The power of data in laboratory medicine

The Association for Diagnostics & Laboratory Medicine’s (ADLM’s) new data science in laboratory medicine certificate program is designed to give clinical laboratorians an edge in a field increasingly driven by data and analytics. Participants will learn how to set up effective data governance structures; request and process lab data appropriately; address issues of data integrity, accuracy, and structure; and evaluate when and how to use machine learning approaches.

Clinical Laboratory News recently spoke to Shannon Haymond, PhD, DABCC, FADLM, former ADLM president and faculty member of the data science certificate program, about this exciting new educational offering and the important role of data science in today’s labs.

Why and how did you end up making data science one of your areas of expertise?

As a clinical lab director, I frequently encountered situations where I felt like we had data that could guide me — but I didn’t have the right access or tools. 

Data science wasn’t really part of my training. As a clinical chemistry fellow, I learned basic statistics, but beyond that I didn’t have the skills I needed. In 2015, when I saw that my dear friends Stephen Master, MD, PhD, FADLM, and Daniel Holmes, MD, FRCPC, were teaching a course on the R statistical programming language, I enrolled. The course, called “Breaking up with Excel,” focused on the power of using R for data preparation, analysis, and visualization, particularly in cases where Excel is unsuitable or limited. It helped me to see that this type of tool and skillset was what I was missing and needed.  

After recognizing that my own laboratory could benefit from using data science, I convinced my boss to make it an area of emphasis across the department. I decided that, if I was going to lead this effort, I needed to expand my education. I enrolled in a master’s program in predictive analytics, where I learned a bunch of stuff I didn’t know I needed to know — about databases, data visualization best practices, and more. 

What really compelled me from that point forward was that knowledge that more people in our field could benefit from learning about data science. This isn’t widely covered or required in current undergraduate clinical laboratory science, PhD clinical fellowship, and pathology residency programs. There’s a pathway for MDs to formally build these skills through informatics fellowships. Increasingly I hear of PhDs completing data science-focused master’s programs, like I did. Given the interest from the community and a lack of available options focused on application in lab medicine, I felt it was important for ADLM to offer a certificate program on this topic. 

What is one way in which your lab is currently using data science to improve testing and/or patient care?

In 2017, we started on a journey to use data science and analytics more broadly throughout the department. We built a culture where people now expect to have data and visualizations to guide their decision-making and problem solving. 

We also have projects under way in genomics interpretation, where we’re looking to augment and automate workflows with artificial intelligence (AI) and large language models (LLMs). And we’re helping our bioinformatics staff and laboratory directors to extract data from clinical notes, which we can use to better understand which variants we should be prioritizing.

As data science advances, what is one thing you hope labs will be able to accomplish with it in the future?

I hope we find ways to develop AI tools like LLMs that provide accurate, reliable, safe, and predictable results. 

On the flipside of that question, is there something people are hoping data science will accomplish that you believe is unrealistic?

I don’t think so. I tend to be pretty optimistic. I do think there’s an underappreciation for the limitations of some of these technologies. We have a lot of work to do to make them safe and reliable. 

What hurdles do you see facing the development of data science in lab medicine over the next 5 to 10 years?

Cybersecurity challenges. There’s a big gap between what the solutions to those need to be and what vendors are currently able to achieve, which can make implementation difficult in a healthcare system.

Software costs and integration capabilities have also been barriers, but the rate at which some of these new tools are improving is quite impressive.

Another limitation has to do with building and upskilling the workforce. It’s not just lab directors who need to be able to interact with these technologies. The folks working in labs must also improve their data literacy and computational thinking skills. Just because people have more access to data doesn’t mean they know what to do with it. We need to stay on top of those skills. 

What will ADLM’s data science certificate program cover?

When we conceived the program, we based it on an expert consensus. In other words, we asked data science experts in our field “what do you think the key learning objectives are for lab medicine professionals and trainees?” We then developed our curriculum around the areas where there was very high agreement.

Topics covered include data management, data security, data visualization, and machine learning. The program both explains important concepts and teaches people how to apply them. For example, participants learn how to evaluate machine learning and how to formulate data requests. We really try to enable people to work with data themselves and communicate effectively with IT and data science experts.

What makes the program unique compared to all the other data science educational programs out there?

We feel great about the content of the program, which was vetted by numerous experts and covers the must-know skills for everyone. It was developed by people who are passionate about data and have a lot of experience teaching it to laboratory medicine professionals. 

While you can find data science education everywhere, and some of it is even free, this certificate program was designed specifically for the laboratory medicine community. The examples provided will be very relatable to this audience. This specificity will help learners feel more engaged with the material and improve the translatability of the skills they learn to their work. 

Once people complete this program, what should they be able to do in practice? 

We don’t expect people to walk away as experts in data science, but they will have a solid foundation of critical skills. After they finish the program, they will be more effective with accessing data, managing data, doing data visualization, and communicating about data. They will also understand the basics of how to assess and evaluate machine learning applications. 

What should their next steps be if they want to learn more?

ADLM has a lot of practical resources they can use to further their skills or practice what they learned. 

After taking the course, they can get started on their own projects. I also recommend attending the ADLM Annual Meeting and other events where people can learn more in a focused area of data science.

Based on your experiences, what advice would you give to lab professionals who are looking to get started in data science?

Learning this stuff is step one. What’s really going to propel you is using it to address your own problems. You need to have strong internal motivation to do this, since applying data science to your everyday work can be hard at the start when you are first gaining access to data and tools and building your skills. Starting with the areas you find most interesting is a great way to accelerate your learning and deepen your enthusiasm. It can even be fun! Data science skills are very sought-after these days, so mastering them can also boost your career. In short, being able to leverage data to solve problems is very powerful.

Jen A. Miller is a freelance writer who lives in Audubon, N.J. +Bluesky: @byjenamiller.bsky.social