Predictive Gait-Based Biomarkers for Fall Risk in Lower-Limb Prosthesis Users - ABSTRACT Falls are the second leading cause of unintentional death worldwide and lower limb prosthesis users (LLPUs) constitute an exceptionally high-risk group. Over half of LLPU fall annually with many falls resulting in injuries.1 Despite these alarming statistics, clinically accessible tools for predicting falls in LLPU are currently lacking. A comprehensive fall prevention research program should address barriers, such as population heterogeneity and specificity of measurements, and identify modifiable biomechanical risk factors across LLPUs subgroups using methods easily translatable into clinical use.4 Gait analysis to as a screening tool for falls has the potential to satisfy these requirements and markerless motion capture makes it possible that routine gait analysis could be feasibly integrated into routine clinical visits. The overall objective of this proposal is to develop a sensitive and specific fall screening protocol for lower limb prosthetic users that utilizes gait and kinematic analysis from markerless motion capture and is applies to a representative population of lower limb prosthetic users. We will work towards this objective by evaluating fall risk predictors from gait analysis and comparing their accuracy to traditional performance-based clinical assessments. To do this, we will collect in clinic markerless motion capture gait and kinematic data as well as common clinical assessments from 150 lower limb prosthetic users. We will then collect prospective fall history from these individuals for 1 year to determine the subset of gait parameters or clinical assessments that best predict an individual’s future fall risk. Finally, during this one-year collection period, we will use wearable sensors to obtain further kinematic data on fall type, directionality, and circumstances to develop a model for incident risk based on fall type. The results from this study will provide important findings to predict fall risk for a representative, heterogeneous population of lower limb prosthetic users. It will determine whether fall risk can be accurately predicted from 3D gait analysis using markerless motion capture. If successful, our proposed work will produce a translatable approach for screening at-risk lower limb prosthetic users and will identify biomechanical risk-factors for specific fall types. All models and tools for this study will be made open source in our continued commitment to open access. Further, we anticipate that our techniques may generalize to other populations at-risk for falls, such as older adults or stroke survivors. Ultimately, we envision these tools being integrated into routine clinical encounters and a subsequent research study testing whether fall-prevention interventions initiated from clinical gait analysis can reduce the frequency of preventable falls.
