Within Heath and Medicine, the research into AI techniques has emerged significantly for clinical decision-making purposes. Clinical decision support systems that are based on intelligent techniques are increasingly used in the health care industry. These systems are intended to help physicians in their diagnostic procedures, making decisions more accurate and effective, minimizing medical errors, improving patient safety, and reducing costs. This book introduces different methods and techniques for clinical decision support systems with the aim of identifying a basic criterion for adequate use of intelligent techniques within such systems.

Traditionally, medical decisions have been determined mainly based on the physician’s unaided memory as well as rapid judgments of the patient’s symptoms and conditions. This has required doctors and physicians to keep up-to-date with the latest research and literature to ensure that they investigate medical cases with up-to-date knowledge.

Medical professional bodies and health care providers have established a realization for the need to develop their own professional training courses and recertification procedures for the purpose of ensuring that doctors have the ability to memorize the most relevant knowledge. However, fundamental limitations of human memory and recall mechanisms, coupled with the exponential growth in knowledge, meant that most of what is known could not be memorized by most individuals.

As such, an opportunity has been presented in developing computer tools and systems to assist in storing and retrieving the appropriate medical knowledge needed by the practitioners and professionals in dealing with some challenging conditions. These systems can also be used and utilized by the professionals to assist in diagnostic decisions and therapeutic and decision-making techniques.

The potential for computing techniques, methods, and tools to support clinicians and health care professionals with assistance, expert advice, and critical decision making was predicted more than four decades ago. Despite decades of research in this field, computer applications, intelligent computer applications, and computer decision support systems receive limited use in health care systems. The various reasons include usability, lack of technology and tools, and problems in integrating such tools with the demanding work that health care institutions provide.

Current advances in technology are providing and making it possible for researchers to develop intelligent applications or refine methods, techniques, and algorithms to develop sophisticated computer applications in various medical fields. In addition, studies show that computer applications and specifically intelligent systems in the field of health and medicine has now become a fast-developing research area that is combining sophisticated computing methods with the insights of expert medical consultants to produce better application tools for refining today’s health care systems. These developments are attributing to better health care provision, as well as lowering the cost of health care, which is a major concern in many European countries and the United States. With the rapid development of new technology, data is becoming available immediately, as a result, pressuring clinicians to keep up with the newest developments in their field. Coupled with time constraints, this can lead clinicians and medical staff to take rapid decisions for diagnosing patients, which could consequently lead to misdiagnoses and unnecessary treatments.

In medical diagnosis, recent work applied fuzzy logic to diagnose pediatric asthma, breast cancer, coronary heart disease, thyroid, diabetes, and ovarian cancer. Case-based reasoning was used for liver disease diagnosis while artificial neural networks were implemented to diagnose heart valve diseases. Expert Systems (ES) are used to diagnose breast cancer, detect lesions in mammograms, and to determine iron deficiency anaemia in women. ES were also developed to diagnose epilepsy from ECG signals, to diagnose and classify dengue patients and in lung cancer diagnosis. Other research shows the implementation of Bayesian networks to discover the human immunodeficiency virus and to predict accurately coronary artery disease. In addition, work presenting a combination of genetic algorithms with other AI techniques to diagnose lung cancer and liver disease is also apparent.

Different AI techniques are applied to diagnose different medical conditions. In literature, it is also evident that AI techniques are used in particular diseases, such as the case to diagnose breast cancer. An interesting finding is that in many medical cases, researchers have utilized more than one AI technique to accomplish their diagnosis. Researchers may use a particular AI technique in their intelligent system to classify a general medical condition while adding another additional technique to accurately predict the diagnosis. Current literature also shows applications of modified or tailored AI techniques to diagnose medical health conditions while in other cases researchers tried to find more accurate intelligent computational methods such as artificial immune systems in addition with other techniques to classify and predict the health condition.

The latest research shows successful diagnosis of diverse medical conditions with the use of different AI computational methods and paradigms. It is clearly evident that the field of medical science is showing an increase in the number of intelligent systems being developed that learn from new cases while in operation. These learning intelligent machines are used to assist clinicians and medical staff to diagnose more precisely their patients’ health conditions. As researchers continue to develop these intelligent methods and techniques, computer applications in medicine will only intensify, get more complex, and perform accurate diagnosis.

