Detection of worsening heart failure

• US 10,271,797 B2
• Filed: 03/29/2017
• Issued: 04/30/2019
• Est. Priority Date: 04/01/2016
• Status: Active Grant

• Current Assignee: Cardiac Pacemakers Incorporated (Boston Scientific Corporation)
• Original Assignee: Cardiac Pacemakers Incorporated (Boston Scientific Corporation)
• Inventors: Zhang, Yi, Thakur, Pramodsingh Hirasingh, Averina, Viktoria A., Thompson, Julie A., An, Qi, Krueger, Jonathan Walter
• Primary Examiner(s): Manuel, George
• Application Number:
  US15/472,445
• Publication Number:
  US 20170281096A1
• Time in Patent Office: 762 Days
• CPC Class Codes: A61B 5/0031 Implanted circuitry
• CPC Class Codes: A61B 5/0215 by means inserted into the body catheters A61M25/00
• CPC Class Codes: A61B 5/0538 invasively, e.g. using a catheter
• CPC Class Codes: A61B 5/11 Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb for measuring pulse A61B5/02 ;A61B5/1038 takes precedence; motion detection to correct for motion artifacts in physiological signals A61B5/721
• CPC Class Codes: A61B 5/318 Heart-related electrical modalities, e.g. electrocardiography [ECG]
• CPC Class Codes: A61B 5/4836 Diagnosis combined with treatment in closed-loop systems or methods A61B5/0036 takes precedence
• CPC Class Codes: A61B 5/7253 characterised by using transforms
• 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: A61N 1/365 controlled by a physiological parameter, e.g. heart potential evoked response A61N1/371
• CPC Class Codes: A61N 1/3956 Implantable devices for applying electric shocks to the heart, e.g. for cardioversion
• CPC Class Codes: A61N 1/3987 characterised by the timing or triggering of the shock
• CPC Class Codes: G16H 40/20 for the management or administration of healthcare resources or facilities, e.g. managing hospital staff or surgery rooms
• CPC Class Codes: G16H 40/63 for local operation
• CPC Class Codes: G16H 40/67 for remote operation
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems

Abstract

Systems and methods for detecting a target cardiac condition such as events indicative of worsening heart failure are described. A system may include sensor circuits for sensing physiological signals and a signal processor for generating a predictor trend indicative of temporal change of the physiological signal. The predictor trend may be transformed into a sequence of transformed indices using a codebook that includes a plurality of threshold pairs each including onset and reset thresholds. The codebook may be constructed and updated using physiological data. The system may detect target cardiac condition using the transformed indices.

Claims

1. A system for detecting a target physiological event, the system comprising:

a physiological sensor circuit including a sense amplifier circuit to sense a first physiological signal;

a signal processor circuit coupled to the physiological sensor circuit, the signal processor circuit including;

a predictor circuit configured to generate a first predictor trend using the first physiological signal, the first predictor trend indicating a change of the first physiological signal over time; and

a signal transformation circuit configured to transform the first predictor trend into transformed indices using a comparison of the first predictor trend to a first plurality of threshold pairs each including an onset threshold and a reset threshold; and

a detector circuit coupled to the signal processor circuit, the detector circuit configured to detect the target physiological event using the transformed indices; and

an output circuit configured to generate a human-perceptible alert in response to the transformed indices satisfying a detection criterion.

2. The system of claim 1, further comprising a therapy circuit configured to deliver a therapy in response to the detection of the target physiological event.

3. The system of claim 1, wherein the signal transformation circuit is to transform the first predictor trend using a codebook including an association between the first plurality of threshold pairs and corresponding code words,

wherein the transformation of the first predictor trend using the codebook includes, for a specified portion of the first predictor trend:

identifying from the codebook a relevant threshold pair with corresponding onset and reset thresholds, the specified portion temporally occurring after the first predictor trend exceeding the corresponding onset threshold and prior to the first predictor trend subsequently falling below the reset threshold; and

assigning a numerical code corresponding to the identified relevant threshold pair to the specified portion of the first predictor trend.

4. The system of claim 3, wherein the signal transformation circuit is further configured to:

identify from the codebook two or more relevant threshold pairs for the specified portion of the first predictor trend;

determine a dominant threshold pair from the two or more relevant threshold pairs, the dominant threshold pair having a larger onset threshold or a larger reset threshold than any other of the identified relevant threshold pairs; and

assign a numerical code corresponding to the dominant threshold pair to the specified portion of the first predictor trend.

5. The system of claim 3, further comprising a codebook formation circuit configured to:

produce a receiver operating characteristic (ROC) using detections of the target physiological event according to a plurality of candidate threshold pairs and, the ROC including operating points indicating sensitivity statistics and false positive rate (FPR) statistics associated with the plurality of candidate threshold pairs;

partition the ROC into a plurality of ROC segments defined by boundary operating points associated with respective boundary threshold pairs;

assign code words for the partitioned ROC segments; and

generate the codebook including the mapping between the boundary threshold pairs and the code words.

6. The system of claim 5, wherein the codebook formation circuit is configured to partition the ROC including one of:

sample the FPR statistics within a specified FPR range at specified sampling intervals;

sample the sensitivity statistics within a specified sensitivity range at specified sampling intervals;

or

sample a specified portion of the operation points within a specified sensitivity range and a specified FPR range at specified sampling intervals.

7. The system of claim 6, wherein the codebook formation circuit is configured to sample from the specified portion of the operation points a specified number of operating points equally spaced along the specified portion of the operation points according to a distance measure.

