Method to Monitor Cough by Employing Piezoelectric Energy Harvesting Configurations

Abstract

Cough is the most common symptom prompting individuals to seek medical advice. However, the widespread adoption of autonomous cough monitoring using wearable devices remains limited. This paper introduces a wireless cough monitoring device utilizing piezoelectric energy harvesting technology. The design emphasizes cost-effectiveness and energy efficiency, allowing simple attachment onto human skin using medical-grade tapes. The device's standout feature lies in its departure from continuously recording real-time acoustic data at a high sampling rate, as commonly employed in prior works. Instead, it capitalizes on the energy harvesting capability, utilizing harvested energy from muscle movements induced by coughing as crucial information. The energy harvested within specific intervals translates into a historical record of cough occurrences during that timeframe. This Energy-as-Data protocol substantially reduces the device's duty cycle, resulting in a remarkable extension of battery life by up to 2100%. Notably, this extension is achieved while maintaining reasonable accuracy in cough monitoring. With this capability, the device can autonomously monitor and analyze cough data from both in- and outpatients, serving daily, research, and clinical purposes. Its potential extends to enhancing prediction and management of severe respiratory diseases.

1 Introduction

Respiratory conditions impact over 700 million patients globally. Despite technological advancements in disease management and treatment, more than half of the affected population remains undiagnosed or experiences inadequate disease management [1,2,3].

Coughing signifies an underlying issue in an individual's respiratory system, yet it remains an underutilized diagnostic tool. Currently, there's a lack of reliable and user-friendly cough monitors. Most existing solutions for smart cough sensing primarily rely on sound detection and recording methods [4, 5]. Prior studies have described the utilization of smartphone microphones to capture cough sounds, followed by the application of specific algorithms within mobile apps for processing and analyzing the recorded sounds. These algorithms were typically designed to differentiate coughs from speech and background noises, utilizing AI-assisted protocols and machine learning. Furthermore, these applications were capable of distinguishing a user's coughs from those of others, providing analyzed statistics and information such as the timing, frequency, and severity of coughs, while also maintaining a relatively extensive and continuous history of the user's coughing behavior [4, 5].

However, research has highlighted that the previously mentioned solution heavily depends on the user keeping the smartphone in close proximity due to the nature of sound propagation. A notable drawback is that as the distance between the user and the smartphone increases, both the accuracy and reliability of cough measurement significantly diminish [6]. Moreover, the application requires the microphone to continuously operate, resulting in rapid battery drain and raising privacy concerns.

In this study, a wearable cough monitoring device that utilizes the piezoelectric energy harvesting concept to accurately measure coughs regardless of the distance between the user and the data receiver is introduced. This device can be directly affixed to the user's throat, enabling on-site storage of all cough-related data. Consequently, it operates independently of any smartphone or digital data receiver. This independence significantly expands its utility to individuals who either do not use smartphones or do not consistently remain in proximity to their phones. This includes demographics such as the elderly, athletes, children, individuals with disabilities, in-patients, and more. Additionally, a substantial extension of battery life is a significant benefit derived from this technology.

2 Methodology

2.1 Architecture of Piezoelectric Cough Sensing System

Generally, piezoelectric materials possess the capability to convert mechanical stress or strain into electric charge, or conversely, convert electric potential into material deformation. Piezoelectric energy harvesters harness the former attribute of these materials. When exposed to ambient kinetic energy like vibrations, impacts, or acceleration/deceleration, the piezoelectric component experiences mechanical stress or strain. In a closed external circuit, this results in the flow of electric charges, generating current, and subsequently producing usable electricity [7].

In fact, the mechanism utilized by piezoelectric sensors for decades involves continuously monitoring the output of the piezoelectric component, reflecting real-time mechanical stimuli on the sensor. However, in this study, employing this sensing mode was unfeasible. Continuous monitoring of the piezoelectric output necessitates incessant readings by the microcontroller in the circuitry. This continuous procedure would rapidly deplete the battery, which proves impractical for long-term cough monitoring, requiring an extended power source lifespan.

