764 USE OF WEARABLE SENSORS TO ASSESS BIOMECHANICAL LEARNING PATTERNS IN ENDOSCOPY TRAINING

Abstract

Objective assessment of motor skills in endoscopy trainees is poorly defined. In US training programs, competency has been regarded as a function of procedural volume and training year. Outcome measures in colonoscopy such as time to cecum, cecal intubation rate, and adenoma detection rate are used as surrogate measures of competency. Remediation of low performance is reliant upon expert experience and intuition. We aimed to use kinematic data from wearable sensors to assess trainee performance of a standardized, validated endoscopic task to identify quantifiable biomechanical signals of procedural competence. Gastroenterology fellows performed the validated EndoBubble abstract psychomotor task on the GI Mentor II virtual reality simulator (Simbionix USA Corp., OH, USA). Flexible wearable sensors (Biostamp RCTM, mc10 Inc., Lexington, MA) were placed on the dorsum of both hands, dorsal section of both lower arms (2/3 distance from wrist to elbow) and forehead. The sensors recorded 3D linear accelerations and angular velocity. Range of changes to linear and angular acceleration were calculated from the raw signal collected during simulation. 19 gastroenterology fellows performed the simulated endoscopy task (14 were novice, first year trainees). Increased head flexion-extension range of velocity correlated significantly with total procedure time (r2 = 0.622, p-value = 0.006). Head flexion-extension inversely correlated with traditional markers of procedural competence such as fellowship training year (r2 = -0.629, p-value = 0.005) and number of colonoscopies performed (r2 = -0.847, p-value = 0.016). Head anterior-posterior acceleration had similar positive correlation with total procedure time (r2 = 0.596, p-value = 0.009) and negative correlations with fellowship year (r2 = -0.757, p-value = 0.049) and number of colonoscopies performed (r2 = -0.623, p-value = 0.006). This study is the first to identify and quantify head motion as specific parameter that correlates with performance outcomes of a validated psychomotor simulation task and traditional measures of endoscopic procedural competence. Excess head motion of novice endoscopy trainees may suggest hand position visualization to make adjustments. The consequent loss of visual focus on the endoscopy monitor may contribute to the prolonged task completion times. Further study may identify additional kinematic parameters to facilitate the development of a personalized endoscopic training curriculum based upon individual trainee profiles with objective biomechanical performance targets.