Critical investment in translating key methods and insights into working systems is required to enable this vision of true evidence-based health care. This is in addition to the realization that key conception opportunities require advances in core computer science research and engineering. The promise to enhance the quality and effectiveness of health care to enhance the quality and longevity of life can be realized by the collection and systematic analysis of data collected on health and illness. This practice would provide new insights on wellness and illness that can be operationalized.

Automation and provision of decision support for accurate triage and diagnosis can be realized within the pipeline of data, to prediction, to action, to generating well-calibrated predictions about health outcomes, to produce effective plans for chronic disease management, and to formulate and evaluate larger-scale health care policies. Data can be transformed into predictive models by utilizing data-centric methods. Forecasts with well-characterized accuracies about the future or diagnoses about states of a patient that we cannot inspect directly can be generated using predictive models. Procedures can harness such forecasts or diagnoses to generate recommendations for actions in the world, and decisions about when it is best to collect more information about a situation before acting, considering the costs and time delays associated with collecting more information to enhance a decision.

Clinical and biomedical discovery is dependent on the collection and systematic analysis of large quantities of data. Methods for learning from data can provide the foundations for new directions in the clinical sciences via tools and analysis that identify subtle but important signals in the fusing of clinical, behavioral, environmental, genetic, and epigenetic data.

Building insights via analysis and visualizations can be enabled by computational procedures, which can only play an active role in framing and designing clinical studies, and in the proposal and confirmation of biomedical hypotheses. Especially, within such methods as those that identify statistical associations among events or observations and help to confirm causal relationships. The cost-effective optimization of health care delivery has many gains, and even more value can come by enabling fundamental scientific breakthroughs in biomedicine.

The advance of intelligent data-driven methods for informed health care decision making will also help support professionals to make informed decisions from current best-practice knowledge bases. In general, CDSS purpose would be used to review and convert the clinical input into a type of information using numerical and logical techniques. This would aid doctors and nurses in making diagnosis and treatment decisions. CDSS use is not yet widespread in hospitals and clinics; however, they have the potential to improve storing and retrieval of medical records, analysis of patient history for many purposes, including diagnosis, evaluation of real-time information collected from the monitors, use of medicines together with the treatment plan, and handling large amount of information and knowledge.

AI techniques have recently seen an increased trend in utilization within the field of CDSS. This is in an effort to enhance the accuracy and effectiveness of disease diagnosis and consequently prevent or at least minimize medical errors. Many studies have shown the effectiveness of applying AI within CDSS, where the intelligent agent as presented in the literature is one of the main components in the information architecture of a CDSS. Many researchers have considered the intelligence module as a fundamental part of CDSS and that the adoption of a CDSS with an embedded intelligence module would have the ability to learn over time. Research has shown that adopting AI within CDSS would support a considerable scope of decision making, particularly the decisions blanketed by uncertainty. It would also manage domains where the decision procedures tend to be more complicated and need specialist knowledge as well as evaluate the consequences of the proposed solution.

A system that combines clinical decision support systems with intelligent techniques would exceed the traditional CDSS and have many necessary characteristics, including an improved and supported decision-making process through enhanced outlook on intelligent behavioral patterns and an increased ability to learn and create new clinical knowledge.

The techniques implemented by systems based on intelligence strategies can either be a single technique that deals with certain problems or utilization of a combination of two or more techniques to tackle complex problems that involve a level of uncertainty and ambiguity. The challenge would be in the adaptation of suitable criteria for the adequate use of intelligent techniques. As such, a revision of current trends become necessary to define a general guideline that deals with issues relating to the evaluation of intelligent CDSSs. This is particularly through the provision of a basic guideline that assists the system designer in making an optimal selection between the various intelligence techniques to be applied within CDSS.

It is very critical to realize that the time is just right for this book to be compiled and published. Health authorities are cutting costs, treatments and medicines are becoming increasingly expensive, and waiting times for procedures and referrals are all escalating. We need to seriously ask ourselves, Why? The answer is not just down to the volume and lack of specialists but more importantly incorrect diagnosis and lack of supporting information. Statistics for misdiagnosis is frightening; patients are seeking second opinions, even traveling to another country for consultations and advice. Although medical research is advancing at a rapid rate, governments and budgets are not. Therefore, it is evident that there is a vital requirement for developing systems that can support health care and streamline the health care process.

