Vitamin D is one of the most important vitamins for human health, and a lack of it is thought to be a precursor to many ailments. Diabetes is a chronic health condition that contributes to the development of several deadly illnesses. The goal of this study is to properly estimate vitamin D levels using parameters in blood test data without doing a laboratory blood test. A decision support system (DSS) is developed to make these predictions. Machine learning predictions produce outcomes incomparably faster than the laboratory vitamin D measurement technique. The model base of the DSS consists of the module developed with the machine learning system. This module is based on version 3.8.6 of the Weka program; support vector machines, the k-nearest neighbor algorithm, decision trees, naive bayes, artificial neural networks, and random forest classifier modules were applied. While the total number of patient blood samples used to train the system is 684, 406 of them are diabetes patient data. While the database is MS SQL, the user interface was prepared using Visual Studio 2017 and the C# programming language. The real data was used to test the validity of the developed DSS software. The validity of the software was tested with a total of 90 random data based on the entire data set, the diabetic patient data set, and the non-diabetic people data set. According to the results, the DSS software made correct predictions on 29 out of 30 data in the entire data analysis. While the DSS software made correct predictions in 30 out of 30 data in the analysis of patients with diabetes, it estimated Vitamin D deficiency status correctly in 27 out of 30 data in the analysis of non -diabetic patients. ORCID NO: 0000-0002-9607-8149
Anahtar Kelimeler: Decision Support Systems, Machine Learning, Vitamin D, Diabetes
|
