Abstract
This paper presents an unsupervised methodology to analyze SeismoCardioGram (SCG) signals. Starting from raw accelerometric data, heartbeat complexes are extracted and annotated, using a two-step procedure. An unsupervised calibration procedure is added to better adapt to different user patterns. Results show that the performance scores achieved by the proposed methodology improve over related literature: on average, 98.5% sensitivity and 98.6% precision are achieved in beat detection, whereas RMS (Root Mean Square) error in heartbeat interval estimation is as low as 4.6 ms. This allows SCG heartbeat complexes to be reliably extracted. Then, the morphological information of such waveforms is further processed by means of a modular Convolutional Variational AutoEncoder network, aiming at extracting compressed, meaningful representation. After unsupervised training, the VAE network is able to recognize different signal morphologies, associating each user to its specific patterns with high accuracy, as indicated by specific performance metrics (including adjusted random and mutual information score, completeness, and homogeneity). Finally, a Linear Model is used to interpret the results of clustering in the learned latent space, highlighting the impact of different VAE architectural parameters (i.e., number of stacked convolutional units and dimension of latent space).
Highlights
Improvements in sensing, networking, and computing technology opened up new scenarios within the Active and Healthy Ageing (AHA) realm: in particular, new technologies foster more effective monitoring, allowing for more extensive and continuous data collection and for higher dimensionality, with new kinds of data entering the monitoring scenario
An active assisted living (AAL) system can help in making services more accessible, by compensating physical or sensory impairments with new smart devices: voice control or even Brain–Computer Interfaces have been integrated within AAL ecosystems to allow severely motor-impaired users to achieve communication and home control [2,3,4]
A missed beat between two R peaks is classified as False Negative (FN); a beat outside the specified tolerance window and between two R peaks is considered as a False Positive (FP), whereas a True Negative (TN)
Summary
Improvements in sensing, networking, and computing technology opened up new scenarios within the Active and Healthy Ageing (AHA) realm: in particular, new technologies foster more effective monitoring, allowing for more extensive and continuous data collection and for higher dimensionality, with new kinds of data entering the monitoring scenario. AAL systems can be oriented to prevention and monitoring; in particular, the daily living patterns and activities, collected by a network of smart home sensors [5], can be analyzed by means of artificial intelligence techniques [6,7,8]: models of users’ habits can be derived, useful for detecting significant behavioral changes through time. Combining such behavioral monitoring with continuous monitoring of vital signs may provide relevant insights [9], with vital signs being contextualized and correlated to activities being carried out. Wrist-worn devices acquiring Heart Rate (HR) information through PhotoPlethysmoGraphy (PPG) are already a commercial reality, whereas clinical-grade heart
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have