Data-Driven Virtual Reality Environment Recommendation for Long-Term Care Residents With Major Neurocognitive Conditions: Feasibility and Proof-of-Concept Study
4 affiliations
- 1School of Design, Faculty of Planning, Architecture, Art and Design, Laval University, 2325, rue de l'Université, Québec, QC, G1V 0A6, Canada
- 2Center of Interdisciplinary Research in Rehabilitation and Social Integration (CIRRIS), Québec, QC, Canada
- 3International Observatory on the Societal Impacts of AI and Digital Technologies (OBVIA), Québec, QC, Canada
- 4School of Rehabilitation Sciences, Faculty of Medicine, Laval University, Québec, QC, Canada
Abstract
Background: Virtual reality (VR) comfort sessions are increasingly used in residential dementia care, but environment selection remains ad hoc—staff have no systematic way to identify which virtual experience best suits a given resident.
Objective: This feasibility and proof-of-concept study aimed to determine whether a pipeline integrating wearable physiological signals (heart rate variability and electrodermal activity), behavioral observations, and semantic environment characteristics can detect a person-specific, environment-level signal usable for between-session VR environment recommendation in long-term care and whether the resulting recommendations can be communicated to care staff and families through plain-language reports.
Methods: Twenty-one residents with major neurocognitive conditions received 420 immersive VR sessions (5 environments, 2 sessions per week, 10 weeks) in a long-term care facility. We defined an operational hybrid calm target requiring behavioral calm (Pittsburgh Agitation Scale vocal intensity score of 0) and root mean square of successive differences at or above the participant’s personal median to address the ceiling effect of behavioral measures alone (sessions already rated calm: 234/358, 65%). A compact 23-feature model (10 physiological, 3 behavioral, and 10 environment) was selected from more than 2100 configurations on the basis of cross-validation error and interpretability; the environment-level signal was tested with an environment permutation test on held-out sessions (weeks 8‐10) and in sensitivity analyses.
Results: Environment features carried a statistically significant but borderline signal for predicting hybrid calm (environment permutation test, P=.045 with 200 permutations; P=.04 with 2000 permutations; Cohen d=1.75); 4 alternative models using 344 to 368 features ignored environment entirely (all P>.79). Person-level effects dominated prediction variance (partial η²=0.42; P<.001); environment-level discrimination was modest (pairwise ranking accuracy 57.9% vs 50% chance; top-1 hit rate 27.8%, 95% CI 19.8%‐35.7%, including chance). Water presence was the most influential environment characteristic. In sensitivity analyses, the calm label was robust to the threshold (κ=0.79‐0.89), but the environment permutation significance was not: under alternative thresholds and a continuous target, the effect kept its direction (d=0.65‐1.60) without reaching significance (P=.06‐.57), and a selection criterion computed on training period data only did not select the compact model. The pipeline generated 60 evidence-traced plain-language reports (30 family and 30 staff); all passed automated safety checks, with staff reports showing higher data fidelity and family reports showing better nonmedical framing.
Conclusions: An end-to-end pipeline from wearable sensor to plain-language report is technically feasible in residential dementia care. The environment-level signal is suggestive—consistent in direction across target definitions and seeds—but its significance was specific to one labeling of one configuration whose selection was partly informed by the test set; it is exploratory evidence requiring independent replication, not a demonstration that the recommendations improve resident outcomes. Prospective validation against resident outcomes is required before any use in care decisions.
Related
- Implementation strategies used in clinical practice guidelines in older adults' mental health and neurocognitive conditions: a scoping review.
- Cross-cultural adaptation and validation of the Indonesian version of the operating room educational environment measure with Indonesian residents.
- Feasibility evaluation of a panvascular health promotion and risk stratification tool: a cross-sectional survey among community residents in a first-tier city in China
- A multicentre immersive virtual-world course for radiology residents: feasibility, educational outcomes, and participant perceptions
- Incidence of Intensive Care Unit Delirium After Heart Surgery: A Feasibility Study of a Forest Environment Intervention.
BibTeX
@article{reyesconsuelo2026datadriven,
title = {{Data-Driven Virtual Reality Environment Recommendation for Long-Term Care Residents With Major Neurocognitive Conditions: Feasibility and Proof-of-Concept Study}},
author = {Alejandro Reyes-Consuelo and Jocelyne Kiss and Julien Voisin},
journal = {JMIR XR and Spatial Computing},
year = {2026},
month = nov,
doi = {10.2196/101192},
url = {https://doi.org/10.2196/101192},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Data-Driven Virtual Reality Environment Recommendation for Long-Term Care Residents With Major Neurocognitive Conditions: Feasibility and Proof-of-Concept Study
AU - Alejandro Reyes-Consuelo
AU - Jocelyne Kiss
AU - Julien Voisin
PY - 2026
DA - 2026/11/12
JO - JMIR XR and Spatial Computing
DO - 10.2196/101192
UR - https://doi.org/10.2196/101192
AB - Background: Virtual reality (VR) comfort sessions are increasingly used in residential dementia care, but environment selection remains ad hoc—staff have no systematic way to identify which virtual experience best suits a given resident. Objective: This feasibility and proof-of-concept study aimed to determine whether a pipeline integrating wearable physiological signals (heart rate variability and electrodermal activity), behavioral observations, and semantic environment characteristics can detect a person-specific, environment-level signal usable for between-session VR environment recommendation in long-term care and whether the resulting recommendations can be communicated to care staff and families through plain-language reports. Methods: Twenty-one residents with major neurocognitive conditions received 420 immersive VR sessions (5 environments, 2 sessions per week, 10 weeks) in a long-term care facility. We defined an operational hybrid calm target requiring behavioral calm (Pittsburgh Agitation Scale vocal intensity score of 0) and root mean square of successive differences at or above the participant’s personal median to address the ceiling effect of behavioral measures alone (sessions already rated calm: 234/358, 65%). A compact 23-feature model (10 physiological, 3 behavioral, and 10 environment) was selected from more than 2100 configurations on the basis of cross-validation error and interpretability; the environment-level signal was tested with an environment permutation test on held-out sessions (weeks 8‐10) and in sensitivity analyses. Results: Environment features carried a statistically significant but borderline signal for predicting hybrid calm (environment permutation test, P=.045 with 200 permutations; P=.04 with 2000 permutations; Cohen d=1.75); 4 alternative models using 344 to 368 features ignored environment entirely (all P>.79). Person-level effects dominated prediction variance (partial η²=0.42; P<.001); environment-level discrimination was modest (pairwise ranking accuracy 57.9% vs 50% chance; top-1 hit rate 27.8%, 95% CI 19.8%‐35.7%, including chance). Water presence was the most influential environment characteristic. In sensitivity analyses, the calm label was robust to the threshold (κ=0.79‐0.89), but the environment permutation significance was not: under alternative thresholds and a continuous target, the effect kept its direction (d=0.65‐1.60) without reaching significance (P=.06‐.57), and a selection criterion computed on training period data only did not select the compact model. The pipeline generated 60 evidence-traced plain-language reports (30 family and 30 staff); all passed automated safety checks, with staff reports showing higher data fidelity and family reports showing better nonmedical framing. Conclusions: An end-to-end pipeline from wearable sensor to plain-language report is technically feasible in residential dementia care. The environment-level signal is suggestive—consistent in direction across target definitions and seeds—but its significance was specific to one labeling of one configuration whose selection was partly informed by the test set; it is exploratory evidence requiring independent replication, not a demonstration that the recommendations improve resident outcomes. Prospective validation against resident outcomes is required before any use in care decisions.
SN - 2818-3045
ER -