Cardiovascular data analytics for real time patient monitoring
- Authors: Allami, Ragheed
- Date: 2017
- Type: Text , Thesis , PhD
- Full Text:
- Description: Improvements in wearable sensor devices make it possible to constantly monitor physiological parameters such as electrocardiograph (ECG) signals for long periods. Remote patient monitoring with wearable sensors has an important role to play in health care, particularly given the prevalence of chronic conditions such as cardiovascular disease (CVD)—one of the prominent causes of morbidity and mortality worldwide. Approximately 4.2 million Australians suffer from long-term CVD with approximately one death every 12 minutes. The assessment of ECG features, especially heart rate variability (HRV), represents a non-invasive technique which provides an indication of the autonomic nervous system (ANS) function. Conditions such as sudden cardiac death, hypertension, heart failure, myocardial infarction, ischaemia, and coronary heart disease can be detected from HRV analysis. In addition, the analysis of ECG features can also be used to diagnose many types of life-threatening arrhythmias, including ventricular fibrillation and ventricular tachycardia. Non-cardiac conditions, such as diabetes, obesity, metabolic syndrome, insulin resistance, irritable bowel syndrome, dyspepsia, anorexia nervosa, anxiety, and major depressive disorder have also been shown to be associated with HRV. The analysis of ECG features from real time ECG signals generated from wearable sensors provides distinctive challenges. The sensors that receive and process the signals have limited power, storage and processing capacity. Consequently, algorithms that process ECG signals need to be lightweight, use minimal storage resources and accurately detect abnormalities so that alarms can be raised. The existing literature details only a few algorithms which operate within the constraints of wearable sensor networks. This research presents four novel techniques that enable ECG signals to be processed within the limitations of resource constraints on devices to detect some key abnormalities in heart function. - The first technique is a novel real-time ECG data reduction algorithm, which detects and transmits only those key points that are critical for the generation of ECG features for diagnoses. - The second technique accurately predicts the five-minute HRV measure using only three minutes of data with an algorithm that executes in real-time using minimal computational resources. - The third technique introduces a real-time ECG feature recognition system that can be applied to diagnose life threatening conditions such as premature ventricular contractions (PVCs). - The fourth technique advances a classification algorithm to enhance the performance of automated ECG classification to determine arrhythmic heart beats based on noisy ECG signals. The four novel techniques are evaluated in comparison with benchmark algorithms for each task on the standard MIT-BIH Arrhythmia Database and with data generated from patients in a major hospital using Shimmer3 wearable ECG sensors. The four techniques are integrated to demonstrate that remote patient monitoring of ECG using HRV and ECG features is feasible in real time using minimal computational resources. The evaluation show that the ECG reduction algorithm is significantly better than existing algorithms that can be applied within sensor nodes, such as time-domain methods, transformation methods and compressed sensing methods. Furthermore, the proposed ECG reduction is found to be computationally less complex for resource constrained sensors and achieves higher compression ratios than existing algorithms. The prediction of a common HRV measure, the five-minute standard deviation of inter-beat variations (SDNN) and the accurate detection of PVC beats was achieved using a Count Data Model, combined with a Poisson-generated function from three-minute ECG recordings. This was achieved with minimal computational resources and was well suited to remote patient monitoring with wearable sensors. The PVC beats detection was implemented using the same count data model together with knowledge-based rules derived from clinical knowledge. A real-time cardiac patient monitoring system was implemented using an ECG sensor and smartphone to detect PVC beats within a few seconds using artificial neural networks (ANN), and it was proven to provide highly accurate results. The automated detection and classification were implemented using a new wrapper-based hybrid approach that utilized t-distributed stochastic neighbour embedding (t-SNE) in combination with self-organizing maps (SOM) to improve classification performance. The t-SNE-SOM hybrid resulted in improved sensitivity, specificity and accuracy compared to most common hybrid methods in the presence of noise. It also provided a better, more accurate identification for the presence of many types of arrhythmias from the ECG recordings, leading to a more timely diagnosis and treatment outcome.
