Interpretable Cognitive Internet of Things for Healthcare

Discussing the use of artificial intelligence (AI) in healthcare, explainability is a highly contentious topic. AI-powered systems may be superior at certain analytical tasks, but their lack of explanation continues to breed distrust. Because the majority of existing AI systems are incomprehensible and opaque, it is unlikely that AI technologies will be properly exploited and incorporated into standard clinical practice. We begin by discussing the present state of XAI development, with a focus on its applications in healthcare. Numerous IoHT-related linked health applications have been examined in XAI to establish their privacy, security, and explainability effectiveness. If we employ clinical decision assistance systems (CDAS) based on artificial intelligence, our approach will combine legal, technological, patient, and medical considerations. To gain a better grasp of the significance of explainability in clinical practice, several disciplines focus on distinct fundamental concerns and values. Explainability must be technically appraised in terms of how it could be attained and what it entails for future development. Important legal checkpoints for explainability include informed consent, certification, and licensing for medical equipment. It is important to look at the relationship between medical AI and people from both the patient’s and the doctor’s points of view.

ARIMA model, an interpretable model, is a statistical machine learning approach that has been used for various predictions in India. From the initial stage of COVID-19, various researchers have been trying to predict with various models. ARIMA is found to be a suitable model for this COVID-19 dataset, though, with little data, no good results were obtained, but having 4-month data the results obtained were promising. ARIMA (4,0,2) model is the best fit for India COVID-19 confirm cases; ARIMA (2,0,2) is a better model for India death cases. Further, a study on four states was carried out using ARIMA, and it was found that ARIMA (2,0,4) is best for Andhra Pradesh, ARIMA (3,0,8) for Delhi confirmed cases. Maharashtra state model was ARIMA (2,0,2), and Tamil Nadu had a better fit model with ARIMA (2,0,0). The models were found to be statistically significant. The predicted case of each state is mapped using GIS for the month September 15, October 15, and November 15, 2020.

Recently, many innovations have been experienced in healthcare by rapidly growing Internet-of-Things (IoT) technology that provides significant developments and facilities in the health sector and improves daily human life. The IoT bridges people and information technology and speeds up shopping. For these reasons, IoT technology has started to be used on a large scale. Thanks to the use of IoT technology in health services, chronic disease monitoring, health monitoring, rapid intervention, early diagnosis and treatment, etc., facilitate the delivery of health services. However, the data transferred to the digital environment pose a threat of privacy leakage. Unauthorized persons have used them, and there have been malicious attacks on the health and privacy of individuals. In this chapter, it is aimed to propose a model to handle the privacy problems based on federated learning. Besides, we apply secure multi-party computation. Our proposed model presents an extensive privacy and data analysis and achieves high performance.

Coronavirus disease in 2019 (COVID19) was caused by severe acute coronavirus syndrome 2. (SARS-CoV-2). It was found in December 2019 in Wuhan, Hubei, China, and has since spread to the rest of the world, causing the latest pandemic. By the 23rd of August 2020, over 23.3 million accidents have been registered in 188 countries and territories, resulting in over 806,000 fatalities. About 15 million people have been rehabilitated. Popular symptoms include coughing, toxins, tiredness, shortness of breath, and a loss of smell and taste. Keeping a healthy distance is the most important way to prevent the spread of this virus. The word “public-health social distance” applies to a category of non-pharmaceutical procedures or programs that are intended to prevent the spread of infectious disease by maintaining a physical distance between individuals. To our knowledge, there is no social distancing tool that can be used to detect social distancing follow-up in real-time images. In this chapter, we introduce our computer-vision-based social distancing tool, which can be used to monitor the follow-up of social distancing in real-time photographs. The findings of the real-time social distancing video can be seen at https://​github.​com/​ahmadusmani12/​Tutorials.

Explainable artificial intelligence (XAI) involves a collection of processes and approaches which enables human users to comprehend and trust the results and output produced by machine learning (ML) approaches. XAI is employed for describing the AI model, its expected impact, and potential biases. At the same time, Internet of Healthcare Things (IoHT) has become a hot research topic in the healthcare sector which assist in the disease diagnostic process. Presently, an efficient computer-aided diagnosis (CAD) model is needed for diagnosing osteoarthritis (OA). This study designs a new particle swarm optimization (PSO) model with deep neural network (DNN), named PSO-DNN technique, for the identification and categorization of osteoarthritis from the knee X-ray images in an IoHT environment. The presented method helps to distinguish between well and diseased knee X-ray images. Here, a guided filter (GF) and adaptive histogram equalization models are correspondingly employed to remove noises and enhance the images. Global thresholding-based segmentation model is employed for extracting the synovial cavity regions from the image, and curvature values are determined. For drawing a good validation, the experimentation takes place on the real-time patient-oriented images gathered from the medical organizations. From the simulation outcome, the presented PSO-DNN model confirmed the superior performance of the applied images.

Cholesterol is an oil-like substance that is found in the membranes of animal cells and also carried in blood plasma, which has some vital functions in the human body, especially in the endocrine and digestive systems. Very low-density lipoprotein (VLDL) is a lipid that is not gained with nutrients, but it is instead produced by the body itself. However, it is considered to be in the bad cholesterol group since this type of cholesterol threatens cardiovascular health. As a result, it is normally expected to be at the lowest levels in the human body. In this study, some interpretable machine learning techniques are applied to estimate VLDL cholesterol value by using some attributes such as age, sex, creatinine, aspartate transaminase (AST), alanine transaminase (ALT), free t4, glucose, and triglyceride. In this, the techniques include the generalized linear model (GLM), decision tree (DT), and gradient boosted trees (GBT). It is computed that GLM has the root-mean-squared-error value of 0.655 and the correlation value of 1.0 so consequently returns the best results compared to others.