Conversely, the conventional piezoelectric energy harvesting approach was not utilized in this study either. Typically, the generated electricity from this method is employed to charge the battery, thereby extending its lifespan. However, in this particular work, the muscle movement induced by coughing is highly intermittent. The power generated per cough, typically ranging from nanowatts to microwatts, is insufficient to effectively charge the battery.

Hence, this study integrates the principles of piezoelectric sensing and energy harvesting. Figure 1 illustrates the operational mechanism of the cough sensing system. The system comprises a piezoelectric energy harvester connected in series with an AC-DC rectifier and an energy storage component. The piezoelectric energy harvester can adopt various structures optimized for off-resonance harvesting of kinetic vibrational signals, such as unimorph and bimorph cantilevers or diaphragms.

Fig. 1.

Schematics of the working mechanism for the cough monitoring system.

The AC-DC rectifier may take the form of a diode or a full diode bridge rectifier. Its role is to convert the AC electric field generated by the piezoelectric energy harvester into a DC field for charging the energy storage. The energy storage element could be any component facilitating rapid charging and discharging, like a capacitor, thin-film battery, supercapacitor, and similar devices. Importantly, it is essential to note that the energy storage utilized here is not designed to serve as the primary power supply for the entire cough sensing device. Instead, it functions as an approach to significantly reduce the system's power consumption, which is elaborated in subsequent sections.

The choice of forward voltage for the rectifier should hinge on the sensitivity of the piezoelectric energy harvester concerning the ratio between the input cough strength and background noise, including body movements, speech, and other acoustic signals. Essentially, the rectifier should effectively distinguish genuine cough signals from non-cough signals.

Regarding the selection of capacitance for the energy storage, it should be based on the intended duration of the monitoring period. For prolonged cough monitoring, a larger capacitance with low leakage is preferable, while shorter-term monitoring favors a smaller capacitance with higher leakage.

2.2 Working Mechanism of Cough Sensing by Piezoelectric Energy Harvester

During the idle stage (Fig. 1), the system operates in standby mode without any energy generation from the piezoelectric energy harvester. However, when the system user coughs, this action triggers the piezoelectric energy harvester to convert the kinetic energy from the muscle movement into electrical energy. This electrical energy then charges the capacitor, constituting what is termed as the ‘Cough-charging’ stage.

Within a defined measurement interval, the voltage level of the energy storage after charging is proportionate to the accumulation in number or intensity of coughs which occurred during that timeframe. Subsequently, the electronics measure the voltage of the charged capacitor at the end of this interval, known as the ‘Measurement-charged’ stage.

In the interval between two consecutive measurements, the energy storage undergoes a discharge process either through natural leakage or intentional discharge via programming. This discharged energy storage prepares for the next operational cycle, termed the ‘Interval-discharging’ stage. Following this stage, the system reverts to the idle stage, and this cyclical process continues over time. Figure 2 illustrates the block diagram of the electronics depicted in Fig. 1, where the energy storage is additionally connected in parallel with the analog input GPIO (general-purpose input/output).

Fig. 2.

Block diagram of the electronics.

2.3 System Integration and Device Packaging

The complete system was devised and compactly encapsulated within a small-sized device. Figure 3 illustrates two examples of the system's packaging. Figure 3(a) demonstrates the application of a piston-type mechanism, while Fig. 3(b) depicts a ball-triggering mechanism implemented onto the piezoelectric component. Biocompatible materials were carefully selected for the packaging of the device. Figure 3(c) demonstrates the outlook of a fabricated device with diameter of 25 mm and thickness of 5 mm.

Fig. 3.

Schematics of two examples for device structures and packaging: a) piston-type and b) ball-induced sensor element; c) Photo of a fabricated device placed on a medical-grade tape.

2.4 Program for Data Acquisition, Storage, and Transmission

Figure 4 illustrates the logical flowchart of the program utilized in this study. At the commencement of the monitoring procedure, the programmed microcontroller manages voltage measurements from the energy storage and subsequently stores the obtained data in a mass data storage unit. Furthermore, the program has the capability to enable wireless data transmission, although this specific function was not within the scope of this study.

