PP1420 - Use of Hearing Aids with Inertial Sensors and Artificial Intelligence for Fall Risk

Abstract:
Hearing aids with inertial sensors (IMU-HAs) may help measure fall risk remotely via an app. We evaluated their usability and accuracy using self-administered assessments from the CDC’s STEADI protocol. Comparing the IMU-HAs' risk scores with clinicians' observations showed high agreement, suggesting their potential for home-based fall risk measurement. However, some participants continued to have difficulty with the app interface. Thus, ensuring the app's usability for those less familiar with technology is crucial to prevent disparities in care.

Summary:
Rationale
As falls are a leading cause of injury and death in elderly adults, the identification of fall risk is critically important. The Center for Disease Control and Prevention (CDC) has created a fall risk protocol called STEADI (Stopping Elderly Accidents, Deaths, and Injuries) that incorporates validated assessments of functional motility to assess the risk of future falls, with reasonably predictive results. Self-administration of these assessments would reduce several barriers that currently deter the clinical implementation of STEADI (e.g., limited resources, increased appointment time, training required) and may increase patient accessibility to a plan of care, shown to reduce falls by 40%. We have been investigating the use of hearing aids embedded with inertial measurement units (IMU-HAs) for self-administered fall risk assessments. In this approach, participants wear the HA and complete the STEADI protocol, which is scored through a phone application. When measured in person and compared with a trained observer, we demonstrated strong agreement between the two, suggesting this approach may be feasible for monitoring fall risk with the STEADI protocol. Subsequent work demonstrated that some participants could complete the procedure at home without supervision, while more than 20% were unable to complete the procedure. To better understand these difficulties and develop procedures to fix them, we repeated the ‘at-home’ protocol in a lab environment while participants were observed remotely by a trained clinician in a different room.

Design
We tested 50 participants aged 55-100 years who self-reported being at risk for falls. Participants were fit with bilateral IMU-HAs and independently completed the CDC STEADI functional mobility assessments, consisting of the Timed Up and Go, The 4-Stage Balance Test, and the Chair Stand Test. All participants were tested in a lab environment, but completed all procedures independently as instructed by the HA app. Participants were simultaneously observed through a telehealth portal by a clinician in another room. The clinician simultaneously scored the STEADI protocol, and identified barriers towards participants being able to complete the procedure independently through the HA application.

Results
Preliminary results again suggest strong agreement between the HA application and the trained observer on all STEADI assessments. In the TUG and chair-stand tests, fewer errors were observed than in the results previously reported, suggesting the updated algorithm is more accurate. Some participants continued to have difficulty utilizing the app to independently complete the STEADI protocol, however. The most common difficulty noted here was ‘confusion with the app.'

Conclusions
Taken together, these results continue to suggest that IMU-HA could be used to identify individuals at risk according to the STEADI protocol. To avoid disparities in care, however, the usability of the app needs to be accessible for individuals less experienced with technology. Moreover, care needs to be taken with the reading level of the app instructions to maximize access to this technology.

Brief Summary of Clinical Takeaways:
New developments in HA technology suggest that hearing aids have the potential to monitor fall risk through self-assessment of the STEADI protocol. However, usability of the app is vital to make this technology accessible to all individuals.