8. The system of claim 3, wherein the signal transformation circuit is configured to produce a signal intensity distribution (SID) of the first predictor trend, and to transform the first predictor trend using the codebook and the SID of the first predictor trend,

wherein the SID represents percentile frequencies of the first predictor trend across a plurality of candidate onset or reset thresholds;

and

wherein the codebook includes an association between the first plurality of threshold pairs including a plurality of percentile frequency thresholds (PFTs) and corresponding code words.

9. The system of claim 3, further comprising a codebook formation circuit configured to:

receive a plurality of first predictor trends from a plurality of patients;

produce a composite signal intensity distribution (cSID) of the plurality of first predictor trends, the cSID representing percentile frequencies across plurality of candidate onset or reset thresholds;

partition the cSID into a plurality of distribution segments defined by boundary percentile frequencies associated with respective boundary onset or reset thresholds;

assign code words for the partitioned distribution segments; and

generate the codebook including the mapping between the boundary onset or reset thresholds and the code words.

10. The system of claim 9, wherein the codebook formation circuit is configured to partition the cSID including one of:

sample the candidate onset or reset thresholds within a specified threshold range at specified sampling intervals;

or

sample the percentile frequencies within a specified percentile frequency range at specified sampling intervals.

11. The system of claim 9, wherein the codebook formation circuit is configured to partition the cSID including:

receive one or more anchor percentile frequencies (APFs); and

partition the cSID into a plurality of distribution segments using the APFs, the partitioned distribution segments associated with respective percentile frequency ranges and onset or rest threshold ranges; and

sample the candidate onset or reset thresholds within each of the onset or reset threshold range.

12. The system of claim 1, wherein:

the physiological sensor circuit is to sense a second physiological signal;

the predictor circuit is to generate a second predictor trend different from the first predictor trend using the second physiological signal;

the signal transformation circuit is to transform the first and second predictor trends jointly into the transformed indices according to a codebook, the codebook including an association between a plurality of multi-dimensional thresholds and a plurality of code words, each multi-dimensional threshold including a first threshold pair for the first predictor trend and a second threshold pair for the second predictor trend; and

the detector circuit is configured to detect the target physiological event using the transformed indices.

13. A method for detecting a target physiological event, the method comprising:

sensing a first physiological signal;

processing the first physiological signal to generate a first predictor trend indicating a change of the first physiological signal over time;

transforming the first predictor trend into transformed indices using a comparison of the first predictor trend to a first plurality of threshold pairs each including an onset threshold and a reset threshold;

detecting the target physiological event using the transformed indices; and

generating a human-perceptible alert in response to the transformed indices satisfying a detection criterion.

14. The method of claim 13, wherein transforming the first predictor trend includes, for a specified portion of the first predictor trend:

identifying from a codebook a relevant threshold pair with corresponding onset and reset thresholds, the specified portion temporally occurring after the first predictor trend exceeding the corresponding onset threshold and prior to the first predictor trend subsequently falling below the reset threshold; and

assigning a numerical code corresponding to the identified relevant threshold pair to the specified portion of the first predictor trend,

wherein the codebook includes an association between the first plurality of threshold pairs and corresponding code words, the code words including numerical codes within a specified range.

15. The method of claim 14, wherein transforming the first predictor trend further includes, for the specified portion of the first predictor trend:

identifying from the codebook two or more relevant threshold pairs;

determining from the two or more relevant threshold pairs a dominant threshold pair having a larger onset threshold or a larger reset threshold than any other of the identified relevant threshold pairs; and

assigning a numerical code corresponding to the dominant threshold pair to the specified portion of the first predictor trend.

16. The method of claim 14, further comprising generating a codebook, the generation of the codebook including:

producing a receiver operating characteristic (ROC) using detections of the target physiological event according to a plurality of candidate threshold pairs, the ROC including operating points indicating sensitivity statistics and false positive rate (FPR) statistics associated with the plurality of candidate threshold pairs;

partitioning the ROC into a plurality of ROC segments defined by boundary operating points associated with respective boundary threshold pairs;

assigning code words for the partitioned ROC segments; and

generating a mapping between the boundary threshold pairs and the code words.

17. The method of claim 16, wherein partitioning the ROC includes one of:

sampling the FPR statistics within a specified FPR range at specified sampling intervals;

sampling the sensitivity statistics within a specified sensitivity range at specified sampling intervals;

or

sampling a specified portion of the operation points within a specified sensitivity range and a specified FPR range at specified sampling intervals.

18. The method of claim 14, further comprising generating a codebook, the generation of the codebook including:

producing a composite signal intensity distribution (cSID) using a plurality of first predictor trends from a plurality of patients, the cSID representing percentile frequencies across plurality of candidate onset or reset thresholds;

partitioning the cSID into a plurality of distribution segments defined by boundary percentile frequencies associated with respective boundary onset or reset thresholds;

assigning code words for the partitioned distribution segments; and

generating a mapping between the boundary onset or reset thresholds and the code words.

19. The method of claim 18, wherein partitioning the cSID includes one of:

sampling the candidate onset or reset thresholds within a specified threshold range at specified sampling intervals;

or

sampling the percentile frequencies within a specified percentile frequency range at specified sampling intervals.

20. The method of claim 13, further comprising:

sensing a second physiological signal;

processing the second physiological signal to generate a second predictor trend different from the first predictor trend; and

transforming the first and second predictor trends jointly into the transformed indices according to a codebook, the codebook including an association between a plurality of multi-dimensional thresholds and a plurality of code words, each multi-dimensional threshold including a first threshold pair for the first predictor trend and a second threshold pair for the second predictor trend; and

wherein detecting the target physiological event includes detecting the target physiological event using the transformed indices.