Sparse Balanced SVM Based Detection of Type2 Diabetes Using FIIMG Dataset

An adequate Type 2 Diabetes unified administration system and regular timely checkup has key role in treatment of Type-2 Diabetes at initial stages. In Recent years there is rapid increase of evolution of Machine learning technique and FIIMG Dataset which is category of Electronic Health Record. Over fitting, Model interpretability and computational cost are the challenges while managing these much of information. Based on these challenges, we proposed a Machine Learning technique called Sparse Balanced Support Vector Machine (SBSVM) Based Type 2 Diabetes discovering by using Electronic Health record dataset named FIMMG dataset. We have collected data for Type 2 diagnosis from uniform age group that related to Electronic Health records such as exemptions, examination and drug prescription. Machine Learning and Deep neural networks are mainly used in solving task. Results proved that Sparse Based SVM provide better predictive performance and computation time when compared to techniques that are present in existing system. To increase model interpretability, we introduced induced sparsity which manages data which have high dimension.