Blood glucose prediction model for type 2 diabetic patients

Abstract

Objectives: 1. To compare means of calories burnt by exercise, compliance, eating behavior score, diabetes knowledge, age, BMI and Fasting Blood Glucose (FBG) between gender. 2. To find correlations between calories burnt by exercise, compliance, eating behavior score, diabetes knowledge, age, BMI and FBG. 3. To estimate Hierarchical Stepwise Multiple Regression Analysis (MRA) model to predict FBG. Method: A cross-sectional deductive survey by telephone interviewing and accessing clinical data were employed to investigate relationship between health behavior factors namely—calories burnt by exercise, compliance, eating behavior score, demographic data and FBG of 200 Type 2 diabetes patients at Saraburi Hospital 2008, randomly generated by computer. Results: Total sample size of 200 (100%) diabetes patients, mostly 138 (69.00%) were female, 62 (31.00%) were male with average age 59.34 ± 11.99 years, average BMI 26.79 ± 2.50, average calories burnt per week 2777.24 ± 2420.89, Average compliance score12.37 ± 7.13, and average FBG 161.25 ± 54.10. Cronbach’s Alpha coefficient of Sorofman’s Compliance scale for constructs “right time” and “right amount” were 0. 8157, and 0.8526 respectively and Auamnoy Eating Behavior Scale was 0.7915. ANOVA confirmed that FBG, compliance, diabetes knowledge, age, eating behavior score and BMI between male and female were not significantly different (p>0.05) however, calories burnt by exercise between male and female patients were significantly different (p<0.05). Pearson’s correlation confirmed that diabetes knowledge, calories burnt by exercise and age were inversely significantly correlated to FBG (r= -0.42**, -0.24**, -0.14* with p<0.01, p<0.01, p<0.05 respectively). MRA model proved that 5 factors; diabetes knowledge, calories burnt by exercise, gender, age, and BMI were the significantly predictors of FBG (Beta= -0.40, -0.25, -0.11, and -0.10 with p< 0.01, 0.01, 0.01, 0.05, 0.05, R 2= 0.26). Conclusion: Diabetes knowledge, calories burnt by exercise and gender were three significantly most influence factors for predicting Fasting Blood Glucose. The diabetes patients should be educated how to cope with diabetes and do the more exercise.