Method and apparatus for monitoring cardio-pulmonary health

• US 10,426,380 B2
• Filed: 05/30/2013
• Issued: 10/01/2019
• Est. Priority Date: 05/30/2012
• Status: Active Grant

• Current Assignee: ResMed Sensor Technologies Limited (ResMed Incorporated)
• Original Assignee: ResMed Sensor Technologies Limited (ResMed Incorporated)
• Inventors: Schindhelm, Klaus Henry, Farrugia, Steven Paul, Colefax, Michael Waclaw, Javed, Faizan, Khushaba, Rami, Heneghan, Conor, De Chazal, Philip, Zaffaroni, Alberto, Fox, Niall, Celka, Patrick, O'Hare, Emer, Redmond, Stephen James
• Primary Examiner(s): Weare, Meredith
• Application Number:
  US14/403,398
• Publication Number:
  US 20150164375A1
• Time in Patent Office: 2,315 Days
• CPC Class Codes: A61B 5/0507 using microwaves or terahertz waves
• CPC Class Codes: A61B 5/08 Detecting, measuring or recording devices for evaluating the respiratory organs A61B5/0205 takes precedence
• CPC Class Codes: A61B 5/0816 Measuring devices for examining respiratory frequency measuring frequency of electric signals G01R23/00
• CPC Class Codes: A61B 5/087 Measuring breath flow
• CPC Class Codes: A61B 5/097 Devices for facilitating collection of breath or for directing breath into or through measuring devices
• CPC Class Codes: A61B 5/103 Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes A61B5/08 takes precedence; measuring aids for tailors A41H1/00; measuring instruments specially adapted for dentistry A61C19/04
• CPC Class Codes: A61B 5/1115 Monitoring leaving of a patient support, e.g. a bed or a wheelchair
• CPC Class Codes: A61B 5/1118 Determining activity level
• CPC Class Codes: A61B 5/1126 using a particular sensing technique
• CPC Class Codes: A61B 5/113 occurring during breathing
• CPC Class Codes: A61B 5/4809 Sleep detection, i.e. determining whether a subject is asleep or not
• CPC Class Codes: A61B 5/4818 Sleep apnoea
• CPC Class Codes: A61B 5/7221 Determining signal validity, reliability or quality A61B5/7207 and A61B5/7217 take precedence
• CPC Class Codes: A61B 5/7246 using correlation, e.g. template matching or determination of similarity
• CPC Class Codes: A61B 5/7264 Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems neural networks per se G06N3/00; expert systems per se G06N5/00
• 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: A61B 5/742 using visual displays A61B5/339 takes precedence
• CPC Class Codes: A61B 5/746 Alarms related to a physiological condition, e.g. details of setting alarm thresholds or avoiding false alarms alarms per se G08B
• CPC Class Codes: G08B 21/02 Alarms for ensuring the safety of persons
• CPC Class Codes: G16H 40/63 for local operation
• CPC Class Codes: G16H 50/20 : for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: G16H 50/30 : for calculating health indices; for individual health risk assessment

Abstract

Disclosed is a cardio-pulmonary health monitoring apparatus. The apparatus comprises a contactless motion sensor configured to generate one or more movement signals representing bodily movement of a patient during a monitoring session; a processor; and a memory storing program instructions configured to cause the processor to carry out a method of processing the one or more movement signals. The method comprises extracting one or more sleep disordered breathing features from the one or more movement signals, and predicting whether a clinical event is likely to occur during a predetermined prediction horizon based on the one or more sleep disordered breathing features.

Claims

1. A cardio-pulmonary health monitoring apparatus comprising:

a contactless motion sensor configured to generate one or more movement signals representing bodily movement of a patient during a monitoring session;

a processor; and

a memory storing program instructions configured to cause the processor to carry out a method of processing the one or more movement signals, the method comprising;

selecting one or more sections of the one or more movement signals, being sections during which the patient was asleep and not performing gross bodily movements,

extracting one or more sleep disordered breathing (SDB) features from said sections of the one or more movement signals, and

predicting whether a clinical event is likely to occur during a predetermined prediction horizon based on the one or more SDB features,

wherein the selecting comprises selecting one or more sections of good quality, and

wherein sections of good quality comprise sections in which a figure of merit for the movement signal over the section exceeds a noise threshold for the section, and

wherein the noise threshold for the section is the smaller of a maximum noise value for the motion sensor and a percentile of a distribution of root mean square (RMS) values of the movement signal over the section.

