Biomedical Signal Analysis for Connected Healthcare

By Sridhar Krishnan

Biomedical Signal Analysis for Connected Healthcare provides rigorous coverage on several generations of techniques, including time domain approaches for event detection, spectral analysis for interpretation of clinical events of interest, time-varying signal processing for understanding dynamical aspects of complex biomedical systems, the application of machine learning principles in enhanced clinical decision-making, the application of sparse techniques and compressive sensing in providing low-power applications that are essential for wearable designs, the emerging paradigms of the Internet of Things, and connected healthcare.

• Provides comprehensive coverage of biomedical engineering, technologies, and healthcare applications of various physiological signals
• Covers vital signals, including ECG, EEG, EMG and body sounds
• Includes case studies and MATLAB code for selected applications

• Language: English
• Publisher: Academic Press
• Release date: Jun 23, 2021
• ISBN: 9780128131732

Sridhar Krishnan

Dr. Sridhar (Sri) Krishnan took degrees in electrical and computer engineering at the University of Calgary before embarking on a long and distinguished career at various research institutions, including Clinical Research Institute of Montreal, University of Western Ontario, University of Toronto, and Ryerson University. Dr. Krishnan, has been the Department Chair of the Department of Electrical and Computer Engineering at Ryerson University, and was the Founding Program Director for Biomedical Engineering at Ryerson University. Dr. Krishnan is currently Affiliate Scientist at the Keenan Research Centre for Biomedical Science at St. Michael’s Hospital and Associate Dean of Research, Development and Graduate Programs in Engineering and Architectural Science at Ryerson University. Dr. Krishnan has developed numerous patents and technical inventions through the course of his career, and has coordinated the establishment of 18 research laboratories in a variety of research areas. In addition, Dr. Krishnan has also played an anchor role in establishing a large research institute (iBEST) to support biomedical sciences and engineering research at Ryerson University. He has been involved in developing institutional partnerships with St. Michael’s Hospital and University Health Network. These partnerships provide strategic access to clinical expertise, health information and experimental test facilities for the biomedical engineering and sciences students and researchers at Ryerson University.

Book preview

Biomedical Signal Analysis for Connected Healthcare

Sri Krishnan

Table of Contents

Cover image

Title page

Copyright

Dedication

About the author

Preface

1. Opportunities for connected healthcare

1. Introduction

2. Internet of things

3. Internet of medical things

4. Wearables for health monitoring

5. Biomedical signals

6. Objectives and organization of the book

2. Wearables design

1. Introduction

2. Wearables survey

3. Wearables design considerations

4. Open hardware design considerations

5. Textile wearables

6. Contactless monitoring

7. Discussions

3. Biomedical signals and systems

1. Introduction

2. Analog to digital conversion

3. Linear systems theory

4. Digital filters design

5. Digital filter realization

6. Applications

7. Discussion

4. Adaptive analysis of biomedical signals

1. Introduction

2. Adaptive filter design

3. Adaptive filter algorithms

4. Linear prediction

5. Time series modeling

6. Applications

7. Discussion

5. Advanced analysis of biomedical signals

1. Introduction

2. Time-domain analysis

3. Frequency-domain analysis

4. Joint time-frequency analysis

5. Signal decomposition analysis

6. Advanced feature extraction and analysis

7. Sparse analysis and compressive sensing

8. Discussion

6. Machine learning for biomedical signal analysis

1. Introduction

2. Machine learning fundamentals

3. Types of machine learning models

4. Challenges with machine learning models

5. Feature analysis

6. Common machine learning techniques

7. Machine learning performance metrics

8. Fairness and ethics in ML

9. Summary

7. Data connectivity and application scenarios

1. Introduction

2. Pulse code modulation

3. Delta modulation

4. Lossless data compression

5. Line coding of waveforms

6. Digital modulation

7. Telecommunication networks

8. Wireless technologies

9. Mobile health

10. Electronic medical records

11. Personal health record

12. Interoperability

13. Health information security and privacy

14. Human factors and user experiences

15. Application scenarios

16. Summary

Index

Copyright

Academic Press is an imprint of Elsevier

125 London Wall, London EC2Y 5AS, United Kingdom

525 B Street, Suite 1650, San Diego, CA 92101, United States

50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States

The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom

Copyright © 2021 Elsevier Inc. All rights reserved.