This book can be used as a reference to help understand the theories that underpin intelligent systems to support health and medicine. The reliance on the knowledge of human experts to build expert computer programs is helpful for several reasons: the decisions and recommendations of a program can be explained to its users and evaluators in terms that are familiar to the experts; additionally, as we hope to duplicate the expertise of human specialists, and in this case for health and medical disciplines, we can measure the extent to which our goal is achieved by a direct comparison of the output behavior to that of the human experts.

To summarize, these are the direct benefits of this book to the reader:

This book is aimed at practitioners and academics whose interest is in the latest developments in applied computing and intelligent systems concepts, strategies, practices, tools, and technologies.

Chapter One presents a framework for the early diagnosis of neurodegenerative diseases (NDDs) using signal processing and signal classification techniques. The problem with NDDs is that they are incurable, hard to detect at an earlier stage because of non-obvious symptoms, and also hard to discriminate at a later stage because of pattern similarities of different NDDs. This work has highlighted the importance of machine learning and signal processing in the early diagnosis of life-threatening diseases such as Alzheimer's, Parkinson's, Huntington's, and ALS. In this chapter, the issues with the early diagnosis of NDDs and also with their possible solutions are presented. The analysis and classification of gait signals is presented using a set of well-known classifiers; linear, nonlinear, and Bayes. Results are presented from various dimensions using more than one performance evaluation technique. In addition, a novel combination of classifiers to improve the accuracy is proposed and demonstrated. The work has presented novel and significant findings that can be used in clinical practices for the early diagnosis of NDDs.

Chapter Two starts by defining wearable technology for lifelogging, which provides an ideal platform for the identification and quantification of negative emotion. Making these data available can be used to encourage reflective thought and facilitate the development of protective coping strategies. This type of technology permits the user to capture multimodal data sources pertaining to everyday behavior. This chapter also explores how lifelogging technologies can be used to monitor the physiological process of inflammation that is associated with negative emotion and deleterious consequences for long-term health. These technologies furnish the user with a platform for self-reflection, enhanced awareness, and the formulation of coping strategies. Development of this technology faces a number of significant challenges, including (1) obtaining measures that are clinically relevant, (2) capturing the context of an event to enable sufficient self-reflection, and (3) designing user interfaces that deliver insight without creating health anxiety or hypochondria. This chapter explores these challenges in detail and offers guidance on the design of a lifelogging technology that promotes effective self-reflection and protective coping.

Chapter Three provides a detailed description of various feature selection algorithms that are suited for gene selection in microarray data. Microarray is the array of DNA molecules that permit many hybridization experiments to be performed in parallel. Both supervised and unsupervised feature selection methods are described in detail. The pros and cons of using supervised versus unsupervised feature selection methods have been illustrated using benchmark micro array data sets. In the category of supervised feature selection algorithms, both filter and wrapper approaches have been dealt with. This chapter presents detailed simulation for each of the filter-based feature selection algorithms (using fivefold cross-validation) on benchmark micro array data sets. Performance of different filter-based feature selection algorithms are also demonstrated on top-ranking genes.

Chapter Four proposes a new and novel approach to correct intensity inhomogeneity in MRI images of the brain. Anatomic structural map, an equivalent of digital brain atlas, guides the algorithm for automatic operation. The structural map is generated directly from the test image. Accurate information from the structural map is combined with distorted intensity-level attributes of the test image to detect outliers in regions of interest (ROIs) generated by K-means clustering. The number of ROIs is the number of tissue classes specified by the user in K-means clustering. Outliers in each ROI are merged with voxels in the appropriate tissue class. Intensity levels of the new set of voxels in each tissue class are rescaled to conform to intensity levels of uncorrupted voxels. A review of current bias field correction strategies is presented by explaining their design techniques, importance, and limitations, respectively. The methodology of the proposed approached is conveyed. Additionally, results of testing the proposal on real magnetic resonance images are displayed, followed by a discussion of the results and concluding remarks.

