Neural network system for the evaluation and the adaptation of antitachycardia therapy by an implantable defibrillator

Patent metadata

Field	Value
Publication number	US-9764151-B2
Application number	US-201514598025-A
Country	US
Kind code	B2
Filing date	Jan 15, 2015
Priority date	Jan 16, 2014
Publication date	Sep 19, 2017
Grant date	Sep 19, 2017

Abstract

The system includes an active medical device with means for delivering defibrillation shocks; means for continuous collection of the patient current cardiac activity parameters; and evaluator means with neuronal analysis comprising a neural network with at least two layers. This neural network comprises upstream three neural sub-networks receiving the respective parameters divided into separate sub-groups corresponding to classes of arrhythmogenic factors; and downstream an output neuron coupled to the three sub-networks and capable of outputting an index of risk of ventricular arrhythmia. The risk index is compared with a given threshold, to enable or disable at least one function of the device in case of crossing of the threshold.

First claim

The invention claimed is:

1. A system for evaluation and adaptation of an antitachycardia therapy, comprising:
at least one lead; and
an active medical device adapted to be implanted in a patient, the active medical device performs at least one function and comprises a neural network with at least two layers and configured to:
deliver defibrillation shocks via the lead;
collect parameters relating to cardiac activity of the patient;
extract three subgroups of descriptors from the collected parameters, wherein the three subgroups of descriptors correspond to classes of arrhythmogenic factors, a first one of the subgroups comprising electrophysiological substrate descriptors, a second one of the subgroups comprising pejorative modulator descriptors, and a third one of the subgroups comprising trigger factor descriptors;
for each of the three subgroups of descriptors, classifying each descriptor based on an ability of the descriptor to label the patient and selecting a descriptor for each of the three subgroups of descriptors having a classification indicating the descriptor is relevant to the patient;
evaluate the selected descriptors using the neural network with at least two layers, the at least two layers comprising:
three neural sub-networks, each configured to process a different one of the selected descriptors, wherein each neural sub-network generates an output; and
at least one output neuron coupled to the three neural sub-networks and configured to generate an index of risk of ventricular arrhythmia based on the output of at least one of the three neural sub-networks; and
compare the index of risk of ventricular arrhythmia to a threshold and activate or disable the at least one function of the active medical device in response to the index crossing the threshold.

2. The system of claim 1 , wherein the function activated or disabled in response to the index crossing the threshold is at least one of:
producing an alarm;
activating or deactivating defibrillation shock therapies;
activating new therapy zones;
adjusting a sensitivity of an arrhythmia detector;
activating or deactivating one or more algorithms; or
modifying one or more therapy settings.

3. The system of claim 1 , further comprising:
a database of reference patients, storing for each reference patient:
a plurality of descriptors developed from the parameters relating to the cardiac activity collected for the reference patient; and
a marker indicating whether a ventricular arrhythmia was detected in the reference patient;

wherein the active medical device is further configured to define a structure of the neural network by learning from the database of reference patients, wherein the active medical device is configured to, for each of the subgroups corresponding to arrhythmogenic factor classes:
determine the structure of the neural sub-networks; and
optimize the neural sub-network.

4. The system of claim 3 , wherein the first one of the subgroups comprises electrophysiological substrate descriptors selected from a group comprising:
a QRS residuum;
a T-Wave residuum;
a QRS-T angle;
QTapex intervals;
QTend intervals;
a downslope of a T wave; or
a ST segment elevation.

5. The system of claim 3 , wherein the second one of the subgroups comprises pejorative modulator descriptors selected from a group comprising:
a heart rate turbulence;
a variability index between successive complexes;
a standard deviation of normal range averages; or
a Poincaré representation of heart rate variability.

6. The system of claim 3 , wherein the third one of the subgroups comprises trigger factor descriptors selected from a group comprising:
a ventricular trigeminy episode;
a ventricular bigeminy episode;
a ventricular tachycardia; or
a supraventricular extrasystole.

7. An active medical device adapted to be implanted in a patient; the device comprising:
at least one function;
a neural network with at least two layers;
a processor configured to:
collect parameters relating to cardiac activity of the patient;
for each of three subgroups of settings, extract three subgroups of descriptors from the collected parameters, wherein the three subgroups of settings correspond to classes of arrhythmogenic factors, a first one of the subgroups comprising electrophysiological substrate descriptors, a second one of the subgroups comprising pejorative modulator descriptors, and a third one of the subgroups comprising trigger factor descriptors;
for each of the three subgroups of descriptors, classifying each descriptor based on an ability of the descriptor to label the patient and selecting a descriptor for each of the three subgroups of descriptors having a classification indicating the descriptor is relevant to the patient;
evaluate the selected descriptors using the neural network with at least two layers, the at least two layers comprising:
three neural sub-networks, each configured to process a different one of the selected descriptors wherein each neural sub-network generates an output; and
at least one output neuron coupled to the three neural sub-networks and configured to generate an index of risk of ventricular arrhythmia based on the output of at least one of the three neural sub-networks; and

compare the index of risk of ventricular arrhythmia to a threshold and activate or disable the at least one function of the active medical device in response to the index crossing the threshold.

8. The device of claim 7 , wherein the function activated or disabled in response to the index crossing the threshold is at least one of:
producing an alarm;
activating or deactivating defibrillation shock therapies;
activating new therapy zones;
adjusting a sensitivity of an arrhythmia detector;
activating or deactivating one or more algorithms; or
modifying one or more therapy settings.

9. The device of claim 7 , further comprising:
a database of reference patients, storing for each reference patient:
a plurality of descriptors developed from the parameters relating to the cardiac activity collected for the reference patient; and
a marker indicating whether a ventricular arrhythmia was detected in the reference patient;

wherein the active medical device is further configured to define a structure of the neural network by learning from the database of reference patients, wherein the active medical device is configured to, for each of the subgroups corresponding to arrhythmogenic factor classes:
determine the structure of the neural sub-networks; and
optimize the neural sub-network.

10. The device of claim 9 , wherein the first one of the subgroups comprises electrophysiological substrate descriptors selected from a group comprising:
a QRS residuum;
a T-Wave residuum;
a QRS-T angle;
QTapex intervals;
QTend intervals;
a downslope of a T wave; or
a ST segment elevation.

11. The device of claim 9 , wherein the second one of the subgroups comprises pejorative modulator descriptors selected from a group comprising:
a heart rate turbulence;
a variability index between successive complexes;
a standard deviation of normal range averages; or
a Poincaré representation of heart rate variability.

12. The device of claim 9 , wherein the third one of the subgroups comprises trigger factor descriptors selected from a group comprising:
a ventricular trigeminy episode;
a ventricular bigeminy episode;
a ventricular tachycardia; or
a supraventricular extrasystole.

13. A method comprising:
collecting, by an active medical device, the active medical device performing at least one function and comprising a neural network with

Assignees

• Sorin Crm Sas

Inventors

• Cappelaere Charles-Henri
• Amblard Amel
• Christophle-Boulard Sylvain
• Dreyfus Gérard
• Dubois Rémi
• Roussel Pierre

Classifications

• G06N3/02

  Neural networks · CPC title

• A61N1/3937 Primary

  Monitoring output parameters · CPC title

• A61N1/3925 Primary

  Monitoring; Protecting · CPC title

• A61N1/3987

  characterised by the timing or triggering of the shock · CPC title

• A61N1/3956

  Implantable devices for applying electric shocks to the heart, e.g. for cardioversion · CPC title