2. A cardio-pulmonary health monitoring apparatus according to claim 1, wherein the contactless motion sensor, the processor, and the memory form part of a contactless sensor unit.

3. A cardio-pulmonary health monitoring apparatus according to claim 2, wherein the method further comprises sending an alert message to an external computing device dependent on the prediction.

4. A cardio-pulmonary health monitoring apparatus according to claim 1, wherein the processor and the memory form part of a computing device that is external to the contactless motion sensor.

5. A cardio-pulmonary health monitoring apparatus according to claim 1, further comprising a display, wherein the method further comprises displaying an alert message on the display dependent on the prediction.

6. A cardio-pulmonary health monitoring apparatus according to claim 1, further comprising an audio output, wherein the method further comprises generating an alarm by the audio output dependent on the prediction.

7. A cardio-pulmonary health monitoring apparatus according to claim 1, wherein the contactless motion sensor is a radio-frequency motion sensor.

8. A cardio-pulmonary health monitoring apparatus according to claim 7, wherein the radio-frequency motion sensor is configured to generate two movement signals based on respective transmitted radio-frequency signals that are in quadrature with each other.

9. A cardio-pulmonary health monitoring apparatus according to claim 8, wherein the method comprises combining the two movement signals into a combined movement signal before the extracting.

10. A cardio-pulmonary health monitoring apparatus according to claim 8, wherein the method comprises combining the SDB features extracted from each movement signal before the predicting.

11. A cardio-pulmonary health monitoring apparatus according to claim 1, wherein the processor is further configured to extract the SDB features from one or more movement signals obtained from a contact motion sensor.

12. A cardio-pulmonary health monitoring apparatus according to claim 1, wherein the processor is further configured to extract the SDB features from data relevant to the cardio-pulmonary health of the patient obtained from a non-motion sensor.

13. A cardio-pulmonary health monitoring apparatus comprising:

a contactless sensor unit comprising a contactless motion sensor configured to generate one or more movement signals representing bodily movement of a patient during a monitoring session,

a processor, and

a memory storing program instructions configured to cause the processor to carry out a method of processing the one or more movement signals, the method comprising;

selecting one or more sections of the one or more movement signals, being sections during which the patient was asleep and not performing gross bodily movements,

computing a correlation feature by correlating a respiratory effort envelope of each selected section of the movement signals with one or more respiratory effort templates,

detecting a sleep disordered breathing (SDB) event whenever the correlation feature exceeds a first threshold for a duration that is greater than a second threshold, and

calculating one or more SDB features from the detected SDB events, the SDB features being indicative of severity of sleep-disordered breathing by the patient during the monitoring session.

14. A cardio-pulmonary health monitoring apparatus according to claim 13, wherein the method further comprises sending an alert message to an external computing device dependent on the calculated SDB features.

15. A cardio-pulmonary health monitoring apparatus according to claim 13, wherein the method further comprises displaying an alert message on a display dependent on the calculated SDB features.

16. A cardio-pulmonary health monitoring apparatus according to claim 13, further comprising an audio output, wherein the method further comprises generating an alarm by the audio output dependent on the calculated SDB features.

17. A cardio-pulmonary health monitoring apparatus according to claim 13, wherein the one or more calculated SDB features comprises a plurality of features from the group consisting of:

apnea/hypopnea index;

total sleep time;

ratio of the total duration of Cheyne-Stokes respiration events to the total sleep time; and

ratio of the total duration of Cheyne-Stokes respiration events to the duration of the monitoring session.

18. A cardio-pulmonary health monitoring apparatus according to claim 13, wherein the contactless motion sensor is a radio-frequency motion sensor.

19. A cardio-pulmonary health monitoring apparatus according to claim 18, wherein the radio-frequency motion sensor is configured to generate two movement signals based on respective transmitted radio-frequency signals that are in quadrature with each other.

20. A cardio-pulmonary health monitoring apparatus according to claim 19, wherein the method comprises combining the two movement signals into a combined movement signal before the extracting.

21. A cardio-pulmonary health monitoring apparatus according to claim 19, wherein the method comprises combining the SDB features extracted from each movement signal before the predicting.

22. A cardio-pulmonary health monitoring apparatus according to claim 13, wherein the processor is further configured to extract the SDB features from one or more movement signals obtained from a contact motion sensor.

