Photo by Bruno Aguirre from Unsplash.com

An Alternative Way to Figure Out Your Blood Glucose Level Using Your Own Smartwatch

Like many of my fitness enthusiasts peers, I recently got a smartwatch (Garmin Vivo Active 4). While I thoroughly love all the different fancy metrics that these watches offer which include sleep, heart rate, body battery, VO2 max estimation, what really caught my eye was the stress metric.

Garmin employs a set of algorithms that imputes your level of stress by using your Heart Rate`s variability. The less variable there is between your beats equals higher stress levels whereas a higher variability means less stress. What I found most amazing was that there was an observed correlation between my nutritional caloric intake and my stress level. In my previous post, I talked about the use of Continuous Glucose Monitoring devices and how they can be potent to monitor my impending insulin resistance. While I have an immense fervor towards their use given their increasing precision, ease of use and accuracy across the years, I also mentioned that there are certain factors that would preclude their use; prime of which is their high cost. The expense of these devices amounts to around 180–200 dollars a month which translates to around 8 dollars a day per patient. Hence, for the management of chronic metabolic syndrome, this can be a substantial burden for the common man. This is also very worsened given the fact that a disproportionate number of people suffering from diabetic occurs in the South East Asian region & developing countries where people might tend to have a lower purchasing power.

In this post, I would like to make a possible suggestion or more precisely a claim to overcome this problem. Does the Garmin stress level have a strong correlation with my CGM level and hence could smartwatches use the metric of HRV or stress to impute your glucose level?

Screenshot by the author: On this day, I had my one meal lunch. Nothing fancy but enough to spike my glucose level. I also did a fast 5 km run which is very glycolytic i.e release lots of glucose. Top picture: picture from Garmin Vivo Active 4. Bottom picture: Corresponding CGM level

General Feeding routine & Fasting routine

I am a strong advocate of fasting with periods of caloric restriction from Mondays-Fridays, I practice OMAD`s (One Meal A Day) which is generally lunch. Once a month, I practice 3 day-water-only-fast. No caloric intake but just water.

My observations.

3 days fast which was followed by a really carb-heavy meal

CGM level (right) & Stress level over three days, the peaks are very correlational. Stress level during my fasts are fairly low & so is my CGM level

Last day observations where I ended my fast with a carb-heavy meal. Both my CGM level and my Stress level are out of whack.

As observed above, there is a general correlation between by HRV values and my corresponding glucose level. More interestingly, I noted that a carb heavy meal tends to create a higher HRV and stress level than a normal lunch. Unsurprisingly, during my 3 days of no caloric intake, my stress values and HRV values were relatively unchanged with a low variance.

Implications and work required

If it is true that by some way your stress level is a proxy for your glucose level and that smartwatches could be used for them, what it implies is that there are possible methods to have glucose monitoring that is a low cost, easy CGM & minimally intrusive and invasive both to your lifestyle and your body. What would further help is together with AI and machine learning, your stress level parameters could be used as input to determine your corresponding CGM level and as a result help as preventative tools for insulin resistance and metabolic syndrome.

Combination of AI and Machine learning to impute CGM level

The necessary framework required would as a primitive but possible example involve the use neural networks and simple binary classification using logistic regression.

Screenshot by author: Deep learning specialization at Coursera which is moderated by DeepLearning.ai. Course taught by Andrew NG

Image adapted by author :Deep learning specialization at Coursera which is moderated by DeepLearning.ai

Concluding remarks.

CGM`s are amazing technologies and the engineers who came up with such a fine piece of invention have might deepest admiration. However, my hope is that they should be made more accessible to the general public as well as more affordable with them being used proactively and well as reactively to combat metabolic syndrome and insulin resistance. Repurposing smart wearable parameters might be a possible way ahead.

Caveat

Neither am I paid or endorsed by any companies. This is just my way of sharing my journey and experience of combatting metabolic syndrome. In no way, would I want you to believe that I am an expert but rather a steward of this idea.

This post is for general information purposes only and not intended for medical advice or treatment.

You can find me on Instagram and Facebook with the FB id here & here and leave me feedback on my posts.