Knowledge distillation under ideal joint classifier assumption.

Knowledge distillation constitutes a potent methodology for condensing substantial neural networks into more compact and efficient counterparts. Within this context, softmax regression representation learning serves as a widely embraced approach, leveraging a pre-established teacher network to guide the learning process of a diminutive student network. Notably, despite the extensive inquiry into the efficacy of softmax regression representation learning, the intricate underpinnings governing the knowledge transfer mechanism remain inadequately elucidated. This study introduces the 'Ideal Joint Classifier Knowledge Distillation' (IJCKD) framework, an overarching paradigm that not only furnishes a lucid and exhaustive comprehension of prevailing knowledge distillation techniques but also establishes a theoretical underpinning for prospective investigations. Employing mathematical methodologies derived from domain adaptation theory, this investigation conducts a comprehensive examination of the error boundary of the student network contingent upon the teacher network. Consequently, our framework facilitates efficient knowledge transference between teacher and student networks, thereby accommodating a diverse spectrum of applications.

A transformer-based diffusion probabilistic model for heart rate and blood pressure forecasting in Intensive Care Unit.

BACKGROUND AND OBJECTIVE: Vital sign monitoring in the Intensive Care Unit (ICU) is crucial for enabling prompt interventions for patients. This underscores the need for an accurate predictive system. Therefore, this study proposes a novel deep learning approach for forecasting Heart Rate (HR), Systolic Blood Pressure (SBP), and Diastolic Blood Pressure (DBP) in the ICU. METHODS: We extracted 24,886 ICU stays from the MIMIC-III database which contains data from over 46 thousand patients, to train and test the model. The model proposed in this study, Transformer-based Diffusion Probabilistic Model for Sparse Time Series Forecasting (TDSTF), merges Transformer and diffusion models to forecast vital signs. The TDSTF model showed state-of-the-art performance in predicting vital signs in the ICU, outperforming other models' ability to predict distributions of vital signs and being more computationally efficient. The code is available at https://github.com/PingChang818/TDSTF. RESULTS: The results of the study showed that TDSTF achieved a Standardized Average Continuous Ranked Probability Score (SACRPS) of 0.4438 and a Mean Squared Error (MSE) of 0.4168, an improvement of 18.9% and 34.3% over the best baseline model, respectively. The inference speed of TDSTF is more than 17 times faster than the best baseline model. CONCLUSION: TDSTF is an effective and efficient solution for forecasting vital signs in the ICU, and it shows a significant improvement compared to other models in the field.

DeScoD-ECG: Deep Score-Based Diffusion Model for ECG Baseline Wander and Noise Removal.

OBJECTIVE: Electrocardiogram (ECG) signals commonly suffer noise interference, such as baseline wander. High-quality and high-fidelity reconstruction of the ECG signals is of great significance to diagnosing cardiovascular diseases. Therefore, this paper proposes a novel ECG baseline wander and noise removal technology. METHODS: We extended the diffusion model in a conditional manner that was specific to the ECG signals, namely the Deep Score-Based Diffusion model for Electrocardiogram baseline wander and noise removal (DeScoD-ECG). Moreover, we deployed a multi-shots averaging strategy that improved signal reconstructions. We conducted the experiments on the QT Database and the MIT-BIH Noise Stress Test Database to verify the feasibility of the proposed method. Baseline methods are adopted for comparison, including traditional digital filter-based and deep learning-based methods. RESULTS: The quantities evaluation results show that the proposed method obtained outstanding performance on four distance-based similarity metrics with at least 20% overall improvement compared with the best baseline method. CONCLUSION: This paper demonstrates the state-of-the-art performance of the DeScoD-ECG for ECG baseline wander and noise removal, which has better approximations of the true data distribution and higher stability under extreme noise corruptions. SIGNIFICANCE: This study is one of the first to extend the conditional diffusion-based generative model for ECG noise removal, and the DeScoD-ECG has the potential to be widely used in biomedical applications.

Coupling analysis of heart rate variability and cortical arousal using a deep learning algorithm.

