Data scientists and clinicians face new urgency to collaborate
While the adoption of electronic health records has produced a flood of new data, finding the best way to apply that information to improving care delivery can be difficult because there’s often a disconnect between clinicians and data scientists.
This insufficiency of collaboration between medical professionals and data scientists results from two factors:
• An absence of education in medical schools that prepares clinicians to collect and use data, as well as to approach analyses from the framework of the scientific method.
• The fact that data scientists are often not exposed to the clinical projects in the medical community and lack insight into the data itself.
A recent study from the Massachusetts Institute of Technology’s Critical Data group concluded that this divide between the two stakeholders creates data waste, which translates to a slow uptake of innovation that might otherwise improve patient outcomes. And while there has been notable progress in increasing collaboration in some areas, the disconnect between clinicians and data scientists within healthcare organizations persists.
To manage the subtle nuances of how data is handled, collaboration and education within healthcare organizations needs to increase.
The gulf between clinicians and data scientists may seem wide and hard to traverse, but it can be accomplished when medical and healthcare organizations embrace a culture of collaboration. As Dr. Alvin Rajkomar, MD, noted in his paper on patient safety, “Health Care Data Science for Quality Improvement and Patient Safety,” there is a dire need for a clinician-data translator, an individual who can take clinical datasets and analyze them; however, the path for training the clinician-data translator just does not exist.
A data scientist may have the technical and scientific skills, but often lacks the specific subject matter expertise needed to develop a clinically practical and useful data project. The inverse is true for a clinician.
However, adopting a three-pronged “best practice” approach can bridge the data divide.
Incorporating data science into the medical curriculum
When I was completing my PhD, I frequently collaborated with pre-med students. Not surprisingly, the students enjoyed designing experiments, collecting and analyzing data, and learning how to test hypotheses.
It was great to see that our future physicians gained exposure to the scientific research methodology, however, I realized that it would be highly unlikely that these students would require this level of data analysis in their careers.
Several clinicians to whom I have spoken confirmed my assumption and said their medical training didn’t involve scientific research methods or statistics. The clinicians also agreed that training in these areas would go far in their fields to create a more data-friendly healthcare landscape. They suggested that if more clinicians understood the data collection and analytic techniques used in formal analyses, they then might be more inclined to trust the findings and use the results to help inform their daily decision-making.
As the MIT study suggests, it is important that medical education includes statistics and methods of experimental design. With fundamental instruction in these areas, clinicians will be more comfortable using healthcare data as a tool for finding solutions to various issues that they face, offering them insights that assist them as they strive to provide excellent patient care.
Fostering appreciation for data science
One major challenge to proper utilization of healthcare data is format, or data architecture. Clinical data systems should not only make sense to those using them, but should also be structured in ways that the data can be extracted and manipulated efficiently.
As natural language processing and text mining become more widely adopted, this type of data will be used to enrich data that is collected in other electronic systems. Furthermore, it is important for patient information and clinical insight to be shared across the organization, and it is imperative to empower clinicians to use technological tools that can lead to more informed decision making.
Such empowerment and information sharing can be achieved through organizational training. My job as a data scientist is to assist the healthcare industry in its quest to explore data, find insights, and give organizations a better understanding of how they may be able to use the information from the data to make decisions and produce practical results.
Regular educational opportunities within any provider organization can help align clinicians and data scientists—whether on methods to analyze complex datasets or deciding how and what new data should be collected. Many organizations seem to resist a “culture of data,” and it behooves all of us to change this through education and real-world projects that will illustrate just how powerful and meaningful the data can be.
Thinking outside the box with teamwork
A multi-disciplinary approach, in which data scientists and clinicians are paired up to understand each profession’s skill sets, can serve as a solid foundation for collaboration. As of late, we’ve seen creative approaches take place in educational institutions.
Earlier this year, Harvard put on its Harvard DataFest conference, a joint effort of multiple schools—including Harvard Medical—to improve understanding of data workflows, curation, visualization and more. Sometimes, when you simply listen to a person who is passionate about their work, the approach they teach you becomes more interesting, and more likely to be adopted.
As healthcare organizations continue to look for cost saving opportunities and ways to improve patient outcomes, data analytics will play an important role in achieving those goals. Clinicians and data scientists will need to work together to glean insights that provide their organizations with meaningful findings that they can use to drive improvements.
Efforts to communicate and collaborate better around data, while crucial at the educational level, must also be ingrained more into organizational culture to solve the data divide. As data continues to be the future, intelligent collection and analysis of data should be the goal that aligns healthcare organizations and brings them the greatest benefit.