Machine Learning Techniques for Healthcare Applications

Abstract

Autism spectrum disorder (ASD) is one of the common disorders in brain. Early detection of ASD improves the overall mental health, which is very important for the future of the child. ASD affects social coordination, emotions, and motor activity of an individual. This is due to the difficulties in getting self-evaluation results and expressive experiences. In the first case study in this chapter, an efficient method to automatically detect the expressive states of individuals with the help of physiological signals is explored. In the second case study of the chapter, the authors explore breast cancer prediction using SMO and IBK. Breast cancer is the second leading cause of cancer deaths in women worldwide and occurs in nearly one out of eight. In this proposed system, the tumor is the feature that is used to identify the breast cancer presence in women. Tumors are basically of two types (i.e., benign or malignant). In order to provide appropriate treatment to the patients, symptoms must be studied properly, and an automatic prediction system is required that will classify the tumor into benign or malignant using SMO and IBK.

Comparison of the accuracy of classification algorithms on three data-sets in data mining: Example of 20 classes

“…One reason for this difference can be related to the size of the data set used because the J48 algorithm performed well on large data sets, especially when the number of attributes was high, the tree would grow bigger and would require a lot of time for calculation (Ozer, 2008). In addition, this study is similar to that of (Rajamohana et al, 2018). According to them, SMO and kNN showed the best performance.…”

“…In this model space, k is a positive integer indicating the number of neighbors and can never be larger than the data set (Arbain & Balakrishnan, 2019). When k = 1, unknown samples in the model space are assigned to the class of the training sample closest to it (Rajamohana et al, 2018). The accuracy of the kNN algorithm is influenced by the magnitude of k because the large value of k reduces the effect of the noise variable in the classification and makes the boundaries between the classes less visible (Kabakchieva, 2013).…”

“…Bagging parallels the approach with multiple classifiers in estimation. The result of each class is passed through a selection process (Rajamohana et al, 2018).…”

“…They also found that the Random Forest algorithm was not suitable for their own work, although its performance seemed appropriate. Rajamohana et al (2018) used Naive Bayes, Random Forest, Bagging, and Multiboost classification algorithms for the early diagnosis of ASD disease. They determined that the algorithm reaching the highest accuracy rate was Multiboost with a 93.18% accuracy rate.…”

“…One reason for this may be the size of the data available. Because for (Rajamohana et al, 2018) since the amount of space required for the training set is linear, SMO can deal with large data sets. In various test problems, SMO is somewhere between linear and quadratic in the training set size because it avoids matrix calculations and SMO is the fastest algorithm for linear SVMs and sparse data sets.…”