Frequent cortical arousal is associated with cardiovascular dysfunction among people with sleep-disordered breathing. Changes in heart rate variability (HRV) can represent pathological conditions associated with autonomic nervous system dysfunction. Previous studies showed changes in cardiac activity due to cortical arousals. However, few studies have examined the instantaneous association between cortical arousal and HRV in an ethnically diverse population. In this study, we included 1,069 subjects' full night ECG signals from unattended polysomnography in the Multi-Ethnic Study of Atherosclerosis dataset. An automated deep learning tool was employed to annotate arousal events from ECG signals. The etiology (e.g., respiratory, or spontaneous) of each arousal event was classified through a temporal analysis. Time domain HRVs and mean heart rate were calculated on pre-, intra-, and post-arousal segments of a 25-s period for each arousal event. We observed that heart rate and HRVs increased during the arousal onsets in the intra-arousal segments, regardless of arousal etiology. Furthermore, HRVs response to cortical arousal occurrence differed according to gender and the sleep stages in which arousal occurred. The more intense HRVs variation due to arousal in females can contribute to a potentially stronger association between arousal burden and long-term mortality. The excessive abrupt sympathetic tone elevation in REM caused by arousal may provide insights on the association between sleep and sudden cardiac death.

BASH-GN: a new machine learning-derived questionnaire for screening obstructive sleep apnea.

PURPOSE: This study aimed to develop a machine learning-based questionnaire (BASH-GN) to classify obstructive sleep apnea (OSA) risk by considering risk factor subtypes. METHODS: Participants who met study inclusion criteria were selected from the Sleep Heart Health Study Visit 1 (SHHS 1) database. Other participants from the Wisconsin Sleep Cohort (WSC) served as an independent test dataset. Participants with an apnea hypopnea index (AHI) ≥ 15/h were considered as high risk for OSA. Potential risk factors were ranked using mutual information between each factor and the AHI, and only the top 50% were selected. We classified the subjects into 2 different groups, low and high phenotype groups, according to their risk scores. We then developed the BASH-GN, a machine learning-based questionnaire that consists of two logistic regression classifiers for the 2 different subtypes of OSA risk prediction. RESULTS: We evaluated the BASH-GN on the SHHS 1 test set (n = 1237) and WSC set (n = 1120) and compared its performance with four commonly used OSA screening questionnaires, the Four-Variable, Epworth Sleepiness Scale, Berlin, and STOP-BANG. The model outperformed these questionnaires on both test sets regarding the area under the receiver operating characteristic (AUROC) and the area under the precision-recall curve (AUPRC). The model achieved AUROC (SHHS 1: 0.78, WSC: 0.76) and AUPRC (SHHS 1: 0.72, WSC: 0.74), respectively. The questionnaire is available at https://c2ship.org/bash-gn . CONCLUSION: Considering OSA subtypes when evaluating OSA risk may improve the accuracy of OSA screening.

Pilot Study for Correlation of Heart Rate Variability and Dopamine Transporter Brain Imaging in Patients with Parkinsonian Syndrome.

BACKGROUND: Parkinsonian syndrome (PS) is a broad category of neurodegenerative movement disorders that includes Parkinson disease, multiple system atrophy (MSA), progressive supranuclear palsy, and corticobasal degeneration. Parkinson disease (PD) is the second most common neurodegenerative disorder with loss of dopaminergic neurons of the substantia nigra and, thus, dysfunction of the nigrostriatal pathway. In addition to the motor symptoms of bradykinesia, rigidity, tremors, and postural instability, nonmotor symptoms such as autonomic dysregulation (AutD) can also occur. Heart rate variability (HRV) has been used as a measure of AutD and has shown to be prognostic in diseases such as diabetes mellitus and cirrhosis, as well as PD. I-123 ioflupane, a gamma ray-emitting radiopharmaceutical used in single-photon emission computed tomography (SPECT), is used to measure the loss of dopaminergic neurons in PD. Through the combination of SPECT and HRV, we tested the hypothesis that asymmetrically worse left-sided neuronal loss would cause greater AutD. METHODS: 51 patients were enrolled on the day of their standard of care I-123 ioflupane scan for the work-up of possible Parkinsonian syndrome. Demographic information, medical and medication history, and ECG data were collected. HRV metrics were extracted from the ECG data. I-123 ioflupane scans were interpreted by a board-certified nuclear radiologist and quantified by automated software to generate striatal binding ratios (SBRs). Statistical analyses were performed to find correlations between the HRV and SPECT parameters. RESULTS: 32 patients were excluded from the final analysis because of normal scans, prior strokes, cardiac disorders and procedures, or cancer. Abnormal I-123 ioflupane scans were clustered using T-SNE, and one-way ANOVA was performed to compare HRV and SBR parameters. The analysis was repeated after the exclusion of patients taking angiotensin-converting enzyme inhibitors, given the known mechanism on autonomic function. Subsequent analysis showed a significant difference between the high-frequency domains of heart rate variability, asymmetry of the caudate SBR, and putamen-to-caudate SBR. CONCLUSION: Our results support the hypothesis that more imbalanced (specifically worse left-sided) neuronal loss results in greater AutD.

