Abstract: We have developed and launched a novel data management approach that readily collates, integrates, validates, and allows direct clinician visualization and complex analytic capabilities of all clinical and research data generated within The Children’s Hospital of Philadelphia (CHOP) Mitochondrial Medicine Frontier Program (MMFP).
We sought to develop a robust data integration tool that efficiently extracts and unifies updated medical, genomic, clinical and research data collected in all potential domains within a single database server. A key concern was maintaining data integrity without duplication or loss, regular streaming updates, selective accessibility to identified vs deidentified data, and connectivity between the electronic medical record (EPIC) and research databases (REDCap, Excel, OnCore, etc).
Our data integration solution adopts a data warehouse built in Alteryx. Alteryx serves as a data staging warehouse that pulls from all desired data sources to enable sophisticated analytics for supervised and unsupervised models, allowing novel algorithms to be developed by our in-house data integration bioinformatics team that support custom analyses of high dimensionality data. These integrated data are then directly exported to a commercial resource, Tableau, which is hosted in-house in a virtual machine (VM) readily accessible via the Web with password protection by clinicians, scientists, and researchers. Tableau supports data visualization in intuitive reports and charts, with user manipulation to rapidly gain desired insights and analytics within the Tableau environment. This unique data integration resource now enables efficient and rapid individual patient or cohort analyses of individuals readily grouped by specific genes, mutations, laboratory values, geographic factors, age, medications, HPO-based phenotypes, procedures and assessments, and outcome measures including survey results, among others.

Describe the new knowledge and additional skills the participant will gain after attending your presentation.: The ultimate goal is to enable prioritization of precision medical care and personalized clinical trial outcome measures that leverage direct clinician mining. This could be aided by machine learning approaches to predict mitochondrial diagnosis, prognoses, biomarkers, and therapeutic response based on complex arrays of molecular, biochemical and clinical outcomes.

Authors:

Ibrahim George-Sankoh (Presenter)
CHILDREN HOSPITAL OF PHILADELPHIA

LAURA MacMullen, CHILDREN HOSPITAL OF PHILADELPHIA
BATSAL DEVKOTA, CHILDREN HOSPITAL OF PHILADELPHIA
DEANNE TAYLOR, CHILDREN HOSPITAL OF PHILADELPHIA
REBECCA GANETZKY, CHILDREN HOSPITAL OF PHILADELPHIA
MARNI FALK, CHILDREN HOSPITAL OF PHILADELPHIA