Automated segmentation of phases, steps, and tasks in laparoscopic cholecystectomy using deep learning

BACKGROUND

Video-based review is paramount for operative performance assessment but can be laborious when performed manually. Hierarchical Task Analysis (HTA) is a well-known method that divides any procedure into phases, steps, and tasks. HTA requires large datasets of videos with consistent definitions at each level. Our aim was to develop an AI model for automated segmentation of phases, steps, and tasks for laparoscopic cholecystectomy videos using a standardized HTA.

METHODS

A total of 160 laparoscopic cholecystectomy videos were collected from a publicly available dataset known as cholec80 and from our own institution. All videos were annotated for the beginning and ending of a predefined set of phases, steps, and tasks. Deep learning models were then separately developed and trained for the three levels using a 3D Convolutional Neural Network architecture.

RESULTS

Four phases, eight steps, and nineteen tasks were defined through expert consensus. The training set for our deep learning models contained 100 videos with an additional 20 videos for hyperparameter optimization and tuning. The remaining 40 videos were used for testing the performance. The overall accuracy for phases, steps, and tasks were 0.90, 0.81, and 0.65 with the average F1 score of 0.86, 0.76 and 0.48 respectively. Control of bleeding and bile spillage tasks were most variable in definition, operative management, and clinical relevance.

CONCLUSION

The use of hierarchical task analysis for surgical video analysis has numerous applications in AI-based automated systems. Our results show that our tiered method of task analysis can successfully be used to train a DL model.

"What brought you in today?": Modeling patient-provider clinic visits to characterize rural providers' antibiotic treatment decisions

BACKGROUND

Designing clinical decision support (CDS) tools is challenging because clinical decision-making must account for an invisible task load: incorporating non-linear objective and subjective factors to make an assessment and treatment plan. This calls for a cognitive task analysis approach.

OBJECTIVES

The objectives of this study were to 1.) understand healthcare providers' decision making during a typical clinic visit, and 2.) explore how antibiotic treatment decisions are made when they arise.

METHODS

Two cognitive task analysis methods were applied - Hierarchical Task Analysis (HTA) and Operations Sequence Diagramming (OSD) - to 39 h of observational data collected at family medicine, urgent care, and emergency medicine clinical sites.

RESULTS

The resulting HTA models included a coding taxonomy detailing ten cognitive goals and associated sub-goals and demonstrated how the goals occur as interactions between the provider and electronic health record, the patient, and the physical clinic environment. Although the HTA detailed resources for antibiotic treatment decisions, antibiotics were a minority of drug classes ordered. The OSD shows the sequence of events and when decisions are made solely at the provider level and when shared decision making occurs with the patient. Qualitative data from the observations informed a constructed vignette case example portraying select tasks from the HTA.

CONCLUSIONS

These findings emphasize that the scope of disease states presenting to a generalist clinical setting is broad and could include acute exacerbations of rare diseases within a time-pressured environment. CDS must be accessible, time efficient, and fit within the resource gathering task before treatment decisions are made.

Development of a mental script for the mental practice of micro suturing: a methodological approach

INTRODUCTION

Motor imagery and mental practice are important for the acquisition and mastery of surgical skills. The success of this technique relies on the use of a well-developed mental script. In this study, we shared how we developed a mental script for basic micro suturing training by using a low-fidelity rubber glove model.

METHODS

This study applied the design and development research framework. Five expert surgeons developed a mental script by performing a cognitive walkthrough to repair a vertical opening in a rubber glove model, followed by hierarchical task analysis. A draft script was created, and its face and content validity assessed with a checking-back process. Twenty-eight surgeons used the Mental Imagery Questionnaire (MIQ) to assess the validity of the final script.

RESULTS

The process of developing the mental script is detailed. The assessment by the expert panel showed the mental script had good face and content validity. The mean overall MIQ score was 5.2±1.1 (standard deviation), demonstrating the validity of generating mental imagery from the mental script developed in this study for micro suturing in the rubber glove model.

