Applications of Machine Learning in Digital Healthcare (Healthcare Technologies)

Description

Machine learning algorithms are increasingly finding applications in the healthcare sector. Whether assisting a clinician to process an individual patient's data or helping administrators view hospital bed turnover, the volume and complexity of healthcare data is a compelling reason for the development of machine learning based tools to aid in its interpretation and use. This edited book focuses on the applications of machine learning in the healthcare sector, both at the macro-level for guiding policy decisions, and at the granular level, showing how ML techniques can be applied to process an individual patient's medical data to swiftly aid diagnosis. Written by an international team of experts, the book presents several applications of machine learning in the healthcare sector, including health system planning, optimisation and preparedness, outlining the benefits and challenges of coordination and data sharing. Machine learning has many applications in processing patient data and topics such as arrhythmia detection, image-guided microsurgery and early detection of Alzheimer's disease are discussed in depth. The book also looks at machine learning applications exploiting wearable sensors for real-time analysis and concepts around enhancing physical performance. Suitable for an audience of computer scientists, healthcare engineers and those involved with digital medicine, this book brings together a plethora of machine learning applications from across the board of the healthcare services. Cover Contents About the editors 1 Introduction 1.1 Why? 1.2 How? 1.3 What is ML? 1.4 The problem 1.5 Gradient descent 1.6 Structural components of the ANN 1.6.1 The fully connected neural network 1.6.2 Convolutional neural network 1.6.3 Pooling layers 1.6.4 The SoftMax function 1.6.5 Putting them together 1.7 Training and evaluating a neural network 1.7.1 Data organisation 1.7.2 Types of errors and useful evaluation metrics 1.7.3 ADAM optimisation for bias reduction 1.7.4 Regularisation for variance reduction 1.8 Conclusion References 2 Health system planning and optimisation – advancements in the application of machine learning to policy decisions in global health 2.1 Model-based decision making 2.2 ML surrogates for prediction from epidemiological models 2.2.1 Gaussian process regression 2.2.2 Action-value function example 2.2.3 Epidemiological model calibration 2.2.4 Bayesian optimisation 2.3 Online learning 2.3.1 Stochastic multi-armed Bandit 2.4 Running epidemiological simulations as Bandits 2.4.1 Time 2.4.2 State 2.4.3 Action 2.4.4 Reward 2.4.5 Bandit approaches for simulated learning 2.4.6 Extensions to online learning 2.5 Reinforcement learning 2.5.1 State 2.5.2 Action 2.5.3 Reward 2.5.4 Markov decision processes 2.5.5 Cumulated return 2.5.6 Policy 2.5.7 Value function 2.5.8 Partially observable MDP (POMDP) 2.5.9 Learning sequential surrogate models from episodic simulators 2.5.10 Prediction – learning a value function 2.5.11 Simulation-based search – decision trees 2.5.12 Monte Carlo tree search (MCTS) 2.5.13 Gaussian process regression with selection inMCTS for learning sequential surrogates (GP–MCTS) 2.6 Control – optimal policies and planning 2.6.1 Optimal policy learning 2.7 Comparing predictions from multi-step and one-step methods with direct experience References 3 Health system preparedness – coordination and sharing of computation, models and data 3.1 Computation 3.1.1 A proposed infrastructure 3.1.2 Platform components 3.1.3 Performance results 3.1.4 Example: technical approach for competitions 3.1.5 Environment web service 3.1.6 Competition API 3.1.7 Example code 3.1.8 Related work 3.2 ML competitions for health system preparedness 3.3 Planning from learnt models 3.4 KDD Cup 2019 and other competitions 3.4.1 Evaluation framework 3.4.2 Submission and scoring 3.4.3 Other competitions 3.5 Collaboration from competition 3.6 Example: analysis of successful competition approaches 3.6.1 Conclusions on competitions for health system planning 3.6.2 Human-in-the-loop References 4 Applications of machine learning for image-guided microsurgery 4.1 Preoperative data collection 4.2 Preprocessing 4.2.1 Intensity histograms 4.2.2 Noise reduction 4.2.3 Contrast adjustment 4.2.4 Preprocessing review 4.3 Segmentation 4.3.1 Thresholding 4.3.2 Region-based thresholding 4.3.3 Edge-based thresholding 4.3.4 Post-processing 4.3.5 Validation 4.4 Registration 4.4.1 Image labeling 4.4.2 Feature identification 4.4.3 Feature matching 4.4.4 Transformation 4.5 