Evaluating the prognostic and clinical validity of the Fall Risk Score derived from an AI-based mHealth app for fall prevention: retrospective Real-World Data Analysis
Background: Falls pose a significant public health concern associated with high mortality rates, increasing occurrence due to the aging population and prevalence of risks such as multimorbidity and frailty. Falls not only lead to physical injuries but also have detrimental psychological and social c...
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article (Journal) |
| Language: | English |
| Published: |
04.12.2024
|
| In: |
JMIR aging
Year: 2024, Volume: 7, Issue: 1, Pages: 1-14 |
| ISSN: | 2561-7605 |
| DOI: | 10.2196/55681 |
| Online Access: | Verlag, kostenfrei, Volltext: https://doi.org/10.2196/55681 Verlag, kostenfrei, Volltext: https://aging.jmir.org/2024/1/e55681 
|
| Author Notes: | Sónia A Alves, PhD, Steffen Temme, BSc, Seyedamirhosein Motamedi, PhD, Marie Kura, MSc, Sebastian Weber, MD, Johannes Zeichen, Prof Dr, Wolfgang Pommer, MD, André Baumgart, PhD |
| Summary: | Background: Falls pose a significant public health concern associated with high mortality rates, increasing occurrence due to the aging population and prevalence of risks such as multimorbidity and frailty. Falls not only lead to physical injuries but also have detrimental psychological and social consequences negatively impacting quality of life. Identifying individuals at high risk for falls is crucial, particularly for those aged 60 years or over and living in residential care settings, and current professional guidelines favor personalized, multifactorial fall risk assessment approaches for effective fall prevention. Objective: The present study aimed to explore the prognostic validity of the Fall Risk Score (FRS), a multifactorial-based metric to assess fall risk, using longitudinal real-world data and establish the clinical relevance of the FRS by identifying threshold values and the minimum clinically important differences (MCID). Methods: This retrospective cohort study involved 617 older adults (857 observations, 615 women, 242 men, 83.3 ± 8.7 years old, 0.49 ± 0.19 m/s, 622 using walking aids) residing in German residential care facilities and employed the LINDERA mHealth application for fall risk assessment. The study focused on the association between FRS at the initial assessment (T1) and the normalized number of falls at follow-up (T2). A quadratic regression model and Spearman’s correlation analysis was utilized to analyze the data, supported by descriptive statistics, and subgroup analyses. Results: The quadratic model exhibited the lowest RMSE (0.015), and Spearman’s correlation analysis revealed that a higher FRS at T1 is linked to an increased number of falls at T2 (Spearman’s correlation coefficient=0.960, P<.001). Sub-groups revealed significant strong correlations between FRS at T1 and falls at T2, particularly for older adults with slower gait speeds (Spearman’s correlation coefficient=0.954, P<.001), and those using walking aids (Spearman’s correlation coefficient=0.955, P<.001). Threshold values revealed that an FRS of 45%, 32%, and 24% corresponded to the expectation of a fall within 6, 12, and 24 months, respectively. Distribution-based MCID values were established, providing ranges for small, medium, and large effect sizes for FRS changes. Conclusions: The FRS exhibits good prognostic validity for predicting future falls, particularly in specific subgroups. The findings support a stratified fall risk assessment approach and emphasize the significance of early and personalized intervention. This study contributes to the knowledge base on fall risk, despite limitations such as demographic focus and potential assessment interval variability. Clinical Trial: Not applicable, due to anonymized data processing and analysis. |
|---|---|
| Item Description: | Gesehen am 18.03.2025 |
| Physical Description: | Online Resource |
| ISSN: | 2561-7605 |
| DOI: | 10.2196/55681 |