No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Details on how to seek permission, further information about the Publisher’s permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions.

This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be noted herein).

Notices

Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary.

Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility.

To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein.

Library of Congress Cataloging-in-Publication Data

A catalog record for this book is available from the Library of Congress

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library

ISBN: 978-0-12-813086-5

For information on all Academic Press publications visit our website at https://www.elsevier.com/books-and-journals

Publisher: Mara Conner

Acquisitions Editor: Chris Katsaropoulos

Editorial Project Manager: Isabella C. Silva

Production Project Manager: Prasanna Kalyanaraman

Cover Designer: Matthew Limbert

Typeset by TNQ Technologies

Dedication

Dedicated to the love and inspiration provided by my wife Mahi, and to our children, Sibi and Sarvi.

Grateful for the constant support and encouragement of my parents, extended family, students, professors, and friends.

About the author

Sridhar (Sri) Krishnan received his B.E. degree in Electronics and Communication Engineering from the College of Engineering, Guindy, Anna University, Chennai, India, in 1993 and M.Sc. and Ph.D. degrees (with student fellowship from Alberta Heritage Foundation for Medical Research) in Electrical and Computer Engineering from The University of Calgary, Calgary, Alberta, Canada, in 1996 and 1999, respectively. Sri Krishnan joined Ryerson University in July 1999 and is currently a professor in the Department of Electrical, Computer, and Biomedical Engineering. From 2011 to 2021, he was an Associate Dean (Research and External Partnerships) for the Faculty of Engineering and Architectural Science. He is also the Founding Co-director of the Institute for Biomedical Engineering, Science, and Technology (iBEST) and an Affiliate Scientist at the Keenan Research Centre in St. Michael's Hospital, Toronto.

Sri Krishnan's research interests include biomedical signal processing, audio signal processing, wearable technology, and machine learning. From 2007 to 2017, Sri Krishnan held a Canada Research Chair position in Biomedical Signal Analysis. He has published more than 340 papers in refereed journals and conferences and has filed 15 invention disclosures/patents. He has presented keynote/plenary/invited talks in more than 60 international conferences and workshops. Sri Krishnan serves in the advisory boards of research institutes, innovation centers, incubator zones, and biomedical tech start-up organizations. Sri Krishnan is a registered Professional Engineer in the Province of Ontario. He was the Founding Chair (2005–2015) of IEEE Signal Processing Society, Toronto Section and Region 7 (Canada), and a Founding Member of the IEEE Engineering in Medicine and Biology Society, Toronto Section. He currently serves as a Technical Committee Member (Biomedical Signal Processing) of IEEE EMBS.

Sri Krishnan is a recipient/awarded Fellow of Canadian Academy of Engineering; Outstanding Canadian Biomedical Engineer Award from the Canadian Medical and Biological Engineering Society; Achievement in Innovation Award from Innovate Calgary; Ryerson University's Sarwan Sahota Distinguished Scholar Award; Ontario Research Innovation Award from Biodiscovery Toronto; Canadian Engineers' Young Engineer Achievement Award from Engineers Canada; New Pioneers Award in Science and Technology; South Asian Community Achiever Award; Exemplary Service Award and Best Chapter Chair Award from IEEE Toronto Section; and IEEE AESS Best Chapter Chair Award.

Preface

Pursuit of knowledge and contributions to research and innovation for the societal benefits have been core aspects of my scientific journey. The interest in biomedical signal analysis started when I was in my undergraduate class on signals and systems and happen to visit my uncle in the emergency ward in which the monitor next to him displayed vital signals. The curiosity of representing those vital signals with the mathematical concepts learnt in the signals and systems course led me to pursue fourth-year capstone design project followed by graduate studies in the area, and eventually as a professor investigating various model-based and data-driven approaches for biomedical signal processing, analysis, and interpretation. Over the course of learning, the ideas from electronics, telecommunication, information theory, machine learning, and human-centered design have shaped my understanding of the domain of biomedical signal analysis for connected healthcare. In the past 10 years, through the various workshops and tutorials that we have conducted, the audience were more interested to learn the concepts systematically so that the techniques could apply to any biomedical signal analysis. Followed by an IEEE Engineering in Medicine and Biology Society (EMBS) conference workshop on the same topic, the Elsevier team approached me to consider writing a book on this topic. During book writing, the pandemic stuck, and that also led to further strengthening certain topics of the book on the aspects of connected and telehealth possibilities.