Visualization 4.5.1 Real-time motion tracking 4.5.2 Overlaying 4.5.3 Image-guided microscopic surgery system 4.5.4 Augmented-reality-based microsurgical systems 4.6 Challenges 4.6.1 Infrastructure challenges 4.6.2 Safety challenges 4.6.3 Cost challenges 4.7 Chapter review References 5 Electrophysiology and consciousness: a review 5.1 Introduction 5.2 Nervous system signals 5.2.1 Central nervous system 5.2.2 ANS 5.2.3 CNS–ANS connection in physiological mechanisms 5.3 Neurophysiological signal recording 5.3.1 Recording the electroencephalogram (EEG) 5.3.2 Recording the ECG 5.4 Applications of biopotentials in health and disease 5.4.1 Neurodegeneration 5.4.2 Anesthesia 5.4.3 Peri-operative stress 5.5 Analysis tools 5.5.1 ECG analysis 5.5.2 EEG analysis methods 5.5.3 Machine learning methods 5.6 Conclusion References 6 Brain networking and early diagnosis of Alzheimer's disease with machine learning 6.1 Background 6.1.1 A brief history of brain study 6.1.2 Modern understanding of the brain 6.2 Laboratory model of brain connectivity 6.3 Problem definition 6.4 Devices used in AD diagnosis 6.5 Data types 6.6 Data preprocessing of MRI data 6.6.1 Median filters 6.6.2 Physiological noise removal by means of deconvolution 6.6.3 Image fusion 6.7 Machine learning for early AD diagnosis 6.7.1 SVMs 6.7.2 Deep learning 6.7.3 SVM techniques 6.7.4 Deep learning techniques 6.8 Conclusion References 7 From classic machine learning to deep learning advances in atrial fibrillation detection 7.1 Physiology essentials 7.1.1 The healthy heart 7.1.2 Atrial fibrillation 7.2 Detection of AF 7.2.1 AF detection based on beat-to-beat irregularities 7.2.2 AF detection based on the ECG waveform morphology and hybrid methods 7.3 Conclusions References 8 Dictionary learning techniques for left ventricle (LV) analysis and fibrosis detection in cardiac magnetic resonance imaging (MRI) 8.1 Introduction 8.2 Basics of dictionary learning 8.2.1 Probabilistic methods 8.2.2 Clustering-based methods 8.2.3 Parametric training methods 8.3 DL in medical imaging – fibrosis detection in cardiac MRI 8.4 HCM and fibrosis 8.4.1 Myocardial fibrosis in HCM 8.5 Cardiac magnetic resonance imaging with LGE-MRI 8.6 The assessment of cardiac fibrosis detection in LGE-MRI: a brief state-of-the-art 8.7 The proposed method 8.7.1 Feature extraction 8.7.2 Clustering 8.7.3 DL-based classification: training stage 8.7.4 DL-based classification: testing stage 8.8 First experiments and results 8.8.1 Study population 8.8.2 Results 8.8.3 Evaluation 8.9 Qualification and quantification of myocardial fibrosis: a first proposal 8.10 Conclusion References 9 Enhancing physical performance with machine learning 9.1 Introduction 9.2 Physical performance and data science 9.2.1 Physical performance overview 9.2.2 The role of data in physical performance 9.2.3 Why ML? 9.3 Contextualise physical performance factors: ML perspectives 9.3.1 Training 9.3.2 Nutrition 9.3.3 Sleep and recovery 9.4 ML modelling for physical performance problems 9.4.1 Choosing ML models for the right physical performance tasks 9.4.2 Contributing ML features and methods 9.4.3 Challenges 9.5 Limitation 9.6 Conclusion References 10 Wearable electrochemical sensors and machine learning for real-time sweat analysis 10.1 Electrochemical sensors: the next generation of wearables 10.2 The mechanisms and content of sweat 10.3 Considerations for on-body sweat analysis 10.3.1 Sweat gland densities and sweat rates 10.3.2 Sweat collection techniques and challenges 10.4 Current trends in wearable electrochemical sweat sensors 10.4.1 Common features of wearable sweat sensors 10.4.2 Opportunities for ISFETs and machine learning in wearable sweat sensing 10.5 The ion-sensitive field-effect transistor 10.5.1 The fundamental theory of ISFETs 10.5.2 ISFETs in CMOS 10.5.3 ISFETs in CMOS for sweat sensing 10.5.4 Existing ISFET-based wearable sweat sensors 10.6 Applications of machine learning in wearable electrochemical sensors 10.6.1 Existing research into ML for biosensors 10.6.2 Existing research into ML for ISFETs 10.6.3 Integration of analogue classifiers with ISFETs in CMOS 10.7 Summary and conclusions Acknowledgements References 11 Last words 11.1 Introduction 11.2 A review of the state-of-the-art 11.3 Implementation and deployment 11.3.1 Traditional computing and the memory hierarchy 11.3.2 Graphics processing unit 11.3.3 Hardware accelerators 11.4 Regulatory landscape 11.4.1 A brief interlude 11.4.2 Software development life cycle 11.4.3 Risk management in medical software development 11.4.4 Challenges specific to ML 11.5 Conclusion References Index Back Cover