Survey on Smart Healthcare Informatics using ML.pptx

Survey on Smart Healthcare Informatics using ML

1. Survey on Smart Healthcare Informatics using ML

2. 1. HealthGuard : A Machine Learning-Based Security Framework for Smart Healthcare Systems[1] Smart Healthcare Systems(SHS)   continuously monitor different vital signs of a patient and automatically detect and prevent critical medical conditions. These increased functionalities of the SHS pose a threat to misuse of vital information Normal functioning of SHS could be impeded by False data insertion, tampering of medical sensors to change the performance during medical emergency. HealthGuard, a novel machine learning based security framework was devised by [ Newaz et al,2019[1]] to detect malicious activities in SHS It correlates the vitals collected by different connected devices in SHS to identify changes in body functions Detects Malicious activities using different ML techniques like Artificial Neural network, Decision Tree, Random Forest, k-Nearest Neighbor techniques. F1 Score -90% Accuracy -90% Research Gap : Accuracy decreases as the number of simultaneous attacks increases

3. Human activity recognition using machine learning methods in a smart healthcare environment [2] A large percentage of elderly population suffer from age-related health problems such as diabetes, cardiovascular disease, osteoarthritis, Alzheimer’s disease, dementia, or other chronic diseases which makes it very difficult for them to lead independent life. A Smart Healthcare Monitoring System(SHMS) enabled by Human Activity Recognition(HAR) systems consisting of wearable sensors and smart phone enabled monitoring is a very good solution for enabling them to live independently. These Smart HAR are context sensitive and detect abnormalities in daily activities and help in fall detection, anomalies in biometrics etc and sends the information to remote server. Data Analysis was performed on the remote server the data was classified using different ML techniques like ANN, SVM, kNN , Naïve Baye’s , C4.5 Decision Tree, Random Forest,etc . Conclusions : For wearable sensor data , SVM gives better accuracy in terms of accuracy, F-measure, ROC area, and Kappa. SVM is better than all classifier methods used in this study For Smart phone based monitored data SVM , RF, and k-NN achieved better accuracy compared to ANN, Naıve Bayes, CART, C4.5, REPTree , and LADTree HAR data taken from body sensors achieve better accuracy than the HAR data taken from the smartphone sensors. Future Prospects : HAR to recognise human interaction and interpersonal relationships, Improvement of accuracy in classification of data collected during activity and movement .

4. Review on Using Blockchain in Healthcare systems[3] Blockchain , a relatively new decentralization mechanism, provides robust solutions for security and efficiency challenges existing in healthcare systems . Advantages of blockchain for the healthcare system include secure data sharing , the consensus in changes, deletion, addition, and decentralized and immutable records modification of records Applications of Blockchain in Healthcare Electronic Medical Data Management Interoperability and Consolidated Healthcare Pharmaceutical Supply Chain Organ Transplantation and Blood Donation Insurance Medical Research/Clinical Trial Other applications like Internet of Medical things( IoMT ), Neuroscience, Patient-Centric Healthcare, Identity Verification Applications in Dental Industry, Medical prescriptions, public health surveillance systems,telesurgery , pervasive social network systems, home based healthcare systems, applications using genomic data. Limitations Scalability : Blockchain stores records of every transaction in every node and thus increases throughput rate and lowers the latency but this will eventually lead to storage limitations Security : Security of EHR systems would be threatened by the spread of IoMT devices Cost of implementation and upgradation to blockchain systems, Processing Power , Lack of Skills

