Computational Intelligence in Early Diabetes Diagnosis: A Review

摘要: The development of an effective diabetes diagnosis system by taking advantage computational intelligence is regarded as a primary goal nowadays. Many approaches based on artificial network and machine learning algorithms have been developed tested against datasets, which were mostly related to individuals Pima Indian origin. Yet, despite high accuracies up 99% in predicting the correct diagnosis, none these reached clinical application so far. One reason for this failure may be that diabetologists or investigators are sparsely informed about, trained use of, tools. Therefore, article aims at sketching out outline wide range options, recent developments, potentials focus supervised unsupervised methods, made significant impacts detection advanced stages. Particular attention paid show promise improving diagnosis. A key advance has more in-depth understanding theoretical analysis critical issues algorithmic construction theory. These include trade-offs maximizing generalization performance, physically realistic constraints, incorporation prior knowledge uncertainty. review presents explains most accurate algorithms, discusses advantages pitfalls methodologies. This should provide good resource researchers from all backgrounds interested intelligence-based allows them extend their into kind research.