Fig. 4.

Logic flow chart of the program used in this work.

3 Results and Discussions

3.1 Optimization of Sensitivity for Cough Detection

The placement of the device on the user's throat skin is a critical factor that determines the sensitivity and selectivity of cough signals. Figure 5 demonstrates the distinction between data collected at an improper position and data collected at an ideal position¹. It should be noted that the data were unrectified and were collected before the energy storage.

Fig. 5.

a) Demonstration of measurement positions and raw output voltage before rectification detected from the piezoelectric energy harvester when the cough sensing device was attached at b) an improper location (position 4. in a)) and c) an ideal location on throat (position 1. in a)).

In Fig. 5 (b), it is evident that when the device was improperly attached to the throat, the piezoelectric energy harvester exhibited sensitivity to various types of signals beyond coughing. This sensitivity was due to muscle movements triggered by diverse actions, leading to the undesired outcome where the intended cough signals couldn't be distinguished from other signals categorized as noise.

Contrastingly, in Fig. 5 (c), when the device was positioned at an ideal location, the majority of unwanted signals could be effectively filtered out, emphasizing the amplified representation of the anticipated cough signal. Determining the correct positioning on the throat may vary among individuals and should be evaluated through trials before implementing the device in clinical use.

Impedance in the voltage measurements of Fig. 5 was fixed at 1 MΩ, giving the maximum output current of approximately 1.1 µA. However, it should be noted that for the methodology used in this study, the output current from the energy harvester was no longer essential. The charged capacitor voltage replaces the instantaneous output voltage and current and then becomes a crucial factor in the measurement procedure.

3.2 Correction of Influence of Self-discharging on Data Acquisition from Energy Storage Unit

Understanding the self-discharge behavior of the chosen energy storage unit is crucial, as depicted in Fig. 1 and Fig. 4. This entails comparing the actual voltage increase after each ‘Cough-charging’ stage against that subsequent to the ‘Interval-discharging’ stage. Figure 6 exhibits an illustration of the self-discharge curve for a 100 µF commercial capacitor (C1210C107M8PAC7800, KEMET), showcasing its behavior without any external contribution of charging from the piezoelectric energy harvester. Additionally, the real-time voltage drop derived from the self-discharge curve is graphically presented.

When programming the system, it is essential to utilize the self-discharge curve as a dynamic baseline and as a reference for voltage calculation and analysis. This analysis involves estimating the expected decrease in voltage over specific periods to ascertain whether the capacitor has indeed been charged during those intervals.

Equation 1 expresses the gained voltage (V_gain) of the capacitor in each operation cycle (Fig. 1) with data acquisition interval (t_i), where t is time, V_t and V_t+ti are capacitor voltages at the beginning and end of a data acquisition cycle, respectively, and V_d is instantaneous voltage drop shown in Fig. 6. V_gain > 0 within t_i indicates that the capacitor has been charged by the piezoelectric energy harvester and the value of V_gain can then be translated to number or intensity of the coughs for the period of t_i.

Fig. 6.

Dependence of measured capacitor voltage and calculated instantaneous voltage drop on time for a 100 µF capacitor connected in the cough sensing system.

$${V}_{\text{gain}}={V}_{t+{t}_{i}}-{V}_{t}+{\int }_{t}^{t+{t}_{i}}d{V}_{d}\cdot {t}_{i}$$

(1)

3.3 Functionality Demonstration of Cough Sensing and Data Interpretation

Figure 5 previously detailed the device's optimization and positioning at an optimal throat location, effectively filtering out most non-cough signals during data collection. Considering that non-cough signals were relatively smaller compared to cough signals, a threshold value for V_gain was established during the programming phase (Fig. 4) to exclude the recording of undesired harvested energy. Figure 7 displays the data gathered from a volunteer wearing the device during routine activities.

Setting V_gain to 100 mV before data collection ensured that only values above 100 mV were exhibited in Fig. 7. Values below this threshold, unlikely to represent a genuine cough, were omitted. In Fig. 7(a), the individual stayed indoors for approximately 6 h during the test. At the test's start and end, notably large V_gain values were recorded due to attaching and detaching the device, causing substantial impacts on the sensor. These anomalies are easily identifiable in practical scenarios and symmetrically appear at both ends, posing no significant concern for the accuracy of cough signal detection.

