German Clinical Trials Register

Acronym/abbreviation of the study

No Entry

URL of the study

No Entry

Brief summary in lay language

This study aims to investigate whether artificial intelligence methods are suitable for detecting and researching arousals. Arousals are brief awakening reactions from sleep, which, as a patient, you generally do not notice. Determining the number of arousals is important because they can significantly disrupt restorative sleep as their number increases. Arousals can be classified into different groups, known as classes, based on their triggering causes. Identifying these classes is a crucial indicator for diagnosing sleep disorders. Until now, this type of arousal diagnostics has been laboriously performed manually through visual evaluation by sleep medicine professionals. Due to the large amount of data to be evaluated, the evaluation is very time-consuming. To investigate the scientific questions, a software system will be developed based on data routinely collected in the sleep laboratory. This system will automatically display these arousals and present the results in a way that is understandable for the sleep physician. Subsequently, it will be examined how high the quality and time expenditure of arousal diagnostics are when a sleep medicine evaluator uses the software system for decision support. The goal is for an evaluator to be faster than before using the system, achieving an accuracy in arousal determination as high as that of the best sleep medicine expert involved in the study, who determined the arousals without system support. Special attention will be paid to whether the system's decisions are transparent and thus verifiable for human evaluators. Furthermore, we hope to identify new markers and rules for determining arousals with the help of pulse wave analysis. This may make arousal diagnostics more reliable, technically simpler, and faster, without the considerable time effort previously required by the examiner. Pulse wave analysis involves examining the pulse waves triggered by contractions of the heart's left ventricle more closely at a finger. These blood pulsations continue from the heart through the large to the smallest vessels in the human body. They are measured technically using a small clip attached to the finger, the pulse oximeter. This pulse oximetry enables continuous measurement of blood pressure, pulse, and blood oxygen saturation. These data also allow for capturing the shape of the pulse wave and its propagation speed. The advantage of the method lies in the continuous, non-invasive measurement of these values, which can also represent short-term changes lasting only a few seconds. Since arousals can cause such short-term changes in these measured variables, pulse wave analysis may represent a diagnostic method to detect such arousals earlier and more accurately. The necessary pulse oximetry is generally used in the nightly routine program in the sleep laboratory, thus representing no additional effort or burden for you during the night in the sleep laboratory. Finally, it will be investigated whether there is a correlation between arousal diagnostics and the medical diagnosis of a sleep disorder or between arousal diagnostics and the patient's statements regarding their daytime sleepiness and sleep quality. Men and women aged 18 and older who are referred for polysomnography to the sleep laboratory of the Klinik für Kardiologie, Angiologie u. Pneumologie of the medical clinic in Esslingen am Neckar, Germany, and who have given their consent will be included in the study. Only purely diagnostic examinations are considered. Examinations in which therapeutic methods in the form of positive pressure therapies are used are excluded.

Brief summary in scientific language

This study aims to investigate the extent to which pulse wave analysis (PWA) is a suitable new method for automating the diagnostics in sleep laboratories in terms of detecting (detection) sleep-related arousals and their causal origin (differentiation) with computer support. In addition to PWA data, measurement and analysis data from the routine polysomnographic (PSG) examination will also be used. Methods of Machine Learning will then be used to develop an arousal detector and classifier. This will form the basis for the development of a decision-support system for medical professionals (hereinafter referred to as "scorers") for arousal diagnostics. Finally, it will be practically evaluated whether the system is usable and whether its use leads to higher efficiency of the scorers in arousal diagnostics and an improvement in the quality of the results. Arousals are brief waking reactions of the body that momentarily interrupt the sleep process or change the sleep stage, leading to sleep fragmentation, which is associated with decreased restfulness of sleep and hypersomnia. According to the criteria of the American Association of Sleep Medicine (AASM), their diagnostics are performed using electroencephalogram (EEG). For their causal differentiation, additional parameters are needed, such as the measurement of limb movements through an electromyogram (EMG) of the tibialis anterior muscle, the breathing movements of the thorax and abdomen, and the airflow. Certain arousals, for example, triggered by pharyngeal obstruction in obstructive sleep apnea, are accompanied by changes in cardiorespiratory parameters of blood pressure, pulse, and oxygen saturation, as well as by hyperventilation phases and obstructive respiratory efforts. This is associated with an increase in sympathetic activity, which can be indirectly measured through the pulse transit time (PTT) of the pulse wave. Important in arousal diagnostics are determining their frequency to assess sleep quality and identifying their triggers for diagnosis, which determines the further therapeutic procedure. For this reason, polysomnographic data analysis is initially automated through a software solution. Since reliable arousal detection is not yet possible today due to the involvement of many parameters, a time-consuming and labor-intensive manual review of a polysomnogram by scorers is subsequently necessary. The capture of the pulse wave, which propagates linearly from the heart to the smallest vessels, is currently done non-invasively through photoplethysmography using a clip attached to the finger. PWA includes the measurement of oxygen saturation, heart rate, and the representation of the pressure curve of the pulse wave over time. Through computational models and a simultaneously derived ECG, the pulse wave velocity (PWV), PTT, and blood pressure can be continuously measured. Since arousals often involve changes in circulation-relevant and respiratory parameters that can also be captured with PWA, it will be examined to what extent PWA data are suitable for supplementing or replacing the current arousal diagnostics according to AASM criteria. In the realization of this system, methods of Machine Learning are used. By evaluating patterns in large amounts of measurement data, a model is initially formed, which is then used on new patient raw data for the determination of arousals. Since Machine Learning makes it possible to uncover unexpected correlations and data patterns, in a further step, sleep medical records, which have not previously been used for the determination of arousals, will also be considered for model formation. Men and women aged 18 and older who are referred for polysomnography to the sleep laboratory of the Klinik für Kardiologie, Angiologie u. Pneumologie of the medical clinic in Esslingen am Neckar, Germany, and who have given their consent will be included in the study. Only purely diagnostic examinations are considered. Examinations in which therapeutic methods in the form of positive pressure therapies are used are excluded.