Chapter Five describes the opportunities and challenges that the method of leveraging Big Data and pervasive analytics for elderly care presents. It is argued that though potentially viable as an approach to raise elderly care to a level where the uniqueness of each elderly person is adequately recognized, it does pose enormous social, ethical, and technical challenges that research cannot afford to ignore. A review of the research challenges in personalized services for elderly care is provided, in addition to an overview of the state-of-the-art and future challenges in the area of Big Data analytics for elderly care. Based on the discussion, ACTVAGE, a lifestyle-oriented context-aware framework for supporting personalized elderly care and independent living, is proposed. The framework combines systematic capture of past lifestyles and knowledge of current activities and user context, and applies rigorous analytics to build a complete picture of the elderly person's lifestyle and needs. A formal representation of the lifestyle concept is built for system design. Based on the representation, required services, including social networking, self-diagnosis and monitoring, advisory, entertainment, exercise and dietary, reminder and local events services, are developed to offer individually tailored and lifestyle-oriented support for active ageing and independent living.

Chapter Six provides a comprehensive discussion of intrapartum hypoxia and presents relevant literature in the area. Uterine contractions produced during labor have the potential to damage a fetus by diminishing the maternal blood flow to the placenta, which can result in fetal hypoxia. In order to observe this phenomenon in practice, labor and delivery are routinely monitored using cardiotocography monitors. The cardiotocography recordings are used by obstetricians to help diagnose fetal hypoxia. However, cardiotocography capture and interpretation is time consuming and subjective, often leading to misclassification that result in damage to the fetus and unnecessary caesarean sections. Therefore, correct classification is dependent on qualified and experienced obstetric and midwifery staff and their understanding of the classification method used, which can be difficult. Alternatively, objective measures may help to mitigate the effects of misclassification. For example, automatic detection of correlates between uterine contractions and fetal heart rate can be used to reduce unnecessary medical interventions, such as hypoxia and caesarean section during the first stage of labor, and is instrumental in vaginal delivery in the second. The challenge is to develop predictive algorithms capable of detecting, with high accuracy, when a child is genuinely compromised before medical intervention is considered. This chapter can be considered as a work in progress that has mapped out a possible work plan for dealing with CTG data and how it might be used in a machine learning environment to predict normal and pathological records in the CTG-UHB data set.

Chapter Seven discusses the dynamical neural network architectures for the classification of medical data. Extensive research indicates that recurrent neural networks, such as the Elman network, generated significant improvements when used for pattern recognition in medical time-series data analysis and have obtained high accuracy in the classification of medical signals. The aim of this chapter is to provide a literature survey of various applications of dynamical neural networks in medical related problems. Medical signals recorded in various applications contain noise that could be the result of measurement error or due to the recording tools. Data preprocessing will be discussed in this chapter to extract the features and to remove the noises. A case study using the Elman, the Jordan, and Layer recurrent networks for the classifications of uterine electrohysterography signals for the prediction of term and preterm delivery for pregnant women are presented.

Chapter Eight imbues clinical decision support with such features as a consequence of a more formal approach to their definition, design, implementation, and testing. Accountability is maintained through a formally design audit process, whereas robustness of decision analysis and system operation is autonomously handled by the system itself. This is extremely important in mobile decision support use by clinicians and in increasing confidence in the use of such systems. It can be noted, from the foregoing discussion, that there are (at least) two separate concerns to be handled, namely, concerns of system operation governance, accountability, and robustness and concerns of decision process governance, accountability, and robustness. Traditionally such concerns would be divided and handled separately in the software development process. In this case, however, there are evidently many cross-cutting concerns that do not fit neatly into one or the other concern. In addition, in a mobile setting, the role of a decision support system is not limited to presenting data analysis; it may also present relevant documentation and online information or provide alerts using the Internet. This chapter proposes, analyses, and assesses a formal representation and reasoning technique for mobile medical decision support systems that handles the separate and cross-cutting concerns of the systems by using a formal calculus of first order logic. The work is evaluated using a Breast Cancer prognosis system previously developed with health care professionals.

Chapter Nine provides and highlights the importance of using mobile technologies in health care management. In addition, the chapter presents the current practices when managing patients diagnosed with hydrocephalus, a medical condition causing headaches. The development of a NeuroDiary application is discussed, which is currently being tested as a software mobile application for collecting data from patients with hydrocephalus. The second phase development of the NeuroDiary is conveyed by adding the intelligent capability to the system, for the purpose of assisting clinicians in the diagnosis, analysis, and treatment of hydrocephalus. The authors are in the process of developing an intelligent system, which will be accessible on Android, Windows, and Apple phones to reach a wider audience of users while at the same time providing a reliable solution that addresses the needs of hydrocephalus patient management.

