Decision support system for management of remotely monitored patients
Telehealth is the provision of health services at a distance. Typically this occurs in unsupervised or remote environments such as a patient's home. Patients record, in an unsupervised fashion, scalar parameter values, such as body temperature, weight and blood glucose; waveform measurements, such as a single channel electrocardiogram (ECG), blood pressure (BP) waveforms (both auscultatory and oscillometric), pulse oximetry photoplethysmogram, ambulation and possible falls events from triaxial accelerometery, and forced and relaxed spirometry signals. Recorded questionnaire data is also recorded which can convey information about patient treatment (including changes in medication), clinical symptoms and general feeling of well-being. From the waveform data, a number of parameters may be automatically estimated; such as heart rate (HR) from the ECG; systolic and diastolic blood pressure, and heart rate from the BP signals; arterial oxygen saturation (SpO2) and heart rate from the photoplethysmogram; energy expenditure and incidence of stumbles and near falls from the triaxial accelerometry; forced vital capacity (FVC) and forced expiratory volume in one second (FEV1) from the forced spirometry flow signal; inspiratory vital capacity (IVC), tidal volume (VT) and respiration rate (RR) from the relaxed spirometry flow signal.
Management of large numbers of remotely monitored patients may be assisted and expedited through the use of a decision support system (DSS). The primary role for such a DSS is to provide an effective means to reduce the data overload and to provide a health risk score to allow appropriate targeting of clinical resources to best manage the health of the patient. In this way, the system may ultimately influence changes in work flow by targeting scarce human resources to patients of most need.
One novel aspect of the our implementation is that raw waveform data are also recorded as part of the measurement process. This allows the design and implementation of biosignal processing algorithms that examine aspects of signal quality and poor measurement technique. Due to the unsupervised measurement environment that is a feature of the telehealth setting, these aspects are crucial to the design of a robust DSS.
An enterprise application server framework, combined with a rules engine and statistical analysis tools, is used to analyze the acquired telehealth data, searching for trends and shifts in parameter values as well as identifying individual measurements and sequences of successive measurements that exceed predetermined or adaptive thresholds, which provides an advanced indication a deterioration in a patient's health.
Contributors: Jim Basilakis, Nigel Lovell, Stephen Redmond, Mas Sahidayana Mohktar, Jumadi Abdul Sukor, Tal Shany, James Zhang, Arni Ariani, Branko Celler.
Multivariate state space analysis of non-invasive physiologic signals for early detection of critical illness
For many years, clinicians at virtually all levels of clinical care have been relying on measurements of blood pressure (BP) and heart rate (HR) for assessing cardiovascular function, despite their notable failure to provide an early sign for a number of potentially life-threatening conditions such as haemorrhage and heart diseases. To circumvent the limitations in the conventional way of diagnosis, which often fails to provide a full picture of the complex cardiovascular system, our laboratory has been researching on a novel multivariate approach of diagnosis that is based on not only BP and HR but also a number of parameters (such as heart rate variability (HRV)) extracted from non-invasive electrocardiogram (ECG) and peripheral volume pulse (PPG) waveforms, and the application of state space modelling and statistical classification techniques. The overall objective is to develop integrated hardware and software system to achieve early diagnosis of critical illness using simple non-invasive physiological measurements.
02/03/2009: The BSL has one Australian
Postgraduate Award Industry (APAI) scholarship available immediately for one
student to pursue a full time PhD degree and participate in a project jointly
supported by the Australian Research Council (ARC) and Telemedcare Pty Ltd. (Read full advertisment here.)
Contributors: Gregory Chan, Qim Yi Lee, Paul Middleton, Branko Celler, Nigel Lovell, Andrey Savkin.
Mathematical models for understanding the cardiovascular system during physiological challenges
Mathematical modelling is a useful tool for enhancing our understanding of the complex cardiovascular system under the influences of autonomic and local control mechanisms and the biomechanical effects of ventricular-vascular coupling. In this project, we aim at developing mathematical models to describe the haemodynamics during physiological challenges such as postural change and exercise. At present, a compartmental human cardiovascular model has been implemented using the Simulink package in Matlab and validated by experimental data obtained from head-up tilting. The developed models will later be extended to the characterisation of heart failure patients, and used for evaluating the potential impact of the installation of implantable rotary blood pumps in those patients under various haemodynamic conditions.
Contributors: Einly Lim, Gregory Chan, Nigel Lovell, Paul Middleton.
Automatic falls management
We describe a distributed falls management system capable of real-time falls detection in an unsupervised
living context and remote longitudinal tracking of falls risk
parameters using a waist-mounted triaxial accelerometer. A
self-administrable falls risk assessment is used to facilitate falls
prevention. A web-interface allows clinicians to monitor the
status of individuals and track their compliance with exercise
interventions. Early identification of increased falls risk allows
targeted interventions to be promptly administered. Real-time
detection of falls allows immediate emergency response
protocols to be deployed, reducing morbidity and increasing
the independence of the community-dwelling elderly
community.
Contributors: Michael Narayanan, Stephen Lord, Marc Budge, Maria Elena Scalzi, Stephen Redmond, Branko Celler, Nigel Lovell.
