Wade L. Schulz
Health care is in the midst of a digital transformation.
The widespread adotpion of electronic health records,
cost-effective genomic testing, and digital consumer
health products now allow us to capture a detailed picture
of an individual’s health, outside of traditional health
care encounters. Using these data to develop digital therapeutics,
analytic models, and real-time clinical decision
support has gained significant interest, but harnessing
these data to produce actionable information remains a
challenge. The use of multiple ontologies, non-standard
methods for real-world data collection and analysis, and
the lack of secure and scalable computing infrastructure
all pose barriers to the routine use of these data. To overcome
these challenges, we have developed Baikal, a data
analytics infrastructure that is able to acquire high-quality
information from these digital health sources and integrate
it with long-term outcomes to support biomedical
research and the implementation of precision medicine
initiatives. Using this infrastructure, which is built on
open-source technologies, we have also developed Donatello,
a framework that can support computational
phenotyping, predictive and prescriptive model training,
and the deployment of clinical analytic models within a
production enviornment. We have used these platforms
to assess real-world data quality, develop computed phenotypes
in cardiology and oncology, and implement artifical
intelligence methods to identify and predict outcomes
in oncology care. The use of such platforms will
allow us to scale these approaches and ensure high quality
data and analytics to advance precision medicine
