An Automated Algorithm To Calculate The Mean Amplitude Of Glycemic Excursion Based On Integer Nonlinear Programming

The blood glucose monitoring is an important part in the management of diabetes.The data of blood glucose monitoring can help the doctor assess the extent of the glucose metabolic disorder,make hypoglycemic scheme,reflect hypoglycemic therapy effect and guidance for treatment of adjustment in patients with diabetes.In recent years,the technology of continuous glucose monitoring(CGM)has been continually improved and optimized.It has been applied gradually in clinical research and diabetic prevention and treatment.Compared with the traditional monitoring method,CGM can provide a continuous data and reliable information throughout the monitoring of blood glucose level.The results of monitoring can more fully and accurately reflect the characteristics of glycemic variability.Therefore,in the clinical studies,it has put forward a variety of index to assess glycemic variability based on CGM data.One of the most representative glycemic indexes is the mean amplitude of glycemic excursions(MAGE)which first described by Service et al.in 1970s.MAGE is the average amplitude of upstrokes or downstrokes that are above a threshold equal to the standard deviation(SD)of measurements for an entire 24-h period.Although several issues of the index have been still disputed recently,and while there are still unresolved questions as to its ability to predict the adverse consequences of high glycemic variability,MAGE is recognized as the "gold standard" for reflecting the fluctuation of the blood glucose level at present.However,in clinical practice,the value of MAGE cannot directly calculate by the analysis program that in the continuous glucose monitoring system(CGMS).It requires manual method to filter the data for analyzing.For large amounts of the CGM data(surveillance at 5-min intervals generates 577 observations over a 48-h period),the task of calculating a MAGE becomes tedious and prone to operator error when the patient data increases.This made the computation need to consume large amounts of time,which reduces the clinical research efficiency.In addition,the clinical researchers who calculate and analyze the index must have relevant professional training to ensure the accuracy of the data screening.If the researchers lack of experience,there might be some errors between the calculation result and the real value.Thus,the accuracy of the data screening is influenced by the experience of the researchers.In this paper,an automated algorithm based on integer nonlinear programming(INLP)method is proposed to calculate the mean amplitude of glycemic excursion.First,the mathematical model is established for the problem that how to automatic calculate and analyze the MAGE index.Second,the index computation problem can be transform to an integer nonlinear programming problem by constructed the objective function and the constraint conditions.Third,the integer nonlinear programming problem is solved by the differential evolution algorithm which for a global optimum point in an n-dimensional hyperspace.In the last step,the result of the INLP can be used to calculate the MAGE index.The algorithm which proposed allows to automated analyze the CGM data and to avoid the imprecision associated with the traditional approach for MAGE assessment in clinical practice.The proposed algorithm can compute MAGE accurately and is expected to be applied in clinic.In addition,a standardized computation computer program is also developed in this paper which by using Visual C#2008.It makes the automated algorithm can actually be applied in clinic to solve the problems in traditional approach for MAGE and improve the efficiency of the clinical scientific research work.The thesis consists of six chapters.Each chapter includes detailed text and graphics instructions.In the algorithm verification tests,the form of calculation results by algorithm is given.Chapter 1 Introduction:Mainly expounds(1)the research background of study,including blood glucose and its control mechanism,the symptoms of diabetes and main classification,the harm of diabetes,the significance of blood glucose monitoring and the basic form,continuous glucose monitoring technology and blood glucose fluctuation and harm of long-term excessive;(2)the research meaning and purpose of study,including importance of mean amplitude of glycemic excursion and the main problem to solve;(3)the structure arrangement in this article.Chapter 2 The research status of algorithm to calculate the mean amplitude of glycemic excursion:Mainly introduces(1)the artificial calculation method of mean amplitude of glycemic excursion;(2)the research status of automated algorithm for calculating the mean amplitude of glycemic excursion;(3)problems existing in the method at present.Chapter 3 An automated algorithm to calculate the mean amplitude of glycemic excursion based on integer nonlinear programming:Introduced(1)nonlinear integer programming;(2)amplitude of glycemic excursions;(3)analysis the problem for solving;(4)mathematical modeling;(5)differential evolution algorithm;(6)solve INLP problem by using differential evolution algorithm;(7)data preprocessing;(8)pseudo code of algorithm.Chapter 4 Algorithm verification and results analysis:Introduced(1)algorithm validation,including chart observation validation and statistical data validation;(2)calculation result analysis,including correlation analysis and Bland-Altman plot.Chapter 5 The software application of algorithm:Detailed introduces the MAGE parameters computer aided analysis software(v 1.0)which developed by Visual C#2008 based on the idea of the proposed algorithm.The software includes four main function modules.In addition,the function of calculation MODD index which added to the software is also introduced.The software of algorithm can really make the algorithm applied to clinical practice.Chapter 6 Summarize and prospect:Generally summarizes and discusses the study in this thesis.The proposed algorithm simplified the process of analyzing MAGE parameters.It makes people who without relevant professional training can also calculate parameter values simple and accurate through the software.Thus,it makes the patients with diabetes monitoring their fluctuations by themselves become possibly.And because of the speed of parameters calculation by software is very fast,the analysis of the dynamic blood glucose monitoring is no longer need to wait until 72 hours or longer,greatly improve the efficiency of calculation the parameters.By using the automated algorithm to calculate the mean amplitude of glycemic excursion based on integer nonlinear programming,it can be faster and more accurate to get MAGE parameter values form CGM data and greatly improve the clinical research work efficiency.