AI-Based Predictive Rehabilitation-Healing Environment Monitoring Package Using Environmental and Health Data

(Purpose) This study aims to establish a new predictive and integrative paradigm for managing rehabilitation and healing environments by linking environmental and health data with AI-based models. (Design/methodology/approach) A short-term eight-week pilot study was conducted with 21 older adults in one nursing hospital and two welfare facilities in Gyeonggi Province. The final selected environmental indicators (PM2.5, CO₂, illuminance, noise) were measured using government-certified sensors, and physiological indicators (heart rate, respiration, HRV, stress index) were monitored. To review the final direction, the data collected in the pilot study underwent preprocessing for missing values and outliers, followed by correlation and regression analyses. Subsequently, AI algorithms (Random Forest, LSTM) were applied to evaluate mutual predictive performance, and SHAP analysis was conducted to interpret the relative contributions of each variable. Through this process, the validity of the integrated environmental-health model was verified, and the study was reviewed in the direction of testing the research hypothesis and developing an AI-based predictive rehabilitation and healing environment monitoring package utilizing environmental-health data.(Findings) Based on the theoretical and case-based reviews and pilot test results, the research hypothesis was empirically validated by confirming significant correlations between environmental factors and health indicators. Specifically, PM2.5 was correlated with heart rate (r = 0.42, p < .01) and respiratory rate (r = 0.38, p < .05), while CO₂ was correlated with the stress index (r = 0.30, p < .05). The AI-based predictive model (LSTM) demonstrated superior performance (ROC-AUC = 0.87, F1 = 0.82) compared to the Random Forest model (ROC-AUC = 0.81, F1 = 0.76). Furthermore, SHAP analysis revealed that PM2.5 and CO₂ were the strongest predictors, noise and HRV had moderate contributions, and illuminance contributed minimally. These findings provide concrete evidence that rehabilitation and healing environment management can transition from equipment-centered approaches to data-driven, subscription-based services. In line with expert consultation and final recommendations, the developmental trajectory of companies engaged in AI-based predictive rehabilitation and healing environment monitoring packages suggests that domestic SMEs can move beyond device-centered businesses and demonstrate high applicability in preventive environmental design and in expanding rehabilitation and healing practices for high-risk facilities such as nursing hospitals and rehabilitation centers. (Research implications or Originality) The study demonstrates the feasibility of shifting from device-centered sales/rental models toward data-driven subscription services with AI-based monitoring dashboards. It provides empirical and strategic evidence for companies such as representative case company to evolve into service-oriented solution providers, while contributing to KS/ISO standardization, public procurement frameworks, and data governance models. Ultimately, the research highlights both industrial competitiveness and public value, supporting predictive rehabilitation-healing environment monitoring as a key tool for reducing health disparities and healthcare costs in a super-aged society.