Lessons from open health data: Three surprisingly human keys to our algorithmic future

Lessons from open health data: Three surprisingly human keys to our algorithmic future

In celebration of Open Data Week and the relaunch of NYC’s open data portal, Luminary Labs Research Strategist Michelle Shevin recently spoke at General Assembly on open data in health. The panel, moderated by NYC Chief Digital Officer Sree Sreenivasan, also featured Andrew Young of GovLab and Rain Henderson, former CEO of the Clinton Health Matters Initiative.

From innovation potential to privacy concerns, the panel surfaced three surprisingly human cornerstones of the “big data” revolution in health.

The human core of automation efforts: A major debate concerns whether automation enabled by data will end up replacing humans with robots—does the future even need us? But the human element is key to unlocking the power of data in health and other industries, since it is human effort training algorithms and underlying most automation. The persistent fantasy of better productivity through machine automation means the importance of teamwork, communication, empathy, and ethics is too often ignored, shaping not only what we build, but who we compensate. Sree Sreenivasan, NYC Chief Digital Officer, spoke of community engagement and open feedback mechanisms as core to New York City’s open data efforts—it’s not a “build it and they will come” endeavor. At our AI Labs Session last fall, Merck VP Sandy Allerheiligen emphasized culture change as an underappreciated key to building analytical businesses.

Narrative—data’s missing human half: What isn’t talked about enough is the importance of narrative, the (very human) stories we use to ascribe meaning to information. Intentionality, bias, metaphors, and myths define how data is used, as our choices embody value judgments. For example, implicit bias is central to why there are racial disparities in health care delivery and health outcomes—disparities we risk designing into automated systems. During the panel, Rain described an example from the Clinton Health Matters Initiative in which data combined with community engagement and storytelling led to radically improved health outcomes in the Coachella Valley, where a culture of silence had contributed to skyrocketing rates of HIV/AIDS.

New methods for a new understanding of reality: Our human perspective on the nature of reality and truth is shifting. Science and medicine are forced to grapple with the irreproducibility of 50% of peer-reviewed results, while facts are out of favor in politics. This is perhaps nowhere more visible than in health, where nutrition advice seems to shift every ten years, and where the legacy of germ theory is shaping up to be a crisis of antibiotic resistance. But before we accept living in a “post-truth society,” we should reevaluate our methodology. At the panel, Andrew Young described efforts at GovLab to develop new methods for training, evaluation, and data stewardship in the public sector to go along with the trend toward evidence-based policymaking.

In health and across industries, adapting to the open data revolution cannot just mean throwing a team of data scientists into the mix, nor should it mean thinking about automation as a replacement for human labor. As emphasized by the panel’s overwhelming focus on humanity, as we automate “cognitive drudgery,” we need to double down on creativity, ethics, security, and collaboration.

Image source: Pexels user Unsplash