Method for establishing heart failure assessment program and method for assessing occurrence of heart failure

1 . A method for establishing a heart failure assessment program, comprising:

providing an ECG (electrocardiogram) signal database, wherein the ECG signal database comprises a plurality of reference ECG signal data of a plurality of heart failure patients, each of the reference ECG signal data comprises a plurality of heartbeat waveform data and a plurality of heart rate data, and each of the plurality of heartbeat waveform data corresponds to one of the plurality of heart rate data;

performing a reference data pre-processing step, wherein each of the plurality of reference ECG signal data is pre-processed so as to obtain a plurality of processed ECG signal data, and the reference data pre-processing step comprises:

performing an offset-correcting step, wherein a baseline of each of the plurality of reference ECG signal data is adjusted based on the plurality of heart rate data corresponding thereto; and

performing a normalizing step, wherein each of the plurality of reference ECG signal data processed by the offset-correcting step is processed by an interpolation method so as to obtain the plurality of processed ECG signal data;

performing a feature analyzing step, wherein the plurality of processed ECG signal data are analyzed by a machine learning algorithm model so as to obtain at least one heart failure assessing feature; and

performing a training step, wherein the machine learning algorithm model is trained by the at least one heart failure assessing feature so as to obtain the heart failure assessment program, and the heart failure assessment program is for assessing whether a subject has a heart failure or not and a severity of the heart failure.

2 . The method for establishing the heart failure assessment program of claim 1 , wherein each of the plurality of reference ECG signal data is a 12-lead electrocardiogram data.

3 . The method for establishing the heart failure assessment program of claim 1 , wherein the plurality of heart rate data of each of the plurality reference ECG signal data are processed by the interpolation method in the normalizing step so as to obtain a normalized heart rate data, and the normalized heart rate data is used to normalize the plurality of heartbeat waveform data of each of the plurality reference ECG signal data.

4 . The method for establishing the heart failure assessment program of claim 3 , wherein the interpolation method is a cubic spline function interpolation method.

5 . The method for establishing the heart failure assessment program of claim 1 , wherein the machine learning algorithm model is a gradient descent algorithm model.

6 . The method for establishing the heart failure assessment program of claim 1 , wherein the plurality of processed ECG signal data are analyzed by the machine learning algorithm model with a grid search method so as to obtain the at least one heart failure assessing feature.

7 . The method for establishing the heart failure assessment program of claim 6 , wherein the plurality of processed ECG signal data are further analyzed by the machine learning algorithm model with a cross validation method so as to obtain the at least one heart failure assessing feature.

8 . The method for establishing the heart failure assessment program of claim 1 , wherein the reference data pre-processing step further comprises:

performing a signal transforming step, wherein each of the plurality of reference ECG signal data is processed with a signal transforming method so as to obtain a plurality of vectorcardiography data; and

performing a cardiac axis correcting step, wherein each of the plurality of vectorcardiography data is analyzed so as to obtain a cardiac axial vector, and then each of the plurality of vectorcardiography data is normalized based on the cardiac axial vector so as to obtain the plurality of processed ECG signal data.

9 . The method for establishing the heart failure assessment program of claim 8 , wherein the signal transforming method uses a 12-lead electrocardiogram data is used as the plurality of reference ECG signal data and a Lead I, a Lead II, a Lead V1, a Lead V2, a Lead V3, a Lead V4, a Lead V5 and a Lead V6 of the 12-lead electrocardiogram data are transformed to the plurality of vectorcardiography data according to the following formulas:

V=AM, wherein V is one of the plurality of vectorcardiography data, which is composed by X vector, Y vector and Z vector, A is a transformation matrix, and M is a matrix composed by the Lead V1, the Lead V2, the Lead V3, the Lead V4, the Lead V5, the Lead V6, the Lead I and the Lead II.

10 . A method for assessing occurrence of heart failure, comprising:

providing the heart failure assessment program established by the method for establishing heart failure assessment program of claim 1 ;

providing a target ECG signal data of the subject, wherein the target ECG signal data comprises a plurality of target heartbeat waveform data and a plurality of target heart rate data, and each of the plurality of target heartbeat waveform data corresponds to one of the plurality of target heart rate data;

performing a data pre-processing step, wherein the target ECG signal data is pre-processed so as to obtain a processed target ECG signal data, and the data pre-processing step comprises:

performing a target offset-correcting step, wherein a baseline of the target ECG signal data is adjusted based on the plurality of target heart rate data corresponding thereto; and

performing a target normalizing step, wherein the target ECG signal data processed by the offset-correcting step is processed by the interpolation method so as to obtain the processed target ECG signal data; and

performing an analyzing step, wherein the processed target ECG signal data is analyzed by the heart failure assessment program so as to obtain a heart failure occurrence assessing result, and the heart failure occurrence assessing result presents a heart failure occurring condition and the severity of the heart failure of the subject.

11 . The method for assessing occurrence of heart failure of claim 10 , wherein the target ECG signal data is a 12-lead electrocardiogram data.

12 . The method for assessing occurrence of heart failure of claim 10 , wherein the plurality of target heart rate data of the target ECG signal data are processed by the interpolation method in the target normalizing step so as to obtain a normalized target heart rate data, and the normalized target heart rate data is used to normalize the plurality of target heartbeat waveform data of the target ECG signal data.

13 . The method for assessing occurrence of heart failure of claim 12 , wherein the interpolation method is a cubic spline function interpolation method.

14 . The method for assessing occurrence of heart failure of claim 10 , wherein the data pre-processing step further comprises:

performing a target signal transforming step, wherein the target ECG signal data is processed so as to obtain a target vectorcardiography data; and

performing a target cardiac axis correcting step, wherein the target vectorcardiography data is analyzed so as to obtain a target cardiac axial vector, and then the target ECG signal data is normalized based on the target cardiac axial vector so as to obtain the processed target ECG signal data.

15 . The method for assessing occurrence of heart failure of claim 14 , wherein the signal transforming method is an inversion equation of Dower uses and a 12-lead electrocardiogram data is used as the target ECG signal data and a Lead I, a Lead II, a Lead V1, a Lead V2, a Lead V3, a Lead V4, a Lead V5 and a Lead V6 of the 12-lead electrocardiogram data is transformed to the target vectorcardiography data according to the following formulas:

V=AM, wherein V is the target vectorcardiography data, which is composed by X vector, Y vector and Z vector, A is a transformation matrix, and M is a matrix composed by the Lead V1, the Lead V2, the Lead V3, the Lead V4, the Lead V5, the Lead V6, the Lead I and the Lead II.