goals, procedures and safety measures, to address questions or concerns and to promote understanding and collaboration among all stakeholders involved in the care process.

This preliminary engagement helped to foster trust and facilitate recruitment and adherence to the research procedures.

Pre-exposure familiarisation phase

Before the experimental sessions, eligible participants who meet inclusion criteria will take part in a brief familiarisation phase to become accustomed to the headset and the procedure in a calm and non-agitated condition.

This phase, lasting approximately 1–2 min, allows participants to experience the headset, explore the visual environment and feel comfortable with the device before its use during episodes of agitation or emotional activation. The purpose of this preliminary exposure is to reduce anxiety, ensure safety and promote a positive first contact with the technology. No data collection for outcome measures will occur during this phase.

VR Exposure sessions

Following the familiarisation phase, participants will undergo three VR exposure sessions featuring personalised virtual environments adapted to individual preferences and real-time adjustments for visual and auditory comfort. The virtual environments have been intentionally designed with different degrees of visual realism and stylistic presentation, including both highly realistic natural settings and stylised or vividly coloured, ‘cartoon-like’ scenes. This variability is intentional and reflects the exploratory nature of the study: it allows researchers to qualitatively investigate which types of visual experiences (eg, realistic vs stylised environments or muted vs saturated colours) are perceived as most engaging, relaxing or comfortable by participants. Findings from this component will help to inform future development of virtual environments for dementia care, optimising realism, sensory stimulation and emotional resonance according to users’ preferences and comfort levels.

Duration and procedure of the VR sessions

The experimental protocol for each participant will consist of at least three VR sessions conducted before, during or after a situation deemed activating (eg, personal hygiene activities). Operators will conduct a baseline assessment to identify specific moments when the participant experiences responsive behaviours. Additionally, operators may administer the Meta Quest 2 headset when responsive behaviours occur outside baseline-identified moments.

The experimental procedure consisted of the following phases:

1. Preliminary assessment (participant selection) (30 min):

   • Participants will be selected based on inclusion and exclusion criteria. Following consent, relevant information will be extracted from clinical records and updated based on the participant’s condition.

   • A preliminary investigation will assess the type, frequency and timing of responsive behaviours in participants who consent to participate (demographic sheet, FAST, NPI, CMAI).

2. Pre-session evaluation (T0) (15 min):

   • Assess pain, responsive behaviours, emotions, anxiety, pleasantness and discomfort before VR use (PAINAD, STAI-Y1).

3. Virtual environment selection (5 min):

   • The operator will choose the virtual environment for the participant using a smartphone, which is WiFi-linked to the Meta Quest 2 headset.

4. Explanation of the Procedure and Contextualisation (5 min).

5. Familiarisation with Meta Quest 2 Headset (1–2 min):

   • The participant will undergo a brief familiarisation period to acclimate to the VR headset.

6. VR session and during-session Evaluation (T1) (8 min within participant’s tolerance level):

   • The VR session is audio-video recorded, and the operator observes and notes participant behaviour and verbal responses (OERS, Observation Sheet, STAI-Y1, Ad Hoc Questionnaire, PAINAD, DataSheet2).

7. Post-session evaluation (T2) (15 min):

   • Post-VR use, pain, responsive behaviours, emotions, anxiety, pleasantness and discomfort are reassessed (PAINAD, DataSheet2, Ad Hoc Questionnaire, STAI-Y1, OERS, NPT-ES, VRSQ).

8. Final evaluation after completion of all sessions (T3) (10 min):

   • Final assessment using the CMAI.

This protocol is designed to measure the impact of VR intervention on pain, anxiety and responsive behaviours and evaluate participant engagement and experience.

Analysis plan

After the entire feasibility study, the expected outcome is to highlight a statistically significant reduction based on the quantitative data analysis concerning the cut-off scores for state anxiety, the frequency of target-responsive behaviours, pain perception and the negative emotions experienced. Additionally, positive results are anticipated regarding reliability and feasibility indicators and the safe use and exposure to the technological devices employed.

