Evidence is rapidly accumulating that multifactorial nocturnal monitoring, through the coupling of wearable devices and deep learning, may be disruptive for early diagnosis and assessment of sleep disorders. In this work, optical, differential air-pressure and acceleration signals, acquired by a chest-worn sensor, are elaborated into five somnographic-like signals, which are then used to feed a deep network. This addresses a three-fold classification problem to predict the overall signal quality (normal, corrupted), three breathing-related patterns (normal, apnea, irregular) and three sleep-related patterns (normal, snoring, noise). In order to promote explainability, the developed architecture generates additional information in the form of qualitative (saliency maps) and quantitative (confidence indices) data, which helps to improve the interpretation of the predictions. Twenty healthy subjects enrolled in this study were monitored overnight for approximately ten hours during sleep. Somnographic-like signals were manually labeled according to the three class sets to build the training dataset. Both record- and subject-wise analyses were performed to evaluate the prediction performance and the coherence of the results. The network was accurate (0.96) in distinguishing normal from corrupted signals. Breathing patterns were predicted with higher accuracy (0.93) than sleep patterns (0.76). The prediction of irregular breathing was less accurate (0.88) than that of apnea (0.97). In the sleep pattern set, the distinction between snoring (0.73) and noise events (0.61) was less effective. The confidence index associated with the prediction allowed us to elucidate ambiguous predictions better. The saliency map analysis provided useful insights to relate predictions to the input signal content. While preliminary, this work supported the recent perspective on the use of deep learning to detect particular sleep events in multiple somnographic signals, thus representing a step towards bringing the use of AI-based tools for sleep disorder detection incrementally closer to clinical translation.

Identification of characteristic points in physiological signals, such as the peak of the R wave in the electrocardiogram and the peak of the systolic wave of the photopletismogram, is a fundamental step for the quantification of clinical parameters, such as the pulse transit time. In this work, we presented a novel neural architecture, called eMTUnet, to automate point identification in multivariate signals acquired with a chest-worn device. The eMTUnet consists of a single deep network capable of performing three tasks simultaneously: (i) localization in time of characteristic points (labeling task), (ii) evaluation of the quality of signals (classification task); (iii) estimation of the reliability of classification (reliability task). Preliminary results in overnight monitoring showcased the ability to detect characteristic points in the four signals with a recall index of about 1.00, 0.90, 0.90, and 0.80, respectively. The accuracy of the signal quality classification was about 0.90, on average over four different classes. The average confidence of the correctly classified signals, against the misclassifications, was 0.93 vs. 0.52, proving the worthiness of the confidence index, which may better qualify the point identification. From the achieved outcomes, we point out that high-quality segmentation and classification are both ensured, which brings the use of a multi-modal framework, composed of wearable sensors and artificial intelligence, incrementally closer to clinical translation.

The interplay between slow-wave sleep and cardiovascular health is increasingly recognized. Our prior research showed that auditory-enhanced slow waves can boost cardiac function, yet the mechanisms behind this remain unclear. Advancing these findings, our current analysis dissected the effects of two slow wave types on cardiovascular function, using data from 18 middle-aged men across three nights. We found that the strength of heart rate and blood pressure responses concurrent with slow waves predicts cardiac function post-sleep. Notably, we identified that highly synchronized type I slow waves, as opposed to lower-amplitude type II slow waves, primarily co-occur with these cardiovascular pulsations. While auditory stimulation enhances both types of slow waves, they exhibit distinct temporal dynamics, pointing to different underlying biological mechanisms. This study crucially addresses how distinct slow wave types can affect cardiovascular function, implying that targeted slow wave stimulation could be a strategic approach to improve heart health.

The importance of sleep as a key component of cardiovascular health has been increasingly recognized. Our previous research demonstrated that auditory‐enhanced slow waves increase cardiac function, but the underlying mechanisms behind these beneficial effects remain uncertain. In this study, we examined the influence of two types of slow waves on cardiovascular functions. 18 healthy middle‐aged men participated in a three‐night sleep lab experiment under three conditions: two auditory slow wave enhancement conditions and one SHAM control. We found that the strength of heart rate and blood pressure responses occurring concurrently with slow waves predicts post‐sleep cardiac function in men as measured through transthoracic echocardiography. Furthermore, we observed that highly synchronized type I slow waves, compared to lower‐amplitude type II slow waves, are primarily associated with increased cardiovascular responses (transient increase in heart rate and blood pressure). While auditory stimulation enhances both types of slow waves, they display distinct temporal dynamics, suggesting different underlying brain and body microstates. These distinct slow wave types may affect cardiovascular function differently, indicating that targeted slow wave stimulation could be a strategic approach to enhancing cardiac function. Trial Registration: This study was registered under clinicaltrials.gov (NCT04166916)

Background and objectives: Bone is a multicellular organ that is the site of complex pathophysiological events (such as cancer). 3D confocal and multiphoton microscopy, followed by manual analysis and quantification, allow the identification of molecular, cellular, and tissue mechanisms in their original context, enabling the investigation of spatial biology at subcellular level. The automation of these analyses is becoming increasingly important due to their time-consuming nature, lack of standardization, and inter-subject variability. Methods: Deep learning applied to image analysis could overcome current limitations and optimize preclinical research. This includes automatic semantic segmentation of bone cells (osteoblasts, osteoclasts, and blood vessels) and mineral component, followed by parameter extraction and quantification. Accordingly, fluorescence microscopy images were generated, pre-processed, and fed (total 21,395 images: 17,104 for training, 4291 for validation) into a neural network-based architecture named Bo-Net. Results: The fully automatic Bo-Net achieved strong performance across the tested metrics and segmentation accuracy similar to experienced biologists (R = 0.81–0.99 for parameters tested) and improved analysis time from days to seconds in a variety of experimental conditions, confirming its relevance in different biologically relevant contexts. Conclusions: As a result, a fully automated, yet reliable, tool to optimize the analysis of multiparametric fluorescence microscopy images available to the bone research community was generated.