Abstrait

Methodology of math-physical medicine (GH-Method)

Gerald C. Hsu

This paper describes the math-physical medicine approach (MPM) of medical research utilizing mathematics, physics, engineering models, and computer science, instead of the current biochemical medicine approach (BCM) that mainly utilizes biology and chemistry. Math-physical medicine starts with the observation of the human body’s physical phenomena (not biological or chemical characteristics), collecting elements of the disease related data (preferring big data), utilizing applicable engineering modeling techniques, developing appropriate mathematical equations (not just statistical analysis), and finally predicting the direction of the development and control mechanism of the disease.  

Methodology of MPM on Diabetes Research: Initially, the author spent four years of self-studying six chronic diseases and food nutrition to gain in-depth medical domain knowledge. During 2014, he defined metabolism as a nonlinear, dynamic, and organic mathematical system having 10 categories with ~500 elements. He then applied topology concept with partial differential equation and nonlinear algebra to construct a metabolism equation. He further defined and calculated two variables, metabolism index and general health status unit. During the past 8.5 years, he has collected and processed 1.5 million data. Since 2015, he developed prediction models, i.e. equations, for both postprandial plasma glucose (PPG) and fasting plasma glucose (FPG). He identified 19 influential factors for PPG and five both wave energy theories, he extended his research into the risk probability of heart attack or stroke. In this risk assessment, he applied structural mechanics concepts, including elasticity, dynamic plastic, and fracture mechanics, to simulate artery rupture and applied fluid dynamics concepts to simulate artery blockage. He further decomposed 1,200 glucose waveforms with 21,000 data and then re-integrated them into 3 distinctive PPG waveform types which revealed different personality traits and psychological behaviors of type 2 diabetes patients between two variables, he used spatial analysis. Furthermore, he also applied Fourier Transform to conduct frequency domain analyses to discover some hidden characteristics of glucose waves. He then developed an AI Glucometer tool for patients to predict their weight, FPG, PPG, and A1C. It uses various computer science tools, including big data analytics, machine learning (self-learning, correction, and simplification), and artificial intelligence to achieve very high accuracy (95% to 99%) mg/dL and A1C is 6.5%.