The analysis of the collected data will primarily focus on assessing the achievement of the study’s objectives. To this end, analyses will be conducted to determine the data distribution, guiding the choice between parametric and non-parametric statistical methods. Continuous data will be expressed as mean ±SD, while categorical data will be summarised as counts and percentages. Subsequently, appropriate multivariable models (eg, analysis of variance, paired-sample t-tests, Wilcoxon signed-rank test) will be used to compare administration times to investigate whether there were changes over time between the pre-VR exposure and postexposure phases. Statistical analyses will be performed using SPSS V.29 and R Studio. In addition, the severity of cognitive impairment, as assessed by the FAST, will be recorded for all participants and considered as an exploratory variable. Descriptive summaries and graphical inspection will be used to examine potential associations between cognitive severity and feasibility or engagement outcomes. Exploratory analyses may include the FAST stage as a single covariate in simple linear or mixed-effects models to evaluate whether the level of cognitive impairment influences feasibility indicators or emotional/behavioural responses. Given the small sample size, these analyses will be treated as hypothesis-generating only, and results will be expressed in terms of effect sizes and bootstrap CIs, rather than inferential p values. This exploratory approach will allow a preliminary assessment of whether the degree of cognitive impairment modulates acceptability or responsiveness to the intervention, thereby informing the refinement of inclusion criteria and study design in future controlled trials. Furthermore, qualitative data will be analysed through a thematic content approach using both inductive and deductive coding strategies. Two independent researchers from the Bruno Kessler Foundation will conduct the qualitative analyses to ensure analytical triangulation and minimise subjective bias. Prior to coding, both researchers will participate in a joint training and calibration session, during which they will review a random subset of approximately 20% of the transcripts. This subset will be jointly coded to refine the preliminary coding framework, resolve ambiguities in code definitions and align interpretative criteria. IRR will be quantitatively assessed using Cohen’s kappa coefficient, calculated on the same 20% subset to measure the degree of agreement beyond chance between the two coders. Kappa values will be interpreted according to standard thresholds proposed by Landis and Koch⁵⁰: values below 0.40 will be considered poor agreement, 0.41–0.60 moderate, 0.61–0.80 substantial, and above 0.80 almost perfect agreement. If the computed κ value falls below 0.70 (the pre-established threshold for satisfactory reliability), the coders will re-examine discrepancies, refine code definitions and repeat the reliability check until the minimum acceptable level is achieved. To complement κ, per cent agreement will be reported as a descriptive indicator of consistency across coding decisions, and Krippendorff’s alpha (α)⁵¹ will be computed as a sensitivity measure, given its applicability to nominal and ordinal data, robustness to small samples and tolerance for missing values. For any continuous or quasi-continuous coding outputs (eg, quantitative indices or rating scores derived from qualitative material), Lin’s concordance correlation coefficient will be calculated to assess both precision (correlation between raters) and accuracy (closeness to perfect agreement). This measure provides a more comprehensive evaluation of rater concordance by integrating both systematic and random components of disagreement. Once adequate reliability (κ≥0.70) is reached, the two coders will proceed to independently code the remaining dataset. Any new codes emerging during analysis will be discussed jointly to maintain consistency and transparency in the analytical process. Regular peer-debriefing meetings will be held throughout the analysis phase to monitor reliability, document analytic decisions and ensure the overall validity and trustworthiness of the qualitative findings. Qualitative data will be transcribed at three time points: before, during and after VR exposure. Audio recordings from the focus group discussions will also be evaluated using the micro interlocutor analysis method, providing insights into participants’ attitudes, perspectives on VR usage and quantitative data on participant group dynamics. Descriptive statistics will be used for both qualitative and quantitative data analysis.

Ethics and dissemination

For this study, the research institutions act as joint data controllers in accordance with Article 26 of the GDPR. From a scientific perspective, the study’s data, parameters and data-management protocols were defined by a technical-scientific committee composed of healthcare professionals and representatives from the participating centres. This committee ensures that data handling and study design comply with ethical standards for patient safety, confidentiality and data integrity. Patients who choose not to participate or who withdraw from the study will continue to receive the same standard of care from their healthcare providers. Given the exploratory nature of the study, exclusion criteria have been implemented to protect participants from potential risks related to VR exposure. Additionally, an initial VR familiarisation session lasting 1–2 min will precede the full VR exposure to enhance safety and participant comfort. The research protocol has been reviewed and approved by the Comitato Etico Territoriale della Provincia Autonoma di Trento per le Sperimentazioni Cliniche, Azienda Provinciale per i Servizi Sanitari–Trento, Italy (Rep. Int. 12090, 15 May 2025). A model participant information sheet and informed consent form (English translation of the original Italian document approved by the Ethics Committee) is provided as online supplemental file 1, in compliance with SPIRIT item #32. All study data will be pseudonymised and stored securely on institutional servers at the Fondazione Bruno Kessler and the APSP ‘Margherita Grazioli’, with controlled access restricted to authorised research staff. Data curation will follow institutional and GDPR-compliant procedures for quality control, version tracking and traceability. In accordance with the approved consent documentation, data will be retained for 3 years after study completion and may be shared externally only in fully anonymised form and for scientific purposes, subject to prior ethical and legal authorisation. Any data shared will exclude direct or indirect identifiers, ensuring that participants cannot be re-identified. The results of this study will be disseminated through peer-reviewed publications, presentations at national and international conferences, and summary reports shared with participating institutions and stakeholders. Dissemination will occur exclusively in aggregated and anonymised form. Where appropriate, anonymised data summaries or metadata may be deposited in open-access repositories to promote transparency and facilitate future research.