Overview
The ultimate objective of our wearable sensors / wearable health monitoring research is to enable adaptive, real-time, continuous and non-invasive healthcare for home and out-of-hospital environments. To this end, we are investigating 1) novel sensor technologies to develop biosensors that can measure high-quality biological signals, and 2) advanced system identification and signal processing algorithms to extract valuable clinical information therein.
Current Members
Professor H. Harry Asada, Ph.D., Ford Professor of Mechanical
Engineering, MIT
Andrew T. Reisner, M.D., Massachusetts General Hospital
Devin B. McCombie, Ph.D. Candidate in Mechanical Engineering, MIT
Wearable Blood Pressure Estimation Using Adaptive Hydrostatic Calibration of Peripheral PTT Measurements
Jin-Oh Hahn, Ph.D. Candidate in Mechanical Engineering, MIT
System Identification Approach to Central Cardiovascular Monitoring
Current Projects
Wearable Blood Pressure Estimation Using Adaptive Hydrostatic Calibration of
Peripheral PTT Measurements
Hypertension
is a major health problem in the Unites States. Wearable blood pressure sensors
may provide better patient feedback and allow more effective treatment of
hypertension. A wearable sensor must be non-obtrusive, compact, light weight,
and low power.
To achieve these goals wearable sensors will require new monitoring technology that both eliminates actuated cuff mechanisms while still allowing adaptive calibration of the non-invasive sensor signals. Altering the vertical height of the hand produces intra-arterial pressure changes (ΔP) that are equivalent to the hydrostatic pressure created by the relative height difference (Δh) between the hand and the heart, ΔP= ρgΔh. For non-invasive circulatory sensors placed on the hand this change in height and the resulting intra-arterial pressure change provides a means to adaptively calibrate sensor signals to arterial blood pressure that is cuff free.
We are working towards the development of an optical based blood pressure sensor that utilizes adaptive hydrostatic calibration to estimate blood pressure from peripheral pulse transit time (PTT) measurements. The device combines not only a unique dual in-line photoplethysmograph device architecture with the adaptive hydrostatic calibration but also novel system identification techniques to accurately estimate the calibration parameters.
Multi-Channel Cardiovascular System Identification for Cardiovascular Health
Monitoring
The
physiologic state of the cardiovascular (CV) system can be most accurately
assessed by using the aortic blood pressure (BP) and flow (BF). However,
standard measurement of these signals, such as Swann-Ganz catheter, entails
costly and risky surgical procedures. Therefore, most of the practically
applicable methods aim to monitor the CV system based on peripheral circulatory
signals, e.g. arterial BP at a limb. Various methods have been developed to
relate the peripheral signals to the CV state. These include group-averaged
transfer function methods for recovering the aortic BP signal from the
upper-limb arterial BP, and the estimation of CV parameters such as left
ventricular elasticity, end diastolic volume, total peripheral resistance (TPR),
and mean aortic flow from arterial BP measurement.
A chronic challenge of these previous methods is that the dynamics of the CV system, which relates the aortic and peripheral signals, is unknown and time-varying as well. So this problem turns out to be an ill-posed system identification problem because we are asked to identify both the unknown system dynamics and input signal using the output signal measurement alone. As a result, most of the previous methods end up with using a predetermined CV system model or transfer function to back out the aortic signal from a peripheral signal measurement, which is not adaptive to different subjects as well as physiologic conditions.
In an attempt to resolve the above mentioned problem and extend the CV monitoring techniques to diverse physiologic conditions, we came up with an innovative idea that by having additional peripheral measurements the CV system can be viewed as a multi-channel system, and this multi-channel CV system can be identified by exploiting the blind system Identification methodology, which is able to identify multi-channel systems without using the unknown input signal.
We have been investigating several aspects related to this multi-channel CV system Identification problem including the model structures, identification / de-convolution algorithms and persistent excitation / model identifiability / asymptotic variance analysis.
Viewing the CV system as a multi-channel system, based on the fact that all the output signal measurements, yi(n), i=1,2,…, are consequences of the same input signal u(n), we have the following correlations between the output signals which do not involve the unknown input signal:
Gi-1(z)yi(n) = Gj-1(z)yj(n),
which we call the "inverse" formulation of the blind system identification, and
Gj(z)yi(n) = Gi(z)yj(n),
which we call the "forward" formulation of the blind system
identification. We can exploit either of the above formulations to identify the
CV dynamics Gi(z), i=1,2,….
Experimental investigation based on a swine subject suggests that the algorithm is able to accurately identify the CV system dynamics (see the upper panel) and stably recover the unknown aortic blood pressure and flow signals (see the lower panel, where red dashed lines are true signals and blue solid lines are recovered signals).
Related Publications
J.O. Hahn, A.T. Reisner, H.H. Asada, "Blind Identification of 2-Channel IIR Wave Propagation Systems for Central Cardiovascular Monitoring," Submitted, Proceedings of Dynamic Systems and Control Conference, Ann Arbor, 2008.
J.O. Hahn, A.T. Reisner, F.A. Jaffer, H.H. Asada, "A New Approach to Reconstruction of Central Aortic Blood Pressure Using Adaptive Transfer Function," Submitted, Proceedings of IEEE Engineering in Medicine & Biology Conference, Vancouver, 2008.
