Hello, my name is Ethan Bond, a second-year medical student at the Texas College of Osteopathic Medicine, and today I am presenting on the relevance of using sway measures to predict fall risk. This first slide shows my poster, which we'll be diving into in more detail in upcoming slides. Falls are the most common cause of injury in adults over the age of 65. If you have ever worked in healthcare, you would know that great measures are taken to prevent falls due to this fact. Natural sway refers to the small, unconscious movements that we perform to maintain an upright position. Essentially, it is a quantifiable metric of balance, and thus has been used in literature as an independent predictor of fall risk. That is, studies have associated greater amounts of sway with a higher risk of falling. Compared to our ongoing study, previous studies failed to collect a large sample size and included a convenience population that did not sample a diverse racial and ethnic mix of participants. This study attempts to feature a larger, more diverse sample population, which will be useful in the future development of a fall risk calculating machine learning algorithm. Our data was collected at the University of North Texas Health Science Center clinics as an additional vital sign that providers could use to inform healthcare decisions. Data collection occurred during the patient's wait time before seeing their provider to avoid any delays in the clinical schedule. After gathering information on the subject's fall history and answering demographic questions, the subject would step onto the force plate, maintaining their hands by their side and looking straight ahead at a pre-adjusted target on the wall. The subject was instructed to remain as still as possible for 30 seconds with their eyes open. Next, the subject was then asked to remain still with their eyes closed for another 30 seconds, completing the encounter. The whole encounter would not take more than two minutes, ensuring clinical flow was not disrupted. Output from the force plate included a stabilogram, which depicts the subject's total sway in the anterior, posterior, and medial lateral directions. From the stabilogram, we were able to calculate 20 variables, all of which attempt to quantify a patient's postural sway. Data from each encounter was then stored in a de-identified data warehouse along with elements of the patient's electronic medical record. Figure 1 is a diagram that shows the data collection process. Note the patient's position with their arms by their side, looking straight ahead. Figure 2 is an example of a stabilogram, which depicts the patient's sway on the force plate in the anterior, posterior, and medial lateral directions. As previously mentioned, 20 variables are then calculated from the stabilogram output to quantify sway. One of the principal goals of this ongoing project is to generate a fall risk predicting machine learning algorithm that could be used to assess a patient's fall risk at point-of-care. As shown in Figure 3, the inputs of the soon-to-be-developed algorithm will include concurrent medical conditions, balance, demographics, and the social determinants of health. The goal of this algorithm is to proactively detect fallers before they fall using the aforementioned inputs. Figure 4 shows the project's dashboard, representing the progress of the study since its beginning. As you can see, 1,729 postural sway measurements were collected on a total of 1,263 subjects between June of 2021 and August of 2024. You can also see the total number of subjects with various health conditions. For example, you can see that 331 patients contained an ICD-10 code related to knee or hip pain. The dashboard also shows the demographic distribution of the sample based on age, gender, ethnicity, race, and patient class. Of note, the current data shows a broad age range, covering patients from 18 years to patients at greater than 80 years, most of which were between the ages of 60 and 79 years. 79 of our patients reported a history of falling, with 42 patients sustaining some injury associated with the fall. The objective of this study is to understand the effects of medications, medical conditions, and the social determinants of health on postural sway to predict fall risk. We believe that the osteopathic significance of the study is rooted in its incorporation of the social determinants of health, emphasizing the core tenet of osteopathic medicine. A person is a unit of body, mind, and spirit. To conclude, we developed a process for measuring bounce and thus fall risk that does not hinder clinical flow. Great efforts were taken to ensure that clinical flow was not disrupted, to ensure that providers did not feel burdened by the collection process. The data collected from the study will be used to develop a machine learning algorithm with fall risk scoring capabilities, which may allow us to flag potential fallers before they experience a fall. Specifically, this algorithm may help us to identify potential fallers who may not have historically been considered high fall risk patients. As the study continues to collect data, we hope to expand our collection process to clinics outside the UNT Health Science Center system to sample a more diverse population. However, there are limitations to expansion, the most concerning being the cost of equipment. Here are the references and acknowledgments for my presentation, and thank you for your attention.

fall risk assessment

postural sway

machine learning

osteopathic principles

elderly adults