Screening for obstructive sleep apnea with novel hybrid acoustic smartphone app technology.

BACKGROUND: Obstructive sleep apnea (OSA) has a high prevalence, with an estimated 425 million adults with apnea hypopnea index (AHI) of ≥15 events/hour, and is significantly underdiagnosed. This presents a significant pain point for both the sufferers, and for healthcare systems, particularly in a post COVID-19 pandemic world. As such, it presents an opportunity for new technologies that can enable screening in both developing and developed countries. In this work, the performance of a non-contact OSA screener App that can run on both Apple and Android smartphones is presented. METHODS: The subtle breathing patterns of a person in bed can be measured via a smartphone using the "Firefly" app technology platform [and underpinning software development kit (SDK)], which utilizes advanced digital signal processing (DSP) technology and artificial intelligence (AI) algorithms to identify detailed sleep stages, respiration rate, snoring, and OSA patterns. The smartphone is simply placed adjacent to the subject, such as on a bedside table, night stand or shelf, during the sleep session. The system was trained on a set of 128 overnights recorded at a sleep laboratory, where volunteers underwent simultaneous full polysomnography (PSG), and "Firefly" smartphone app analysis. A separate independent test set of 120 recordings was collected across a range of Apple iOS and Android smartphones, and withheld for performance evaluation by a different team. An operating point tuned for mid-sensitivity (i.e., balancing sensitivity and specificity) was chosen for the screener. RESULTS: The performance on the test set is comparable to ambulatory OSA screeners, and other smartphone screening apps, with a sensitivity of 88.3% and specificity of 80.0% [with receiver operating characteristic (ROC) area under the curve (AUC) of 0.92], for a clinical threshold for the AHI of ≥15 events/hour of detected sleep time. CONCLUSIONS: The "Firefly" app based sensing technology offers the potential to significantly lower the barrier of entry to OSA screening, as no hardware (other than the user's personal smartphone) is required. Additionally, multi-night analysis is possible in the home environment, without requiring the wearing of a portable PSG or other home sleep test (HST).

A pilot study of the nocturnal respiration rates in COPD patients in the home environment using a non-contact biomotion sensor.

Nocturnal respiration rate parameters were collected from 20 COPD subjects over an 8 week period, to determine if changes in respiration rate were associated with exacerbations of COPD. These subjects were primarily GOLD Class 2 to 4, and had been recently discharged from hospital following a recent exacerbation. The respiration rates were collected using a non-contact radio-frequency biomotion sensor which senses respiratory effort and body movement using a short-range radio-frequency sensor. An adaptive notch filter was applied to the measured signal to determine respiratory rate over rolling 15 s segments. The accuracy of the algorithm was initially verified using ten manually-scored 15 min segments of respiration extracted from overnight polysomnograms. The calculated respiration rates were within 1 breath min(-1) for >98% of the estimates. For the 20 subjects monitored, 11 experienced one or more subsequent exacerbation of COPD (ECOPD) events during the 8 week monitoring period (19 events total). Analysis of the data revealed a significant increase in nocturnal respiration rate (e.g. >2 breath min(-1)) prior to many ECOPD events. Using a simple classifier of a change of 1 breath min(-1) in the mode of the nocturnal respiration rate, a predictive rule showed a sensitivity of 63% and specificity of 85% for predicting an exacerbation within a 5 d window. We conclude that it is possible to collect respiration rates reliably in the home environment, and that the respiration rate may be a potential indicator of change in clinical status.

Real time breathing rate estimation from a non contact biosensor.

An automated real time method for detecting human breathing rate from a non contact biosensor is considered in this paper. The method has low computational and RAM requirements making it well-suited to real-time, low power implementation on a microcontroller. Time and frequency domain methods are used to separate a 15s block of data into movement, breathing or absent states; a breathing rate estimate is then calculated. On a 1s basis, 96% of breaths were scored within 1 breath per minute of expert scored respiratory inductance plethysmography, while 99% of breaths were scored within 2 breaths per minute. When averaged over 30s, as is used in this respiration monitoring system, over 99% of breaths are within 1 breath per minute of the expert score.

A portable automated assessment tool for sleep apnea using a combined Holter-oximeter.

