Avdagic Aja, Begic Fazlic Lejla
Faculty of Medicine-Ludwig Maximilian University of Munich.
University of Sarajevo - Faculty of Electrical Engineering.
Stud Health Technol Inform. 2017;235:116-120.
The aim of this study is to present novel algorithms for prediction of dermatological disease using only dermatological clinical features and diagnoses collected in real conditions. A combination of the Adaptive Neuro-Fuzzy Inference Systems (ANFIS) and Genetic algorithm (GA) for ANFIS subtractive clustering parameter optimization has been suggested for the first level of fuzzy model optimization. After that, a genetic optimized ANFIS fuzzy structure is used as input in GA for the second level of fuzzy model optimization. We used double 2-fold Cross validation for generating different validation sets for model improvements. Our approach is performed in the MATLAB environment. We compared results with the other studies. The results confirm that the proposed model achieves accuracy rates which are higher than the one with the previous model.
本研究的目的是提出仅使用在实际情况下收集的皮肤病临床特征和诊断结果来预测皮肤病的新算法。对于模糊模型优化的第一级，已提出将自适应神经模糊推理系统（ANFIS）与用于ANFIS减法聚类参数优化的遗传算法（GA）相结合。之后，将遗传优化的ANFIS模糊结构用作GA中模糊模型优化第二级的输入。我们使用双重2折交叉验证来生成不同的验证集以改进模型。我们的方法在MATLAB环境中执行。我们将结果与其他研究进行了比较。结果证实，所提出的模型实现的准确率高于先前模型。
