Efficient k-NN Implementation for Real-Time Detection of Cough Events in Smartphones

Abstract

The potential of telemedicine in respiratory health care has not been completely unveiled in part due to the inexistence of reliable objective measurements of symptoms such as cough. Currently available cough detectors are uncomfortable and expensive at a time when generic smartphones can perform this task. However, two major challenges preclude smartphone-based cough detectors from effective deployment namely, the need to deal with noisy environments and computational cost. This paper focuses on the latter, since complex machine learning algorithms are too slow for real-time use and kill the battery in a few hours unless specific actions are taken. In this paper, we present a robust and efficient implementation of a smartphone-based cough detector. The audio signal acquired from the device's microphone is processed by computing local Hu moments as a robust feature set in the presence of background noise. We previously demonstrated that pairing Hu moments and a standard k-NN classifier achieved accurate cough detection at the expense of computation time. To speed-up k-NN search, many tree structures have been proposed. Our cough detector uses an improved vantage point (vp)-tree with optimized construction methods and a distance function that results in faster searches. We achieve 18× speed-up over classic vp-trees, and 560× over standard implementations of k-NN in state-of-the-art machine learning libraries, with classification accuracies over 93%, enabling real-time performance on low-end smartphones.

Automated, multiparametric monitoring of respiratory biomarkers and vital signs in clinical and home settings for COVID-19 patients

“…Point-of-care or semicontinuous methods for quantifying coughing or other vocal activities rely on electromyography, respiratory inductive plethysmography, accelerometry, or auditory recordings captured with one or several sensors, sometimes with other exploratory approaches (e.g., the nasal thermistor or the electrocardiography) (36)(37)(38)(39)(40)(41). Digital signal processing followed by machine learning algorithms often serves as the basis for classification (42)(43)(44)(45)(46)(47)(48)(49)(50)(51)(52)(53). Microphone-based methods prevail due to their widespread availability and their alignment with large crowd-sourced datasets (e.g., COUGHVID, HealthMode, DetectNow, VoiceMed).…”

Novel Coronavirus Cough Database: NoCoCoDa

“…implemented a cough event detection tool in real time using smartphone recordings. Their approach used a k-NN classifier with accuracies of 93% [29]. In [30] coughs were classified based on 8-dimentional number octonions.…”

Efficient computation of image moments for robust cough detection using smartphones

“…The Discrete Cosine Transform (DCT-II) is then computed and coefficients 2nd-14th are kept to constitute the feature vector [18]. This feature vector gets passed as an input to an optimized k-NN classifier to detect cough events in real time [24].…”

“…This becomes patent from the dramatic improvements achieved when the classifier can be skipped. Optimization of the classifier has been addressed in our previous work [24] in order to enable real-time processing even in the worst case scenario. With the optimizations proposed in this paper, we achieve 12× speed-up, which allow real-time computation with significant battery savings even in those cases.…”

“…However, to warrant seamless functionality with low battery drainage in regular use, the system also needs to be implemented in an efficient manner. Our work in [24] proposed an efficient implementation of the pattern recognition module proposed in [18] so that detection could also be performed in real time and battery consumption reduced. This paper focuses on the feature extraction part, which for local invariant image moments applied to cough detection, can significantly be optimized.…”