Socratic Artificial Intelligence Learning (SAIL): The Role of a Virtual Voice Assistant in Learning Orthopedic Knowledge

Abstract

Objective

We hypothesized that learning through multiple sensory modalities would improve knowledge recall and recognition in orthopedic surgery residents and medical students.

Design

We developed a virtual study assistant, named Socratic Artificial Intelligence Learning (SAIL), based on a custom-built natural language processing algorithm. SAIL draws from practice questions approved by the American Board of Orthopaedic Surgery and quizzes users through a conversational, voice-enabled Web interface. We performed a randomized controlled study using a within-subjects, repeated measures design.

Setting

Participants first took a pretest to assess their baseline knowledge. They then underwent 10 days of spaced repetition training with practice questions using 3 modalities: oral response, typed response, and multiple-choice. Recall and recognition of the practiced knowledge were assessed via a post-test administered on the first day, first week, and 2 months after the training period.

Participants

Twenty-four volunteers, who were medical students and orthopedic surgery residents at multiple US medical institutions.

Results

The oral, typed, and multiple-choice modalities produced similar recall and recognition rates. Although participants preferred using the traditional multiple-choice modality to study for standardized examinations, many were interested in supplementing their study routine with SAIL and believe that SAIL may improve their performance on written and oral examinations.

Conclusions

SAIL is not inferior to the multiple-choice modality for learning orthopedic core knowledge. These results indicate that SAIL can be used to supplement traditional study methods.

Competencies

medical knowledge; practice-based learning and improvement.