The book comprises seven chapters, and with a people-centered approach, biomedical signal analysis topic has the largest coverage followed by wearables design and then machine learning and connectivity topics. The book covers these topics in a logical flow as per the following sequence. The context of current global healthcare situation and the role of digital technology in improving people's lives are covered in Chapter 1. An understanding of the applicability of biomedical signals in real-world application can inspire hardware design considerations, and the topic of wearables design is covered in Chapter 2. Biomedical preprocessing techniques of sampling, quantization, and filtering are covered in Chapter 3. In Chapter 4, details of adaptive analysis of biomedical signal analysis are covered. Chapter 5 deals with various feature extraction and analysis techniques. The role of biomedical signals in decision-making applications will benefit from using machine learning techniques, and some requirements of those are covered in Chapter 6. Approaches for digital transmission of biomedical signals for connected healthcare, telemedicine, and electronic medical records are covered in Chapter 7.

The contents of the book clearly fit an undergraduate course in biomedical signal analysis, biomedical systems and signals, digital health technologies, telemedicine, and also for an advanced course at the graduate level. With the availability of a plethora of open datasets with various forms of biomedical signals and computer programs for signal processing, feature analysis, machine learning, and embedded systems programming, the students can experience the concepts covered in the book in individual or collaborative laboratory exercises and projects.

The book was possible because of the many help I received from my graduate students, clinical collaborators, international research collaborators, industry partners, and peers in the area. Specifically, I would like to thank my graduate students and research assistants, Yashodhan Athavale, Fayez Qureshi, Garima Sharma, and Magdaleen Singarajah, for helping with the content of various chapters in this book. I am also grateful to Ryerson University Signal Analysis Research (SAR) group members for their research contributions on various aspects of the book topic. Elsevier publications team have been diligent in their work, and I thank them for their help and support throughout the book writing phase. I would like to place my gratitude for the unconditional support provided by my family and friends throughout this journey of knowledge dissemination. May we see a connected world with good health and happiness!

Sri Krishnan

Toronto, Canada

June 2021

1: Opportunities for connected healthcare

Abstract

Digital technologies and Internet have made significant transformations to the way humans interact and work. In this chapter, the potentials of these technologies to harness healthcare advancements are introduced. Specifically, how different sensing technologies, signal processing and machine learning techniques, and Internet of Things could be integrated in providing a connected healthcare framework that could be applied in a wide variety of applications for telemedicine, virtual healthcare, remote monitoring, and continuous monitoring of health and wellness are covered. Appropriate overview of the topics and organization of this book are discussed to facilitate the readers' choice to get into the details of a particular topic related to biomedical signal analysis for connected healthcare.

Keywords

Affordable health technology; Bioacoustics; Digital health; ECG; EEG; EMG; Internet of things (IoT); Internet of medical things (IoMT); PPG; Remote monitoring; Telemedicine; Wearables

1. Introduction

Healthcare is an inevitable and core requirement of humanity and its very own existence and sustainability. In this ever resource constraint world, healthcare is under constant strain of lack of adequate and appropriate medical attention at various levels of the system. The requirements manifest in the form of long-term needs for chronic healthcare, care for aging population, remote delivery of various health services, escalation of hospitalization costs, and increased chances of infection due to prolonged hospital stays. These challenges could be addressed if the healthcare is able to harness the technology connectivity we are all enjoying as part of the digital revolution and a conscious effort to bridge the digital divide gap thereby providing a platform for anywhere, anytime connectivity and seamless and timely health services. As shown in Fig. 1.1, among the Top 10 diseases globally, for most of these diseases, if not all, the role of technology in diagnosing, managing, and intervening is huge. Inspite of the prevalence of these diseases worldwide, the good news is the fact that the average life expectancy of humans continues to increase worldwide. Fig. 1.2 shows the average life expectancy map of the countries, and it is obvious that socio-economic factors and access to high-end technologies play a significant role in influencing these statistics. The ever-increasing prevalence of affordable digital health technologies and the penetration of information and communication technologies (ICTs) have great potential to bridge the digital divide and contribute to an equitable access of healthcare services.

