A Hybrid approach for an analysis of diabetes and prediction using machine learning techniques

Healthcare data management and analysis applications are constructed to handle huge volume of data items. Clinical and analytical methods are applied for the Diabetes prediction process. The machine learning techniques are adapted to analyze the diagnosis data values. The classification methods are applied to discover the patterns and apply the patterns for disease prediction process. Late diagnosis of diabetes decreases the risk of coronary, kidney, nerve injury and blindness. The diabetes prediction is carried out with the medical bioinformatics analytics. The resampling methodology for bootstrapping is paired with the classification approaches for Naive Bayes, Decision Trees and KNN. Continuous Glucose Monitoring Systems (CGMS) are very essential to monitor the blood glycaemic levels. The potential estimation of glycaemic rates facilitates the avoidance of adverse hyperglycaemic or hypoglycaemic situations. To order to boost efficiency of diabetic treatment, glycaemic thresholds are taken into consideration. The prediction models have been developed for the same stage of patient diagnosis. The time series data analysis model is very efficient in the diabetes prediction and control applications. The multi patient glucose level data is applied to construct the prediction model with large volume of sample data. The prediction model is applied to forecast the blood glucose level for the other patients. The bootstrapping resampling technique and Support Vector Machine (BRSVM) are integrated to construct the prediction model and future level identification process. The benchmark diabetes diagnosis data values are used in the analysis.