- Description: Doctor of Philosophy
- Authors: Allami, Ragheed
- Date: 2017
- Type: Text , Thesis , PhD
- Full Text:
- Description: Improvements in wearable sensor devices make it possible to constantly monitor physiological parameters such as electrocardiograph (ECG) signals for long periods. Remote patient monitoring with wearable sensors has an important role to play in health care, particularly given the prevalence of chronic conditions such as cardiovascular disease (CVD)—one of the prominent causes of morbidity and mortality worldwide. Approximately 4.2 million Australians suffer from long-term CVD with approximately one death every 12 minutes. The assessment of ECG features, especially heart rate variability (HRV), represents a non-invasive technique which provides an indication of the autonomic nervous system (ANS) function. Conditions such as sudden cardiac death, hypertension, heart failure, myocardial infarction, ischaemia, and coronary heart disease can be detected from HRV analysis. In addition, the analysis of ECG features can also be used to diagnose many types of life-threatening arrhythmias, including ventricular fibrillation and ventricular tachycardia. Non-cardiac conditions, such as diabetes, obesity, metabolic syndrome, insulin resistance, irritable bowel syndrome, dyspepsia, anorexia nervosa, anxiety, and major depressive disorder have also been shown to be associated with HRV. The analysis of ECG features from real time ECG signals generated from wearable sensors provides distinctive challenges. The sensors that receive and process the signals have limited power, storage and processing capacity. Consequently, algorithms that process ECG signals need to be lightweight, use minimal storage resources and accurately detect abnormalities so that alarms can be raised. The existing literature details only a few algorithms which operate within the constraints of wearable sensor networks. This research presents four novel techniques that enable ECG signals to be processed within the limitations of resource constraints on devices to detect some key abnormalities in heart function. - The first technique is a novel real-time ECG data reduction algorithm, which detects and transmits only those key points that are critical for the generation of ECG features for diagnoses. - The second technique accurately predicts the five-minute HRV measure using only three minutes of data with an algorithm that executes in real-time using minimal computational resources. - The third technique introduces a real-time ECG feature recognition system that can be applied to diagnose life threatening conditions such as premature ventricular contractions (PVCs). - The fourth technique advances a classification algorithm to enhance the performance of automated ECG classification to determine arrhythmic heart beats based on noisy ECG signals. The four novel techniques are evaluated in comparison with benchmark algorithms for each task on the standard MIT-BIH Arrhythmia Database and with data generated from patients in a major hospital using Shimmer3 wearable ECG sensors. The four techniques are integrated to demonstrate that remote patient monitoring of ECG using HRV and ECG features is feasible in real time using minimal computational resources. The evaluation show that the ECG reduction algorithm is significantly better than existing algorithms that can be applied within sensor nodes, such as time-domain methods, transformation methods and compressed sensing methods. Furthermore, the proposed ECG reduction is found to be computationally less complex for resource constrained sensors and achieves higher compression ratios than existing algorithms. The prediction of a common HRV measure, the five-minute standard deviation of inter-beat variations (SDNN) and the accurate detection of PVC beats was achieved using a Count Data Model, combined with a Poisson-generated function from three-minute ECG recordings. This was achieved with minimal computational resources and was well suited to remote patient monitoring with wearable sensors. The PVC beats detection was implemented using the same count data model together with knowledge-based rules derived from clinical knowledge. A real-time cardiac patient monitoring system was implemented using an ECG sensor and smartphone to detect PVC beats within a few seconds using artificial neural networks (ANN), and it was proven to provide highly accurate results. The automated detection and classification were implemented using a new wrapper-based hybrid approach that utilized t-distributed stochastic neighbour embedding (t-SNE) in combination with self-organizing maps (SOM) to improve classification performance. The t-SNE-SOM hybrid resulted in improved sensitivity, specificity and accuracy compared to most common hybrid methods in the presence of noise. It also provided a better, more accurate identification for the presence of many types of arrhythmias from the ECG recordings, leading to a more timely diagnosis and treatment outcome.
- Description: Doctor of Philosophy
ECG reduction for wearable sensor
- Allami, Ragheed, Stranieri, Andrew, Balasubramanian, Venki, Jelinek, Herbert
- Authors: Allami, Ragheed , Stranieri, Andrew , Balasubramanian, Venki , Jelinek, Herbert
- Date: 2016
- Type: Text , Conference proceedings
- Relation: 2016 12th International Conference on Signal-Image Technology & Internet-Based Systems (SITIS); Naples, Italy; 28th November-1st December 2016 p. 520-525
- Full Text:
- Reviewed:
- Description: The transmission, storage and analysis of electrocardiogram (ECG) data in real-time is essential for remote patient monitoring with wearable ECG devices and mobile ECG contexts. However, this remains a challenge to achieve within the processing power and the storage capacity of mobile devices. ECG reduction algorithms have an important role to play in reducing the processing requirements for mobile devices, however many existing ECG reduction and compression algorithms are computationally expensive to execute in mobile devices and have not been designed for real-time computation and incremental data arrival. In this paper, we describe a computationally naive, yet effective, algorithm that achieves high ECG reduction rates while maintaining key diagnostic features including PR, QRS, ST, QT and RR intervals. While reduction does not enable ECG waves to be reproduced, the ability to transmit key indicators (diagnostic features) using minimal computational resources, is particularly useful in mobile health contexts involving power constrained sensors and devices. Results of the proposed reduction algorithm indicate that the proposed algorithm outperforms other ECG reduction algorithms at a reduction/compression ratio (CR) of 5:1. If power or processing capacity is low, the algorithm can readily switch to a compression ratio of up to 10: 1 while still maintaining an error rate below 10%.
