Artificial neural network cardiopulmonary modeling and diagnosis

• US 5,680,866 A
• Filed: 03/29/1996
• Issued: 10/28/1997
• Est. Priority Date: 03/29/1996
• Status: Expired due to Fees

• Current Assignee: Battelle Memorial Institute
• Original Assignee: Battelle Memorial Institute
• Inventors: Keller, Paul E., Kangas, Lars J.
• Primary Examiner(s): Kamm, William E.
• Assistant Examiner(s): SCHAETZLE, KENNEDY
• Application Number:
  US08/625,025
• Time in Patent Office: 578 Days
• Field of Search: 395/21, 395/23, 128/670, 128/668, 128/671, 128/713, 128/672
• US Class Current: 600/483
• CPC Class Codes: G16H 10/60 for patient-specific data, e.g. for electronic patient records
• CPC Class Codes: G16H 15/00 ICT specially adapted for medical reports, e.g. generation or transmission thereof
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: Y10S 706/924 Medical

Abstract

The present invention is a method of diagnosing a cardiopulmonary condition in an individual by comparing data from a progressive multi-stage test for the individual to a non-linear multi-variate model, preferably a recurrent artificial neural network having sensor fusion. The present invention relies on a cardiovascular model developed from physiological measurements of an individual. Any differences between the modeled parameters and the parameters of an individual at a given time are used for diagnosis.

Claims

1. A method of diagnosing a cardiopulmonary condition of a test individual, comprising the steps of:

(a) cataloging at least one demographic parameter for the test individual;

(b) subjecting the test individual to a progressive multi-stage test and collecting at least one actual physiological parameter as a function of time, together with at least one progressive multi-stage test operating parameter as a function of time;

(c) entering the demographic parameter(s) together with progressive multi-stage test operating parameter(s) as a function of time into a non-linear multi-variate cardiopulmonary model for a healthy individual and obtaining at least one modeled physiological parameter as a function of time for the healthy individual;

(d) obtaining a result vector as a function of time of a combination of the actual physiological parameter(s) from the test individual mathematically combined with the modeled physiological parameter(s); and

(e) comparing the result vector to at least one vector of results known to represent at least one specific cardiopulmonary condition of at least one diseased individual having at least one demographic parameter similar to the test individual and determining the cardiopulmonary condition of the test individual.

2. The method as recited in claim 1, wherein constructing the non-linear multi-variate cardiopulmonary model comprises the steps of:

(a) cataloging at least one demographic parameter for at least one individual;

(b) subjecting the at least one individual to a progressive multi-stage test and collecting at least one physiological parameter as a function of time, and obtaining at least one progressive multi-stage test operating parameter as a function of time;

(c) entering the demographic parameter(s) and physiological parameter(s) together with the progressive multi-stage test operating parameter(s) into a non-linear multi-variate modeling system; and

(d) building the non-linear multi-variate model for the tested individual(s).

3. The method as recited in claim 2, wherein said at least one vector of results known to represent at least one specific cardiopulmonary condition is obtained comprising the steps of:

(a) cataloging at least one demographic parameter for a diseased individual having the at least one specific cardiopulmonary condition;

(b) subjecting the diseased individual(s) to a progressive multi-stage test and collecting at least one physiological parameter as a function of time together with at least one progressive multi-stage test parameter;

(c) entering the demographic parameter(s) together with the progressive multi-stage test parameter(s) into the non-linear multi-variate modeling system;

(d) obtaining a result vector of a mathematical combination of physiological parameters from the diseased individual combined with the cardiopulmonary multi-variate model for at least one healthy individual of similar demographic parameters; and

(e) storing the result vector.

4. The method as recited in claim 1, wherein said non-linear multi-variate cardiopulmonary model is a trained artificial neural network.

5. The method as recited in claim 1, wherein said comparing is done within a trained artificial neural network.

6. The method as recited in claim 5, wherein said artificial neural network is a feed forward artificial neural network with a recurrent layer.

7. The method as recited in claim 1, wherein the non-linear multi-variate cardiopulmonary model is a generic model.

8. The method as recited in claim 1, wherein the non-linear multi-variate cardiopulmonary model is an individual model.

9. The method as recited in claim 1, wherein the result vector is a signed difference vector of mathematical differences of a measured physiological parameter subtracted from a modeled physiological parameter.

10. The method as recited in claim 10, wherein the measured and modeled physiological parameters are functions of time.

11. A method of diagnosing a cardiopulmonary condition of a test individual, comprising the steps of:

(a) cataloging at least one demographic parameter for the test individual;

(b) subjecting the test individual to a progressive multi-stage test and collecting at least one tested physiological parameter as a function of time, together with at least one progressive multi-stage test operating parameter as a function of time;

(c) entering the demographic parameter(s) together with progressive multi-stage test operating parameter(s) as a function of time into a non-linear multi-variate cardiopulmonary model for a healthy individual and obtaining at least one modeled physiological parameter as a function of time for the healthy individual, said non-linear multi-variate cardiopulmonary model constructed by

(i) cataloging at least one demographic parameter for at least one model individual;

(ii) subjecting the at least one model individual to a model progressive multi-stage test and collecting at least one model physiological parameter as a function of time, and obtaining at least one model progressive multi-stage test operating parameter as a function of time;

(iii) entering the model demographic parameter(s) and model physiological parameter(s) together with the model progressive multi-stage test operating parameter(s) into a non-linear multi-variate modeling system; and

(iv) building the non-linear multi-variate model for the modeled individual(s)

(d) obtaining a signed difference vector as a function of time of differences of the actual physiological parameter(s) from the test individual subtracted from the modeled physiological parameter(s); and

(e) comparing the signed difference vector to at least one vector of differences known to represent at least one specific cardiopulmonary condition of at least one diseased individual having at least one demographic parameter in common with the test individual and determining the cardiopulmonary condition of the test individual, said at least one vector of differences obtained by

(i) cataloging at least one demographic parameter for a diseased individual having the at least one specific cardiopulmonary condition;

(ii) subjecting the diseased individual(s) to a progressive multi-stage test and collecting at least one diseased physiological parameter as a function of time together with at least one diseased progressive multi-stage test parameter;

(iii) entering the diseased demographic parameter(s) together with the diseased progressive multi-stage test parameter(s) into the non-linear multi-variate modeling system;

(iv) obtaining a difference vector of differences of diseased physiological parameters subtracted from the cardiopulmonary multi-variate model for at least one healthy individual of similar demographic parameters; and

(v) storing the difference vector.

12. The method as recited in claim 11, wherein said non-linear multi-variate cardiopulmonary model is a trained artificial neural network.

13. The method as recited in claim 11 wherein said comparing is done within a trained artificial neural network.