Ambulatory methods for recording cough

https://doi.org/10.1016/j.pupt.2006.10.016Get rights and content

Abstract

Recording cough sounds to objectively quantify coughing was first performed using large reel-to-reel tape recorders more than 40 years ago. Coughs were counted manually, which is an extremely laborious and time-consuming process. Current technologies including digital recording techniques, data compression and improvements in digital storage capacity should make the process of recording and counting coughs suitable for automation; however, to date no accurate, objective cough monitoring device is available. Cough sounds are easily distinguishable from other vocalizations by the human ear and hence it is reasonable to assume that coughs sounds should have characteristic, identifying acoustic properties. However, the acoustic features of spontaneously occurring cough sounds are extremely variable. Furthermore, in even the worst cases of cough, the time spent speaking is an order of magnitude greater than the time spent coughing. It follows that even an algorithm that mistakes only a very small proportion of speech as cough will still have an unacceptable false positive rate. There is a clear need for an objective measure of cough for use in clinical practice, clinical research and trials of novel treatments. In the near future automated ambulatory systems with sufficient accuracy to be of clinical use should be available.

Section snippets

Why do we need and objective measure of cough?

Cough is consistently the commonest symptom for which patients’ seek medical care [1], [2] and yet despite this there is no universally accepted objective measure of cough. The study of cough generally relies upon subjective reporting, measurement of the sensitivity of the cough reflex and more recently cough related quality of life [3], [4].

A patients’ perception of the severity of a symptom and its effect on their well-being are unquestionably important. The processes involved in the

The history of cough monitoring

Recording cough sounds in order to objectively quantify coughing is not a new concept, with the first publications in the 1960s [12]. Using large reel-to-reel tape recorders, hospital in-patients were studied and coughs were counted manually, which is a laborious and time-consuming process (Fig. 1).

As technology progressed ambulatory recording devices became available using sound in combination with chest wall EMG signals to identify coughs [13], [14]. The signals still had to be reviewed and

Challenges in automated cough detection

There are three main challenges in automatically detecting and counting coughs:

  • 1.

    differentiation from ambient noise,

  • 2.

    differentiation from other patient sounds, especially speech, laughing sneezing and

  • 3.

    variability in the acoustics of cough sounds both within and between individuals.

Detection of coughing has most commonly been attempted from sound recordings [15], [16], [17], [18], [19] or sound in combination with a second signal such as EMG [13], [14]/chest wall movement [20], [21]. One of the

Quantifying cough

There is no universally agreed unit of cough. The most intuitive way to quantify cough is to count the expulsive (first phase) of cough sounds (Fig. 3). If long bursts (or peals) of expulsive cough sounds are present, then to identify each expulsive phase can be very difficult. We have devised a new method for quantifying cough, the ‘cough second’ which is the number of seconds containing at least one expulsive cough phase [22]. This gives a representation of the time spent coughing. The main

Performance of objective cough monitors

Any cough monitor will identify some cough events correctly (true positives), mistake non-cough events as cough (false positives), miss some cough events (false negatives) and correctly ignore non-cough (true negatives). The sensitivity is the proportion of coughs correctly detected by the cough monitor (Eq. (1)) as Sensitivity=truepositivestruepositives+falsenegatives.

However, the number and nature of false positive results is also of importance if the monitor is to give clinically useful

Recent developments in cough monitoring

Several groups are currently developing and publishing work in the field of automated cough monitoring and a summary of some of this recent work follows. The soon-to-be-published European Task Force Guidelines on the Measurement of Cough will help to further standardize the reporting of the performance of cough monitors.

Other endpoints in cough monitoring

Cough frequency is the most basic measure of coughing but the objective study of cough signals has the potential to identify further features which may be clinically relevant and hence useful endpoints to study.

The temporal patterns of coughing vary both in the short term (peals or epochs of coughs versus single coughs), from hour to hour and from day to night [24], [25], and can only be appreciated using objective ambulatory measures. Short term patterns could be an indication of individual

Conclusions

In summary, automated detection of coughing is required to make the objective measurement of cough a useful clinical and research tool. If this can be achieved, there are a number of other endpoints which may be important and require exploration. The testing of new cough monitoring systems needs to be rigorous and current systems have not quite achieved the levels of accuracy and consistency in performance necessary. With current progress, ambulatory cough monitoring should be possible in the

Acknowledgements

I would like to thank Alyn Morice, Samantha Barry, Surrinder Birring and Alex Derchak for information about their respective works.

References (41)

  • C.W. Burt et al.

    Ambulatory care visits to physician offices, hospital outpatient departments, and emergency departments: United States, 1999–2000

    Vital Health Stat 13

    (2004)
  • S.S. Birring et al.

    Development of a symptom specific health status measure for patients with chronic cough: leicester cough questionnaire (LCQ)

    Thorax

    (2003)
  • Decalmer S, Webster DA, Kelsall AA, McGuinness K, Woodcock A, Smith JA. Chronic cough: how do cough reflex sensitivity...
  • A.B. Chang et al.

    Relation between measurements of cough severity

    Arch Dis Child

    (2003)
  • P.V. Dicpinigaitis et al.

    Antitussive effect of the leukotriene receptor antagonist zafirlukast in subjects with cough-variant asthma

    J Asthma

    (2002)
  • C.R. Woolf et al.

    Objective assessment of cough suppressants under clinical conditions using a tape recorder system

    Thorax

    (1964)
  • J.Y. Hsu et al.

    Coughing frequency in patients with persistent cough: assessment using a 24 h ambulatory recorder

    Eur Respir J

    (1994)
  • A.B. Chang et al.

    A new use for an old Holter monitor: an ambulatory cough meter

    Eur Respir J

    (1997)
  • Y. Hiew et al.

    DSP algorithm for cough identification and counting

    IEEE Proc Int Conf Speech Signal Process (ICASSP)

    (2002)
  • S. Matos et al.

    Detection of cough signals in continuous audio recordings using hidden Markov models

    IEEE Trans Biomed Eng

    (2006)
  • Cited by (0)

    View full text