5. 4. Intelligent Fog-Enabled Smart Healthcare System for Wearable Physiological Parameter Detection [4] Detection and analysis of the physiological data from wearable devices on the Cloud based servers is an essential process in smart healthcare. Cloud computing introduces latency in the analysis which can be tackled by using Fog Computing or Edge Computing Fog and Edge Computing   in help leveraging the computing capabilities within a local network to carry out computation tasks that would ordinarily have been carried out in the cloud and thus increase the efficiency of the system. This may introduce the problems of overloading of the data in the Fog systems which would again lead to latency. A novel tri-fog health architecture proposed by Ijaz et al,2020 proposes a 3 layer architecture consisting of wearable layer, intelligent fog layer, and cloud layer for physiological parameter detection which in addition to the above mentioned issues also tries to address elimination of faulty data accumulated due to environmental factors. The novel features of this system are mentioned below Fault Data elimination is done by using rapid kernel principal component analysis Data redundancy/duplication is handled by using fuzzy assisted objective optimization by ratio analysis ( FaMOORA ) algorithm T he two-level health hidden Markov model (2L-2HMM ) perrforms the timely prediction of user’s health status by analysis the temporal variations in the data collected from wearable data. The fog layer categorizes the user’s status by using SpikQ -Net , a hybrid algorithm and takes necessary action while the multi-objective spotted hyena optimization ( MoSHO ) algorithm takes care of offloading between overloaded and underloaded fog nodes to ensure lower response time in service . Future Prospects : Extending this systems features to diagnose specific diseases as this currently recognises general abnormalities, also to improve the data security in the system.

6. 5. A smart healthcare monitoring system for heart disease prediction based on ensemble deep learning and feature fusion[5] This SHMS attempts to accurately predict the heart disease which is essential for the efficient treatment of cardiac patients before a heart attack occurs by using ML models This SHMS tries to overcome the problems of high dimensional data handling and tries to reduced performance in detection faced by the contemporary models which use conventional techniques to select feature data by using ensemble deep learning and feature fusion approaches. T he feature fusion method combines the extracted features from both sensor data and electronic medical records(EMR) to generate valuable healthcare data and the information gain technique eliminates irrelevant and redundant features , and selects the most relevant features and thus improves the performance of the system by feeding this data to the ensemble deep learning model which is trained for heart disease prediction Future Prospects : T his method can be connected to different feature extraction , feature fusion, attribute selection, feature weighting, and disease–risk prediction systems, since it can extract valuable features from both structured and unstructured data. Feature fusion technique in this system can be enhanced using data mining techniques Feature reduction techniques can be redesigned to handle huge numbers of features and large volumes of healthcare records More sophisticated method can be investigated for removing irrelevant features and managing the missing values and noise to achieve efficient results.

7. 6. Vision-based personalized Wireless Capsule Endoscopy for smart healthcare[6] Wireless Capsule Endoscopy (WCE) is a patient-friendly approach for digestive tract monitoring to support medical experts towards identifying any anomaly inside human’s Gastrointestinal (GI) tract This paper reviews different methods used in detection/classification of diseases by video analysis of WCE data, analysis of public WCE datasets , trends ,open issues and future prospects as this research area is still significantly less explored by computer vision experts than other fields related to medical imaging Future prospects : Automatic/Adaptive brightness adjustment while the WCE capsule is moving through the GI tract to prevent over/under illumination of the images Controlling the movement (speed, orientation, position) of the capsule as per the physician’s choice to focus of region of interest by adopting activity recognition algorithm Capsule’s size in GI tract: Implications in Power and Movement Dynamics Ensuring the security, smartness and energy efficiency by offloading computation on cloud by encrypted data transfer to ensure security eXplainable Artificial Intelligence (XAI) techniques for image-based Deep Learning methods where in AI results are validated by the medical experts to improve the performance and trust on the WCE techniques.

8. 7. Smart materials for smart healthcare– moving from sensors and actuators to self-sustained nanoenergy nanosystems Smart materials are the state-of-the-art micro/ nano -systems that are suitable for implantable and wearable diagnostic, therapeutic and treatment applications which are also suitable for harvesting biomechanical energies from human motions , environment or body heat, or shaping of biofuel powered devices and thus make them self sustainable Powering of medical devices specially for implantable ones is the main bottleneck that is facing the healthcare technology , and this could be solved by designing devices for harvesting biomechanical energies from the body. Future Prospects : A rtificial intelligence and Machine learning techniques used in combination with these self sustainable nano sensors in various medical /non-medical applications .

