University of Twente Student Theses

Login

Towards the prediction of bronchoscopic skill acquisition on a low-fidelity endoscopic prototype

Mührmann, Lisa (2018) Towards the prediction of bronchoscopic skill acquisition on a low-fidelity endoscopic prototype.

[img] PDF
1MB
Abstract:Introduction: A great development in the field of medicine is the admittance of minimally invasive surgery, as it allows operations with reduced blood loss, pain, hospitalization time, and improved cosmetic. One well-known method is the flexible bronchoscopy, in which the lungs could be examined for abnormalities. However, the risk for patients is increased, as the intervention takes place on a vitally necessary organ. In order to minimize the risk, professional surgeons are needed, who are selected for their suitability during different training programs, such as the virtual reality (VR) simulator training. Nevertheless, adequate methods have to be further developed. The original goal of this study was to test, if training on a low-fidelity endoscopic prototype (boxtrainer), which simplify the real bronchoscopic procedure, can improve the VR-simulator task performance. Due to occurring technical problems with the VR-simulator, we focused now only on the boxtrainer task-performance by approaching the performance variables time on task, wall contacts and task success. Another unexpected problem arose, as the estimation of learning curves failed. However, this allowed us to concentrate on different aspects of performance, such as the speed-accuracy tradeoff, without the difficulty to appreciate learning curves. The resulting goal of this research was then to explore the association between the performance variables time on task and wall contacts. Method: Twenty four students of the University of Twente participated. A one-hour training on a low-fidelity boxtrainer is administerd from an allocentric and an egocentric perspective. All participants did the same tasks. Stopping rule was time. A time series design was applied. The original goal of estimating learning curves with a non-linear mixed effect model based on the performance variable time on task, wall contact and task success, failed. Insteed we used the linear multi-level model in order to obtain the association between the performance variables time on task and wall contact. Results: The estimation of learning curves failed. Performance did not improved after prolonged training on the box tainer. Therefore, the predictor trial could be disrepected. The problem simplifies to a multi-level linear model where trials become exchangeable repeated measures. Through employing a generalized linear model (GML) with a poisson distribution a linear association between the performance varbiables time on task and wall contact could be noticed on a population level, as well as on a participant level. Participants made more mistakes, the more time they needed for completing a task. Conclusion: The low-fidelity boxtrainer is not an adequate substitute for the high-fidelity VR-simulators, as the estimation of learning curves is not possible. However, instead, training on the box can measure certain other important aspects of performance, such as the speed accuracy tradeoff during the execution of bronchoscopic tasks. Future research should further examine the possibility of the speed-accuracy tradeoff in performing bronchoscopic tasks, in order to obtain a realistic assessment of the potential of a person.
Item Type:Essay (Bachelor)
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:77 psychology
Programme:Psychology BSc (56604)
Link to this item:http://purl.utwente.nl/essays/75187
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page