Data and the World in Which I Live
There’s a mass, she signed, in your colon.
I watched her, as she explained to my friend the diagnosis. Colon cancer. I watched the dynamic, my friend asking for more detail, the doctor struggling with how much to share, and the interpreter attempting to reconcile the vast differences in between.
American Sign Language exists in 3D – it allows the use of space in ways linear languages cannot even fathom. For example, if we knew, we could detail very easily the exact location in the colon of the mass. We could give clear idea of its size embedded in the word for mass, nodule, or tumor. We could even include, with minimal changes to the script, an idea of how hard or gelatinous the mass itself was. English speakers rarely consider these details. Often, tumor is enough and a proximal location is sometimes sought, but not frequently.
This event happened years ago. In the time since, I’ve stopped interpreting, moved into data work, and helped numerous others close to me navigate their health affairs. The gift of ASL allows me to think differently about space, abstraction, and how we communicate.
I’m often handed sets of data. Sometimes, it comes in the form of a tidy connection already plumbed into Tableau. More frequently, I’m handed some level of access to the database and a fishing rod to find what I need. On good days, there’s someone else to help me navigate the lake. On bad days, I’m in a boat by myself, sometimes fishing at night.
Data is an abstraction of our world, an indication of what we value because somewhere, someone thought to collect it and archive it in some (hopefully) meaningful manner. My job is to make it accessible and understandable.
Some things aren’t comprehensible though, such as our understanding of cancer, the brain, and so forth. Oh, sure, we understand a lot, but a lot continues to change. For example, the lymphatic system doesn’t exist in the brain, but scientists may have found something else equivalent, the glymphatic system, as its being called. I don’t know about you, but glymphatic was never in my high school health texts (don’t you dare tell me I’m old).
These systems are essential to get the junk out. Cells die and they need to go. Enter the need for lymphatic/glymphatic systems. We seek to understand our biology and ourselves, but these discoveries are still evolving.
Just like biology, data is highly abstract. It becomes more valuable when we look at it in integrated ways (ethics call-out here). In the last decade, we’ve seen significant changes in how we handle data, what we expect from data, and – most critical here – how we analyze it.
The World My Clients See
I worked in healthcare when the (US) Affordable Care Act came into play. I was part of conversations with administrators trying to find patient populations that were likely to readmit.
Communication is an essential part of healthcare. We have to understand symptoms, treatment, and what’s reasonable to expect after episodic care. Medication is a key complicator in care. We expect patients to sometimes follow an almost robotic means of administering it. If you take something 4 times a day, 6 hours seems reasonable until you factor in sleep. Do you really want to stay up until midnight or wake up at 6 to comply with a schedule? We don’t see this unless it’s presented in a way that supports it.
Here’s what my clients saw: there’s a million reasons why people end up back in the hospital. Many of these can be prevented. My clients don’t often see the data in a digestible format. Most don’t care directly about the data. No, they care about the people – the patients – under their care. The data helps, but often, it’s an almost mystical abstraction and, more frequently, distraction, if they’re the ones inputting it.
We have to keep this in mind as data people. Our clients don’t care if we call it analyzing the data, doing data science, or putting in machine learning. Most care about the result and the proof we’ve done it right. Sure, some may be curious, but I haven’t found those to be the norm. Most people are too focused on their side of the work. And, we data people, are here to make that easier, better, or more accurate in some way.
My clients often know the end result they want. They know what fits into their processes and what doesn’t. They’re tired – exasperated, really – of solutions that aren’t. You know, the workflows that add more work…we have them too and we hate them. Our clients are no different.
A Glimpse into my World
A friend and colleague recently shared this video with me. Take a bit, and watch it. I was blown away by this.
In this video, Dr. McCormick discusses how we problem-solve mazes. For some reason, too many of us start the maze at the beginning, where it’s most complex. At some point, a chunk of us learn to start at the end. And, that’s how he approaches converting high density images to data – with the endpoint in mind.
Perhaps it’s a quirk, but I’ve always started mazes at the end. That translates to data as well. I can see beyond its abstraction to the tangibles. What if it could help us find reasons people admit? Or common locations of lung cancer (see his research) that improve screening and outcomes?
Too often, we approach data in a very soloed fashioned. We need a data warehouse. We want some dashboards. Maybe we’ll throw some data science at this.
We’re approaching a convergence of data needs. We’re facing abstraction that requires a multitude of skills. For example, Dr. McCormick has his degree in biomedical engineering and uses Python to work with this data. From images, he’s able to build a geospatial assessment of tumor locations. I see this, I think of my experiences, of ASL and conveying information about the tumor. I see shape, density, and the world in which I know from ASL rendered in data.
I could potentially restructure this data, parse it through Alteryx, and do my own analysis in Tableau. If we wanted to examine this longitudinally, we’ll need a database and we’ll probably want to pull in some epidemiologists to get a look at social influences on health. We could begin to understand so much more.
The world in which I live doesn’t draw clear lines around who does what with data. Instead, it’s interdisciplinary and fuzzy. Developments in biomedical engineering (I did not know this was a thing) excite me. We need to ensure data work remains a team sport and draws from a wide variety of influences and expertise to hold true to ethics, fair use, and accuracy. It takes into account the lives and stories of people and helps make the world a bit clearer and hopefully easier.
