Many senior living and care operators are turning to AI-enabled fall prevention tools. The ideal behind this slate of software and sensors is not just to anticipate an immediate risk of falling, but to create a risk profile for residents that can prevent emergencies long-term.
But although almost everyone — including older adults themselves, buys into the value of falls prevention tech — a debate exists over which blend of data or process produces the most accurate results.
A new study shows that using AI to analyze how residents conduct various standing and balance exercises produces a superior fall-risk profile (compared with using AI to extrapolate from past falling incidences). One of the study’s aims was to parse the nuances of monitoring tools.
Researchers refer to the AI analysis of posture as “computerized posturography.” Study participants were asked to conduct a physical “up-and-go test,” with data captured by HTC VIVE headsets, to produce a fall risk stratification profile.
“The prevention of falls is challenging for the complexity and dynamic nature of contributing factors,” the study authors wrote. “Older adults may still function well in community settings despite gradually declined balance function. The findings of our study indicated that the incorporation of feature selection techniques can significantly improve the accuracy and overall performance [of monitors].”
The AI-analysis of posture was superior to both pencil-and-paper questionnaire screeners and human-only evaluation of mobility, the researchers said.
The study report was careful to include an explanation of the AI techniques and process. Currently one of the biggest issues around trusting AI tools is the lack of transparency around how these models work, the McKnight’s Tech Daily reported Monday.
AI remains a “black-box framework,” the study authors acknowledged, and “can be a significant drawback for the underlying trust issue and lead to low use in practice, especially in healthcare.”
The study was published Monday in the Journal of NeuroEngineering and Rehabilitation.
