Submit Manuscript  

Article Details

Spline Longitudinal Multi-Response Model for the Detection of Lifestyle-Based Changes in Blood Glucose of Diabetic Patients


Anna Islamiyati*  


Background: Blood sugar and lifestyle problems have long been problems in diabetes. There has also been a lot of research on that. However, we see that diabetic patients are still increasing even though many patients are not aware of the start of the disease occurrence. Therefore, we consider it very important to examine these two main problems of diabetes by using a more flexible statistical approach to obtain more specific results regarding the patient's condition.

Objective: The form of data for type 2 diabetes patients is repeated measurements so that it is approached through longitudinal studies. We investigated various intervals of pattern change that can occur in blood glucose, namely fasting, random, and 2 hours after meals based on blood pressure and carbohydrate diets in diabetic patients in South Sulawesi Province, Indonesia.

Methods: This research is a longitudinal study proposing a flexible and accurate statistical approach. It is a weighted spline multi-response nonparametric regression model. This model is able to detect any pattern of changes in irregular data in large dimensions. The data were obtained from Hasanuddin University Teaching Hospital in South Sulawesi Province, Indonesia. The number of samples analyzed was 418 from 50 patients with different measurements.

Results: The optimal spline model was obtained at 2 knots for blood pressure and 3 knots for carbohydrate diets. There are three blood pressure intervals that give different patterns of increase in patient blood glucose levels, namely below 126.6 mmHg, 126.6-163.3 mmHg, and above 163.3 mmHg. It was found that blood sugar rose sharply at blood pressure above 163.3 mmHg. Furthermore, there are four carbohydrate diet intervals that are formed, which are below 118.6 g, 118.6-161.8 g, 161.8-205 g, and above 205 g. The result is that blood sugar decreased significantly at intervals of carbohydrate diet 161.8-205 g.

Conclusion: Blood glucose increases with a very high increase in blood pressure, whereas for a carbohydrate diet, there is no guarantee that a high diet will be able to reduce blood glucose significantly. This may be affected by the patient's saturation of a very high carbohydrate diet.


Blood glucose, blood pressure, carbohydrate diet, multi-response, nonparametric regression, spline longitudinal


Anna Islamiyati Department of Statistics, Faculty of Mathematics and Natural Sciences, Hasanuddin University, Makassar

Full Text Inquiry