Intelligent Personalized Abnormality Detection for Remote Health Monitoring

Intelligent Personalized Abnormality Detection for Remote Health Monitoring

Poorani Marimuthu (Madras Institute of Technology, Anna University, India), Varalakshmi Perumal (Madras Institute of Technology, Anna University, India) and Vaidehi Vijayakumar (Vellore Institute of Technology, India)
Copyright: © 2020 |Pages: 23
DOI: 10.4018/IJIIT.2020040105

Abstract

Machine learning algorithms are extensively used in healthcare analytics to learn normal and abnormal patterns automatically. The detection and prediction accuracy of any machine learning model depends on many factors like ground truth instances, attribute relationships, model design, the size of the dataset, the percentage of uncertainty, the training and testing environment, etc. Prediction models in healthcare should generate a minimal false positive and false negative rate. To accomplish high classification or prediction accuracy, the screening of health status needs to be personalized rather than following general clinical practice guidelines (CPG) which fits for an average population. Hence, a personalized screening model (IPAD – Intelligent Personalized Abnormality Detection) for remote healthcare is proposed that tailored to specific individual. The severity level of the abnormal status has been derived using personalized health values and the IPAD model obtains an area under the curve (AUC) of 0.907.
Article Preview
Top

Introduction

Abnormality detection in health plays a main role in various fields of medicines like smart healthcare and intelligent remote healthcare systems. Many early warning score systems have been developed and being implemented for abnormality detection in human well-being by many countries. Initially, data mining techniques were used to detect the unknown pattern in the health data which shows the abnormality in health status. Today the data mining techniques have been evolved into data analytics which tries to take intelligent decision based on the useful and new patterns. The survey (Acampora et al., 2013; Aranki et al., 2016; Priyatharshini and Chitrakala, 2019) shows that many research works have been done for diagnosing various diseases with the help of medical images like MRI, CT, Ultrasound, X – rays etc. and these research outcomes are very useful in the field of medicine. Abnormality detection models including machine learning, evolutionary computation and fuzzy based approaches are continuously arriving in this field. The support of IoT (Hassanalieragh et al., 2015; Pang, 2013), cloud and fog for medical diagnosis (Mutlag et al., 2019) again increases the demand for highly optimized abnormality detection models and the field is continuously growing.

Most of the abnormality detection approaches considers machine learning techniques to learn the patterns acquired from the vital signs. Supervised learning classification models such as Neural Network, Decision Trees and Hidden Markov Model etc., can be used to improve classification accuracy for better decision making. All these decision-making approaches has the drawbacks such as architecture complexity, poor generalization, less efficiency and large model size. In real time scenario, the design of high accuracy early alert generation is feasible only if the abnormality is categorized based on the severity level. This forces the need for highly efficient dynamic learning classification algorithms. Recently, data analysis with soft computing learning algorithms is preferably used for anomaly detection and decision making. Health parameters are nonlinear and dynamic as the environment is subject to continuous changes. Hence, it is hard to describe the input and output relationship of the health data apriori. Dynamic learning process is necessary to adjust the model behavior according to the online health data stream which reduce the false alarm. Soft computing approach is highly recommendable for handling these dynamic and complex real-world applications.

Need for the Work

Remote Health Monitoring (RHM) (Brown, 2001) falls under two categories, Vision Based Remote Health Monitoring System (V-RHMS) and Wearable Sensor based Remote Health Monitoring System (WS-RHMS) (Huang et al., 2009; Majumder et al., 2017; Patel et al., 2012). Although, vision-based health monitoring system provides a good abnormality detection accuracy, no person wants to be monitored all the time using camera which leads the researchers to have more attention on the development of wearable sensor-based health monitoring models. In general, the vital health parameters show a considerable deviation in their values with respect to activity of the person, age, gender, pre / post-surgery etc. These deviations can be noted and early diagnosis of abnormality could be done which paves the way for rehabilitation and timely medical care. This leads the proposed work to be based on wearable sensor-based health monitoring system.

Mostly, the abnormality detection methodologies using wearable sensor readings are done based on general or normal health values, but in reality, the values are different for each person. For the same individual itself there will be a change in the health parameter values as per his / her age and environmental conditions. Taking a decision based on generalized values leads to a high false positive and false negative rate which is not encouraged in the medical field.

Complete Article List

Search this Journal:
Reset
Open Access Articles: Forthcoming
Volume 17: 4 Issues (2021): Forthcoming, Available for Pre-Order
Volume 16: 4 Issues (2020): 3 Released, 1 Forthcoming
Volume 15: 4 Issues (2019)
Volume 14: 4 Issues (2018)
Volume 13: 4 Issues (2017)
Volume 12: 4 Issues (2016)
Volume 11: 4 Issues (2015)
Volume 10: 4 Issues (2014)
Volume 9: 4 Issues (2013)
Volume 8: 4 Issues (2012)
Volume 7: 4 Issues (2011)
Volume 6: 4 Issues (2010)
Volume 5: 4 Issues (2009)
Volume 4: 4 Issues (2008)
Volume 3: 4 Issues (2007)
Volume 2: 4 Issues (2006)
Volume 1: 4 Issues (2005)
View Complete Journal Contents Listing