JMIR Publications

ABSTRACT

Background:

There is nowadays an increased demand for accurate and personalized healthcare monitoring due to the different challenges facing healthcare care systems, namely rising costs, and a shortage of clinicians. Nonintrusive monitoring of vital signs is a potential solution to close current gaps in patient monitoring. As an example, bed-embedded ballistocardiogram (BCG) sensors can provide information about abnormal cardiac patients (e.g., patients with arrhythmias) and abnormal breathing (viz., obstructive sleep apnea) non-intrusively without disturbing the patient’s daily activities. Detecting obstructive sleep apnea via BCG sensors is gaining increased attention from many researchers due to their simple installation and accessibility, i.e., their non-wearable nature. In the field of nonintrusive vital sign monitoring; microbend fiber optic sensor (MFOS), among other sensors, has proven suitable. Yet, few studies looked into apnea detection.

Objective:

This study aimed at assessing the capabilities of an MFOS for nonintrusive vital signs and sleep apnea detection during an in-lab sleep study. Data were collected from patients having sleep apnea in the sleep lab of Khoo Teck Puat Hospital.

Methods:

10 participants underwent full polysomnography (PSG) and the MFOS was placed under the patient’s bed-mattress for our BCG data collection. The predictive capabilities of the suggested sensor for apnea detection against the gold-standard PSG were assessed based on the accuracy (Acc), sensitivity (Sens), and specificity (Spec). The sleep apnea events were detected through a histogram based-thresholding method and the detection was performed on a minute by minute basis. The apneic events detection algorithm was evaluated against the manually scored events obtained from the PSG study. Further, normalized mean absolute error (NMAE) and normalized root-mean-square error (NRMSE) were employed to assess the sensor capabilities for vital signs measurement, comprising the heart and respiratory rates. Heart rates were detected form derived BCG signals via the multi-resolution analysis of the maximal overlap discrete wavelet transform. Respiratory rates were detected from derived respiratory effort signals using a sliding window peak detection approach. Vital signs were evaluated based on a 30-second time window with an overlap of 15 seconds. In this study, electrocardiogram and thoracic effort signals were used as references to assess the proposed vital signs detection algorithms.

Results:

Across the 10-patients recruited for the study, the system achieved impartial results to PSG for sleep apnea detection such as an accuracy of 50.36%±6.61%, a sensitivity of 56.14%±13.40%, and a specificity of 46.47%±10.59%. Besides, the system achieved close results for heart and respiratory rates such as an NMAE of 4.92%±2.58% and an NRMSE of 8.15%±2.72% for heart rate, while an NMAE of 11.16%±3.15% and an NRMSE of 15.10%±3.4% for respiratory rate.

Conclusions:

Overall, the recommended system produced average results for apneic events detection considering the complexity of sensors required to diagnose this syndrome in a clinical setting, whereas satisfying results were obtained for vital signs detection compared to the PSG outcomes. These results provide preliminary support for the potential use of the MFOS for sleep apnea detection. The proposed sensor is not a replacement for the PSG. However, it can be thought of as a complementary monitoring method for clinicians, in the case where they don’t have access to the patient’s health status (e.g., out-of-hospital/patients’ home).