Data Republic powers Anthem’s latest Hackathon
Earlier this month in Palo Alto, Anthem Inc. tackled some of healthcare’s hefty challenges in a two-day Hackathon, with Data Republic behind them as a key prize sponsor and platform provider.
Competing student teams were tasked, by invitation, to find innovative applications of Anthem data and discover correlations between patients diagnosed with diabetes, lower back pain, obesity, and heart disease. In order to complete this challenge, they were given secure access to de-identified patient data about prescriptions, labs, patient eligibility, and medical events. Sampled through Data Republic’s Senate Platform, in a secure sandbox environment, contestants’ delivered innovative solutions under tense time restraints.
But they weren’t alone. During the event, data scientists and solution architects were on site at all times to help contestants explore possibilities and answer their questions about the environment.
Data Republic’s General Manager USA, Steve Prestidge said,
We are so pleased to have partnered with Anthem’s AI team to facilitate this hackathon. There is huge potential for innovation when secure, governed access to data is enabled, and our platform was able to help Anthem and the competing students unlock the power of data. It was incredibly inspiring to see these smart minds at work to solve real healthcare challenges with data-driven insights.
The energy levels were high, the competition was close and the calibre of work exceeded everyone’s expectations. While the Anthem.ai team provided some ideas for data applications – such as, the probability of being diagnosed with a disease after being treated for another disease, or the likelihood of developing a disease in 12 months – teams also came up with their own alternatives. Stand outs include a model for the materialization of a diabetes mellitus diagnosis based on an elevated A1C level and other medical indicators, and a model for the propensity to develop a psychiatric condition based on medications.
Judges based their assessments on novelty, anticipated clinical impact, proof of correctness, and innovative use of data provided. In the end, the winning team – a father-son duo from University of California, San Francisco – built an impressive model to predict the likelihood of having a heart attack after a specific health event and depending on comorbidities like diagnosis of pre-diabetes.