The idea of connecting human civilization on a global scale gave birth to telephony in the 1920s. This was also augmented by advancements in other communication modalities such as Morse code, telegraphy, radar, and sonar techniques. Though communication forms one aspect of human civilization, it is indeed the most primal medium of survival, and through the 20th and 21st centuries, communication technologies have boosted a rapid growth in most verticals including infrastructure, transportation, food, and healthcare. Up until the late 1960s, most communication methods which were applied for commercial and military applications involved extensive manual labor, in order to collect, store, analyze, and transmit information or decisions. Fig. 1.3 shows the history of Internet Connectivity and its evolution.

Figure 1.1 Top 10 diseases globally in 2017. 

Source: IHME, Global Burden of Disease, Our World in Data, BBC [1].

Figure 1.2 World life expectancy map (2016). 

Source: Wikipedia [2].

Figure 1.3 History of internet connectivity [3].

With the evolution of the Internet in early 1970s, human civilization has immensely progressed in developing advanced wireless communication protocols, cloud analytics and services, smart phones, tablets, computers, and wearables. Following the development of commercial websites and systems in the early 1990s, we have witnessed rapid growth in the development of mobile phones, music players, laptops, tablets, smart phones, and their related data encoding and communication protocols. This has also started a trend to standardize communication pathways between devices and users and sensor design, depending on specific applications.

In a recent survey conducted by Wollschlaeger et al. [4], it was found that by 2015, 15 billion devices were already connected to the internet, and this number is now projected to grow over 65 billion toward the end of this decade. Development of new communication protocols, such as Bluetooth 4.0, IEEE R 802.11 a/b/c/g/n, LTE (Long Term Evolution), NFC (near field communication), and wireless sensor networks (WSNs), has extensively promoted the development of customer-specific smart devices for usage in communications, health monitoring, home automation, transportation, and virtual and augmented environments. To cite a few, we have seen the following recent technological advancements in recent times:

• Smart phones, computers, tablets, laptops, and wearables

• Cloud computing services and Smart energy grids

• Smart home appliances for temperature control, cleaning, cooking, and refrigeration

• Advanced gaming consoles, VR (virtual reality), and AR (augmented reality) devices

• Autonomous (or self-driving) transportation

• Advanced healthcare services such as digital pathology, robotic surgeries, and physiological data analysis using wearables

The Internet of Things (IoT) framework ensures a seamless connectivity and friendly interaction of the users with their surroundings, devices, and sensors. The model addresses gaps in device interoperability, internet connectivity, and data analytics and ensures the development of a smart, connected environment to the end user. Following section describes the IoT in more details.

2. Internet of things

The IoT can be defined as a complex, intelligent, interconnection of computers, phones, wearables, home appliances, vehicles, and infrastructure via the Internet, to collect, store, and exchange data [5,6]. In simple terms, any electronic device or equipment with an ON–OFF switch can be connected to the IoT network. This framework allows an individual to engage in a friendly interaction with their surroundings, thereby improving quality of life. A one-on-one interaction with different devices within an IoT structure generates structured and unstructured big data, which could be then further analyzed to develop user-specific feedback and decisions. In order to enable these features, it is imperative that all the connected devices or appliances are embedded with sensors, actuators, and communication protocols.

In order to understand the IoT framework, we briefly cover two main aspects of this internet connectivity model along with the functioning of its layered architecture. Fig. 1.4 depicts two aspects of the IoT architecture, namely,- hardware and software [3].

