INESC TEC research highlights need to improve synchronisation of physiological data from different devices

In biomedical research and clinical monitoring, teams increasingly rely on the simultaneous use of multiple wearable and physiological sensors, which make it possible to collect different types of data using various devices – and this is where the real challenge begins. To ensure a reliable analysis of the information, the data streams need to be properly synchronised. An INESC TEC study found that 58% of multi-device systems reported in the literature have never assessed synchronisation latency – a gap that R&D can help close.

EEG, ECG and EMG (or electroencephalogram, electrocardiogram and electromyogram) are types of exams that collect physiological signals, namely brain, heart and muscle activity. In biomedical research and clinical monitoring, this data collection is often carried out simultaneously, thanks to the use of multiple wearable and physiological sensors. “Obtaining data from multiple devices is only half the challenge. Ensuring that these data streams are properly synchronised is essential for reliable analysis, and this is a challenge that is often overlooked,” said João Paulo Cunha, adding that “synchronising multimodal physiological data streams is a critical and increasingly relevant challenge in biomedical engineering.”

The INESC TEC researcher, who is also a lecturer at the Faculty of Engineering of the University of Porto (FEUP), explained that one of his PhD students carried out a systematic review covering five databases and 1,176 publications, from which 60 articles were selected for in-depth analysis. “We mapped out the landscape of synchronisation techniques, devices and physiological modalities, and assessed the technology readiness level of each approach,” said Francisco Vieira, PhD student at FEUP and Carnegie Mellon University in Pittsburgh (U.S.A.) whose work focuses on developing a solution to improve data acquisition synchronisation in multimodal analyses.

The researchers focused on several aspects, including the types of physiological data streams, the devices used to collect data, the methods used to measure temporal alignment latency, the synchronisation techniques used by the authors of the publications, and the technology readiness level of each technique.

The team found that 58% of studies using multiple devices never assessed synchronisation latency, which is a key performance indicator for comparing and validating different approaches. “This gap has real consequences for reproducibility and data quality across the field of physiological monitoring,” mentioned João Paulo Cunha.

These findings –  available in the paper Synchronization of Multimodal Physiological Data Streams: State-of-the-Art, Trends, and Future Challenges, published by IEEE Access – reinforce the growing importance of robust synchronisation in multimodal physiology; i.e., the simultaneous analysis of different physiological signals to get a more complete understanding of how the human body functions, and point to future directions for R&D in this area.

The study demonstrates “the urgency” of further research and development of synchronisation techniques in these contexts, as well as the need to improve current synchronisation methods. Moreover, the researchers also emphasised the importance of thorough latency verification to optimise data acquisition, analysis and the overall quality of research into multimodal physiological data streams.

“This paper is the result of work carried out by our research and development team, namely PhD student Francisco Vieira. We hope it will become a useful reference for researchers and engineers working at the intersection of biomedical engineering, wearable sensors and data acquisition,” he concluded.