Learning Analytics to Enhance Dermatopathology Education Among Dermatology Residents

December 2019 | Volume 18 | Issue 12 | Original Article | 1231 | Copyright © December 2019


Euphemia W. Mu MD,a,b Martin Pusic MD PhD,c,d Matt Coneybeare BA MBA,e Shane A. Meehan MDa,f

aThe Ronald O. Perelman Department of Dermatology, New York University (NYU) School of Medicine, New York, NY
bPiedmont Plastic Surgery & Dermatology, Charlotte, NC
cThe Ronald O. Perelman Department of Emergency Medicine, NYU School of Medicine, New York, NY
dDivision of Learning Analytics, Institute for Innovations in Medical Education, NYU School of Medicine, New York, NY
eDepartment of Computer Science, University of California, Berkeley, Berkeley, CA
fDermatology of Pathology, NYU School of Medicine, New York, NY

Abstract
BACKGROUND: With the advent of digital microscopy, learning analytics can be leveraged to advance teaching of dermatopathology in dermatology residency.

OBJECTIVE: To analyze the acquisition and decay of dermatopathology visual recognition skills and areas of diagnostic confusion amongst residents using learning metrics generated by a web-based learning tool.

METHODS: This was a prospective, longitudinal study of dermatology residents who studied digital photomicrographs of 18 routine diagnostic entities using an online software platform. Residents at different years of training were given 60 minutes to complete assessments on three occasions (initial test with follow-ups at one and three months).

RESULTS: 4,938 responses were analyzed. Accuracy and time to diagnosis improved within each assessment and between the first and second assessments. First year residents showed knowledge decay when tested at three months (67% vs 64%; P=0.002) while third year residents retained knowledge and continued to improve upon their accuracy after three months (83% to 91%, <0.001). Learning analytics highlighted diagnostically challenging cases for residents that contradicted experts’ predictions (R=0.48).

CONCLUSIONS: The use of learning analytics and interactive digital platforms enhances dermatopathology curriculum development by identifying challenging diagnostic entities, assessing mastery of subject material, and optimizing review schedules.

J Drugs Dermatol. 2019;18(12):1231-1236.

INTRODUCTION

Traditional methods for instructing dermatopathology rely heavily on approaches such as didactics and standard microscopy. In response to shortcomings of con- ventional medical education techniques, advances in learning sciences offer promising new tools for learners and instructors with a goal of more efficiently acquiring knowledge.1 Prior studies have utilized digital test-enhanced learning to evaluate proficiency in clinical dermatology,2 histology,3 and radiography.4 The use of learning analytics in dermatopathology has not been well studied.

We present a web-based dermatopathology learning application to demonstrate that resident education can be enhanced using data-driven approaches. We developed an electronic platform to assess the acquisition of dermatopathology skills by dermatology residents over time with the goal of optimizing their dermatopathologic educational curriculum.

METHODS

Participant Recruitment and Setting
For this study, we recruited 24 dermatology residents in the Ronald O. Perelman Department of Dermatology at the New York University School of Medicine.To compare with a reference standard, we also invited three board-certified dermatopathologists from three separate institutions to take the survey under study conditions. All participants provided informed consent, and the study was approved by the NYU Institutional Review Board.

Histologic Image Selection and Diagnostic Classification
Selected histologic images of 90 different cases comprised 18 diagnostic entities in the spindle cell neoplasm differential. Each case was chosen from a slide teaching collection created by the Ronald O. Perelman Department of Dermatology, and three magnifications were captured using the Olympus DP70 camera and Aperio slide scanner. The cases included had each been previously diagnosed by an experienced board-certified dermatopathologist and included: atypical fibroxanthoma, an- giofibroma, angioleiomyoma, blue nevus, dermatofibroma, dermatofibrosarcoma protuberans, dermatomyofibroma, desmoplastic melanoma, hypertrophic scar, Kaposi sarcoma, leiomyoma, neurofibroma, neurothekeoma, nodular fasciitis, normal skin, scar, schwannoma, and traumatic neuroma.