CLN - Focus on Data Science
Sponsored By
In laboratory medicine, the strategic use of data can transform everything from operational efficiency to groundbreaking research discoveries. It can lead to insights in everything from identifying how to deploy phlebotomists more efficiently across a healthcare system to using statistics to identify patterns that lead to breakthroughs for better patient care.
But a large volume of data is only useful if you know how to use it. That’s especially true in clinical laboratories, which produce reams of data.
CLN spoke to Shannon Haymond, PhD, Senior Vice President and Chief of Pathology and Laboratory Medicine and Arthur C. King Professor of Pathology and Laboratory Medicine at Ann & Robert H. Lurie Children’s Hospital of Chicago, about what clinical laboratory directors should know about data, and how they can work with other departments across their institutions to make sure they’re using data to the benefit of both patients and the institution.
To me, data literacy is the ability to interpret and communicate data so that you can make decisions or inform others’ decision making.
It’s important because there are so many things we can be informed about with data, and having this awareness helps you make better decisions. If you can understand what happened or why it happened, it can help you determine what you should do next, or what’s likely to happen next. Those are all very powerful things that help you take the appropriate action.
I think this should be a fundamental concept. Data science is transforming many aspects of our life. Our laboratories are no different, because laboratory directors are decisionmakers and innovators who will increasingly apply data-centric approaches to operational, quality, and clinical workflows. Additionally, more of the applications and instruments we use today will rely on data-driven technologies. So, data analytics and data literacy are critical skills for now and in the future. We must continue to advance our ability to collect, analyze, and consume data so we can glean valuable insights from data and effectively incorporate predictive and prescriptive methods to help automate our decision-making.
There are many approaches to building pipelines for informatics and data analytics. One we found works best is a partnership with our institutional information technology group. That allows us to leverage institutional policies, processes, technologies, and people resources to do things most efficiently and, I think, effectively.
In this model, laboratory professionals are the experts on our workflows and challenges, and we can articulate our needs, and evaluate or develop tools and solutions. But we rely on our information technology or information management group and embedded analysts to help us implement those solutions within the hospital’s informatics frameworks and governance policies. This also allows for better integration, so we’re not limited to accessing only our own laboratory data or applications.
I’ve always encouraged people to start by finding out what is going on at their institutions already. There is an ongoing digital transformation happening in all industries, including healthcare. I’ve found that many institutions are viewing data as an asset, and they want to make sure people are maximizing that value. They’ve recognized that this requires resources and infrastructure and are investing in these areas. Laboratories can take advantage of work that is in process or that already exists at their institutions to get started.
Next, you have to find out who the decisionmakers are and then be able to articulate your use cases. How do you want to use the data? What data and tools do you have? Will these meet your needs? How is it going to impact your clinical and research initiatives? Once you know all that’s happening in the institution, then you can start to figure out how these projects are funded and prioritized. Is it through operational funds? Some mix of operational and research funds? Or other sources of funding?
I personally feel that data science or data analytics — this falls under computational pathology for us — is a necessary function and a specialty of laboratory medicine that should be a part of all departments. Computational pathology at Lurie Children’s is composed of bioinformatics, pathology informatics, data science, data literacy, and digital pathology. Just like the laboratory builds infrastructure for a point-of-care testing program, I feel every laboratory needs infrastructure for these computational efforts that is in addition to the traditional support we’ve had for laboratory informatics. Maturity in data science is becoming a basic tenets of running an efficient and cutting-edge clinical laboratory.
High-quality data is a very important part of informatics and data science. That’s why it is so important for clinical laboratories to collaborate with other professionals who have data science expertise. I’ve seen both scenarios: situations where people only have a laboratory medicine background but don’t have necessary technical data skills; and those where people have tremendous data science skills but don’t understand the laboratory medicine workflows or regulatory constraints. They’re missing that context or subject matter experience.
We need teams of people, composed of those who understand best practices for data collection, data management, and data preparation and those who understand the clinical laboratory’s workflows and regulatory considerations.
One of the major roles of ADLM is to help educate and build not only data literacy but also technical skills in laboratory directors and trainees. This way, they are poised to lead these types of efforts that are going to become foundational to laboratory medicine.
ADLM is also making sure to publish and disseminate cutting-edge science and innovation in this area, which has been very exciting to see. The other thing ADLM does well is work on advocacy. We’re making sure that members who have clinical laboratory expertise are part of larger initiatives to really help steward and optimize the use of data and its integration at a very broad and even national level for clinical care and research.
We wanted to deploy phlebotomists more effectively within our hospital to balance their blood collection workloads, particularly in times of staffing shortages. So, we built a model that would look at factors like the number of requests in a day and the types of patients to understand the difficulty or complexity of the phlebotomy required. With that data, we figured out how we could better utilize our people to improve patient care through more timely blood collections and thus faster result availability.
One area that is very promising is the question of augmenting human workflows, particularly in areas where we have a shortage of people or shortage of very specialized expertise. Are there ways we can improve access to substantial expertise, but also where we can better utilize our human resources? These kinds of solutions will enable us both to handle more samples and perform even higher quality testing more efficiently. In addition, there also are a lot of tedious or repetitive and low risk tasks that we should automate.
So, I see laboratory medicine using data science not only to provide new insights that guide care, but also to enhance the efficiency and effectiveness of laboratories themselves, both of which enable us to improve patient care.
Jen A. Miller is a freelance journalist who lives in Audubon, New Jersey. +X: @byJenAMiller