Using "big data" to guide implementation of a web and mobile adaptive learning platform for medical students

Area of Science:

• Educational Technology
• Medical Education Research
• Data Mining in Education

Background:

• Adaptive learning platforms (ALPs) offer potential for efficient medical education.
• Student persistent use of ALPs is crucial but currently limited.
• Understanding factors influencing ALP persistence is vital for educational improvement.

Purpose of the Study:

• To investigate factors associated with persistent use of the Osmosis ALP among U.S. medical students.
• To apply educational data mining techniques to a large dataset of student interactions with an ALP.
• To identify predictors of sustained engagement with digital learning tools in medical training.

Main Methods:

• Utilized educational data mining on data from 6787 U.S. medical students using Osmosis (Aug 2014-Jul 2015).
• Employed multivariate logistic regression to model persistence based on platform-collected variables.
• Analyzed item response data, mobile device usage, subscription status, and group affiliations.

Main Results:

• 31.5% of students demonstrated persistent use of the ALP.
• Item count, mobile use, subscription payment, and group membership significantly predicted persistence (p < 0.001).
• Persistent users reported higher quality ratings but similar confidence; accuracy was slightly lower (OR 0.93) after adjustment.

Conclusions:

• Big data research in medical education can guide ALP implementation strategies.
• Factors like engagement metrics and user characteristics influence persistent ALP use.
• Further research is needed to optimize ALPs for medical student learning and retention.