9. 8. A dynamic and interoperable communication framework(DICF) for controlling the operations of wearable sensors in smart healthcare applications[8] DICF proposes to regulate the operations of wearable healthcare devices by designing a fully operative and automated seamless working of the sensing devices. The sensing devices operate in both autonomous and interconnected manner depending upon the sensed information and the observed body conditions of the patient to improve the performance of the system. A classification and regression based decision making aids event detection and emergency interval identification for improving the performance of the wearable sensor operations. DICF harmonizes data collection, event detection and analysis and communication features of WS along with decision-making system for improving the benefits of sensor dependent personal healthcare system. Future prospects of this framework : Implement on smart waste management and smart e-waste system for smart city applications

10. 9. An accurate and dynamic predictive model for a smart M-Health system using machine learning[9] Smart Mobile Health monitoring using ML techniques proposes a UI based interface in the front end on the mobile app where a user can access biometric data, share it with relevant medical personnel and receive notifications At the backend , it provides a secure content storage with user authentication performs data analysis for the early Cardiovascular diseases and is enabled with big data handling as it is Cloud based Smart Monitoring system Future prospects Extension of analysis to detect other diseases like Traumatic Brain Injury (TBI ) To perform 10 fold validation (current model has 4 fold validation)

11. 10. Consistency of Medical Data Using Intelligent Neuron Faster R-CNN Algorithm for Smart Health Care Application[10] Medical data are stored as structured and unstructured data. However, many medical data contain errors , omissions and mistakes and sometimes this leads to medical accidents due to physician errors A faster R-CNN intelligent agent cloud architecture is proposed to reduce the error rule when reading medical image data by analysis of the actual existing medical data through Conv feature map using deep ConvNet and ROI Projection This system processed and verified 12,000 medical imaging data and categorised 1000 of these as side effects, 40% of the images were labeled negative . Future prospects : extract and research more data for medical imaging in the future by introducing more cloud systems to study cancer data among these health care data for early detection

12. 11. The Internet of Things for Dementia Care[11] Dementia affects more than 46 million people around the world and patients need round the clock monitoring and TIHM aims to develop innovative living environments which empower people with dementia and their carers to enjoy better health and quality of life, with reduced dependence on institutional care. They propose to use ( IoT ) enabled solutions with a set of machine learning and data analytics algorithms generate notifications regarding the well-being of the patients. The information is monitored around the clock by a group of healthcare practitioners who take appropriate decisions according to the collected data and generated notifications A number of algorithms have been developed to learn the patients ’ daily patterns and find possible pattern deviations, detect if patients are agitated and/or irritated, and detect the possibility of Urinary Tract Infections (UTI). This provides early identification of need which has the potential to enable smart deployment of healthcare resources, better quality of care , and savings to the health and care economies Future prospects : Extension of this system to monitor and detect other specific diseases.

13. 12. Intelligent edge computing based on machine learning for smart city A mobile edge server is taken as the focus, and the available resources around the mobile edge server are used for collaborative computing to further improve the computing performance of a mobile edge computing system. Machine learning is applied to the distributed task scheduling algorithm and distributed device coordination algorithm in Mobile Edge Computing devices(MEC) The distributed task scheduling algorithm and distributed device coordination algorithm are tested by experiments . A distributed task scheduling algorithm based on an ADMM Stackelberg game algorithm and a distributed device system algorithm based on ADMM are proposed. These algorithms provide faster convergence , good stability in a large scale network, good scalability and it can avoid losing the diversity of the data L imitations: This study only considers changes in an MEC server. When there are multiple MEC servers, their scheduling policies may change.