STUDY OBJECTIVES: Resource limitations have raised interest in portable monitoring systems that can be used by specialist sleep physicians as part of an overall strategy to improve access to the diagnosis of sleep apnea. This study validates a combined electrocardiogram and oximetry recorder (Holter-oximeter) against simultaneous polysomnography for detection of sleep apnea. DESIGN: Prospective study. SETTING: A dedicated sleep disorders unit. PARTICIPANTS: 59 adults presenting for evaluation of suspected sleep apnea. INTERVENTIONS: NA. MEASUREMENTS AND RESULTS: An automated algorithm previously developed for sleep apnea detection was applied to the electrocardiogram and oximetry measurements. The algorithm provides (a) epoch-by-epoch estimates of apnea occurrence and (b) estimates of overall per-subject AHI. Using separate thresholds of AHI > or =15 and AHI <5 for defining clinically significant and insignificant sleep apnea, sensitivity, specificity, and likelihood ratios, conditional on positive or negative (but not indeterminate) test results were used to assess agreement between the proposed system and polysomnography. Sensitivity of 95.8% and specificity of 100% was achieved. Positive and negative likelihood ratios were >20 and 0.04 respectively, with 16.7% of subjects having intermediate test results (AHI 5-14/h). Regardless ofAHI, 85.3% of respiratory events were correctly annotated on an epoch-by-epoch basis. AHI underestimation bias was 0.9/h, and the antilogs of log-transformed limits of agreement were 0.3 and 2.7. Correlation between estimated and reference AHI was 0.95 (P <0.001). CONCLUSION: Combined Holter-oximeter monitoring compares well with polysomnography for identifying sleep apnea in an attended setting and is potentially suitable for home-based automated assessment of sleep apnea in a population suspected of having sleep apnea.

Automated detection of paroxysmal atrial fibrillation from inter-heartbeat intervals.

An automated method for detecting episodes of probable paroxysmal atrial fibrillation based on processing blocks of inter-heartbeat intervals is considered. The method has very low computational requirements making it well-suited to near real-time, low power applications. A supervised linear discriminant classifier is used to estimate the likelihood of a block of inter-heartbeat intervals containing paroxysmal atrial fibrillation (PAF). Per block accuracies in separating normal from PAF were 92%, 94%, 100% and 100% when the method was used to process the Physionet MITDB, AFDB, NSRDB and NSR2DB databases respectively.

An evaluation of a non-contact biomotion sensor with actimetry.

Actimetry is a widely accepted technology for the diagnosis and monitoring of sleep disorders such as insomnia, circadian sleep/wake disturbance, and periodic leg movement. In this study we investigate a very sensitive non-contact biomotion sensor to measure actimetry and compare its performance to wrist-actimetry. A data corpus consisting of twenty subjects (ten normals, ten with sleep disorders) was collected in the unconstrained home environment with simultaneous non-contact sensor and ActiWatch actimetry recordings. The aggregated length of the data is 151 hours. The non-contact sensor signal was mapped to actimetry using 30 second epochs and the level of agreement with the ActiWatch actimetry determined. Across all twenty subjects, the sensitivity and specificity was 79% and 75% respectively. In addition, it was shown that the non-contact sensor can also measure breathing and breathing modulations. The results of this study indicate that the non-contact sensor may be a highly convenient alternative to wrist-actimetry as a diagnosis and screening tool for sleep studies. Furthermore, as the non-contact sensor measures breathing modulations, it can additionally be used to screen for respiratory disturbances in sleep caused by sleep apnea and COPD.

Detection of obstructive sleep apnea in pediatric subjects using surface lead electrocardiogram features.

STUDY OBJECTIVES: To investigate the feasibility of detecting obstructive sleep apnea (OSA) in children using an automated classification system based on analysis of overnight electrocardiogram (ECG) recordings. DESIGN: Retrospective observational study. SETTING: A pediatric sleep clinic. PARTICIPANTS: Fifty children underwent full overnight polysomnography. INTERVENTION: N/A. MEASUREMENTS AND RESULTS: Expert polysomnography scoring was performed. The datasets were divided into a training set of 25 subjects (11 normal, 14 with OSA) and a withheld test set of 25 subjects (11 normal, 14 with OSA). Features, calculated from the ECG of the 25 training datasets, were empirically chosen to train a modified quadratic discriminant analysis classification system. The selected configuration used a segment length of 60 seconds and processed mean, SD, power spectral density, and serial correlation measures to classify segments as apneic or normal. By combining per-segment classifications and using receiver-operator characteristic analysis, a per-subject classifier was obtained that had a sensitivity of 85.7%, specificity of 90.9%, and accuracy of 88% on the training datasets. The same decision threshold was applied to the withheld datasets and yielded a sensitivity of 85.7%, specificity of 81.8%, and accuracy of 84%. The positive and negative predictive values were 85.7% and 81.8%, respectively, on the test dataset. CONCLUSIONS: The ability to correctly identify 12 out of 14 cases of OSA (with the 2 false negatives arising from subjects with an apnea-hypopnea index less than 10) indicates that the automated apnea classification system outlined may have clinical utility in pediatric patients.