An Overview and Analysis of Machine Learning Classification Algorithms in Healthcare

1. Introduction

These increasing technological developments of the 21st century have made our lives much more comfortable and stunning. Diverse data types have played the most critical role in this technological development. This advancement in technology in distinct fields uses different types of data depending on the task and judging the desired solution. Due to this promotion and spread of technology, the number of data in the world has increased, the use of data has increased, and a large amount of data is constantly being generated to solve various day-to-day practical problems. For example, in medical science, a considerable amount of data is being generated continuously as the types of diseases and variants of these types increase alarmingly. A wide range of software platforms and apps are available, and some are explicitly designed to provide information to academics (Lobe et al., 2020). Collecting data from a suitable medium in the healthcare industry with the help of some revolutionary inventions like X-ray Machines, MRI (Magnetic Resonance Imaging) Machine, and CT (Computed Tomography) Scanner, making the data useful enough, and selecting the appropriate data to solve the specific health issues makes it possible to solve any complex problem efficiently. Nowadays, everything around us is wired up to a data source since we live in the data era (Sarker, 2021). Data collected from various sources are used to solve any tricky and complex problems around us. The widely used machine learning algorithms make a significant, influential, and indispensable contribution to solving these problems.

One of the main reasons for using machine learning algorithms to solve a variety of challenging and complex biological problems is that these algorithms can work straightforwardly, using relatively less time and less space, and can work towards achieving the desired results by solving a variety of existing internal problems and in most cases the best results can be achieved. For instance, using machine learning algorithms, predictions of anxiety, sadness, and stress were generated relatively simply in medical research (Priya et al., 2020). It can also be used to simultaneously select the all-important and irrelevant instances for a classification problem (Saha et al., 2022). Machine learning algorithms are divided into categories depending on the type and nature of the problem, the specific course of action to solve the problem, the nature of the data collected to solve the problem, and the target variables. The most important types of machine learning algorithms are supervised machine learning algorithms, unsupervised machine learning algorithms, semi-supervised machine learning algorithms, and reinforcement machine learning algorithms. Although there are various reasons, the type of machine learning algorithm to be used mainly depends on the type of problem being solved and the type of result to be achieved after solving it. Supervised machine learning algorithms are relatively more used than other algorithms when considering existing problems. One of the best and most used fields of the supervised machine learning algorithm is medical science, where for any specific disease prediction, multiple supervised algorithms are used to be comparable and generate better results on every occasion (Uddin et al., 2019).