Distributed Full Synchronized System for Global Health Monitoring Based on FLSA.

In modern medicine, smart wireless connected devices are gaining an increasingly important role in aiding doctors' job of monitoring patients. More and more complex systems, with a high density of sensors capable of monitoring many biological signals, are arising. Merging the data offers a great opportunity for increasing the reliability of diagnosis. However, a huge problem is constituted by synchronization. Multi-board wireless-connected monitoring systems are a typical example of distributed systems and synchronization has always been a challenging issue. In this paper, we present a distributed full synchronized system for monitoring patients' health capable of heartbeat rate, oxygen saturation, gait and posture analysis, and muscle activity measurements. The time synchronization is guaranteed thanks to the Fractional Low-power Synchronization Algorithm (FLSA).

Obstructive sleep apnea predicts 10-year cardiovascular disease-related mortality in the Sleep Heart Health Study: a machine learning approach.

STUDY OBJECTIVES: Obstructive sleep apnea (OSA) is considered to be an important risk factor for the development of cardiovascular disease (CVD). This study aimed to develop and evaluate a machine learning approach with a set of features for assessing the 10-year CVD mortality risk of the OSA population. METHODS: This study included 2,464 patients with OSA who met study inclusion criteria and were selected from the Sleep Heart Health Study. We evaluated the importance of potential features by mutual information. The top 9 features were selected to develop a random forest model. RESULTS: We evaluated the model performance on a test set (n = 493) using the area under the receiver operating curve with 95% confidence interval and confusion matrix. A random forest model awarded the highest area under the receiver operating curve of 0.84 (95% confidence interval: 0.78-0.89). The specificity and sensitivity were 73.94% and 81.82%, respectively. Sixty-three years old was a threshold for increased risk of 10-year CVD mortality. Persons with severe OSA had higher risk than those with mild OSA. CONCLUSIONS: This study demonstrated that a random forest model can provide a quick assessment of the risk of 10-year CVD mortality. Our model may be more informative for patients with OSA in determining their future CVD mortality risk. CITATION: Li A, Roveda JM, Powers LS, Quan SF. Obstructive sleep apnea predicts 10-year cardiovascular disease-related mortality in the Sleep Heart Health Study: a machine learning approach. J Clin Sleep Med. 2022;18(2):497-504.

A deep learning-based algorithm for detection of cortical arousal during sleep.

STUDY OBJECTIVES: The frequency of cortical arousals is an indicator of sleep quality. Additionally, cortical arousals are used to identify hypopneic events. However, it is inconvenient to record electroencephalogram (EEG) data during home sleep testing. Fortunately, most cortical arousal events are associated with autonomic nervous system activity that could be observed on an electrocardiography (ECG) signal. ECG data have lower noise and are easier to record at home than EEG. In this study, we developed a deep learning-based cortical arousal detection algorithm that uses a single-lead ECG to detect arousal during sleep. METHODS: This study included 1,547 polysomnography records that met study inclusion criteria and were selected from the Multi-Ethnic Study of Atherosclerosis database. We developed an end-to-end deep learning model consisting of convolutional neural networks and recurrent neural networks which: (1) accepted varying length physiological data; (2) directly extracted features from the raw ECG signal; (3) captured long-range dependencies in the physiological data; and (4) produced arousal probability in 1-s resolution. RESULTS: We evaluated the model on a test set (n = 311). The model achieved a gross area under precision-recall curve score of 0.62 and a gross area under receiver operating characteristic curve score of 0.93. CONCLUSION: This study demonstrated the end-to-end deep learning approach with a single-lead ECG has the potential to be used to accurately detect arousals in home sleep tests.