Chapter Ten provides for the first time a detailed step-by-step breakdown of the implementation process for mobile health (mHealth) in developing countries. A vast array of research exists that focuses on barriers of mHealth adoption in such domains. However, the majority of these papers embrace the concept of adoption to cover the entire process of implementation. This chapter acknowledges that various phases of implementation exist. As a result, the researchers identify potential barriers for each phase of mHealth implementation in developing countries. By examining existing literature, this study reveals that various sociocultural and technological factors across individuals and organizations collectively can hinder mHealth implementation in developing regions. Existing research indicates that the focus of mHealth in these constituencies, a nascent area of research, places too much emphasis on the benefits associated with mHealth implementation. Subsequently, this chapter endeavors to outline the barriers that should assist with overcoming common obstacles in the successful implementation of mHealth initiatives in developing countries.

Chapter Eleven presents different computer-aided algorithms channeled at disease diagnosis to solve the problems associated with the voluminous diseases reported and recorded. Although these algorithms have proven successful, the negative selection algorithm provides a new pathway to adequately distinguish disease-impaired patients from the healthy ones. The Variable Detector (V-Detector) by generating sets of detectors randomly with the aim of maximizing the coverage area of each detectors, is compared with Sequential Minimal Optimization (SMO), Multi-Layer Perceptron (MLP), and Non-Nested Generalized Exemplars (NNGE) on the detection of disease, and experimental results shows that the V-Detector generated the highest detection rates of 98.95% and 74.44% for Breast Cancer Wisconsin and BUPA Liver Disorder data sets, and also performed significantly with the Biomedical data at detection rate of 71.64%. Thus, the V-Detector can achieve the highest detection rate and lowest false alarm rates.

Chapter Twelve covers the application of IT for the food processing industry in the fields of smart packaging and materials, automation and control technology, standards and their application scenarios, and production management principles. Although some field data can be automatically acquired and transmitted by sensor networking, most agricultural activity information is recorded by manual handwriting for the traceability information systems. An end-to-end mobile application system that records the farming activities by using smart devices to capture information of farming operations is developed. The information for farming activities is coded in two-dimensional labels of Quick Response (QR) codes. By scanning the proper operation labels, the corresponding farming data can be captured and uploaded simultaneously to the back-end web server. The proposed mobile farming information can be implemented either as the mobile data collection tool for public traceability or a private traceability system. These two applications are verified through implementation projects in Tainan and Tianjin. The results showed that the mobile farming information can be successfully implemented. Food traceability can be more credible because of the reliability of collected farming data.

Chapter Thirteen describes and analyzes the process of incorporating a telehealth system that was proposed as part of a 6-month pilot research project within the NHS Liverpool UK. The proposed system was developed to offer remote management facility patients with Long Term Conditions (LTC). The implementation relied upon clinicians who otherwise operate a conventional care delivery method which is more hospital based. The study reflects the aim of the pilot study, which sought to strengthen the quality of primary care delivered to patients, but also to educate clinicians on the use of telehealth systems. In addition, the study provides detailed insight into user experience of the system. For both patients (including carers) and clinicians involved in the pilot project, a quantitative and qualitative exercise was carried out through paper-based questionnaire and one-to-one semistructured interviews, respectively. The sources for this analysis include information from related research and documents from the funding institutions who worked closely with the researchers to structure and deliver the overall project.

Chapter Fourteen proposes two hybrid algorithms, namely GGABC and HGABC, for breast cancer classification tasks. The simulation results from the breast cancer data set demonstrate that the proposed algorithms are able to classify women's breast cancer disease with a high accuracy and efficiently from standard algorithms. Furthermore, the proposed GGABC and HGABC algorithms are successfully used to balance the high amount of exploration and exploitation process. Besides breast cancer classification, the proposed methods may also be used for breast cancer diagnosis and medical image processing. The two proposed bio-inspired learning algorithms are explained; the experimental setup and results for cancer prediction are also conveyed and discussed.