Method and system for detection of physiological conditions

• US 5,810,014 A
• Filed: 03/25/1997
• Issued: 09/22/1998
• Est. Priority Date: 03/25/1997
• Status: Expired due to Fees

• Current Assignee: Dennis W. Davis, Jaime R. Roman, James H. Michels
• Original Assignee: Dennis W. Davis, Jaime R. Roman, James H. Michels
• Inventors: Davis, Dennis W., Michels, James H., Roman, Jaime R.
• Primary Examiner(s): Kamm, William E.
• Assistant Examiner(s): EVANISKO, GEORGE ROBERT
• Application Number:
  US08/827,032
• Time in Patent Office: 546 Days
• Field of Search: 600/300, 600/301, 600/481, 600/508, 600/507, 128/901, 128/920, 128/923, 128/924, 128/925
• US Class Current: 600/508
• CPC Class Codes: G06F 18/00 Pattern recognition
• CPC Class Codes: G06F 2218/00 Aspects of pattern recognition specially adapted for signal processing

Abstract

The invention described herein is directed to a method and associated system for the detection of various physiological conditions using multichannel sensor data. Truth data organized in the form of a vector sequence is used to generate a multichannel state variable model for a given known physiological condition. Such state variable models are obtained using one or more of a host of system identification algorithms. Upon identification of the model parameters, a multichannel inverse filter is derived from the model parameters. The inverse filter can be used to whiten or decorrelate sensor data sequences that are associated with the given physiological condition. The degree of whitening is a measure of the match of an unknown physiological condition with that physiological condition represented by the data used to generate the filter. On this basis, a number of such filters can be produced for a variety of physiological conditions. In a preferred embodiment, these filters operate in parallel on a given multichannel data sequence that represents an unknown physiological condition. The outputs of these filters are used in a hypothesis testing formulation to determine the identity of the physiological condition represented by the data under test.

Claims

1. A method of determining the presence or absence of physiological abnormalities in multichannel test data from a physiological sensor suite, said method comprising the steps of:

(a) processing truth data to create physiological condition detection filters, said step of creating said physiological condition detection filters further comprising the steps of;

(i) extracting said truth data containing physiological information from a sensor suite;

(ii) pre-processing said truth data;

(iii) further processing said pre-processed data to obtain parameters of a multichannel state variable model to represent a physiological condition associated with said pre-processed data;

(iv) forming physiological condition detection filters using said parameters of said state variable model; and

(b) determining the presence or absence of physiological abnormalities in test data that represents an unknown physiological condition by processing said test data using said physiological condition detection filters, said step of determining the physiological condition associated with said test data further comprising the steps of;

(i) extracting said test data containing physiological information from a sensor suite;

(ii) pre-processing said test data;

(iii) processing said pre-processed test data with a plurality of said physiological condition detection filters so as to provide output data residual sequences that are preferentially whitened when a said physiological condition detection filter processes said test data that corresponds to the condition represented by same said physiological condition detection filter,

(iv) calculating statistical metrics for each said output data residual sequence output from each said physiological condition detection filter;

(v) ascertaining the identity of the physiological condition represented by said test data by performing a diagnostic decision based on the plurality of said statistical metrics.

2. A method as recited in claim 1, wherein pre-processing of said truth data and said test data is for the purposes of segmentation, trend removal, and normalization.

3. A method as recited in claim 1, wherein said physiological abnormalities are cardiac abnormalities.

4. A system for determining the presence or absence of physiological abnormalities in multichannel test data from a physiological sensor suite, said system comprising:

(a) means for processing truth data to create physiological condition detection filters, said means for creating said physiological condition detection filters further comprising;

(i) means for extracting truth data containing physiological information from a sensor suite;

(ii) means for pre-processing said truth data;

(iii) means for further processing said pre-processed truth data to obtain parameters of a multichannel state variable model to represent a physiological condition associated with said pre-processed truth data;

(iv) means for forming physiological condition detection filters by obtaining coefficients of an inverse filter derived from said state variable model parameters; and

(b) means for determining the presence or absence of physiological abnormalities in test data that represents an unknown physiological condition, said means of determining the physiological condition associated with said test data further comprising;

(i) means of extracting said test data containing physiological information from a sensor suite;

(ii) means of pre-processing said test data;

(iii) means of processing said pre-processed test data with a plurality of said physiological condition detection filters so as to provide output data residual sequences that are preferentially whitened when a said physiological condition detection filter processes a said test data that corresponds to a physiological condition represented by same said physiological condition detection filter,

(iv) means for calculating statistical metrics for each said output data residual sequence output from each said physiological condition detection filter;

(v) means of ascertaining the identity of the condition represented by said test data, said means of ascertaining performing a diagnostic decision based on the plurality of said statistical metrics.

5. A system as recited in claim 4 including annunciation means for informing users of said system of the determination of said physiological condition.

6. A system as recited in claim 4 which includes data storage means for storage of said truth data, said test data, and said diagnostic decisions.