Automated operative workflow analysis of endoscopic pituitary surgery using machine learning: development and preclinical evaluation (IDEAL stage 0)

OBJECTIVE Surgical workflow analysis involves systematically breaking down operations into key phases and steps. Automatic analysis of this workflow has potential uses for surgical training, preoperative planning, and outcome prediction. Recent advances in machine learning (ML) and computer vision have allowed accurate automated workflow analysis of operative videos. In this Idea, Development, Exploration, Assessment, Long-term study (IDEAL) stage 0 study, the authors sought to use Touch Surgery for the development and validation of an ML-powered analysis of phases and steps in the endoscopic transsphenoidal approach (eTSA) for pituitary adenoma resection, a first for neurosurgery. METHODS The surgical phases and steps of 50 anonymized eTSA operative videos were labeled by expert surgeons. Forty videos were used to train a combined convolutional and recurrent neural network model by Touch Surgery. Ten videos were used for model evaluation (accuracy, F1 score), comparing the phase and step recognition of surgeons to the automatic detection of the ML model. RESULTS The longest phase was the sellar phase (median 28 minutes), followed by the nasal phase (median 22 minutes) and the closure phase (median 14 minutes). The longest steps were step 5 (tumor identification and excision, median 17 minutes); step 3 (posterior septectomy and removal of sphenoid septations, median 14 minutes); and step 4 (anterior sellar wall removal, median 10 minutes). There were substantial variations within the recorded procedures in terms of video appearances, step duration, and step order, with only 50% of videos containing all 7 steps performed sequentially in numerical order. Despite this, the model was able to output accurate recognition of surgical phases (91% accuracy, 90% F1 score) and steps (76% accuracy, 75% F1 score). CONCLUSIONS In this IDEAL stage 0 study, ML techniques have been developed to automatically analyze operative videos of eTSA pituitary surgery. This technology has previously been shown to be acceptable to neurosurgical teams and patients. ML-based surgical workflow analysis has numerous potential uses—such as education (e.g., automatic indexing of contemporary operative videos for teaching), improved operative efficiency (e.g., orchestrating the entire surgical team to a common workflow), and improved patient outcomes (e.g., comparison of surgical techniques or early detection of adverse events). Future directions include the real-time integration of Touch Surgery into the live operative environment as an IDEAL stage 1 (first-in-human) study, and further development of underpinning ML models using larger data sets.

Using a Clinical Workflow Analysis to Enhance eHealth Implementation Planning: Tutorial and Case Study

eHealth apps often fail to improve clinical outcomes due to poor integration with clinical workflow—the sequence and personnel needed to undertake a series of tasks for clinical care. Our central thesis is that eHealth interventions will be more effective if the clinical workflow is studied and taken into consideration for intervention implementation. This paper aims to provide an introductory tutorial on when and how to use a clinical workflow analysis to guide the implementation of eHealth interventions. The tutorial includes a step-by-step guide to conducting a clinical workflow analysis in planning for eHealth implementation. We began with a description of why a clinical workflow analysis is best completed before the implementation of eHealth interventions. Next, we described 4 steps needed to perform the clinical workflow analysis: the identification of discrete workflow components, workflow assessment, triangulation, and the stakeholder proposal of intervention implementation. Finally, we presented a case study of a clinical workflow analysis, which was conducted during patient visits of patients aged 11 or 12 years from 4 diverse pediatric or family medicine clinics to plan the implementation of a tablet-based app for adolescent vaccination. Investigators planning the implementation of new eHealth interventions in health care settings can use the presented steps to assess clinical workflow, thereby maximizing the match of their intervention with the clinical workflow. Conducting a prospective workflow study allows for evidence-based planning, identifying potential pitfalls, and increasing stakeholder buy-in and engagement. This tutorial should aid investigators in increasing the successful implementation of eHealth interventions.

Clinician Perspectives on Unmet Needs for Mobile Technology Among Hospitalists: Workflow Analysis Based on Semistructured Interviews (Preprint)

BACKGROUND The hospitalist workday is cognitively demanding and dominated by activities away from patients’ bedsides. Although mobile technologies are offered as solutions, clinicians report lower expectations of mobile technology after actual use. OBJECTIVE The purpose of this study is to better understand opportunities for integrating mobile technology and apps into hospitalists’ workflows. We aim to identify difficult tasks and contextual factors that introduce inefficiencies and characterize hospitalists’ perspectives on mobile technology and apps. METHODS We conducted a workflow analysis based on semistructured interviews. At a Midwestern US medical center, we recruited physicians and nurse practitioners from hospitalist and inpatient teaching teams and internal medicine residents. Interviews focused on tasks perceived as frequent, redundant, and difficult. Additionally, participants were asked to describe opportunities for mobile technology interventions. We analyzed contributing factors, impacted workflows, and mobile app ideas. RESULTS Over 3 months, we interviewed 12 hospitalists. Participants collectively identified chart reviews, orders, and documentation as the most frequent, redundant, and difficult tasks. Based on those tasks, the intake, discharge, and rounding workflows were characterized as difficult and inefficient. The difficulty was associated with a lack of access to electronic health records at the bedside. Contributing factors for inefficiencies were poor usability and inconsistent availability of health information technology combined with organizational policies. Participants thought mobile apps designed to improve team communications would be most beneficial. Based on our analysis, mobile apps focused on data entry and presentation supporting specific tasks should also be prioritized. CONCLUSIONS Based on our results, there are prioritized opportunities for mobile technology to decrease difficulty and increase the efficiency of hospitalists’ workflows. Mobile technology and task-specific mobile apps with enhanced usability could decrease overreliance on hospitalists’ memory and fragmentation of clinical tasks across locations. This study informs the design and implementation processes of future health information technologies to improve continuity in hospital-based medicine.