Computational Intelligence and Healthcare Informatics. Группа авторов
Читать онлайн книгу.
on accuracy, precision and recall is also presented with an emphasis on the future direction of research.
– Chapter 3 discusses various application domains in need of feature selection techniques and also the way to deal with feature reduction problems occurring in large, voluminous datasets.
– Chapter 4 presents a detailed analysis of the available ML and ANN models conducted vis-à-vis the data considered, and the best one is applied for training and testing the neural networks developed for the present work to detect and predict a disease based on the symptoms described.
– Chapter 5 presents an approach for heart sound classification using the time-frequency image texture feature and support vector machine classifier.
– Chapter 6 proposes a novel approach for selecting a prototype without dropout for the accuracy of the multi-label classification algorithm.
– Chapter 7 introduces an intelligent computational predictive system for the identification and diagnosis of diabetes. Here, eight machine learning classification hypotheses are examined for the identification and diagnosis of diabetes. Numerous performance measuring metrics, such as accuracy, sensitivity, specificity, AUC, F1-score, MCC, and ROC curve, are applied to inspect the effectiveness and stability of the proposed model.
– Chapter 8 proposes hyperparameter optimization for ensemble learners as it has a lot of hyperparameters. The optimized ensemble learning model can be built by tuning the hyperparameters of ensemble learners. This chapter applies a grid search and random search algorithms for tuning the hyper-parameters of ensemble learners. Three ensemble learners are used in this proposed work: two boosting models (AdaBoost and Gradient boosting algorithms) and one bagging model (Random Forest algorithm).
– Chapter 9 presents a detailed analysis of the different types of healthcare simulations—from discrete event simulation (DES) and agent-based methods (ABM) to system dynamics (SD).
– Chapter 10 focuses on the application of Wolfram’s cellular automata (CA) model in different domains of health informatics, medical informatics and bioinformatics. It also reports on the analysis of medical imaging for breast cancer, heart disease, tumor detection and other diseases using CA. Augmenting the machine learning mechanism with CA is also discussed, which provides higher accuracy, precision, security, speed, etc.
– Chapter 11 considers the global dataset of 204 countries for the period of December 31st 2019 to May 19th 2020 from the Worldometer website for study purpose and data from May 20th to June 8th is considered to predict the evaluation of the outbreak, i.e., three weeks ahead. Three of the most prominent data mining techniques—linear regression (LR), association rule mining (ARM) and back propagation neural network (BPNN)—are utilized to predict and analyze the COVID-19 dataset.
– Chapter 12 proposes a hybrid support vector machine (SVM) with chicken swarm optimization (CSO) algorithm for efficient sentiment analysis. Part-of-speech (POS) tagged text is used in this algorithm for extracting the potential features.
– Chapter 13 discusses the primary healthcare model for remote areas using a self-organizing map network.
– Chapter 14 proposes a real-time face mask detection approach using VGG19 from the video stream recorded using a webcam that achieved 100 percent training accuracy with logloss 0.00 and a validation accuracy of 99.63 percent with logloss 0.01 in just 20 epochs.
– Chapter 15 focuses on different types of machine and deep learning algorithms like CNN and SVM for skin disease classification. The methods are very helpful in identifying skin diseases very easily and in fewer time periods.
– Chapter 16 discusses a program developed for collecting heart rhythm, pulse rate, body temperature, and inclination data from patients.
– Chapter 17 describes a proposed automatic COVID-19 detection system that can be used as an alternative diagnostic medium for the virus.
– Chapter 18 presents an innovative approach for deriving interesting patterns using machine learning and graph database models to incorporate the preventive measures in an earlier state. A graph-based statistical analysis (GSA) model to study the COVID-19 pandemic outbreak’s impact is also proposed.
– Chapter 19 discusses the conceptualization of tomorrow’s healthcare through digitization. The objective of this chapter is to utilize the latest resources available at hand to design the case studies.
– Chapter 20 provides a systematic procedure for the development of the POS tagger trained on general domain corpus and the development of biomedical corpus in Hindi.
– Chapter 21 studies the concepts of neuro-linguistic programming (NLP) used in healthcare applications, and also examines the NLP system and the resources which are used in healthcare. Along with the challenges of NLP, different aspects of NLP are studied as well as clinical methods.
Finally, we would like to sincerely thank all the chapter authors who contributed their time and expertise and for helping to reach a successful completion of the book. Second, the editors wish to acknowledge the valuable contributions of the reviewers regarding the improvement of quality, coherence, and content presentation of chapters.
The Editors July 2021
1
Machine Learning and Big Data: An Approach Toward Better Healthcare Services
Nahid Sami* and Asfia Aziz
School of Engineering Science and Technology, Jamia Hamdard, New Delhi, India
Abstract
Artificial Intelligence has been considered the biggest technology in transforming society. The transformation is highly influenced by the tools and techniques provided by Machine Learning and Deep Learning. One latest discovery is Robotic Surgical Tools like Da Vinci robot which helps surgeons to perform surgeries more accurately and detect distances of particular body parts to perform surgery precisely over them using computer vision assisted by machine learning. The maintenance of health records is a tedious process and ML in association with Big Data has greatly saved the effort, money and time required while processing such records. MIT is working on intelligent smart health record learning method using machine learning techniques to give suggestions regarding diagnosis and medication. Deep mining techniques are used in medical image