A Novel Multi-Centroid Template Matching Algorithm and Its Application to Cough Detection

Abstract

Cough is a major symptom of respiratory-related diseases. There exists a tremendous amount of work in detecting coughs from audio but there has been no effort to identify coughs from solely inertial measurement unit (IMU). Coughing causes motion across the whole body and especially on the neck and head. Therefore, head motion data during coughing captured by a head-worn IMU sensor could be leveraged to detect coughs using a template matching algorithm. In time series template matching problems, K-Nearest Neighbors (KNN) combined with elastic distance measurement (esp. Dynamic Time Warping (DTW)) achieves outstanding performance. However, it is often regarded as prohibitively time-consuming. Nearest Centroid Classifier is thereafter proposed. But the accuracy is comprised of only one centroid obtained for each class. Centroid-based Classifier performs clustering and averaging for each cluster, but requires manually setting the number of clusters. We propose a novel self-tuning multi-centroid template-matching algorithm, which can automatically adjust the number of clusters to balance accuracy and inference time. Through experiments conducted on synthetic datasets and a real-world earbud-based cough dataset, we demonstrate the superiority of our proposed algorithm and present the result of cough detection with a single accelerometer sensor on the earbuds platform.Clinical relevance-Coughing is a ubiquitous symptom of pulmonary disease, especially for patients with COPD and asthma. This work explores the possibility and and presents the result of cough detection using an IMU sensor embedded in earables.

CoughTrigger: Earbuds IMU Based Cough Detection Activator Using An Energy-efficient Sensitivity-prioritized Time Series Classifier

“…Zhang et al proposed a template matching algorithm [9], called Multi-Centroid Classifier, which aims at iteratively creating an increasing number of clusters, each of which has its own centroid and radius and all together cover all the positive samples while include as few negative samples as possible. When a satisfying accuracy is achieved, the derived centroids and radii will be used in the test set as templates and thresholds to classify positive samples from negative ones.…”

“…(3) When increasing the number of clusters, instead of adding one random centroid seed, we select the sample that brings the greatest increase to the cost function. Inference Phase The inference steps are unchanged [9]. The distance between a test sample and each template is calculated and compared against the threshold.…”

“…In recent decades, wearable devices such as smartphones and earbuds are becoming prevalent in our daily life. A body of work has recently emerged, with wearable sensors showing promise in automatically detecting coughs [1][2][3][4][5][6][7][8][9]. Earbudsbased devices have been shown to be effective in cough detection [7][8][9].…”

“…A body of work has recently emerged, with wearable sensors showing promise in automatically detecting coughs [1][2][3][4][5][6][7][8][9]. Earbudsbased devices have been shown to be effective in cough detection [7][8][9]. Fig.…”

“…For example, Samsung Galaxy Buds2 has a 61 mAh battery that supports up to 7.5 hours of play time per charge [10], and adding extra features will cause a drop in battery life, and ultimately utility and user-interest. Some studies explored the use of an inertial measurement unit (IMU), due to its low battery footprint and computational load, to detect coughs [7,9]. It was soon realized that using a traditional classifier (XGBoost) to detect coughs could only yield a 47% sensitivity and 54% F1-Score [7].…”