Simulation and control of an implantable rotary blood pump
One of the most critical design aspects for rotary blood pumps used as left ventricular assist devices is to control the pump to meet the body's metabolic demand and to do this without the need for implanting additional sensors. The controller must also detect and avoid dangerous states associated with over and under pumping.
In this research a pump control algorithm has been derived and implemented to facilitate the non-invasive prediction of real-time pump flow, automatic detection of pumping states and adjustments for the metabolic demand of the body. Several algorithms have been validated on the bench and in vivo studies. Further exercise tests on implant recipients will be performed in the future to assess the functional improvement obtained via use of an automated pump control strategy.
Contributors: Dean Karantonis, Einly Lim, Nigel Lovell.
Remote patient monitoring and data fusion
We are developing an advanced remote patient monitoring system for health care providers. Such a system includes developing appropriate wireless sensors, comprehensive knowledge management, as well as sophisticated data analysis techniques. Using this system, detailed patient activities can be traced, and data analysed in order to aid health care providers in maintaining a better understanding of the actual need of the patient. The fundamental goal of the research is to increase the quality of service for health care providers and to have a more accurate level of patient care classification.
The chosen approach for achieving the research goals outlined above is a two step solution. Firstly, we must construct a comprehensive sensor network that is robust enough to be merged into the actual nursing home environment so that patient data can be captured in real-time. To validate this technology, a series of trails will be conducted to evaluate the effectiveness and accuracy of the developed monitoring system. Base on these trial results the sensor network design will be refined.
The second stage is to analyse the captured data, and based on existing and novel data fusion techniques, provide the patient with decision support and vital illness prevention.
Contributors: James Zhang, Leroy Chan, Arni Ariani, Nigel Lovell.
Classification of human movement patterns from a triaxial accelerometer for home telecare
Unsupervised monitoring of human movement, especially the elderly living independently, with the use of a wearable triaxial accelerometer has become a major research focus in the recent years. Monitoring clinically sensitive parameters of movement in order to identify early changes in fall risks and health status is essential. The aim of the project is to develop digital signal processing algorithms to process the data from a waist mounted single triaxial accelerometer so that real time movement monitoring can be achieved in an unsupervised home setting. These algorithms will identify clinically significant parameters, detect adverse events and raise an alarm. The resulting robust falls monitoring system will assist independent living.
Contributors: Ning Wang, Ronny Ibrahim, Felicity Allen, Eliathamby Ambikairajah, Nigel Lovell, Branko Celler, Nelly Laydrus.
Modelling of central cardiorespiratory variables response to exercise
It is of considerable physiological interest to investigate the steady state relationships between cardiovascular variables, such as Heart rate (HR), Stroke Volume (SV), Cardiac Output (CO), Total peripheral resistance (TPR), and Oxygen Consumption (VO2) during exercise. However, because of the complexity and diversity of physiological responses, the relationships are nonlinear in general, and nonlinear mathematical models are then necessary for modelling the relationships. Therefore, we have applied the Support Vector Machine based regression or the so-called Support Vector Regression (SVR) method to model the non-linear characteristics of cardiovascular variables to exercise. Our models demonstrated to perform more superior than the models obtained from the traditional regression methods, enabling us to describe some of the intrinsic physiological behaviours that typical models fail to do.
Contributors: Steven Su, Teddy Cheng, Branko Celler, Paul Middleton, Andrey Savkin, Greg Chan.
Dynamic modeling of exercise HR and oxygen consumption kinetics
During physical exercise, our cardiovascular systems increase the delivery of blood and oxygen to working muscles as the metabolic demand increases, resulting in an increase of HR and oxygen consumption. Dynamic models that describe the responses of HR and oxygen consumption to exercise are therefore important for us to understand how our cardiovascular systems behave during exercises. However, a major challenge in modelling physiological systems, including the HR and oxygen consumption responses to exercise, is that they all exhibit some degree of nonlinear dynamic behaviours. To overcome this, we have derived nonlinear dynamic models to describe the kinetics of these variables in response to exercise, in particular treadmill exercise, using Hammerstein systems and nonlinear State-space systems. Our models are able to capture the macro underlying nonlinearities of the physiological behaviours, yet they are simple enough to be applied in practice, for example, in controlling and regulating these variables, since general physiological models, such as Gordins model, are often considered to be too complicated for practical use.
Contributors: Steven Su, Teddy Cheng, Branko Celler, Andrey Savkin.
Control of Heart Rate Response to Exercise
The control of HR kinetics during exercise is of paramount importance in, for example, the design of exercise protocols for patients with cardiovascular diseases, and in developing rehabilitation exercises to aid patients recovering from cardiac-surgery. The main purpose of control is to avoid the cardiovascular system being overstressed and hence, to reduce the risk of cardiac failure. Unfortunately, as mentioned above, the HR responses to exercise exhibit significant nonlinear dynamic behaviours, and the traditional linear control techniques cannot be applied. Based on our derived nonlinear models, we have developed a number of nonlinear controllers for the control and regulation of HR responses during exercise. The controllers were designed using the modern control techniques, such as Robust H infinity control, Model Predictive Control, and combined piecewise Optimal Linear Quadratic and Robust H infinity control. All the controllers have been experimentally verified on treadmill exercise and they demonstrated to perform well in various types of exercising HR profiles.
Contributors: Steven Su, Teddy Cheng, Branko Celler, Andrey Savkin.