14. References A. I. Newaz , A. K. Sikder , M. A. Rahman and A. S. Uluagac , "HealthGuard: A Machine Learning-Based Security Framework for Smart Healthcare Systems,"  2019 Sixth International Conference on Social Networks Analysis, Management and Security (SNAMS) , Granada, Spain, 2019, pp. 389-396, doi : 10.1109/SNAMS.2019.8931716 Subasi , Abdulhamit (2020).  Innovation in Health Informatics , Human activity recognition using machine learning methods in a smart healthcare environment. , 123-144.   doi:10.1016/B978-0-12-819043-2.00005-8  L . Soltanisehat , R. Alizadeh , H. Hao and K. R. Choo , "Technical, Temporal, and Spatial Research Challenges and Opportunities in Blockchain -Based Healthcare: A Systematic Literature Review," in  IEEE Transactions on Engineering Management , doi : 10.1109/TEM.2020.3013507 . Ijaz , M., Li, G., Wang, H., El- Sherbeeny , A. M., Moro Awelisah , Y., Lin, L., … Noor, A. (2020).  Intelligent Fog-Enabled Smart Healthcare System for Wearable Physiological Parameter Detection. Electronics, 9(12), 2015.   doi:10.3390/electronics9122015 Farman Ali, Shaker El- Sappagh , S.M. Riazul Islam, Daehan Kwak , Amjad Ali, Muhammad Imran, Kyung-Sup Kwak,A smart healthcare monitoring system for heart disease prediction based on ensemble deep learning and feature fusion,Information Fusion,63,2020, 208-222,https ://doi.org/10.1016/j.inffus.2020.06.008 . Khan Muhammad, Salman Khan, Neeraj Kumar, Javier Del Ser , Seyedali Mirjalili,Vision -based personalized Wireless Capsule Endoscopy for smart healthcare: Taxonomy, literature review, opportunities and challenges,Future Generation Computer Systems, 113,2020,266-280,https :// doi.org/10.1016/j.future.2020.06.048 Faezeh Arab Hassani , Qiongfeng Shi, Feng Wen, Tianyiyi He, Ahmed Haroun , Yanqin Yang, Yuqin Feng , Chengkuo Lee,Smart materials for smart healthcare– moving from sensors and actuators to self-sustained nanoenergy nanosystems , Smart Materials in Medicine , 1,2020,92-124,ISSN 2590-1834,https ://doi.org/10.1016/j.smaim.2020.07.005 . S. Baskar , P. Mohamed Shakeel , R. Kumar, M.A. Burhanuddin , R. Sampath,A dynamic and interoperable communication framework for controlling the operations of wearable sensors in smart healthcare applications,Computer Communications,149, 2020,17-26,ISSN 0140-3664,https ://doi.org/10.1016/j.comcom.2019.10.004 . Naseer Qureshi , Kashif ; Din, Sadia ; GwanggilJeon , ; Piccialli , Francesco (2020). An Accurate and Dynamic Predictive Model for a Smart M-Health System Using Machine Learning. Information Sciences, (), S0020025520306113–doi:10.1016/j.ins.2020.06.025 Kim, Seong-Kyu ; Huh, Jun-Ho (2020). Consistency of Medical Data Using Intelligent Neuron Faster R-CNN Algorithm for Smart Health Care Application. Healthcare, 8(2), 185–. doi:10.3390/healthcare8020185  S. Enshaeifar   et al ., "The Internet of Things for Dementia Care," in  IEEE Internet Computing , vol. 22, no. 1, pp. 8-17, Jan./Feb. 2018, doi : 10.1109/MIC.2018.112102418. Zhihan Lv , Dongliang Chen, Ranran Lou, Qingjun Wang, Intelligent edge computing based on machine learning for smart city, Future Generation Computer Systems, Volume 115,2021,Pages 90-99,ISSN 0167-739X,https ://doi.org/10.1016/j.future.2020.08.037.