CONCLUSION

The methodological approach described in this study is based on a design and development research framework to teach surgical skills. This model is inexpensive and easily accessible, addressing the challenges of reduced opportunities to practise surgical skills. However, although motor skills are important, the surgeon's other non-technical expertise is not addressed with this model. Thus, this model should act as one surgical training approach, but not replace it.

Hierarchical task analysis of endoscopic sleeve gastroplasty

BACKGROUND

Endoscopic sleeve gastroplasty (ESG) is a minimally invasive endoscopic weight loss procedure used to treat obesity. The long-term goal of this project is to develop a Virtual Bariatric Endoscopy (ViBE) simulator for training and assessment of the ESG procedure. The objectives of this current work are to: (a) perform a task analysis of ESG and (b) create metrics to be validated in the created simulator.

METHODS

We performed a hierarchical task analysis (HTA) by identifying the significant tasks of the ESG procedure. We created the HTA to show the breakdown and connection of the tasks of the procedure. Utilizing the HTA and input from ESG experts, performance metrics were derived for objective measurement of the ESG procedure. Three blinded video raters analyzed seven recorded ESG procedures according to the proposed performance metrics.

RESULTS

Based on the seven videos, there was a positive correlation between total task times and total performance scores (R = 0.886, P = 0.008). Endoscopists expert were found to be more skilled in reducing the area of the stomach compared to endoscopists novice (34.6% reduction versus 9.4% reduction, P = 0.01). The mean novice performance score was significantly lower than the mean expert performance score (34.7 vs. 23.8, P = 0.047). The inter-rater reliability test showed a perfect agreement among three raters for all tasks except for the suturing task. The suturing task had a significant agreement (Inter-rater Correlation = 0.84, Cronbach's alpha = 0.88). Suturing was determined to be a critical task that is positively correlated with the total score (R = 0.962, P = 0.0005).

CONCLUSION

The task analysis and metrics development are critical for the development of the ViBE simulator. This preliminary assessment demonstrates that the performance metrics provide an accurate assessment of the endoscopist's performance. Further validation testing and refinement of the performance metrics are anticipated.

Modelling the use of variable rate intravenous insulin infusions in hospitals by comparing Work as Done with Work as Imagined

BACKGROUND

Variable rate intravenous insulin infusions (VRIIIs) are widely used to treat elevated blood glucose (BG) in adult inpatients who are severely ill and/or will miss more than one meal. VRIIIs can cause serious harm to the patient if used incorrectly. Recent safety initiatives have embraced the Resilient Health Care (RHC) approach to safety by understanding how VRIIIs are expected to be used (Work as Imagined, 'WAI') and how it is actually used in everyday clinical care (Work as Done, 'WAD').

OBJECTIVES

To systematically compare WAI and WAD and analyse adaptations used in situ to develop a model explaining VRIII use.

METHODS

A qualitative observational study video-recording healthcare practitioners using VRIII. The video data were transcribed and inductively coded to develop a hierarchical task analysis (HTA) to represent WAD. This HTA was compared with a HTA previously developed to represent WAI. The comparison output was used to develop a model of VRIII use.

RESULTS

While many of the tasks in the WAD HTA were aligned with the tasks presented in the WAI HTA, some important ones did not. When misalignment was observed, permanent adaptations (e.g. signing as a witness for a changed VRIII's rate without independently verifying whether the new rate was appropriate) and temporary workarounds (e.g. not administering intermediate-acting insulin analogues although the intermediate-acting insulin prescription was not suspended) were the most frequently observed adaptations. The comparison between WAI and WAD assisted in developing a model of VRIII use. The model shed light on strategies used to imagine everyday work (e.g. incident reports, VRIII guidelines), how everyday work was accomplished (e.g. context-dependent adaptations) and how these contributed to both successful and unsuccessful outcomes.

CONCLUSIONS

This study provided in-depth understanding of the tasks required while using VRIIIs, and responses and adaptations needed to achieve safer care in a complex environment.

