In this post, I cover a system I created to rank my various meals in terms of their glucose response; it is like having a virtual Continuous Glucose Monitor (CGM) without actually having one on.
As I have covered in previous posts, I have had a CGM on and off for more than 8 years and have seen which of my meals gave me large spikes (not good) to the ones that gave me very low spikes (very good). This allowed me to optimize my meals, sometimes by making minor tweaks like replacing one fruit with another in my fruit salads to sometimes cutting off certain meals like the Subway veggie delight because of the excessive added sugar in the bread. The result of making these tweaks was a very significant reduction in my LDL-Cholesterol, Triglycerides and HbA1c levels.
However, it is not practical to always have a CGM on and so I wanted to create an algorithm that would effectively predict what glucose response any meal would have on me even before I consumed it. I am sharing some examples of the results in the image below what the actual CGM data looked like for the 2 hours before and after a few different meals I consumed. The color of the box with the food names (green, orange or red) is what my algorithm predicted while the color of the border on the CGM graphs is what the actual glucose response was. Ideally, if my predictions matched the actual results perfectly, the colors would be the same. As you can see, it does a reasonably good job for many of the foods but there is still room for improvement (the last food was actually a Red but my algorithm predicted it as Orange).
I should also note that because I have incorporated a 20 minute walk after each meal, my algorithm takes that into account. This means that if I didn’t do the walk after each of these meals, my glucose response will be worse and I would have to recalibrate my algorithm. This model will need to be tweaked for each individual based on having a CGM on at least once for the 10 to 14 day period.
Another point worth mentioning is that the algorithm is primarily focused on optimizing for glucose response but it is also important to get a well balanced meal in terms of carbohydrate, fat, protein and fiber content, especially with your goal in mind (weight gain or loss, reduction in LDL-C levels, etc.). I manage that in parallel so that I don’t end up with foods that are great in terms of glucose response but poor in terms of balance of the macronutrients and micronutrients that my body needs to meet my goals.
Now for some interesting takeaways:
- The Chipotle burrito bowl (graph #3) contains a salad base, black beans, chicken, veggies, spicy salsa, corn, cheese, sour cream, guacamole and is one of my best meals. It is well balanced in macros and the glucose response curve is amongst the best I have ever seen.
- One of my favorite meals is the California Crepe (graph #2) at Crepevine. It is a savory crepe with a salad on the side, chicken, sour cream, olives, avocado, salsa and some veggies. It is also nutritionally well balanced and is a Green in terms of glucose response.
- However, at Crepevine, I sometimes (often!) also indulge in a Sienna crepe (nutella, nuts, strawberries, whipped cream). I have this right after the California Crepe and this combination (graph #6) causes a spike, making it an Orange in terms of glucose response. Had I not walked for the 20 minutes, this would definitely have been a Red; something I will hopefully be able to demonstrate in a future post.
- The home cooked breakfast (graph #1) is one of the better meals while the other home cooked meal (graph #4) had some pani puri in it that caused it to be an Orange.
- Unsurprisingly, pizza is not a winning formula (graph #5) but interestingly, another home cooked meal (graph #7) was red. On closer inspection, it seems like the parathas were more of an issue than I expected. I will be repeating that food again to investigate it further.