23. A method of monitoring cardio-pulmonary health of a patient, the method comprising:

selecting one or more sections of one or more movement signals representing bodily movement of the patient, the movement signals being generated by a contactless motion sensor directed at the patient during a monitoring session, the selected sections being sections during which the patient was asleep and not performing gross bodily movements,

extracting one or more sleep disordered breathing (SDB) features from said sections of the one or more movement signals, and

predicting whether a clinical event is likely to occur during a predetermined prediction horizon based on the one or more SDB features,

wherein the selecting comprises selecting one or more sections of good quality, and

wherein sections of good quality comprise sections in which a figure of merit for the movement signal over the section exceeds a noise threshold for the section, and

wherein the noise threshold for the section is the smaller of a maximum noise value for the motion sensor and a percentile of a distribution of root mean square (RMS) values of the movement signal over the section.

24. A method according to claim 23, wherein the extracting comprises:

detecting SDB events in said sections, and

calculating SDB features from the detected SDB events.

25. A method according to claim 24, wherein detecting SDB events comprises:

computing a correlation feature by correlating a respiratory effort envelope of each said section of the movement signal with one or more respiratory effort templates, and

detecting an SDB event whenever the correlation feature exceeds a first threshold for a duration that is greater than a second threshold.

26. A method according to claim 24, wherein detecting SDB events comprises

computing a mean value of a respiratory effort envelope of each said section over a main window,

calculating a baseline as the mean respiratory effort envelope value over two windows surrounding the main window, and

detecting an SDB event if the mean respiratory effort envelope value over a main window satisfies an apnea criterion or a hypopnea criterion with respect to the calculated baseline.

27. A method according to claim 24, wherein the extracting comprises first pre-processing the movement signals.

28. A method according to claim 27, wherein the pre-processing comprises de-trending each movement signal.

29. A method according to claim 27, wherein the number of movement signals is two, and the pre-processing comprises first combining the two movement signals.

30. A method according to claim 29, wherein the combining comprises:

removing the mean of each movement signal,

computing the minimum value of each movement signal,

computing the length of a sample vector whose components at each sample are the distances of the respective movement signals from their respective minimum values, and

removing the mean of the length at each sample.

31. A method according to claim 23, wherein the figure of merit is a root mean square (RMS) value of the movement signal over the section.

32. A method according to claim 23, wherein the figure of merit is the 75^th percentile of the distribution of RMS values of the movement signal over the section.

33. A method according to claim 23, wherein said sections comprise epochs in which the patient is present and a waveform length value of the movement signal in the epoch is less than a threshold.

34. A method according to claim 33, wherein epochs in which the patient is present are epochs in which a root mean square (RMS) value of the movement signal within the epoch exceeds a noise-only threshold.

35. A method according to claim 23, wherein said sections in which the patient was asleep and not performing gross bodily movements comprise epochs in which the patient was asleep and a variance of the movement signal within the epoch exceeds a threshold for the monitoring session.

36. A method according to claim 23, wherein the predicting comprises applying a classifier to classify the SDB features into a “

no-event-occurring”

class or an “

event-occurring”

class.

37. A method according to claim 36, wherein the classifier is a linear discriminant analysis classifier.

38. A method according to claim 36, wherein the classifier is applied to one or more principal components of the SDB features.

39. A method according to claim 23, wherein the predicting comprises applying a heuristic rule to one or more principal components of the SDB features.

40. A method according to claim 23, further comprising:

triggering generation on a display of at least one query based on the prediction, the query being configured to prompt the patient for a response; and

triggering generation of a clinical alert based on a patient response to the query.

41. A method according to claim 23, further comprising generating a potential relapse alert based on the prediction.

42. A method of monitoring cardio-pulmonary health of a patient, the method comprising:

selecting one or more sections of one or more movement signals representing bodily movement of the patient, the movement signals being generated by a contactless motion sensor directed at the patient during a monitoring session, the selected sections being sections during which the patient was asleep and not performing gross bodily movements,

extracting one or more sleep disordered breathing (SDB) features from said sections of the one or more movement signals, and

predicting whether a clinical event is likely to occur during a predetermined prediction horizon based on the one or more SDB features,

wherein said sections in which the patient was asleep comprise epochs labelled by a linear discriminant analysis (LDA) classifier as asleep, wherein the LDA classifier is configured to label each epoch based on an activity count series and a movement flag series for the epoch, and

wherein the movement flag series is obtained by;

computing a noise feature, a correlation feature, and a power feature for each sample of the movement signal; and

setting a movement flag for each sample to true if a noise feature for the sample is above a high threshold and the corresponding correlation and power features are below a low threshold, otherwise setting the movement flag to false.

