Exploratory: Sports Data Science

With the help of wearable devices that are able to record huge quantities of physiological data 24 hours a day, 7 days a week, we can obtain a clear view of an individual’s health status and behaviour. The data provided in this dataset were collected and provided by BioBeats (https://biobeats.com) in collaboration with researchers from the University of Pisa. BioBeats operates in the health science industry that produces IoT wearable devices aiming to detect people’s psychophysiological stress. The data were recorded by sport and health scientists, psychologists and chemists with the objective of assessing psychophysiological response to stress stimuli and sleep.

The Multilevel Monitoring of Activity and Sleep in Healthy people (MMASH) dataset provides 24 hours of continuous psycho-physiological data, i.e., beats-to-beats heart data, triaxial accelerometer data, sleep quality, physical activity, psychological characteristics (e.g., anxiety status, stress events and emotions) and sleep hormone levels for 22 University of Pisa students. MMASH dataset is released under the Open Database License (ODbL) v1.0 and is publicly available on Physionet (https://bit.ly/MMASH_data).

MMASH is the first dataset providing several aspects of people's everyday life such as cardiovascular responses, psychological perceptions (e.g., stress, anxiety, and emotions), sleep quality, movement information (e.g., wrist accelerometer data and steps) and hourly activity descriptions. Due to the complexity of this data, experts from several research fields could use this dataset to investigate the relationship between several aspects of psychophysiological responses having a complete overview of the users' daily life. For example, it is possible to investigate the relationship between perceived (PSQI) and observed sleep quality (e.g., melatonin, cortisol, sleep fragmentation index and sleep length) by individual characteristics such as daily stress, anxiety status, emotion perceived throughout the previous day and daily activities.

Moreover, machine learning algorithms could be developed to detect daily activities, moods, emotions, individual predisposition to react toward aversive or positive events and stress following cardiovascular responses (e.g., heart rate and heart rate variability) and/or actigraphy data. These algorithms could be used to predict people's routine by using accelerometers data and cardiovascular responses that are nowadays continuously recorded by wrist-worn devices that have become more and more popular thanks to the technological advent of the last two decades. These are only a few examples of all the possible research topics that could be rise by using this dataset.

The main reason to release MMASH is the difficulty to record this kind of data for a long period. This dataset would give researchers and companies the chance to have a ground truth of several psychophysiological responses to develop predictive models and thus passively assess people's everyday life following wrist-worn devices estimating their well-being.

Written by: Alessio Rossi
Revised by: Luca Pappalardo