2.1. Hardware

Sensors: These include sensing chips embedded within smart devices and appliances for capturing various types of data such as vibrations, motion, direction, orientation, light, temperature, and pressure. Sensors could be in the form of accelerometers, gyroscopes, pressure films, magnetometers, altimeters, and light detectors. These modules capture analog information from the user and surroundings, denoise and digitize it and transmit it to embedded microcontrollers for further analysis. In a lot of cases, the sensed data are also transmitted to cloud-based services for analysis.

Figure 1.4 IoT architecture [3].

Actuators: These components are primarily responsible for controlling a mechanism or system within an appliance or device. Unlike a sensor, an actuator requires the triggering of a control signal and voltage to start performing a specific function. Examples of actuators include energy grids, water supply, temperature control units, vehicle gears, and heating sources. Actuators could be pneumatic, electric, or mechanical in nature, depending on the equipment and its application. It should be noted that sensors and actuators could be directly connected to the Internet within a device and are usually not included as secondary or supporting components of the IoT structure.

Secondary Components: These are the modules which are indirectly affected or controlled in an IoT-based mechanism. For example, a garage door which is controlled by a mobile application via the internet or a power supply outlet to a bulb which controls the intensity of light in a room.

Sensor Networks: These networks could be wired or wireless in nature depending on the type of device or appliance used. For example, home appliances such as refrigerators could have Ethernet-based internet connectivity to control cooling and water supply or door entry points could be controlled by wireless, closed-loop NFC or RFID (radio-frequency identification) tags. Another example would be a smart home whose individual users, devices, and appliances are connected either using wired or wireless connections, governed by device-specific as well as universal communication protocols to store and exchange information.

2.2. Software

• Event-based: These applications are user specific and cater to small tasks such as calendars, reminders, notes, and alarms.

• Semantics-based: These applications are usually deployed for enabling interoperability between devices. For example, Google's Chromecast plug-in device, when connected to a TV, enables wireless video playback via an Android or iOS device having the YouTube application installed.

• Service-based: Service-oriented applications usually cater toward billing a user for accessing different applications such as music, movies, mobile phone plans, internet subscriptions, and cable TV. These applications do not interact much with user inputs, and mostly work in background for subscription validation and tracking. Service-oriented applications can also be free of cost and could include services for healthcare, emergencies, and police.

• Database-oriented: These applications are generally not visible to the end user and function in the background within the device, external storage, or cloud services. The primary role of these applications is to store user- and device-generated data in tabular formats in order to ensure integrity and easy access to analytical applications. For example, physiological data captured using wearables are stored on the device's on-board memory or onto a supporting cloud storage in the form of standard formats such as CSV (comma separated values) files.

• Middleware: These applications could have multiple functionalities such as the following:

– Create and deploy new IoT services between devices to exchange and process data.

– Enable device interoperability using a single universal code, protocol, or small application.

– Define standard networking protocols.

Middleware applications usually act as a bridge between a device's operating system and its supporting applications with the user and other equipment in a smart environment.

In order to ensure a seamless functioning of various hardware and software applications within an IoT model, one must follow a layered architecture model as illustrated in Fig. 1.5.

Each layer has been designed to meet specific functions and interoperates with layers above and below it, depending on user interaction and information generation [3,5]. We briefly describe the functioning of each layer as follows:

Figure 1.5 IoT layers [5].

1. Physical or Perception Layer: This layer enables the hardware connectivity of users with surrounding devices and equipment using operating systems and middleware applications. It also includes sensors and actuators which gather data for information processing and analytics.

2. Network Layer: The Transport and Processing layers together form the networking layer of the IoT model and include various functions such as follows:

• Connecting devices using networking protocols to the internet

• Data transmission and exchange from sensors

• Store and analyze incoming data using DBMS (database management systems), cloud computing, and big data modules

• Enable semantic and middleware applications for inter- and intradevice operability and connectivity

3. Application Layer: This layer is usually visible to the user in the form of service-based, event-based, and user-specific applications. It includes software installed on devices to interact with the user, as well as gather, process, and transmit sensory information in an IoT environment.

