A static archive for my projects and research.

CSC 2228 Course Project: A Survey of Cough Detection in Mobile and Wearable Device - University of Toronto

Project Proposal

Coughing is a common symptom in many diseases [1]; however, an objective measurement of cough frequency and severity is lacking in the literature [2][3]. Although researchers have proposed methods for cough detection, the implementations of the developed algorithms are not publicly available. Furthermore, some of the approaches require patients to wear an accelerometer sensor or other devices, which disturbs the users’ daily activities which makes them impractical in real-world scenarios. In this project, we try to identify the existing baseline of the proposed algorithms to facilitate further research on the cough detection and related topics. Moreover, we investigate design solutions that run on handheld and wearable devices (i.e., smartphone, smartwatch) that perform cough detection with a minimum user's intervention.

In this survey study, we also plan to implement the state-of-the-art of cough detection algorithms [4] [5] on edge server, smartphone, and smartwatch. Furthermore, we seek to build a docker container of our system along with a fully functional mobile application (perhaps iOS) or Docker image powered by the existing models. Finally, we plan to provide an evaluation of the proposed algorithms to measure their performance using the collected data in our previous study (WearCOPD) [6].

(PDF download)



Tentative Timeline

  • Background Reading (October 4)
  • CI Environment Setup (October 11)
  • Data Annotation and Verification(October 18)
  • Model Selection(November 8)
  • Final Report (December 13)




[1] Peter G Gibson, Anne B Chang, Nicholas J Glasgow, Peter WHolmes, Andrew S Kemp, Peter Katelaris, Louis I Landau,Stuart Mazzone, Peter Newcombe, Peter Van Asperen, et al.Cicada: Cough in children and adults: Diagnosis and assess-ment. australian cough guidelines summary statement.MedicalJournal of Australia, 192(5):265–271, 2010.

[2] Sergio Matos, Surinder S Birring, Ian D Pavord, and H Evans.Detection of cough signals in continuous audio recordings us-ing hidden markov models.IEEE Transactions on BiomedicalEngineering, 53(6):1078–1083, 2006.

[3] Steve S Kraman, George R Wodicka, Gary A Pressler, andHans Pasterkamp. Comparison of lung sound transducers usinga bioacoustic transducer testing system.Journal of AppliedPhysiology, 101(2):469–476, 2006.

[4] Eric C. Larson, TienJui Lee, Sean Liu, Margaret Rosenfeld,and Shwetak N. Patel. Accurate and privacy preserving coughsensing using a low-cost microphone.InProceedings ofthe 13th International Conference on Ubiquitous Computing,UbiComp'11, pages 375–384, New York, NY, USA, 2011.ACM.

[5] Xiao Sun, Zongqing Lu, Wenjie Hu, and Guohong Cao.Symdetector: Detecting sound-related respiratory symptomsusing smartphones.InProceedings of the 2015 ACM In-ternational Joint Conference on Pervasive and UbiquitousComputing, UbiComp'15, pages 97–108, New York, NY,USA, 2015. ACM.

[6] Daniyal Liaqat, Ishan Thukral, Parco Sin, Hisham Alshaer,Frank Rudzicz, Eyal De Lara, Robert Wu, and Andrea Ger-shon. Poster: Wearcopd-monitoring COPD patients remotelyusing smartwatches.InProceedings of the 14th AnnualInternational Conference on Mobile Systems, Applications, andServices Companion, pages 139–139. ACM, 2016.