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.

Cuffless estimation of arterial blood pressure (ABP) is an ongoing topic of research and development that may revolutionize home monitoring. Into this path, innovative artificial intelligence (AI) tools, especially deep neural networks based on end-to-end computation, have gained much attention as they can leverage the bundle of signals acquired by integrated wearable devices to estimate directly the ABP, avoiding the assessment of intermediate features. In this work, we performed a feasibility analysis testing different neural architectures to process in bundle ECG and PPG signals to estimate continuously the BP. Data were collected from an already processed version of the MIMIC-II dataset from physionet.org. The reconstructed ABP was only partially accurate (mean absolute error in the range of 10 mmHg) due to the questionable quality of the data, despite extensive noise and outliers removal. This poses questions about the role of end-to-end approaches that, while saving effort in feature-engineered detection, appears to be very sensitive to the input data quality.

No abstract