Determining respiratory deterioration and decision support tool
Decision support technology is provided for use with patients who may experience respiratory deterioration. A mechanism is provided to determine an indicator of a patient's probability of deterioration of respiratory functioning, which may include calculating a probability of a respiratory deterioration event in patients from whom serial digital recordings of respiratory sounds are acquired. A forecast or score representing a likelihood of deterioration may be generated and used for pulmonary disease prognosis, diagnosis, and/or determining or implementing an appropriate response action such as automatically issuing an alert or notification to a caregiver associated with the patient.
1. A decision support tool for monitoring or treating respiratory deterioration, comprising:
a computer processor;
computer memory storing non-transitory computer-readable instructions that when executed by the computer processor perform operations comprising:
receiving respiratory information comprising a plurality of recorded measurements of respiratory cycles, each recorded measurement of a respiratory cycle associated with a time, wherein the respiratory information received was collected using a digital stethoscope;
determining a power spectra for the respiratory information by transforming the respiratory information into a frequency domain;
determining spectral roughness of the determined power spectra, the spectral roughness determined for each recorded measurement of the plurality of recorded measurements of respiratory cycles;
based on the spectral roughness determined for each recorded measurement, determining a time series of spectral roughness determinations;
performing a changepoint analysis on the time series to determine a changepoint event; and
in response to determining a changepoint event, automatically initiating an intervening action.
2. The decision support tool of claim 1 , further comprising trimming the plurality of recorded measurements to exclude audio not indicative of a respiratory cycle prior to determining the power spectra for the respiratory information.
3. The decision support tool of claim 1 , wherein the respiratory information is transformed into the frequency domain using Fourier Transform or Wavelet Transform.
4. The decision support tool of claim 1 , wherein the spectral roughness is determined by Hausdorff Dimension.
5. The decision support tool of claim 1 , wherein the spectral roughness is determined within a passband between 600 Hz and 6.0 KHz.
6. The decision support tool of claim 1 , wherein the changepoint event is determined based on time series of spectral roughness determination exceeding a predetermined probability threshold.
7. The decision support tool of claim 6 , wherein the threshold is at least 0.5.
8. The decision support tool of claim 6 , wherein the intervening action comprises issuing a notification.
9. One or more non-transitory computer storage media having computer-executable instructions embodied thereon that, when executed, by one or more processors, cause the one or more processors to perform a method for monitoring or treating respiratory deterioration, the method comprising:
receiving respiratory information comprising a plurality of recorded measurements of respiratory cycles, each recorded measurement of a respiratory cycle associated with a time;
determining a power spectra for the respiratory information;
determining spectral roughness of the determined power spectra;
determining a time series of spectral roughness determinations;
performing a changepoint analysis on the time series to determine a changepoint event; and
in response to determining a changepoint event, automatically initiating an intervening action.
10. The media of claim 9 , wherein the respiratory information received was collected using at least one of a digital stethoscope, and a smartphone.
11. The media of claim 9 , wherein the spectral roughness is determined for a set of recorded measurements pruned from the plurality of recorded measurements of respiratory cycles.
12. The media of claim 9 , wherein the spectral roughness is determined by Hausdorff Dimension.
13. The media of claim 9 , wherein the changepoint event is determined based on time series of spectral roughness determination exceeding a predetermined probability threshold.
14. The media of claim 9 , wherein the intervening action is one of modifying a care program, scheduling an appointment, and generating a notification.
15. A computer-implemented method for monitoring or treating respiratory deterioration, the method comprising:
receiving respiratory information comprising a plurality of recorded measurements of respiratory cycles, each recorded measurement of a respiratory cycle associated with a time;
determining a power spectra for the respiratory information;
determining a time series of spectral roughness determinations, each of the spectral roughness determinations of the time series determined from the power spectra;
from the time series of spectral roughness, determining a changepoint event; and
automatically initiating an intervention action in response to the changepoint event.
16. The method of claim 15 , wherein the power spectra is determined using Fourier Transform or Wavelet Transform.
17. The method of claim 15 , wherein the respiratory information received was collected using at least one of a digital stethoscope and a smartphone.
18. The method of claim 15 , wherein the spectral roughness is determined by Hausdorff Dimension.
19. The method of claim 15 , wherein the changepoint event is determined based on time series of spectral roughness determination exceeding a predetermined probability threshold.
20. The method of claim 15 , wherein the intervening action is one of modifying a care program, scheduling an appointment, and generating a notification.