Identification of elders at higher risk for fall with statewide electronic health records and a machine learning algorithm
International Journal of Medical Informatics
More than a third of community-dwelling older adults are estimated to fall each year, with around 20% of them requiring medical attention and 5% experiencing fractures or severe head wounds. It can often spur a downward spiral – fear of falling, inactivity, decreased strength and mobility, increased risk of falling.
Recognized as a major cause of injury, hospitalization and mortality worldwide, the U.S. alone experienced total fall-related medical costs more than $50 billion in 2015, and expect to reach $55 billion by 2020.
Current risk screening and assessment methods are often manual, include a limited number of intrinsic factors, lack environmental factors and/or lack discriminatory capabilities that a robust machine-learned model can provide on a richer set of data.
In our study, we constructed and validated an electronic health record-based fall risk predictive tool to identify those elders at risk and support personalized interventions for fall reduction and its corresponding morbidity, mortality and medical costs.
The one-year fall prediction model was developed using the machine-learning-based algorithm, XGBoost, and tested on an independent validation cohort. The data were collected from electronic health records comprising 265,225 older patients (≥65 years of age) over two years.
This model attained a validated C-statistic of 0.807, where 50% of the identified high-risk true positives were confirmed to fall during the first 94 days of next year. The model also captured in advance 58.01% and 54.93% of falls that happened within the first 30 and 30–60 days of next year.
The XGBoost algorithm captured 157 impactful predictors into the final predictive model – both intrinsic and environmental. We found cognitive disorders, abnormalities of gait and balance, Parkinson’s disease, fall history and osteoporosis were identified as the top-5 strongest predictors of the future fall event.
This model is available now for clients in HBI Spotlight 2.0 Population Risk Management.
We invite you to read the study in its entirety or contact us for more information