A clinical working group at a “hackathon” exercise designed and implemented a research study using electronic health record data from 3 open Translator Clinical Knowledge Sources that replicated suspected or recognized associations between sex, diabetes, obesity, and fine particulate matter exposure in people with severe asthma, according to a report published in the Journal of Biomedical Informatics.
The 5-day hackathon event, known as The Biomedical Data Translator Consortium, was held March 4 to 8, 2019, in Chapel Hill, North Carolina. As part of a larger program sponsored by the National Center for Advancing Translational Sciences, the hackathon aimed to promote team science efforts through the use of open clinical data sources to examine the aforementioned relationships. Investigators were interested in offering a proof of concept that provided evidence of the potential utility of data sharing within their collaboration model for the purposes of advancing science and producing clinically meaningful results.
The electronic health record open access data sources included Clinical Profiles (Johns Hopkins Medicine), Integrated Clinical and Environmental Exposures Service (ICEES; University of North Carolina Health Care System) and Columbia Open Health Data (COHD; Columbia University Medical Center). Jupiter Python notebooks, based in GitHub repositories, were used for integration and analysis of the results.
Asthma was chosen as the focus because of the wide availability of relevant clinical and demographic information for this disease across the 3 databases. Relationships among sex, obesity, diabetes, and exposure to airborne particulate matter ≤2.5 microns in diameter (PM2.5) were explored in people with severe asthma.
After stratification by sex, obesity was significantly more common in women vs men across all data sources (P <.001 for Clinical Profiles; P <.0001 for ICEES and COHD). When stratified by obesity, the data revealed a significantly higher proportion of females among obese vs non-obese patients (P <.001 for Clinical Profiles and COHD; P <.0001 for ICEES and Clinical Profiles + ICEES). Compared with non-obese individuals, obese patients demonstrated a significantly higher prevalence of diabetes (P <.001 for Clinical Profiles and COHD; P <.0001 for ICEES and Clinical Profiles + ICEES). Stratification by diabetes saw a greater prevalence of obesity in those with diabetes vs those who did not have diabetes across sources (P <.001 for Clinical Profiles and COHD; P <.0001 for ICEES and Clinical Profiles + ICEES).
When ICEES was used to explore relationships between PM2.5 exposure and sex, obesity, and diabetes among those with severe asthma, the group found that people exposed to higher PM2.5 levels were more likely to be obese (P =.0593) and have diabetes (P <.01) than those exposed to lower levels. There was no correlation detected between PM2.5 levels and sex.
These findings successfully replicated suspected or established connections among sex, obesity, diabetes, PM2.5 exposure, and severe asthma. At the same time, the 3 different Translator Clinical Knowledge Sources produced distinct differences that represented variations in environment- and/or cohort-specific factors related to the services provided or the covered catchment area.
“Collectively, this special communication demonstrates the power and utility of intense, team-oriented hackathons and offers general technical, organizational, and scientific lessons learned,” noted the authors.
Fecho K, Ahalt SC, Arunachalam S, et al; Biomedical Data Translator Consortium. Sex, obesity, diabetes, and exposure to particulate matter among patients with severe asthma: scientific insights from a comparative analysis of open clinical data sources during a five-day hackathon. J Biomed Inform. 2019;100:103325.