Using Math-Physics Medicine to Control Type-2 Diabetes

The author has had type 2 diabetes, hyperlipidemia, and hypertension for 20 years. His health data in 2010 vs 2017 are listed as follows:

Weight: 205/172 lbs
Waistline: 44/34 inches
FPG/PPG: 185/380 vs 119/116 mg/dL
90-days daily glucose: 279/117 mg/dL
A1C: 10.0/6.1 %
ACR: 116/12 mg/mmol
Triglycerides: 1161/67 mg/dL
LDL/HDL: 174/28 vs 74/48
SBP/DBP: 250/113 vs 105/65 mmHG

He applied multiple disciplines, including advanced mathematics, big data analytics, cloud mobile computing, nonlinear engineering modeling, signal processing, artificial intelligence (AI) to conduct his research for 18,000 hours in seven years. He simulated the human organic metabolic system using 10 categories including four outputs (weight, glucose, blood pressure, lipid), and six inputs (food, exercise, stress, sleep, water, life pattern regularity) and 500 elements. He defined two new terms: metabolism index (MI) and General Health State Unit (GHSU) to obtain health status at any time. He collected >1M data and developed 4 prediction models with ~20 influential factors, weight, fasting plasma glucose (FPG), postprandial glucose (PPG), hemoglobin A1C, to provide an early warning.

He performed statistical analyses for durations between 900 and 1,450 days with 25,000 – 60,000 data to identify basic characteristics of glucose predictions. Primary factors, such as medication, weight, carbs & sugar, and exercise contributed >80% of glucose. Secondary factors, such as weather, measurement time, stress, sleep, illness, and traveling contributed <20% of glucose. Predicted results for weight, FPG, and PPG have reached 97% – 99% linear accuracy and >80% correlation with actual data.

Figure 1: Period of 4/1/17 to 8/31/17 results
Figure 2: Period of 9/1/17 to 1/28/18 results
Table 1: Health Examination results (2010-2017)