Identifying fall risk using machine learning algorithms

Abstract

A person's fall risk may be determined based on machine learning algorithms. The fall risk information can be used to notify the person and / or a third party monitoring person (e.g. doctor, physical therapist, personal trainer, etc.) of the person's fall risk. This information may be used to monitor and track changes in fall risk that may be impacted by changes in health status, lifestyle behaviorsor medical treatment. Furthermore, the fall risk classification may help individuals be more careful on the days they are more at risk for falling. The fall risk may be estimated using machine learning algorithms that process data from load sensors by computing basic and advanced punctuated equilibrium model (PEM) stability metrics.

Description

technical field [0001] This disclosure relates to machine learning algorithms. More specifically, portions of this disclosure relate to applying machine learning algorithms to determine patient balance or identify patient fall risk. Background technique [0002] Among the U.S. population, accidental falls are responsible for more than 30,000 deaths per year. Older adults are the most prone to falls and thus suffer more than 300,000 hip fractures each year. 50% of people who break a hip never return home. The poor balance that causes these falls often declines decades before the fall, but the conventional approach to addressing poor balance is to seek medical diagnosis and intervention after a fall occurs or the patient has very serious balance problems. In fact, the current best predictor of falls is whether or not the person has fallen. [0003] To truly improve national falls statistics, preventive interventions should be implemented before the first fall. Balance, li...

AI Technical Summary

Problems solved by technology

[0007] While the methods described above improve the understanding of postural stability, it is generally understood that the multifactorial nature of falls means that prediction of falls falling outside real-time and near-real-time ranges is difficult

Despite developments to date, there is still a need to improve pose stability representations

Images