A device and method for non-invasive left ventricular end diastolic pressure (lvedp) measurement

Patent metadata

Field	Value
Publication number	US-2018160917-A1
Application number	US-201615571771-A
Country	US
Kind code	A1
Filing date	May 5, 2016
Priority date	May 5, 2015
Publication date	Jun 14, 2018
Grant date	—

Abstract

The present application relates to systems and methods for non-invasively determining at least one of left ventricular end diastolic pressure (LVEDP) or pulmonary capillary wedge pressure (PCWP) in a subject's heart, comprising: receiving, by a computer, a plurality of signals from a plurality of non-invasive sensors that measure a plurality of physiological effects that are correlated with functioning of said subject's heart, said plurality of physiological effects including at least one signal correlated with left ventricular blood pressure and at least one signal correlated with timing of heartbeat cycles of said subject's heart; training a machine learning model on said computer using said plurality of signals for periods of time in which said plurality of signals were being generated during a heart failure event of said subject's heart; determining said LVEDP or PCWP using said machine learning model at a time subsequent to said training and subsequent to said heart failure event.

First claim

We claim:

1 . A system for non-invasively determining at least one of left ventricular end diastolic pressure (LVEDP) or pulmonary capillary wedge pressure (PCWP) in a subject's heart, comprising:
a signal processor; a pressure sensor operatively connected to said signal processor to communicate therewith; and a timing sensor operatively connected to said signal processor to communicate therewith, wherein said pressure sensor is structured to be brought into mechanical connection with an external surface region of said subject so as to provide peripheral pressure signals corresponding to a peripheral artery blood pressure, wherein said timing sensor is structured to non-invasively measure a physical property of said subject's heart that is correlated with said subject's heart beat so as to provide timing signals comprising timing information with respect to heartbeat cycles, and wherein said signal processor is configured to receive said peripheral pressure signals and said timing signals and to non-invasively determine an estimated value of at least one of said subject's LVEDP or PCWP based at least partially thereon.

2 . The system according to claim 1 , wherein said timing sensor is at least one of an acoustic sensor capable of providing a phonocardiogram, an electrical sensor capable of providing an electrocardiogram, or optical sensor, or an impedance sensor.

3 . The system according to claim 2 , further comprising a blood pressure sensor operatively connected to said signal processor to communicate therewith,
wherein said blood pressure sensor is structured to be brought into mechanical connection with an external surface region of said subject so as to provide signals corresponding to diastolic blood pressure of said subject, and wherein said signal processor is further configured to receive said diastolic blood pressure signals and to non-invasively determine an estimated value of said subject's LVEDP based at least partially on said peripheral pressure signals, said timing signals and said diastolic blood pressure signals.

4 . The system according to claim 3 , wherein said peripheral pressure signals, said timing signals and said diastolic blood pressure signals are taken at times while said subject is undergoing a heart failure event and at times subsequent to said heart failure event.

5 . The system according to claim 1 , wherein said signal processor is further configured to non-invasively determine said estimated value of said subject's at least one of LVEDP or PCWP using a machine learning model.

6 . The system according to claim 5 , wherein said machine learning model is trained on said subject during a heart failure event.

7 . The system according to claim 6 , wherein said machine learning model is a neural network machine learning model.

8 . The system according to claim 7 , wherein said neural network machine learning model is a feed forward neural network that provides at least said estimated at least one of LVEDP or PCWP.

9 . The system according to claim 7 , wherein said neural network machine learning model is a recurrent neural network that provides at least one of an estimated LVEDP waveform or PCWP.

10 . The system according to claim 5 , wherein said machine learning model is calibrated to said subject comprising:
taking said peripheral pressure signals and said timing signals during a catheterization procedure; measuring a left ventricular end diastolic pressure in said subject's heart during said catheterization procedure; and correlating said peripheral pressure signals and said timing signals to said left ventricular end diastolic pressure.

11 . The system according to claim 10 , further comprising altering said left ventricular end diastolic pressure in said subject during said catheterization procedure by manipulating fluid status in said subject, performing a Valsalva maneuver on said subject, exposing said subject to an exercise, or a combination thereof.

12 . A computer-implemented method for non-invasively determining at least one of left ventricular end diastolic pressure (LVEDP) or pulmonary capillary wedge pressure (PCWP) in a subject's heart, comprising:
receiving, by a computer, a plurality of signals from a plurality of non-invasive sensors that measure a plurality of physiological effects that are correlated with functioning of said subject's heart, said plurality of physiological effects including at least one signal correlated with left ventricular blood pressure and at least one signal correlated with timing of heartbeat cycles of said subject's heart; training a machine learning model on said computer using said plurality of signals for periods of time in which said plurality of signals were being generated during a heart failure event of said subject's heart; determining said at least one of LVEDP or PCWP using said machine learning model at a time subsequent to said training and subsequent to said heart failure event.

13 . The method of claim 12 , wherein said at least one signal correlated with left ventricular blood pressure corresponds to peripheral artery blood pressure, and
wherein said at least one signal correlated with timing of heartbeat cycles of said subject's heart corresponds to at least one of an acoustic signal or an electrical signal.

14 . The method according to claim 12 , wherein said machine learning model is a neural network machine learning model.

15 . The method according to claim 14 , wherein said neural network machine learning model is a feed forward neural network that provides at least one of said estimated LVEDP or said PCWP.

16 . The method according to claim 14 , wherein said neural network machine learning model is a recurrent neural network that provides at least one of an estimated LVEDP waveform or PCWP.

17 . The method according to claim 12 , wherein said machine learning model is calibrated to said subject comprising:
taking said plurality of physiological effects during a catheterization procedure; measuring a left ventricular end diastolic pressure in said subject's heart during said catheterization procedure; and correlating said plurality of physiological effects to said left ventricular end diastolic pressure.

18 . The method according to claim 17 , further comprising altering said left ventricular end diastolic pressure in said subject during said catheterization procedure by manipulating fluid status in said subject, performing a Valsalva maneuver on said subject, exposing said subject to an exercise, or a combination thereof.

19 . A computer-readable medium comprising computer-executable code for non-invasively determining at least one of left ventricular end diastolic pressure (LVEDP) or pulmonary capillary wedge pressure (PCWP) in a subject's heart, said computer-readable medium, when executed by said computer causes said computer to:
receive a plurality of signals from a plurality of non-invasive sensors that measure a plurality of physiological effects that are correlated with functioning of said subject's heart, said plurality of physiological effects including at least one signal correlated with left ventricular blood pressure and at least one signal correlated with timing of heartbeat cycles of said subject's heart; train a machine learning model on said computer using said plurality of signals for periods of time in which said plurality of signals were being generated during a heart failure event of said subject's heart; determine said at least one of LVEDP or PCWP using said machine

Assignees

• Univ Johns Hopkins
• Boston Scient Scimed Inc

Inventors

• Liu Qian
• Leartprapun Nichaluk
• Wanjala Jackline
• Acharya Soumyadipta
• Bicek Andrew
• Barodka Viachaslau
• Anand Umang
• Alghatrif Majd
• Kass David
• Bernier B Westbrook
• Hwang Chao-Wei
• Johnston Peter
• Langston Trent

Classifications

• A61B2562/0204

  Acoustic sensors · CPC title

• A61B5/02108 Primary

  from analysis of pulse wave characteristics · CPC title

• A61B5/0535

  Impedance plethysmography (for measuring blood flow A61B5/0295) · CPC title

• A61B5/7267

  involving training the classification device · CPC title

• G16H50/70

  for mining of medical data, e.g. analysing previous cases of other patients · CPC title