Human reliability analysis of bronchoscope-assisted percutaneous dilatational tracheostomy: implications for simulation-based education

BACKGROUND

Teaching and assessing clinical procedures requires a clear delineation of the individual steps required to successfully complete the procedure. For decades, human reliability analysis (HRA) has been used to identify the steps required to complete technical procedures in higher risk industries. However, the use of HRA is uncommon in healthcare. HRA has great potential supporting simulation-based education (SBE) in two ways: (1) to support training through the identification of the steps required to complete a clinical procedure; and (2) to support assessment by providing a framework for evaluating performance of a clinical procedure. The goal of this study was to use HRA to identify the steps (and the risk associated with each of these steps) required to complete a bronchoscope-assisted percutaneous dilatational tracheostomy (BPDT). BPDT is a potentially high-risk minimally invasive procedure used to facilitate tracheostomy placement at the bedside or in the operating theatre.

METHODS

The subgoals, or steps, required to complete the BPDT procedure were identified using hierarchical task analysis. The Systematic Human Error Reduction and Prediction Approach (SHERPA) was then used to identify potential human errors at each subgoal, the level of risk and how these potential errors could be prevented.

RESULTS

The BPDT procedure was broken down into 395 subgoals, of which 18% were determined to be of high-risk. The most commonly identified remediation strategies for reducing the risk of the procedure included: checklist implementation and audit, statutory and mandatory training modules, simulation training, consultant involvement in all procedures, and fostering a safety-focused hospital culture.

CONCLUSION

This study provides an approach for how to systematically identify the steps required to complete a clinical procedure for both training and assessment. An understanding of these steps is the foundation of SBE. HRA can identify 'a correct way' for teaching learners how to complete a technical procedure, and support teachers to give systematic and structured feedback on performance.

Going solo: Hierarchical task analysis of the second driver in "two-up" (multi-person) freight rail operations

While many countries have dual-driver ("two-up") modes in freight rail, driver shortage creates increasing pressure to move to single-driver operations. While this change has implications for workload and safety, the roles of the primary and second drivers have not been systematically mapped. This mapping is the focus of this paper, which presents a hierarchical task analysis (HTA) from a multi-methods study (n = 40). Results indicated that transitioning from two-up to single driver operations will result in substantial changes in physical and cognitive workload for the remaining driver. These changes go much further than the simple loss of a crew-mate to double-check or verify actions and cues. This HTA can form the basis of an evidence-based safety case for the change from two-up to single-driver operations, as well as a platform for considering mechanisms to maintain safety and productivity for the now solo train driver.

Systematic engineering tools for describing and improving medication administration processes at rural healthcare facilities

This study demonstrates a series of systematic methods for mapping medication administration processes and for elaborating violations of work standards at two rural hospitals. Thirty-four observational periods were conducted to capture the details of clinical activities, and hierarchical task analysis (HTA) was used to demonstrate the current medication administration process. Facility nurse managers in five units across the two facilities participated in focus group discussions to validate the observational data and to generate a reliable context-appropriate medication administration process. The potential errors or misconduct when passing the drugs were identified, such as unsafe storage and transportation of drugs from room to room. Those hazards would cause drug contamination, loss, or access by unauthorized individuals. Hospitals without 24-hour pharmacy coverage and other interruptions would hinder the medication administration process. Preparing drugs for more than one patient at a time would increase the risk of passing the drugs to the wrong patient. This study shows the use of observation and focus groups to describe and identify violations in the medication administration process. A clear road map for continuous clinical process improvement obtained from the current study could be used to help future health information technology implementation.

Applying hierarchical task analysis to improving the patient positioning for direct lateral interbody fusion in spinal surgery

The present study brings together for the first time the techniques of hierarchical task analysis (HTA), human error identification (HEI), and business process management (BPM) to select practices that can eliminate or reduce potential errors in a surgical setting. We applied the above approaches to the improvement of the patient positioning process for lumbar spine surgery referred to as 'direct lateral interbody fusion' (DLIF). Observations were conducted to gain knowledge on current DLIF positioning practices, and an HTA was constructed. Potential errors associated with the practices specific to DLIF patient positioning were identified. Based on literature review and expert views alternative practices are proposed aimed at improving the DLIF patient positioning process. To our knowledge, this is the first attempt to use BPM in association with HEI/HTA for the purpose of improving the performance and safety of a surgical process - with promising results.