Machine Learning for Healthcare

Handling and Managing Data

Rashmi Agrawal, Jyotir Moy Chatterjee, Abhishek Kumar, Pramod Singh Rathore, Dac-Nhuong Le, Rashmi Agrawal, Jyotir Moy Chatterjee, Abhishek Kumar, Pramod Singh Rathore, Dac-Nhuong Le

• Publisher: Chapman and Hall/CRC
• Year: 2020
• Topic: Computer Science
• eBook ISBN:
  9781000221886
• Subtopic: Computer Science General
• Index: Computer Science

Machine Learning for Healthcare: Handling and Managing Data provides in-depth information about handling and managing healthcare data through machine learning methods. This book expresses the long-standing challenges in healthcare informatics and provides rational explanations of how to deal with them.

Machine Learning for Healthcare: Handling and Managing Data provides techniques on how to apply machine learning within your organization and evaluate the efficacy, suitability, and efficiency of machine learning applications. These are illustrated in a case study which examines how chronic disease is being redefined through patient-led data learning and the Internet of Things. This text offers a guided tour of machine learning algorithms, architecture design, and applications of learning in healthcare. Readers will discover the ethical implications of machine learning in healthcare and the future of machine learning in population and patient health optimization. This book can also help assist in the creation of a machine learning model, performance evaluation, and the operationalization of its outcomes within organizations. It may appeal to computer science/information technology professionals and researchers working in the area of machine learning, and is especially applicable to the healthcare sector.

The features of this book include:

• A unique and complete focus on applications of machine learning in the healthcare sector.
• An examination of how data analysis can be done using healthcare data and bioinformatics.
• An investigation of how healthcare companies can leverage the tapestry of big data to discover new business values.
• An exploration of the concepts of machine learning, along with recent research developments in healthcare sectors.

1

Fundamentals of Machine Learning

Rashmi Agrawal

CONTENTS

1.1 Introduction

1.2 Data in Machine Learning

1.3 The Relationship between Data Mining, Machine Learning, and Artificial Intelligence

1.4 Applications of Machine Learning

1.4.1 Machine Learning: The Expected

1.4.2 Machine Learning: The Unexpected

1.5 Types of Machine Learning

1.5.1 Supervised Learning

1.5.1.1 Supervised Learning Use Cases

1.5.2 Unsupervised Learning

1.5.2.1 Types of Unsupervised Learning

1.5.2.2 Clustering

1.5.2.3 Association Rule

1.5.2.4 Unsupervised Learning Use Case

1.5.3 Reinforcement Learning (RL)

1.6 Conclusion

References

1.1 Introduction

Machine learning is a discipline in which algorithms are applied to help mine knowledge out of large pools of existing information. It is the science that gives power to computers to perform without being openly programmed. “It is defined by the ability to choose effective features for pattern recognition, classification, and prediction based on the models derived from existing data” (Tarca and Carey 2007). According to Arthur L Samuel (1959), “machine learning is the ability of computers to learn to function in ways that they were not specifically programmed to do”. Many factors have contributed to making machine learning a reality. These include sources of data that are generating vast information, improved computational control for processing large amounts of information in fractions of time, and algorithms which are now more reliable and efficient.

Machine learning is one of the most exciting technologies one could come by. As is apparent from its name, machine learning offers the computer the ability to learn, meaning it can become more like a human. It is being vigorously used today, perhaps in many more ways than one would expect.

1.2 Data in Machine Learning

The data required for analysis is gathered from various sources such as web pages, emails, IoT sensors, text files, etc. This data serves as the input needed for machine learning algorithms to generate insights. Without data, we can’t train any models and all contemporary research and automation would be ineffective. Large initiatives are spending masses of money just to collect as much specific data as possible. Data uncertainty is common in real-world applications. Various factors like physical data generation and collection processes, unreliable data transmission, transmission bandwidth, measurement errors, and decision errors contribute to the uncertainty in data. This may apply for both numerical data and categorical data (Agrawal and Ram 2015).

After collecting data, it is preprocessed and used for extracting information and knowledge (Figure 1.1).

FIGURE 1.1
Data, information and knowledge.

Now the question of how the data is used in machine learning arises. As shown in Figure 1.2, the data is split into three parts – testing, training, and validation data.

FIGURE 1.2
Types of data.

Training data is applied to train machine learning models and, after completion of the training part, testing data is used for unprejudiced valuation of the model. Validation data is used for frequent evaluation of the model thereafter. Thus, the data plays an important role in the model building and selection. Data has a lot of potential for organizations and almost all large- and mid-level organizations are therefore continuously looking for ways to utilize it (Agrawal 2020). Some of the important dimensions of big data are described here:

1. 1) Volume: the main characteristic feature or dimension of big data is its sheer volume. The term volume refers to the amount of data an organization, or an individual, collects and/or generates. Currently, to qualify as big data, a minimum of 1 terabyte is the threshold for big data which stores as much data as would fit on 1,500 CDs or 220 DVDs, (or enough to store approximately 16 million Facebook images). The vast amounts of data are generated every second. E-commerce, social media, and various sensors produce high volumes of unstructured data in the form of various audio, images, and video files. Today big data is also generated by machines, networks, and human interaction on systems and the volume of data to be analyzed is massive.
2. 2) Variety: is one of the most attractive dimensions in technology, as almost all information is digitized nowadays. Traditional data types or structured data include information (such as date, amount, and time) in a structured way which can easily fit neatly in a relational database. Structured data is augmented by unstructured data. Modern day data sources include Twitter feeds, YouTube videos, audio files, MRI images, web pages, web logs, and anything else that can be captured and stored and does not require any meta model for its structure to access it later on.
3. Unstructured data i...

Table of contents

• Cover
• Half-Title
• Title
• Copyright
• Contents
• Preface
• Acknowledgments
• Editors
• List of Contributors
• 1. Fundamentals of Machine Learning
• 2. Medical Information Systems
• 3. The Role of Metaheuristic Algorithms in Healthcare
• 4. Decision Support System to Improve Patient Care
• 5. Effects of Cell Phone Usage on Human Health and Specifically on the Brain
• 6. Feature Extraction and Bio Signals
• 7. Comparison Analysis of Multidimensional Segmentation Using Medical Health-Care Information
• 8. Deep Convolutional Network Based Approach for Detection of Liver Cancer and Predictive Analytics on Cloud
• 9. Performance Analysis of Machine Learning Algorithm for Healthcare Tools with High Dimension Segmentation
• 10. Patient Report Analysis for Identification and Diagnosis of Disease
• 11. Statistical Analysis of the Pre- and Post-Surgery in the Healthcare Sector Using High Dimension Segmentation
• 12. Machine Learning in Diagnosis of Children with Disorders
• 13. Forecasting Dengue Incidence Rate in Tamil Nadu Using ARIMA Time Series Model
• Index