With regards to user data processing acquired from sensors, applications, and devices, IoT provides many sources such as follows [3]:

a) User's personal computer

b) Personal mobile device for local storage and small analysis

c) Enabling gateway for a network server which stores data from different connected devices

d) Cloud storage and computing services for big data analytics and feedback generation to the user via personal computer or mobile device

e) Fog computing algorithms and applications which enable on-device data processing and analysis, thereby reducing computational constraints on local computer or cloud

It is evident that for a seamless functioning of the IoT framework, its three components, namely, hardware, software, and supporting architecture layers must coordinate and function in synchronization with each other [5,6]. This being said, current niche markets retailing smart devices for IoT environments pose one or many of the following challenges in order to ensure the robustness and integrity of a connected environment:

• Ensuring user data privacy, security, and confidentiality. As per our survey, although currently available tools and devices for an IoT model do support this clause, there still exists a huge gap in designing data formatting, security, and standardization, which must address inter- and intraoperability concerns.

• An IoT model must be able to handle smart device data traffic in a mobile environment without any significant information loss, which could impact the intelligence and thereby the feedback given to a user.

• IoT frameworks must ensure reliability and real-time actuation with respect to user inputs and data.

• The cloud services employed for big data analytics must address the scalability factor by being able to handle multiple device requests for data storage, transmission, and processing.

• IoT models must be designed by considering power and memory requirements of each connected device, as well as that of available computing resources such as local computers or cloud services.

• The communication and networking protocols must adhere to multihop methods in order to ensure secure and robust interdevice connectivity.

• As a layered model, it is also expected in an IoT model to have middleware applications which could be used by an individual for remote device management.

• Lastly, it is highly imperative that sensor and device design must focus on context detection, and the associated data processing resources must include machine learning models for analytics and feedback generation.

Meeting these challenges will not only help us develop a robust IoT environment for an individual, but a conglomeration of many such models could potentially lead to development of a connected human population. Recent developments in smart technologies, coupled with ubiquitous sensor design, have provided a significant number of opportunities in the IoT domain, as highlighted in Fig. 1.6.

As evident from these opportunities, current IoT models enable the networking of devices with limited CPU, memory, and power with the Internet and end user. Our literature review indicates that home automation, advanced healthcare, and autonomous transportation are some of the niche markets within the IoT context, and have spurred a massive retailing of smart home appliances, wearables, devices for security, lighting and air conditioning, and self-driven vehicles [3]. A classic example of one such technology would be the Nest, which is a self-learning, Wi-Fi-enabled, self-driven, intelligent thermostat which optimizes home temperatures by gathering and analyzing human inputs and presence data. This device has a built-in machine learning capability for analyzing streaming temperature and ambient data, in order to make appropriate heating or cooling adjustments. It is one of the few devices in the market to employ a fog or edge computing approach to data analytics, thereby reducing the need to maintain constant connectivity with its cloud services for feedback generation.

Figure 1.6 Opportunities in IoT [3].

IoT-based home automation technologies have greatly enabled and motivated a healthy lifestyle adoption for elderly citizens, disabled individuals, chronically ill subjects, and the general population as well. Home automation tools include devices which provide assistive living, thus improving the quality of life. This being said, it is evident that apart from infrastructure and home appliances, an IoT-based home would also feature user-specific devices for better living, such as follows:

• Voice-based security and temperature control

• Assistive devices such as Google's Home or Amazon's Echo for setting reminders, playing music, news, and general interaction with the user

• Smart cooking and refrigeration systems

• Wearables for continuous monitoring of daily activities of an individual. These would be helpful for remote health consultations, fitness, and alarm generation

In the next section, we will cover various aspects of a substructure of the IoT framework, the Internet of Medical Things (IoMT) model, which will address the use of remote health monitoring technologies in a connected healthcare and smart environment context.

3. Internet of medical things

Before we delve into the working of an IoMT framework, we must first understand the basics of Telehealth and how IoMT fits in it [7]. Telehealth refers to the remote delivery of clinical and nonclinical healthcare services in a given geographical area. In olden times, when there were no computing or connectivity technologies, telehealth was practiced subtly in the form of visiting doctors and home care nurses who would monitor a patient's health and prescribe medicine accordingly. Over the years