Overview

Details

This session contains two 40 minute panels.

Healthcare data, such as electronic health records and administrative claims, can be a valuable source of evidence on the effects of medical treatments. Recent advances in health information systems and data standards, and increased adoption of clinical systems, has led to a growth in data networks and in observational studies worldwide. However, such observational research for the purpose of causal inference is often criticized because of the potential for bias. Previously, we have proposed the LEGEND (Large-scale Evidence Generation and Evaluation in a Network of Databases) principles [2] embodying a new paradigm for observational research aimed at addressing these concerns. These principles prescribe pre-specifying which clinical questions to answer and the methods used to generate the evidence. Here we propose to extend this work, by pre-specifying how to decide whether the generated evidence can be trusted. We formulate a set of objective diagnostics, such as minimum power, minimum equipoise, and minimum covariate balance. Using the millions of results of the LEGEND Hypertension study,[1] we demonstrate the effect of systematically applying these diagnostics, both in terms of the number of results meeting each diagnostic, and the distribution of estimates for questions where we believe the null to be true.

This panel comprises five individuals who will discuss the various aspects mentioned above. George Hripcsak will highlight the need for improving the rigor of observational research. Martijn Schuemie will review the LEGEND principles and LEGEND Hypertension study. Patrick Ryan will review the proposed objective observational study diagnostics. Yong Chen will present the evaluation of the diagnostics using LEGEND Hypertension results. Marc Suchard will close with a discussion of findings, overall conclusions, and a path forward.

Digital transformation of healthcare services is often slow and fails to meet expectations. Yet most digital health research focuses on developing and evaluating new technologies, while translational methods are understudied. The International Centre for Translational Digital Health (ICTDH) is a collaboration between the Universities of Toronto (Canada), Melbourne (Australia) and Manchester (UK) that seeks to address these challenges. This panel session will: (1) raise awareness of the challenges associated with translation of digital health technologies to health care practice, (2) familiarise the IMIA community with ICTDH and its mission, (3) inform future activities in translational digital health of ICTDH and IMIA and (4) examine opportunities that emerge from international collaboration on this front. Four ten-minute presentations will be followed by forty-five minutes of audience discussion of how the IMIA community can work together to improve and accelerate the uptake and impact of digital health technologies through international collaboration.

Speaker