Since 7/1/2015, the author has utilized his collected data of finger pierced glucose readings 4 times daily, carbs/sugar intake amount, and post-meal walking steps for each meal to calculate the predicted glucose values. He then utilized his developed software calculated daily HbA1C values (the “daily finger A1C”).
Starting from 5/5/2018, along with finger glucose levels, he has been collecting 96 glucose data each day using a continuous glucose monitoring (CGM) sensor device until present day. Based on the collected CGM sensor glucoses, he further developed two extra HbA1C prediction models, the “sensor-1” A1C model using the combination of both average sensor glucoses and daily glucose fluctuations, and the “sensor-2” A1C model using the average sensor glucoses (eAG). Both sensor-1 and sensor-2 Predicted A1C models have utilized a simple but different conversion factor (CF) of the value for eAG/A1C.
This article presents the Comparison between the lab-tested A1C versus three predicted A1C: finger, sensor-1, and sensor-2.
In conclusion, both finger A1C and sensor-2 A1C models have yielded the same predicted HbA1C values of 6.6% as the lab-tested HbA1C value. However, the sensor-1 model produces a slightly higher A1C of 6.8% (103%) compared to the lab-tested A1C of 6.6% due to its heavy contribution (71%) from glucose fluctuation (GF).
In addition, all of these three predicted A1C datasets have reasonable high correlation (66%-68%) versus the lab-tested A1C dataset.
The objective is to provide some simple yet useful A1C prediction tool to other diabetes patients for their diabetes control efforts. If we can predict the future outcomes of A1C on a daily basis, then diabetes control will not be a difficult task.
Both glucose and HbA1C involve many influential factors. Although the medical community lacks a precise definition for the term HbA1C (mathematically), it loosely defines HbA1C as being the 90-days average glucose value. However, the actual life-span of red blood cells (RBC) range between 90 to 120 days, where some documents even stated as 115 days. In reality, a lab-tested HbA1C is also affected by many other non-biomedical influential factors, including but not limited to its operational procedures, possible human errors, testing environment differences (even the altitude of the laboratory), etc.
The author spends his time and efforts on developing several highly accurate HbA1C prediction models in order to provide an “early and preventive warning” to diabetes patients on a daily basis. Therefore, they do not have to wait until the actual lab-test day to find out their HbA1C value. By that time, it will be too late to make any modifications for past behaviors in order to control their diabetes.
The author strongly believes that an accurate prediction offers a better chance in preventing the disease, which is always superior to treating it, including medications, injections, surgeries, chemotherapy, or radiation.