| Theme: eLearning | |||
 |
 |
|||||
| Learning Analytics in Screen Based Simulation of Radiograph Interpretation |
 |
|||||
|
||||||
Radiology simulations allow deliberate practice using hundreds of image-based clinical cases. Learning analytics can be defined as "the use of learner-produced data and analysis models for predicting and advising people's learning." In this study, we apply learning analytics to a screen-based simulation of radiograph interpretation.
Objective: To investigate candidate learning analytic parameters for radiograph interpretation using an expert-novice comparison.
We recruited low experience (LE) medical learners including 20 medical students and 18 residents, and a high experience (HE) group which included 5 attending emergency physicians and 3 radiologists. Using a web-based program that simulated the clinical presentation of 234 ankle radiographs in an emergency department, participants classified cases as normal or abnormal; if “abnormal” was selected, they specified the location of the abnormality. Immediate feedback on the diagnosis was provided. The system recorded the following process measures: total time on case, time on each radiograph view, number of radiograph views examined, and frequency of re-review of the case history.
The mean (SEM) time on each case for the LE and HE groups were 35.8 (0.45) and 52.6 (1.3) seconds, respectively (p<0.0001). The LE spent an average of 4.0 (0.09) seconds on each view, while the HE group spent 7.2(0.23) seconds, p=0.02. The LE aggregate toggled amongst the views an average of 4.00 (0.02) times per case, while the HE group performed this 4.9 (0.05) times per case, p=0.04. The HE group was 1.7 times as likely as the LE group to re-review a case, although this was not found to be statistically significant (p=0.13).
Simulation environments have the advantage of providing rich process information which, when combined with performance measures, can provide insight into the learner's interpretation process. This information could be used to enhance simulations by adjusting instructional designs according to either developmental level or known markers of expert level performance.
The investigators wish to gratefully acknowledge funding from the Royal College of Physicians and Surgeons of Canada.
Post your comments
Send Email