43. A method according to claim 42, wherein the activity count series is obtained by integrating an envelope of the movement signal over one or more intervals within the epoch.

44. A method according to claim 42, wherein said sections in which the patient was asleep and not performing gross bodily movements comprise sections in which the patient was asleep and a movement flag series indicating gross bodily movement is false.

45. A method of monitoring cardio-pulmonary health of a patient, the method comprising:

selecting one or more sections of one or more movement signals representing bodily movement of the patient, the movement signals being generated by a contactless motion sensor directed at the patient during a monitoring session, the selected sections being sections during which the patient was asleep and not performing gross bodily movements, and

computing a correlation feature by correlating a respiratory effort envelope of each said selected section of the movement signals with one or more respiratory effort templates,

detecting a sleep disordered breathing (SDB) event whenever the correlation feature exceeds a first threshold for a duration that is greater than a second threshold, and

calculating one or more SDB features from the detected SDB events, the SDB features being indicative of severity of sleep-disordered breathing by the patient during the monitoring session.

46. A method according to claim 45, wherein the correlation feature is the maximum of the correlation values over the respiratory effort templates.

47. A method according to claim 45, wherein the respiratory effort templates are respective generic templates whose duration has been scaled by a mean breathing modulation cycle length estimated for the patient over the monitoring session.

48. A method according to claim 47, wherein the breathing modulation cycle length is estimated as a weighted average of the reciprocal of the peak frequencies of the power spectral density of the respiratory effort envelope computed over one or more macro-epochs within the section.

49. A method according to claim 48, wherein the weighting is dependent on the heights of the respective peaks.

50. A method according to claim 45, further comprising confirming each detected SDB event as a valid SDB event.

51. A method according to claim 50, wherein the confirming comprises verifying that hyperpnea sections adjacent the minimum of a respiratory effort envelope corresponding to each detected event have a duration greater than a minimum value.

52. A method according to claim 51, wherein the confirming further comprises verifying that the respiratory effort envelope dips below a hypopnea threshold for a time greater than one fifth of a modulation cycle length for the patient.

53. A method according to claim 50, wherein the confirming comprises:

computing one or more features from the movement signal corresponding to each candidate SDB event, and

applying a rule-based inference engine to the one or more computed features to determine whether to confirm each detected SDB event.

54. A method according to claim 45, wherein the respiratory effort envelope is calculated for the section by lowpass filtering the modulus of a complex-valued signal whose real part is the movement signal of the section and whose imaginary part is the Hilbert transform of the movement signal of the section.

55. A method according to claim 45, wherein the respiratory effort envelope is calculated for the section as the empirical mode decomposition of the Hilbert transform of the movement signal of the section.

56. A method according to claim 45, wherein the respiratory effort envelope is calculated for the section by applying a double max and hold filter to the movement signal of the section to compute a positive envelope and a negative envelope, and subtracting the negative envelope from the positive envelope.

57. A method according to claim 45, wherein the calculating comprises computing statistics of the detected SDB events for the monitoring session.

58. A method according to claim 57, wherein the calculating further comprises calculating the statistics of the detected SDB events for each of multiple sub-sessions of the monitoring session.

59. A method according to claim 58, wherein the sub-sessions are halves of a main sleep period of the monitoring session.

60. A method according to claim 58, wherein the sub-sessions are quartiles of the monitoring session.

61. A method according to claim 57, wherein the calculating comprises computing statistics of the detected SDB events for multiple monitoring sessions.

62. A method according to claim 57, wherein the calculating further comprises normalising the calculated SDB features to zero mean and unity standard deviation over a sliding window comprising multiple monitoring sessions preceding and including the monitoring session.

63. A method according to claim 62, wherein the sliding window comprises all monitoring sessions preceding and including the monitoring session.

64. A computer-readable medium storing program instructions configured to cause a processor to carry out a method of monitoring cardio-pulmonary health of a patient, the method comprising:

selecting one or more sections of one or more movement signals representing bodily movement of the patient, the movement signals being generated by a contactless motion sensor directed at the patient during a monitoring session, the selected sections being sections during which the patient was asleep and not performing gross bodily movements, and

computing a correlation feature by correlating a respiratory effort envelope of each said selected section of the movement signals with one or more respiratory effort templates,

detecting a sleep disordered breathing (SDB) event whenever the correlation feature exceeds a first threshold for a duration that is greater than a second threshold, and

calculating one or more sleep disordered breathing features from the detected SDB events, the SDB features being indicative of severity of sleep-disordered breathing by the patient during the monitoring session.