Measuring the learning outcomes of datathons

Purpose Healthcare datathons are events in which cross-disciplinary teams leverage data science methodologies to address clinical questions using large datasets. The aim of this research was to evaluate participant satisfaction and learning outcomes of datathons. Methods A multicentre cross-sectional study was performed using survey data from datathons conducted in Sydney, Australia (April 2018) n=98, Singapore (July 2018) n=169 and Beijing, China (December 2018) n=200. Participants (n=467) completed an online confidential survey at the end of the datathons which contained the Affective Learning Scale, and measures of event satisfaction, perceived knowledge gain, as well as free text responses, and participants’ demographic background. Data analysis used descriptive statistics and multivariate analysis of variance (MANOVA). Thematic analysis was performed on the text responses. Results The overall response rate was 64% (301/467). Participants were mostly male (70%); 50.2% were health professionals and 49.8% were data scientists. Based on the Affective Learning Scale (7-point Likert type scale), participants reported a positive learning experience (M = 5.93, SD = 1.21), satisfaction for content and subject matter of the datathon (M = 5.81, SD = 1.17), applying behaviours (M = 4.71, SD =2.02), instruction from mentors (M = 6.01, SD = 1.18), and intention to participate in future datathons (M = 6.03, SD = 1.23). The MANOVA showed significant differences between health professionals and data scientists in perceived knowledge gain from the datathons. Themes from text responses emerged: (1) cross-disciplinary collaboration; (2) improving healthcare using data science and (3) preparations for big data analytics. Conclusions Datathons provide a satisfying learning experience for participants and promote affective learning, cross-disciplinary collaboration and knowledge gain in health data science.