- Advances in data analytics offer great potential for researchers to move beyond correlations to mechanisms and cause-and-effect relationships (Khalil).
- Incorporating patient-reported outcomes, clinical care records, and data from personal devices such as smartphones will further enrich the data available for mining (Lovestone).
- Patient consent to access and use data remains a thorny issue for citizens, researchers, and legislators (Bynum, Lovestone, Powell, Rocca).
- Researchers must respect the intentions of the persons from whom the data were derived, but at the same time ensure that data are used as widely as possible for the best possible purposes (Lovestone).
NOTE: These points were made by the individual speakers identified above; they are not intended to reflect a consensus among session participants.
The Century Cures Act, which, among other goals, aims to advance interoperability in health information technology as a pathway to improved biomarkers and therapies for AD and other diseases.
Beyond the potential of exploiting these clinical data, researchers now measure things with greater granularity, marry data across types, and combine these data with computational and machine learning
techniques to move beyond correlations to mechanisms and cause/effect relationships, said Iya Khalil, executive vice president and co-founder of GNS Healthcare. For example, computer algorithms combining anthropometric and laboratory measures are now capable of predicting risk of developing metabolic syndrome and model response to various interventions, providing the capability for personalized medicine. Similarly, companies such as GNS Healthcare are working to integrate multiple layers of data from AD patients to develop an algorithm that can predict who is at risk for AD as well as the optimal treatment protocols. These data go beyond simple clinical and laboratory measures to include linkage of genetic and molecular markers and imaging readouts to longitudinal clinical outcomes, providing insight into how things work from the biological level to the health care system level. Big data may also be useful as a means of providing real-world data about areas of unmet need, noted Khalil.
Simon Lovestone added that the ability to incorporate patient-reported outcomes, clinical care records, and measures from personal devices such as smartphones will further enrich the data available for mining. The key, he said, is to enable patients to have some control over their data. Tia Powell noted that much less research is done on low-income and minority groups, so working to build trust among these populations will be especially important.
However, a few participants noted that consent to access and use data from DHRs remains a challenge for citizens, researchers, and legislators alike. For example, Minnesota law requires patients to provide general consent before their medical records can be used for Institutional Review Board–approved research, said Walter Rocca. Regulatory decisions may also be open to reinterpretation over time. Julie Bynum said that an older regulation about access to information on substance abusers and substance abuse was recently reinterpreted to require removal of all substance abuse claims from claims data. The result of that change is that if a researcher tries to examine claims data to assess longitudinal change in risk of substance abuse, claims for a segment of high-risk individuals will be missing, compromising the results of the analysis.
Multinational studies may be especially impacted by regulations limiting the sharing and reuse of data, said Lovestone. As a research community, he said, we have a responsibility to respect the researcher’s intentions for data utilization, but ensure in tandem that data are used as widely as possible for optimal purposes.