- Authors: Allami, Ragheed , Stranieri, Andrew , Balasubramanian, Venki , Jelinek, Herbert
- Date: 2016
- Type: Text , Conference proceedings
- Relation: 2016 12th International Conference on Signal-Image Technology & Internet-Based Systems (SITIS); Naples, Italy; 28th November-1st December 2016 p. 520-525
- Full Text:
- Reviewed:
- Description: The transmission, storage and analysis of electrocardiogram (ECG) data in real-time is essential for remote patient monitoring with wearable ECG devices and mobile ECG contexts. However, this remains a challenge to achieve within the processing power and the storage capacity of mobile devices. ECG reduction algorithms have an important role to play in reducing the processing requirements for mobile devices, however many existing ECG reduction and compression algorithms are computationally expensive to execute in mobile devices and have not been designed for real-time computation and incremental data arrival. In this paper, we describe a computationally naive, yet effective, algorithm that achieves high ECG reduction rates while maintaining key diagnostic features including PR, QRS, ST, QT and RR intervals. While reduction does not enable ECG waves to be reproduced, the ability to transmit key indicators (diagnostic features) using minimal computational resources, is particularly useful in mobile health contexts involving power constrained sensors and devices. Results of the proposed reduction algorithm indicate that the proposed algorithm outperforms other ECG reduction algorithms at a reduction/compression ratio (CR) of 5:1. If power or processing capacity is low, the algorithm can readily switch to a compression ratio of up to 10: 1 while still maintaining an error rate below 10%.
A count data model for heart rate variability forecasting and premature ventricular contraction detection
- Allami, Ragheed, Stranieri, Andrew, Balasubramanian, Venki, Jelinek, Herbert
- Authors: Allami, Ragheed , Stranieri, Andrew , Balasubramanian, Venki , Jelinek, Herbert
- Date: 2017
- Type: Text , Journal article
- Relation: Signal Image and Video Processing Vol. 11, no. 8 (2017), p. 1427-1435
- Full Text:
- Reviewed:
- Description: Heart rate variability (HRV) measures including the standard deviation of inter-beat variations (SDNN) require at least 5 min of ECG recordings to accurately measure HRV. In this paper, we predict, using counts data derived from a 3-min ECG recording, the 5-min SDNN and also detect premature ventricular contraction (PVC) beats with a high degree of accuracy. The approach uses counts data combined with a Poisson-generated function that requires minimal computational resources and is well suited to remote patient monitoring with wearable sensors that have limited power, storage and processing capacity. The ease of use and accuracy of the algorithm provide opportunity for accurate assessment of HRV and reduce the time taken to review patients in real time. The PVC beat detection is implemented using the same count data model together with knowledge-based rules derived from clinical knowledge.
- Authors: Allami, Ragheed , Stranieri, Andrew , Balasubramanian, Venki , Jelinek, Herbert
- Date: 2017
- Type: Text , Journal article
- Relation: Signal Image and Video Processing Vol. 11, no. 8 (2017), p. 1427-1435
- Full Text:
- Reviewed:
- Description: Heart rate variability (HRV) measures including the standard deviation of inter-beat variations (SDNN) require at least 5 min of ECG recordings to accurately measure HRV. In this paper, we predict, using counts data derived from a 3-min ECG recording, the 5-min SDNN and also detect premature ventricular contraction (PVC) beats with a high degree of accuracy. The approach uses counts data combined with a Poisson-generated function that requires minimal computational resources and is well suited to remote patient monitoring with wearable sensors that have limited power, storage and processing capacity. The ease of use and accuracy of the algorithm provide opportunity for accurate assessment of HRV and reduce the time taken to review patients in real time. The PVC beat detection is implemented using the same count data model together with knowledge-based rules derived from clinical knowledge.
- «
- ‹
- 1
- ›
- »