Abstract
BACKGROUND:
Traditional healthcare is centred around providing in-hospital services using hospital owned medical instruments. The COVID-19 pandemic has shown that this approach lacks flexibility to insure follow-up and treatment of common medical problems. In an alternative setting adapted to this problem, participatory healthcare can be considered centred around data provided by patients owning and operating medical data collection equipment in their homes.
OBJECTIVE:
In order to trigger such a shift reliable and price attractive devices need to become available. Snoring, as a human sound production during sleep, can reflect sleeping behaviour and indicate sleep problems as an element of the overall health condition of a person.
METHODS:
The use of off-the-shelf hardware from Internet of Things platforms and standard audio components allows the development of such devices. A prototype of a snoring sound detector with this purpose is developed.
RESULTS:
The device, controlled by the patient and with specific snoring recording and analysing functions is demonstrated as a model for future participatory healthcare.
CONCLUSIONS:
Design of monitoring devices following this model could allow market introduction of new equipment for participatory healthcare, bringing a care complementary to traditional healthcare to the reach of patients, and could result in benefits from enhanced patient participation.
Introduction
Where speech is primarily intended, snoring is an unintended sound production mechanism, resulting from the interaction of the respiratory airflow with the vocal pathway of a sleeping person. The produced sound originates from the flow induced vibrations of different relaxed obstructions in this pathway, functioning as acoustical sources [1], like the nasopharynx, the velum, the oropharyngeal area, the tongue base and the epiglottis as is known in the VOTE classification scheme [2]. As snoring is related to a degree of respiratory airflow obstruction, severe levels of obstruction can lead to an intermittent shortage in oxygen supply, a physiological dangerous condition, labeled as obstructive sleep apnoea (OSA). Patient monitoring techniques and related therapy support including a telemedicine approach have been applied in the treatment of such patients [3]. The recording and analysis of snoring sounds can be part of a sleep-tracking technology giving opportunities in assessment, e.g. of the upper airway anatomy during sleep [4]. Snoring sounds are important elements of the nocturnal sounds, which have been shown to provide a good estimation of sleep architecture and quality [5]. In a new snoring detection device the snoring event extraction was based on a new hysteresis algorithm [6]. In addition, machine learning is offering additional opportunities in specific use of snoring characteristics, as the snoring frequency has been shown to relate to sleep apnoea [7].
According to the technological evolution, connected portable devices with specialized functionality have become commodity products, giving a boost in collecting and storing medical data of patients. Due to their attractive characteristics e-health and m-health devices find their use in, amongst others, insonia management [8], cardiovascular medicine [9], screening of elderly cognitive impairment [10]. They can consist of standard devices (e.g. smart-phones, tablets) or can be implemented as wearable devices [11], realized as flexible hybrid electronics [12]. These products, acting as Internet of Things (IoT)-devices, are receiving acceptance from medical professionals and governments as additional modalities of practice in the healthcare ecosystem. The e-health ecosystem itself consists mainly of data providers (measurement devices), resource providers (operating the communication platforms for data exchange) and brokering parties levering secure services to manage private data and requests. In view of the healthcare ecosystem, participatory healthcare [13, 14, 15] is considered as an modus operandus for deploying healthcare where the patient is engaged and activated through participation in decision making. Considering the provision of data, in traditional healthcare a patient visits a hospital or specialized center in order to obtain the desired measurement results from highly functional specific medical equipment, as decided by the physician accordingly to clinical symptoms and signs. In this way, the hospital as owner of the medical devices is providing its traditional services. The recent COVID-19 pandemic has shown that the traditional modality of providing medical care has been disrupted with indications of a large number of patients with undiagnosed conditions [16] and with an important shift towards different types of telemedicine [17].
Where device ownership is common with wearables (e.g. smartwatches to follow heartrate, activity level and sleep stages [18]), it is limited in case of non-intrusive devices. In this study a low cost IoT snoring sound detector to be placed at a fixed place in the sleeping room was developed with the focus on such use at home. The system allows for patient controlled snoring sound recording and automated signal processing, resulting in patient specific reports of the personal sleep pattern.
Conceptual illustration of the device application showing the device with the external database for user administration and the implemented user access roles, including a photograph of the assembled device.
