Looking at Data Science Through a Personal Wellness Lens

Being a CEO of a bootstrapped innovation agency comes with its fair share of stress. That, coupled with raising a two-year-old, day-to-day running of the household, cooking meals, and doing laundry - it is a lot of work and left me exhausted at the end of every day. 

I had stopped paying attention to my wellness even while advocating that my team put their wellness and families at the top of their priority list. I gained a lot of weight, weighing the most I ever had. I experienced terrible insomnia and a host of other issues.

Realizing that my own wellness was suffering, II set out to change the trajectory. Being a software engineer and having a love for data and analytics, I invested in several tools I could use to capture data about my health.

Here is a list of my favorites:

Lumen measures CO2 levels in your exhaled breath to determine whether your body is burning primarily carbohydrates, fat, or somewhere in between. 

Libre2 is a device for non-diabetics that one wears on the back of their arm to track glucose levels ongoing. Using this data, I can tell how different foods affect my glucose levels throughout the day and helps me make smarter dietary decisions. It also helped me kick my awful refined sugar addiction.

Apple Watch to track activities like weightlifting and runs - calculating calories burned and time in heart rate zones and VO2 Max - the maximum oxygen consumption rate attainable during physical exertion - a good indicator of overall cardiovascular health.

Oura Ring tracks blood oxygen levels, body temperature, heart rate throughout the day, and sleep patterns, including time awake, REM, and light sleep. There is an overlap in the sensors with the Apple Watch, but the ring is much less noticeable throughout the night.

WeightGuru scale to track body fat, water weight, muscle mass, and BMI. I use this information to track trends.

So far, I’ve learned a lot about my body, but one thing stood out like a sore thumb. At night, I fell out of deep sleep often; when I did, my blood sugar had dropped below 60 mg/dl, nearing dangerous waters. What I was experiencing unknowingly for some time was nocturnal hypoglycemia. 

Access for the general consumer to track their biometrics is becoming increasingly common, making the data more accurate. It provides a massive amount of information that one can use to understand better how one's body operates and what changes one can make to improve their overall health.

Unfortunately, there are few tools for analyzing the data holistically. Even Apple Health isn’t all that helpful in this aspect. I am optimistic that will change.

From the Individual to the Population

Imagine, by my permission, the aforementioned data is shared as part of a broader pool of anonymized data for researchers. This includes the foods I consume, my macros, and other medical information like pharmaceuticals I take or existing medical conditions. 

A few things could happen - data science can help find anomalies in people with similar conditions, glucose patterns, and meals. Another could find patterns that are early indicators of conditions so that individuals can be notified about their risk.

This is in the range of possibilities but does beg the question: how reliable is the data in aggregate? This is a significant concern in data science. Data must be as clean as possible to make the best predictions, or else false predictions and inaccurate results will surface.

Uncover Opportunities in Your Organization

To uncover opportunities to leverage data science in your organization, consider considering your data sources (databases, spreadsheets, or unstructured documents like pdfs or Word Docs) as the data produced from Lumen, Libre2, or Apple Watch telemetry. 

Another way to put it is that you want to think about the data you have and what questions you want to ask of it. In our population health example, the questions are “what patterns were exhibited by individuals that ultimately are diagnosed with diabetes?”

If you’re a supply chain company, you might want to ask, “what is causing the bottlenecks and delays in shipments of drug X from the manufacturer to a hospital?”

The more focused your questions, the better likelihood that a data scientist will be able to provide you with the answers. 

Tell a Story

Data is powerful but becomes actionable when people can relate to it - versus numbers on a screen. This can be achieved by creating a narrative that tells the story of the data and provides real-world examples of the problems that can be solved and the next best actions to leverage the findings. Data science initiatives don’t end with the analysis, they should end with actions like addressing supply chain bottlenecks or predicting indicators of a diabetes diagnosis.

In my personal case, I asked, "How does my glucose level impact my general health? The narrative is about my experiencing nocturnal hypoglycemia. This is the story I plan to share with my primary care physician to confirm my suspicions clinically, and the data drastically impacts the narrative I will use with my doctor.

Granted, data science was optional for me to understand my personal data. Still, you can now see how aggregated data could be incredibly powerful and benefit an entire population.

What’s Next

In my next article in this data science series, I’ll break down all of the various sub-disciplines that fall under the data science umbrella. For example, artificial intelligence, natural language processing, anomaly detection, linear regression analysis, machine learning, etc. in a format accessible to the layman.

Let's talk if you are part of an organization, particularly in health tech, interested in building digital products and finally realizing digital transformation. In my experience, internal innovation programs tend to fail nearly 90% of the time. At EIG, we have a process that can help avoid the pitfalls and help you actually deliver new and meaningfully useful products.