J.O. Hahn, A.T. Reisner,
H.H. Asada, “Physiologic-State-Adaptive Recovery of Aortic Blood Pressure and Flow Using Blind 2-Channel IIR Cardiovascular System Identification,” Submitted, Proceedings of American Control Conference, Seattle, 2008.
J.O. Hahn, A.T. Reisner, H.H. Asada, "Identification of Multi-Channel
Cardiovascular Dynamics Using Dual Laguerre Basis Functions for Noninvasive
Assessment of Aortic Flow and TPR," Proceedings of ASME International Mechanical
Engineering Congress and Exposition, Seattle, No. IMECE-2007-41186, 2007.
J.O. Hahn, A.T. Reisner,
H.H. Asada, “Sensitivity and Variance Analysis of Arterial Pressure Transfer
Dynamics Estimated from Adaptive Multi-Channel System Identification,”
Proceedings of American Control Conference, New York, pp. 613-618, 2007.
D.B. McCombie, A.T.
Reisner, H.H. Asada, “Adaptive Blood Pressure Estimation from Wearable PPG
Sensors Using Peripheral Artery Pulse Wave Velocity Measurements and
Multi-Channel Blind Identification of Local Arterial Dynamics,” Proceedings of
IEEE Engineering in Medicine and Biology Conference, New York, pp. 3521-3524,
2006.
J.O. Hahn, A.T. Reisner, H.
Hojman, H.H. Asada, “Reconstruction of Central Aortic Pressure Waveform Using
Adaptive Multi-Channel Identification,” Proceedings of IEEE Engineering in
Medicine and Biology Conference, New York, pp. 3377~3380, 2006.
J.O Hahn, A.T. Reisner, H.H.
Asada, “A Blind Approach to Reconstruction of Aortic Blood Pressure Waveform
Using Gray-Box Identification of Multiple Pressure Transfer Channels,”
Proceedings of American Control Conference, Minneapolis, pp. 3415~3420, 2006.
D.B. McCombie, A.T. Reisner,
H.H. Asada, “Laguerre Model Blind System Identification: Cardiovascular Dynamics
Estimated from Multiple Peripheral Circulatory Signals,” IEEE Transactions on
Biomedical Engineering, vol. 52(11), pp. 1889-1901, November 2005.
D.B. McCombie, A.T. Reisner, H.H. Asada, "Identification of
Vascular Dynamics and Estimation of the Cardiac Output Waveform from Wearable
PPG Sensors," Proceedings of IEEE Engineering in Medicine and Biology
Conference, Shanghai, China, pp. 3490-3493, 2005.
J.O. Hahn, D.B. McCombie,
A.T. Reisner, H.H. Asada, H. Hojman, R. Mukkamala, “Adaptive Left Ventricular
Ejection Time Estimation Using Multiple Peripheral Pressure Waveforms,”
Proceedings of IEEE Engineering in Medicine and Biology Conference, Shanghai,
China, pp. 2383~2386, 2005.
H.H. Asada, A.T. Reisner,
P.A. Shaltis, D.B. McCombie, "Towards the Development of Wearable Blood Pressure
Sensors: A Photo-Plethysmograph Approach," Proceedings of IEEE Engineering in
Medicine and Biology Conference, Shanghai, China, pp. 4156-4159, 2005.
D.B. McCombie, A.T.
Reisner, H.H. Asada, “Identification of Cardiovascular Dynamics from Peripheral
Cardiovascular Waveforms Using Laguerre Model Blind System I.D.,” Proceedings of
ASME International Mechanical Engineering Congress and Exposition, Anaheim,
IMECE2004-60509, 2004.
D.B. McCombie, A.T.
Reisner,H.H. Asada, “Identification of Cardiovascular Dynamics from Peripheral
Circulatory Waveform Signals Using Two Sensor Blind System I.D.,” Proceedings of
IEEE Engineering in Medicine and Biology Conference, San Francisco, pp. 972-975,
2004.
A.T. Reisner, D.B. McCombie,
H.H. Asada, “Estimation of Cardiac Output from Peripheral Pressure Waveforms
using Laguerre Model Blind System Identification,” Proceedings of IEEE
Engineering in Medicine and Biology Conference, San Francisco, pp. 913-916,
2004.
A.T. Reisner, P.A. Shaltis,
D.B. McCombie, H.H. Asada, “A Critical Appraisal for Opportunities for Wearable
Medical Sensors,” Proceedings of IEEE Engineering in Medicine and Biology
Conference, San Francisco, pp. 2149-2152, 2004.
D.B. McCombie, H.H. Asada, “Multi-Channel Blind System
Identification of the Arterial Network using a Hemodynamic Wave Propagation
Model,” Proceedings of American Control Conference, Boston, pp.1645-1646, 2004.
Past Projects
Wearable Blood Pressure Sensor Using
Oscillometric PPG and Micro Accelerometers
Adaptive Motion Artifact Reduction for Wearable Health Monitoring Sensors Using MEMS Accelerometers
Conductive Fabric Garment for a Cable-Free Body Area Network
Wearable Conductive Fiber Sensors for Multi-Axis Human Joint Angle Measurement