The dashed line in Fig. 7(a) represents the minimum level (150 mV) for successful cough signal detection. Among the 13 points surpassing this level, 10 were identified as true cough signals. If this minimum threshold was increased to 200 mV, all nine points above this level were accurate cough signals, but one genuine cough signal fell below the threshold, resulting in a missed detection. Therefore, basic post-data acquisition statistical analysis yielded a detection accuracy range of approximately 77% (10 out of 13) to 90% (9 out of 10) for indoor cough monitoring. It should be noted that the detection accuracy may also be affected by the frequency of cough during the wearing period of the user.

However, detection accuracy notably declined during outdoor activities, as seen in Fig. 7(b). Besides the substantial signals at the test's start and end due to device attachment/detachment, numerous non-cough signals were recorded. Regardless of the threshold level set for statistical analysis, either more non-cough data points were recorded than true cough data points, or a majority of authentic cough data points were disregarded.

In Fig. 7(b), non-cough signals were primarily from clothing adjustments and outdoor activities such as walking and driving. The major reason for these false signals was a scarf worn by the volunteer rubbing against the sensor. As a scarf is likely the closest possible object to the wearer’s throat during outdoor activities, the case in Fig. 7(b) hence represents the worst possible scenario that may appear in practice.

The volunteer deliberately tapped the sensor during the test, with the finger-tapping signals reaching comparable levels to non-cough signals, indicating that outdoor clothes impacting the sensor generated a similar energy amount and transferred it to the capacitor.

While advanced data analysis, like deep learning, might differentiate between cough and non-cough data by analyzing data shapes or correlations [8, 9], this aspect exceeds the study's scope. Despite potential advanced analytical tools, optimizing the piezoelectric energy harvester and device structure to be less susceptible to external stimuli is pivotal. Future works should focus on designing a more sensitive piezoelectric energy harvester, possibly functioning in resonance mode, for improved cough detection.

Another possible challenge could be defining a standard set of thresholds that can be applied to a particular demographic based on gender, age, size, etc. Nevertheless, through clinical trials, it was proven that the position of the device did not need to be adjusted throughout the day for the same device wearer.

Despite the need for enhancements in outdoor cough monitoring, the primary advantage exhibited by the Energy-as-Data protocol implemented in the device—where muscle movement energy harvested through piezoelectric means served as data regarding cough history within specific intervals—was the remarkable extension of battery life. The Li-ion battery (CR1225, Reneta Batteries, Switzerland) powering the entire monitoring system would typically last only overnight when operating in the traditional piezoelectric sensing mode with a high sampling rate. However, in the Energy-as-Data mode employed in this study, the battery could sustain operation for over a week, marking a 2100% increase in battery life thanks to the significantly lowered duty cycle. Considering the fact that possible addition of data analysis protocols can increase the energy consumption, the above-mentioned device structure optimization will be preferred to using on-site data analysis since there is a good chance that an optimized device structure is already able to screen cough data from the input end.

Fig. 7.

Data captured by the cough monitoring device worn by a volunteer during regular activities, covering a) indoor sessions exclusively and b) combined indoor and outdoor scenarios. The V_gain threshold for data acquisition was established at 100 mV.

4 Conclusions

This research has introduced the operational mechanism, architectural design, data acquisition methodology, and performance evaluation of an independent cough monitoring system using a wearable piezoelectric energy harvesting device. Leveraging the Energy-as-Data concept, the battery life has been magnified by a factor of 21. Given the reasonably accurate cough detection rates of 77–90% for indoor monitoring, this novel device has showcased suitability for clinical trials aimed at forecasting and handling respiratory illnesses in patients predominantly indoors. Future endeavors will focus on refining cough detection accuracy in both indoor and outdoor settings. Additionally, study should be conducted for comfort level of wearing the device and hence possible strategies for improving the practicality of wearing the device for cough monitoring in real scenarios.