Название: Artificial Intelligence in Healthcare
Автор: Lalit Garg, Sebastian Basterrech
Издательство: Springer
Год: 2022
Страниц: 157
Язык: английский
Формат: pdf (true), epub
Размер: 17.6 MB

This book highlights the analytics and optimization issues in healthcare systems, proposes new approaches, and presents applications of innovative approaches in real facilities. In the past few decades, there has been an exponential rise in the application of swarm intelligence techniques for solving complex and intricate problems arising in healthcare. The versatility of these techniques has made them a favorite among scientists and researchers working in diverse areas. The primary objective of this book is to bring forward thorough, in-depth, and well-focused developments of hybrid variants of swarm intelligence algorithms and their applications in healthcare systems.

Automated Approaches. Systems that depend of training of data on Machine Learning techniques. They have fast performance but sometimes they have problems of overfitting and underfitting. They don’t rely on human set of rules but on the Machine Learning techniques. Firstly, during the training process, the model learns a specific input present in the form of text provided by the training data and gives output based on the learning the made from the data. Some of the algorithms that use machine learning techniques are Naive Bayes, Support Vector Machine (SVM), Linear Regression. Naive Bayes: Naive Bayes is an uncomplicated and straightforward but efficacious classification algorithm. This is widely used for document level categorization. It simply calculates the probabilities of groups passed as a test document by utilizing the collective probabilities of words and categories. It is optimal for specific problems with highly dependent characteristics. Naive Bayes technique gives computationally fast results during taking decisions. Large amounts of data are not required before training can begin.

Random Forest: This is a method that trains various decision trees. Every single tree is trained by taking a stochastic subset of the vector. Each decisions of tree are joined using a voting process that delivers the outcome. The main objective is to reduce the cost function that calculates the performance of the trees.