Sleep medicine relies on the measurement of overnight multidimensional biosignals and audiovisual recordings. For inpatients, a diagnostic procedure is called polysomnography, which consists of several biosignal recordings typically performed in a certified sleep lab within a hospital.
The most prevalent sleep disorder is obstructive sleep apnea (OSA), where patients stop breathing due to an obstruction of the upper airways. A recent review reports values between 9% and 38% in developed countries, in 2000, OSA was the 7th most often primary coded diagnosis in male inpatient (104.763) according to the German Federal Statistical Office. The default diagnosis procedure is an unattended home sleep testing. After instructions by the technical staff of the hospital, the patient attaches the sensors to him-/herself at home and starts the recording device. The data are stored within the device and are returned by the patient the next day. In both cases, data transfer and remote execution of analysis methods are not employed for patient data in sleep medicine in the context of daily care, neither for inpatient, nor for outpatient scenarios. Solutions for storing and processing sleep-related data online are limited due to the inherent data privacy and are currently used only for research. In a similar manner, the use of self-measurements for professional health assessment is currently difficult to reach, as data cannot be processed and shared securely over the internet.
The vision of Asclepios is to store data encrypted in a cloud, while data still can be searched and analysed, and access can be provided dynamically, based on attributes. Asclepios can enable secure cloud storage of the biosignals, which would facilitate efficient and attribute based data sharing with multiple actors. In the inpatient scenario, data could be shared with other experts to provide better medical assessment. This is particularly important, as sleep disorders may be caused by a variety of reasons, spanning virtually all medical fields. Similar cases could be searched without actually revealing the data. In the outpatient scenario, many more actors may be involved. Secure sharing of data would enable to send the devices directly to the patients without the need to go to the clinics. This would be facilitated by attribute based encryption. But even more important: Home sleep monitoring could be evaluated remotely by staff in the sleep laboratory, and may intervene if the measurement is not correctly carried out. Furthermore, automatic data analysis can be carried out within the cloud and remote visual inspection by the physician can be realized without the need that the patient brings the device. This would enable home sleep testing even in regions where no sleep laboratory is in the vicinity, such as many rural areas.
ASCLEPIOS demonstratorThe demonstrator will implement a data management solution with integrated analysis methods within the ASCLEPIOS framework. The core will be dynamic access policies based on attributes rather than personal identification, that would allow efficient but still GPDR compliant remote processing of home sleep testing as well as inpatient recordings. Furthermore, the demonstrator will explore the ASCLEPIOS capabilities to search on the encrypted metadata with standard mechanisms, again to facilitate effective patient care.
Last but not least, the functional encryption module of the ASCLEPIOS framework will be used to provide different automatic signal analysis tools for assessment of the signal quality and the patient‘s health status. The methods explored are
- Detection of various types of artefacts that may occur, such as movement artefacts, sensor detachment, channel interference (in particular ECG).
- Heart rate and sleep apnea detection on ECG: A robust heart rate detection is crucial for the correct subsequent sleep apnea detection. Different heart beat detection methods are executed and an overall rating is performed along the agreement of the methods.
- Respiratory event detection on the individual respiratory channels: Different patterns of respiratory events, such as complete and incomplete obstructive apneas, central apneas, wheezing, snoring and tossing are detected on different respiratory channels and the audio channel.
Impact
The sleep medicine demonstrator built in ASCLEPIOS will illustrate how modern encryption methods for sharing and processing data on clouds can be used to facilitate innovations in patient care. The demonstrator will pave the way for optimal health care in sparsely populated regions by GPDR compliant remote data sharing of monitoring data.