GH-METHODS

Math-Physical Medicine

NO. 033

Quantitative Analysis of Relationship Between Postprandial Plasma Glucose and Food/Meal (Math-Physical Medicine)

Corresponding Author: Gerald C. Hsu, eclaireMD Foundation, USA.

Abstract
The author has been diagnosed with three chronic diseases including type 2 diabetes (T2D), hypertension, and hyperlipemia.  Since 2010, he focused on T2D research to save his life. He collected and processed approximately 1.5 million data regarding his health and life details. In 2014, he developed a mathematical model of the metabolic system known as the math-physical medicine (MPM) approach by applying mathematics, physics, engineering modeling, and computer science such as big data analytics and artificial intelligence.  This paper focuses on the quantitative relationship between postprandial plasma glucose and food/meal.

Key Words
Type 2 diabetes, chronic diseases, metabolism, metabolic conditions, lifestyle data, artificial intelligence, postprandial plasma glucose, nutrition, food, meals, and math-physical medicine.

Introduction
The author used math-physical medicine to research and identify the quantitative relationship between postprandial plasma glucose (PPG) and food/meal.

Methods
Food is the most important factor of PPG, but it is also difficult to regulate eating habits.  He created an artificial intelligent (AI) based software to collect his meal data by utilizing optical physics, signal processing, mathematics, statistics, and machine learning.  He then developed a PPG prediction model by combining 6M food nutrition data from the United States Department of Agriculture (USDA) and his ~4,000 meal photos as his food database.  Each meal picture links with data, including nation, meal location, food type, menu/dish name, and nutritional ingredients.  The system can estimate consumed carbs/sugar amount and then predict PPG value prior to eating.

Results
He selected a period of 1,194 days (6/1/2015-9/7/2018) with 3,721 meals (including snacks) and ~100,000 data for his analysis.  There were 86 airline meals consumed during his 94 trips during this period.  The summary results are listed by both nation and meal location; then, they were sorted by PPG value with the format of PPG (mg/dL) & carbs/sugar (gram).

  • By Nation (Figure 1):
    Airlines – Cross nations: (137.3, 26.0g), Other Nations: (123.7, 19.8g)
    Taiwan: (123.0, 14.9g)
    USA: (117.6, 13.0g)
    Japan: (117.4, 15.6g)
    Canada: (115.1, 14.3g)

In summary, he had 58% of meals within the USA and 42% in other nations.

  • By Location (Figure 2):
    Airlines: (137.3, 26.0g)
    Supermarket: (130.3, 25.7g)
    Individual Restaurant: (127.7, 20.6g)
    Chain Restaurant: (121.2, 11.7g)
    Home Cooking: (113.8, 11.5g)

In summary, he had 59% of meals at home and 41% outside.

Figure 1: Nation Summary Results
Figure 2: Eating Location Summary Results

Conclusion
The analysis (Figures 3, 4, 5) and predicted PPG model (99.9% accuracy) assisted the author to lower his PPG from 279mg/dL to 119mg/dL.

Figure 3: Detailed Meal Analysis
Figure 4: Using AI Glucometer to Predict Glucose Value via Meal Photos
Figure 5: Accuracy Comparison between Nutritional Intelligence (NI) and Artificial Intelligence (AI)