IoT device
In order to obtain a reliable device with flexible functionality, the widespread open source embedded platform Raspberry Pi (
Cloud services
Different cloud storage providers offer fully automated cloud services with best practices for security and operational aspects, which can be used to provide data exploration, visualisation and maintenance. In order to demonstrate our current approach MongoDB (
System performance
IoT device
The realized device combined with a conceptual illustration of the cloud based application is shown in Fig. 1. The device is assembled from the following parts: Raspberry Pi 3 model B (Raspberry Pi Foundation, Cambridge, UK; price 35
The Wifi enabled device is operated in the sleeping room, with remote access possible through an available internet connection. The device is accessible from the graphical user interface (GUI) of the Node.js web server (OpenJS Foundation,
Timing analysis of a 5 minute snoring sequence: (upper) audio signal; (middle) envelope signal (blue) with indication of snoring onsets (red solid line) and snoring offsets (red dashed line); (lower) combined histogram of snoring durations (left) and snoring onset-onset intervals (right).
Snoring and breathing follow in general a respiratory periodic pattern (with a period normally ranging between 3 and 5 seconds), resulting in a specific temporal pattern which can be used to eliminate non-snoring sound events. In order to illustrate the functional performance of the present device, sound recordings from a regular breathing and clear acoustic dominant snoring conditions were considered. Snoring event detection was implemented using an algorithm similar to the hysteresis extraction algorithm [6] and is designed to determine the onset and offset time of a snoring event. It is based on an envelope threshold passing condition, with an appropriate threshold level well above the predetermined background level. With this purpose, the envelope of the audio signal is calculated by applying a moving average filter (with a time window of about 0.15 s) on the absolute values of the recorded signal. In order to obtain the onset time a time frame of about 0.3 s is shifted with sample step incrementation over the envelope signal. The time frame is divided symmetrical around the current sample in a left part (about 130 ms), a middle part (about 40 ms) and a right part (about 130 ms). The following condition is tested: if the maximum value in the left frame part is below the threshold value and the minimum in the right frame part is above the threshold value, then the current sample is considered as the snoring onset point. Reversing the conditions is then applied to determine in a similar way the snoring offset point. The middle part of the frame is not used in order to improve the noise robustness of the algorithm.
As a result, the snoring duration and the snoring onset-onset intervals were obtained as standard features. Based on this snoring sound segmentation, additional features (e.g. snoring frequency center of mass, acoustic energy, crest factor) of the snoring sounds can be calculated for snoring characterization [19]. The functional performance was validated from recordings obtained during private home use as part of operational testing and demonstration of the system. Figure 2 illustrates the processing showing a typical 5 min recorded audio signal extracted from a nightly recording.
Discussion and conclusion
As demonstrated, the realized device can integrate data collection, data processing and data distribution in a single system. The use of a dedicated non-intrusive device is, in our view, preferred to guarantee identical conditions for the long term monitoring with minimal effort in operation and maintenance. It offers the flexibility to consistently select a single specific measurement position (below the bed, next to the bed, above the bed, or even integrated in the sleeping room) to guarantee the quality of follow-up measurements. This is in contrast to the use of a smartphone with a ‘snoring sound detection’ app, which could offer similar functionality [8]. The use can be tailored to the different user roles of the stakeholders by enabling the authorization levels for accessing the different functions. Where the patient (and his home mates) in combination with the medical specialist (and his assistants) are the main stakeholders, it could be expanded to additional professionals like home care givers. The flexibility of the device makes that different analysis options can be provided, adapted towards the specific needs of a patient. Such a personalized approach can consist in the follow-up of snoring features [7], or in the determination of a person’s architecture of the night [5].
In order to promote the device in view of participatory healthcare, its use should be beneficial for both the patient and the medical specialist. In using the device the measurements are performed in the home setting of a patient ensuring reliable relevant data collection, which is essential for decision making. In addition long term repeated monitoring can be obtained. In case of simple snoring the device can measure snoring to obtain longitudinal information over weeks in view of the patient’s (or partners) complaining.
Footnotes
Acknowledgments
The authors would like to thank Arthur Van de Velde, Dries Vangansbeke and Simon Adams for their contribution in the development of the device. There was no support in the form of funding for this study.
Conflict of interest
None to report.