Wearable sensors for modeling and assessing nontechnical skills in surgery - Abstract Non-technical (NT) skills encompass teamwork, communication, decision-making, leadership, and situational awareness and are critical aspects for safe care delivery in the operating room (OR). Non-technical skills of surgery have been identified as the most frequent causes for sentinel events that impact patient mortality, post- operative pain, and quality of life. Lapses in communication, teamwork, and leadership have been shown to account for 60% of major perioperative complications, most of which are related to communication failures. Given this strong link between NT and team skills and patient safety, several NT skills assessment tools have been validated specifically for surgical teams. However, they require the presence of trained observers or retrospective review of videos. Such methods are time-consuming and resource intensive to obtain. Importantly, these assessments are subject to multiple assessment biases of the observer which threatens their objectivity and value. Automated tools that do not require an expert observer to conduct real-time observations, do not need time-intensive audio-video analysis, and could provide objective, quantifiable, and continuous measurements of NT skill are, thus, highly needed and could allow for the more widespread implementation of NT skills training. The objectives of the current project are to 1) investigate and develop predictive models linking objective sensor- based data streams with individual and team non-technical skills, 2) test non-intrusive sensing devices in simulated surgical team simulation training for practicing professionals, and 3) test sensor's accuracy in distinguishing teams with and without TeamSTEPPS training. To achieve these objectives, three specific aims are proposed. Aim 1 will validate our multi-modal sensing system for measuring NT skills in surgical simulation. A collaborative team of human factors engineers and surgeon educators will perform summative usability evaluation and determine usability with the System Usability Scale and Technology Acceptance Model. For Aim 2, our team will implement sensors in well-validated surgical simulations with experienced and inexperienced teams. As sensors continuously collect data, observer-based non-technical skill assessment tools will be performed real-time and video recorded. Analytics, data fusion, and regression analysis will be performed to model sensor metrics with gold-standard ratings. Finally, we will evaluate whether sensors can distinguish differences in team skills between teams with/without TeamSTEPPS training. The expected deliverables will include usability validation of a non-intrusive sensing system and analytics framework for automated skill assessment that can be implemented in surgical simulation. If proven valid and acceptable through the proposed work, our objective NT skill assessments could be used in the future to assess effectiveness of team training and provide structured, individually-tailored feedback to enhance healthcare providers' NT skills. Thus, this work could ultimately reduce errors due to poor NT skills and enhance patient safety.