Medical monitoring systems in hospitals and many other medical centers are experiencing significant growth, and portable healthcare monitoring systems with the latest technology are of great concern in many countries around the world today. These type of monitoring systems are mainly based on the Inter net of Things (IoT). IoT provides the integration of multiple sensors and objects that can directly communicate with each other without human intervention. This chapter discusses the smart IoT devices with the current web technologies that are used for monitoring purposes. It also discusses smart personal healthcare systems with issues and challenges of such systems.

The integration of Big Data Analytics (BDA) and the Internet of Medical Things (IoMT) has brought a significant transformation in the healthcare industry. The emergence of Explainable Artificial Intelligence (XAI) has further revolutionized the healthcare sector by providing insights into complex machine learning models. This survey aims to explore the application of BDA and IoMT in healthcare with a view on XAI.

The survey highlights the benefits of BDA and IoMT in healthcare, such as improved patient outcomes, reduced healthcare costs, and enhanced personalized medicine. It also discusses the challenges associated with the use of BDA and IoMT, including data privacy, security, and regulatory compliance. The survey provides an overview of the latest research and development in the field of XAI, with particular focus on its application in healthcare.

Furthermore, the survey presents a detailed analysis of the existing literature on the integration of BDA, IoMT, and XAI in healthcare. It discusses the various applications of BDA, IoMT, and XAI in healthcare, such as medical imaging, drug discovery, diagnosis, and treatment planning. The survey also highlights the potential benefits of XAI in healthcare, including transparency, interpretability, and fairness.

Finally, the survey concludes by discussing the future research directions in the field of BDA, IoMT, and XAI in healthcare. It emphasizes the need for ethical guidelines and best practices for the responsible use of BDA, IoMT, and XAI in healthcare to ensure patient safety and privacy. The survey provides valuable insights into the integration of BDA, IoMT, and XAI in healthcare and their potential to revolutionize the healthcare industry.

Today, as a result of the development of remote sensing techniques, the importance of using sensing sensors on unmanned ground vehicles (UGVs) has increased. Developing sensor technologies are used in many different areas, from natural disasters to the defense industry. Here, among many different scenarios, it is too important to provide immediate medical first aid at the time of disasters. The unmanned ground vehicle, thanks to its sensors, is able to recognize its environment and transfer the correct data about the environment to the relevant people or institutions, preventing possible bad scenarios. Autonomous robots used today have insufficient mobility or sensing techniques and are costly for individuals or institutions. In this study, it is aimed that the developed unmanned ground vehicle can be easily accessed in environments where it is planned to detect people needing first aid, thanks to the sensor techniques to be used and to perform its task more effectively by recognizing the relevant environment. However, the low cost of the developed unmanned ground vehicle is very important. In the developed system, a LIDAR laser scanner sensor is used to model the environment where the unmanned ground vehicle is located. Within the scope of the study, 3D environmental modeling was carried out using 2D LIDAR. In the system design, the environment definition has been enriched by using the image processing technique and infrared camera. The motor driving operations of the unmanned ground vehicle and the control of various peripherals are provided by the Arduino microcontroller. LIDAR and camera are controlled on Raspberry Pi embedded system computer. All data from the LIDAR sensor, camera, motor driver, and other peripherals are displayed and controlled in a single interface via the developed mobile application. As a result of the study, an ergonomic, safe, integrated robot design that will reduce financial resources for the organizations targeted to use unmanned ground vehicles, where the user can monitor the dangerous environments remotely and recognize the relevant environment, has been created. Furthermore, thanks to the formed IoT synergy, the detection of people needing first aid in unreachable places can be ensured easily with an unmanned solution.

The emerging communication and technologies have affected manifolds across all industries and businesses. However, the medical industry and healthcare have embodied changes in patient care, medical innovations, personal care, remote consultation, diagnosis, and clinical trials.

We have witnessed a technological transformation in healthcare in terms of remotely accessible, cost-effective solutions for doctors, patients, pharmacists, and hospitals, such as smart wearable Internet of Things (IoT) devices, remote surgery using IoT devices, virtual consultation, remote medical services, intelligent clinics, EMR, and EHR to name a few. The present market study shows the rise in IoT-based diverse application-specific and smart healthcare solutions to combat the present pandemic COVID-19. The IoT provides noninvasive healthcare solutions, real-time patient data acquisition through sensors, and quality of care and remote physiological monitoring of patients. IoT in healthcare market includes components, such as sensor-based medical equipment, applications such as telemedicine, medical imaging, and remote monitoring; and end users like hospitals, doctors, and patients. The market study shows that more than 75% of top-notch companies such as IBM, SAP, Google, Microsoft, and Oracle are ready to use Cognitive IoT-based systems in healthcare, the blend of IoT and artificial intelligence techniques.

This chapter includes the concept of cognitive IoT in healthcare and its key drivers, opportunities, challenges, and limitations. The further section presents the market analysis and key market players in cognitive IoT in healthcare. The chapter concludes with the impact of pandemic on cognitive IoT.