Sleep patterns and sleep deprivation recorded by actigraphy in 4th-grade and 5th-grade students.

OBJECTIVE: This study investigates sleep patterns of fourth- and fifth-grade students using actigraphy. METHODS: The study included 257 students enrolled in a Southwestern US school district who participated in a novel sleep science curriculum during the Spring 2016-17 and Fall 2017-18 semesters and met the study inclusion criteria. As part of this curriculum, participants underwent 5-7 days of continuous wrist actigraphy and completed an online sleep diary. RESULTS: Approximately two-thirds of the 9-11-year-old fourth- and fifth-grade students slept less than the minimum 9 h per night recommended by both the American Academy of Sleep Medicine/Sleep Research Society and the National Sleep Foundation. The sleep midpoint time on weekends was about 1 h later than on weekdays. There was a significant effect of age on sleep duration. Compared to 9-year old students, a larger proportion of 10-year old students had a sleep duration less than 8.5 h. Boys had shorter sleep duration than girls, and a larger percentage of boys obtained less than 9 h of sleep compared to girls. CONCLUSIONS: Insufficient sleep is a highly prevalent condition among 9-11-year-old fourth- and fifth-grade elementary students. Importantly, there is a difference between sleep patterns on weekdays and weekends which may portend greater problems with sleep in adolescence and young adulthood.

A Novel Artificial Neural Network Based Sleep-Disordered Breathing Screening Tool.

STUDY OBJECTIVES: This study evaluated a novel artificial neural network (ANN) based sleep-disordered breathing (SDB) screening tool incorporating nocturnal pulse oximetry with demographic, anatomic, and clinical data. The tool was compatible with 6 categories of apnea-hypopnea index (AHI) with 4% oxyhemoglobin desaturation threshold, ≥ 5, 10, 15, 20, 25, and 30 events/h. METHODS: Using a general population dataset, the training set included 2,280 subjects, whereas the test set included 470 subjects. The input of this tool was a set of 22 variables. The tool had six neural network models for each AHI threshold. Several metrics were explored to evaluate the performance of the tool: area under the receiver operating characteristic curve (AUC), sensitivity, specificity, positive predictive value, negative predictive value, and 95% confidence interval (CI). RESULTS: The AUC was 0.904, 0.912, 0.913, 0.926, 0.930, and 0.954, respectively, with models of AHI ≥ 5, 10, 15, 20, 25, and 30 events/h thresholds. The sensitivities of all neural network models were higher than 95%. The AHI ≥ 30 events/h model had the maximum sensitivity: 98.31% (95% CI: 95.01%-100%). CONCLUSIONS: The results of this study suggested that the ANN based SDB screening tool can be used to identify the presence or absence of SDB. Future validation should be performed in other populations to determine the practicability of this screening tool in sleep clinics and other at-risk populations.

Recent Design Development in Molecular Imaging for Breast Cancer Detection Using Nanometer CMOS Based Sensors.

As one of the key clinical imaging methods, the computed X-ray tomography can be further improved using new nanometer CMOS sensors. This will enhance the current technique's ability in terms of cancer detection size, position, and detection accuracy on the anatomical structures. The current paper reviewed designs of SOI-based CMOS sensors and their architectural design in mammography systems. Based on the existing experimental results, using the SOI technology can provide a low-noise (SNR around 87.8 db) and high-gain (30 v/v) CMOS imager. It is also expected that, together with the fast data acquisition designs, the new type of imagers may play important roles in the near-future high-dimensional images in additional to today's 2D imagers.