Research

The representation of complex systems as networks whose units are functionally connected is ubiquitous in many fields of science, ranging from climatology to econometrics and the study of brain and physiological systems. At BIT lab we develop tools rooted in the general framework of information theory to quantify how information is generated at each node of a complex network, transferred between pairs of nodes, and shared in redundant and synergistic ways among several nodes. Applied to both discrete and continuous time processes studied in the time or frequency domains, we use these tools to describe how network function emerges from network structure and changes across diverse states and conditions.

Based on the view of the human body as an integrated network composed by several organ systems, each with its own internal dynamics but also functionally connected to the other organs, we apply novel methods for multivariate time series analysis to the nonlinear, multi-scale and time-variant output signals of brain, heart and peripheral physiological systems. This integrated unconventional approach aims at providing new insight on the functional structure of the human physiological networks and on its evolution across different physiological states and pathological conditions.

The brain and the heart are complex systems, whose interaction shows emergent properties that cannot be highlighted by the analysis of each system independently. While clinical evidence points to the functional interplay between brain and body, a solid comprehensive data modelling/processing framework linking the very different physiological dynamics is still missing. We use electroencephalographic (EEG) and electrocardiographic (ECG) data to measure brain and heart activities and we develop and apply data-driven methodologies to identify brain-heart interactions in several different physiological conditions (e.g. enteroception and sleep stages).

Brain connectivity refers to the intricate network of connections that exist within the human brain. It is a fundamental aspect of neuroscience and is crucial for understanding how the brain processes information, performs various functions, and gives rise to cognition and behavior. Among the several definitions of brain connectivity, the concept of functional connectivity is the most known and spread in the literature. It is based on the idea that regions of the brain that are active together, even though anatomically, are likely to be functionally connected. We use functional magnetic resonance imaging (fMRI) and electroencephalography (EEG) data to measure brain activity and identify patterns of synchronous neural activity. 

In recent years, the application of machine learning (ML) algorithms to the analysis of medical data has increased significantly in healthcare research and industry. ML algorithms have been used to help classify physiological and pathological states related to, for example, different types of stress. The use of time series in data analysis allows the temporal dynamics of pathophysiological states to be captured, enabling machine learning models to identify hidden correlations and complex patterns. These algorithms can thus provide important decision support tools for the early diagnosis, monitoring, and treatment of stress-related disorders, enabling more effective and personalised management of patient well-being.