Combining predictive capabilities of Transcranial Doppler (TCD) with Electrocardiogram (ECG) to predict hemorrhagic shock

• US 8,762,308 B2
• Filed: 03/17/2010
• Issued: 06/24/2014
• Est. Priority Date: 04/08/2009
• Status: Expired due to Fees

• Current Assignee: Virginia Commonwealth University
• Original Assignee: Virginia Commonwealth University
• Inventors: Najarian, Kayvan, Ward, Kevin R., Ji, Soo-Yeon, Hakimzadeh, Roya
• Primary Examiner(s): CHANG, LI WU
• Application Number:
  US13/255,549
• Publication Number:
  US 20120136224A1
• Time in Patent Office: 1,560 Days
• Field of Search: None
• US Class Current: 706/45
• CPC Class Codes: A61B 5/026 Measuring blood flow A61B3/1233, A61B3/1241 take precedence
• CPC Class Codes: A61B 5/0295 using plethysmography, i.e. measuring the variations in the volume of a body part as modified by the circulation of blood therethrough, e.g. impedance plethysmography
• CPC Class Codes: A61B 5/316 Modalities, i.e. specific diagnostic methods
• CPC Class Codes: A61B 5/318 Heart-related electrical modalities, e.g. electrocardiography [ECG]
• CPC Class Codes: A61B 5/346 Analysis of electrocardiograms
• CPC Class Codes: A61B 5/7203 for noise prevention, reduction or removal
• CPC Class Codes: A61B 5/726 using Wavelet transforms
• CPC Class Codes: A61B 5/7267 involving training the classification device
• CPC Class Codes: A61B 5/7275 Determining trends in physiological measurement data; Predicting development of a medical condition based on physiological measurements, e.g. determining a risk factor
• CPC Class Codes: A61B 8/06 Measuring blood flow measuring volume flow in general G01F, e.g. G01F1/66, G01F1/72; measuring speed of fluids in general G01P5/00
• CPC Class Codes: A61B 8/0808 for diagnosis of the brain
• CPC Class Codes: G06F 2218/08 Feature extraction
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: G16Z 99/00 Subject matter not provided for in other main groups of this subclass

Abstract

A real-time decision-support system predicts hemorrhagic shock of a patient by analysis of electrocardiogram (ECG) signals and transcranial Doppler (TCD) signals from the patient. These signals are subject to signal decomposition using Discrete Wavelet Transform (DWT) to sets of wavelet coefficients and selecting significant signal features. Machine learning is applied to the significant features to evaluate and classify hypovolemia severity based on the input ECG and TCD signals from the patient. The classification of blood loss severity is displayed in real-time. An extension of the decision-support system integrates Arterial Blood Pressure (ABP) signals and thoracic electrical bio-impedance (DZT) signals with the ECG and TCD signals from the patient to evaluate severity of hypovolemia.

Claims

1. A real-time decision-support system for predicting hemorrhagic shock of a patient comprising:

means for receiving electrocardiogram (ECG) signals from the patient;

means for receiving transcranial Doppler (TCD) signals from the patient;

first bandpass filter for filtering the ECG signals;

second bandpass filter for filtering the TCD signals;

first means using Discrete Wavelet Transform (DWT) to decompose filtered ECG signals to generate a first set of wavelet coefficients and selecting significant signal features from the first set of wavelet coefficients;

second means using DWT to decompose filtered TCD signals to generate a second set of wavelet coefficients and selecting significant signal features from the second set of wavelength coefficients;

data processing means receiving significant signal features and, using machine learning, evaluating and classifying hypovolemia severity based on the selected significant signal features generated from the wavelet coefficients of input ECG and TCD signals from the patient; and

a display for displaying a classification of blood loss severity.

2. The real-time decision-support system according to claim 1, wherein the first and second means using DWT uses the Daubechies 4 (db4) form of DWT.

3. The real-time decision-support system according to claim 1, wherein the data processing means extracts features for hypovolemia severity prediction from TCD and ECG signals using statistical calculations and significance testing.

4. The real-time decision-support system according to claim 1, wherein the data processing means uses support vector machine (SVM) to generate a predictive model for hypovolemia severity.

5. The real-time decision-support system according to claim 1, further comprising:

means for generating arterial blood pressure (APB) signals from the patient; and

means for generating thoracic electrical bio-impedance (DZT) signals from the patient;

and wherein the data processing means extracts features for hypovolemia severity prediction from APB and DZT signals as well as TCD and ECG signals using statistical calculations and significance testing.

6. A computer implemented real-time decision-support method for predicting hemorrhagic shock of a patient, comprising the steps of:

receiving electrocardiogram (ECG) and transcranial Doppler (TCD) signals from the patient;

decomposing the ECG and TCD signals by a computer using discrete wavelet transformation (DWT) to generate a first set of wavelet coefficients for the ECG and selecting significant signal features from the wavelet coefficients of the ECG signals and to generate a second set of wavelet coefficients for the TCD and selecting significant signal features from the wavelet coefficients of the TCD signals;

combining by a computer wavelet coefficients into a raw feature set;

extracting by a computer statistical features from the raw feature set to produce a new feature set;

testing a significance of each new feature in the new feature set;

applying machine learning to new features derived from significant features for classification based on said testing step; and

displaying a prediction of blood loss severity.

7. The computer implemented real-time decision support method according to claim 6, wherein prior to the step of decomposing the ECG and TCD signals, filtering the ECG and TCD signals to eliminate noise from the signals.

8. The computer implemented real-time decision support method according to claim 6, wherein the step of extracting statistical features is done via analysis of variance (ANOVA) to generate a set of significant signal features.

9. The computer implemented real-time decision support method according to claim 6, wherein the step of applying machine learning to significant features is via Support Vector Machine (SVM) to generate a predictive model for hypovolemia severity.

10. The computer implemented real-time decision support method according to claim 6, further comprising the step of receiving Arterial Blood Pressure (ABP) and thoracic electrical bio-impedance (DZT) signals from the patient, wherein the step of decomposing includes decomposing the ABP and DZT signals using discrete wavelet transform (DWT) to generate a third set of wavelet coefficients and selecting significant signal features from the ABP signals and to generate a fourth set of wavelet coefficients and selecting significant signal features from the DZT signals.