Intelligent System Algorithms and Applications in Science and Technology

Sunil Pathak, Pramod Kumar Bhatt, Sanjay Kumar Singh, Ashutosh Tripathi, Pankaj Kumar Pandey, Sunil Pathak, Pramod Kumar Bhatt, Sanjay Kumar Singh, Ashutosh Tripathi, Pankaj Kumar Pandey

• Publisher: Apple Academic Press
• Year: 2022
• ISBN:
  9781000406870
• Edition: 1
• Topic: Computer Science
• Subtopic: Digital Media
• Index: Computer Science

The 21st century has witnessed massive changes around the world in intelligence systems in order to become smarter, energy efficient, reliable, and cheaper. This volume explores the application of intelligent techniques in various fields of engineering and technology. It addresses diverse topics in such areas as machine learning-based intelligent systems for healthcare, applications of artificial intelligence and the Internet of Things, intelligent data analytics techniques, intelligent network systems and applications, and inequalities and process control systems. The authors explore the full breadth of the field, which encompasses data analysis, image processing, speech processing and recognition, medical science and healthcare monitoring, smart irrigation systems, insurance and banking, robotics and process control, and more.

PART I Machine Learning-Based Intelligent Systems for Healthcare

CHAPTER 1 BIG DATA ANALYTICS AND MACHINE LEARNING PARADIGM: PREDICTIVE ANALYTICS IN THE HEALTHCARE SECTOR

PRATIYUSH GULERIA

National Institute of Electronics and Information Technology, Shimla, Himachal Pradesh, India, E-mail: [email protected]

ABSTRACT

Big data analytics is the emerging field of data mining where the major challenge is to discover meaningful information from raw data. Big data comprises of unstructured, semi-structured, and structured forms. The supervised and unsupervised learning techniques, i.e., classification and clustering algorithms of data mining helps to retrieve meaningful information. The big data managed using mining techniques have increased computational intelligence and effective decision-making. In this chapter, the author has proposed the patient-centric healthcare frameworks in the context of big data and has discussed the Naïve Bayes, Decision tree data mining techniques, and their solutions to the healthcare sector. The supervised and unsupervised machine learning (ML) algorithms are implemented on the dataset for data analysis and information retrieval.

1.1 INTRODUCTION

Big data analytics has received attention from different fields of computer science which involves data mining, artificial intelligence (AI), machine learning (ML), and mathematical analytics. The big data term is based on the characteristics, which are (a) volume, (b) velocity, and (c) variety. The first characteristics focus on the volume of data which may be in terabytes, petabytes, or more. The second characteristics are the velocity of data where the data emerges in every second, and the third is a variety of data. The data can be in any format, i.e., images, pdf files, office automation, or it can be data from internet resources, software, and database application files. The Big data analytics field utilization in medical and healthcare sectors is the demand of present research. The ML and data mining techniques have been working in close interrelation with big data analytics to provide solutions in (a) computer-aided medical diagnosis, (b) intelligent healthcare informatics system, (c) predictive analytics in healthcare, (d) electronic record maintenance of patients, (e) smart systems for identifying diseases and diagnosis, (f) IoT enabled devices for medical diagnostics, (g) intelligent web semantics for healthcare sector, (h) knowledge data discovery for healthcare sector. ML algorithms help in handling healthcare challenges and can discover the ethical implications of healthcare data, patient health optimization, etc.

Authors Liang and Kelemen [1] have discussed the healthcare functionalities, i.e., (a) clinical decision support, (b) diseases surveillance, (c) healthcare management, etc., concerning big data.

1.1.1 CHALLENGES OF BIG DATA IN HEALTHCARE SECTOR

The challenges in handling big data related to the healthcare sector involves:

1. Data generation from multiple sources;
2. Diversified data formats;
3. Huge voluminous data;
4. Symptoms vary from patient to patient;
5. Patients pertain to different geographical areas;
6. Non-availability of data in electronic form;
7. Hidden and missing values in the datasets;
8. Non-availability of diagnostic data;
9. Missing historical record of patients;
10. Diversified healthcare areas;
11. Challenges in the collection of clinical reports and drug prescribed information;
12. Drug prescriptions to patients are in unstructured and non-electronic form.

The vital role of Big data analytics in convergence with ML is to give useful insight into these challenges. The tasks of transforming the raw data into meaningful form, preprocess the data to detect outliers, and prediction of fruitful results can be done using big data analytics, ML. The healthcare data available in digital format enables the patients to communicate with the patients having similar problems and gain information like (a) disease symptoms, (b) side-effects of medicines, (c) clinical reports, etc. [2, 3]. The predictive analytics can be done using data mining techniques. With the help of mining and statistical techniques, future events can be predicted [4].

The big data consists of the MapReduce programming model which uses the Hadoop distributed file system for analyzing the big data. In this framework, the data is divided into <key, value> pair to filter the meaningful information. Authors in Ref. [5] have proposed the architectural framework for healthcare systems. The framework consists of layers where data extraction and transformation take place in Transformation Layer and then the data goes through staging processes. In the Big data platform layer, the Hadoop ecosystem works using MapReduce Programming models. After the Big data platform layer, the analytical layer work through data mining techniques, reporting, etc.

1.1.2 ELECTRONIC HEALTH CARE FRAMEWORK

In the traditional approach, the patient approaches the doctor with some symptoms, and in response, after the check-up of the patient, the doctor recommends him to take some tests and diagnose the patients with some medicines. The major challenge in this approach is that the data related to the symptoms and diagnosis is not maintained electronically. In such a scenario, big data analytics coul...