The aim of this study is to obtain scientifically valid findings from wearable devices. Our methodology involves determining cause-and-effect linkages between the variables being examined, rather than just identifying relationships based on correlation. This entails determining the causal impacts between the exposure and result variables to address causal inquiries. Specifically, we aim to clarify how the diagnostic category (e.g., neurodegenerative disease) affects the accelerometer data collected from wearable devices.
The research involves analysing ambient temperature data from a data file produced by continuous wear of the accelerometer-based activity monitor wrist-worn technology, comparing it with the UK’s National Meteorological Service (MET Office) weather data. This comparison aims to determine whether the data can show if a data point at a specified time was taken while the user was indoors or outdoors.
The study will also explore the correlation between wrist-worn ambient temperature readings and official weather data. This will enhance the understanding of how wearable technology can be used in conjunction with environmental data to provide insights into the wearer’s location, this is crucial for understanding user behaviour and health patterns. Additionally, by analysing accelerometer data alongside temperature data, we can extract significant features to develop robust models with new contexts, achieving more accurate interpretations and reliable predictions from these ML models.
This information can be valuable for various applications, including the development of advanced ML algorithms to better understand the user and their environment. This project not only enhances data analysis and machine learning skills but also has practical implications in healthcare and other fields. It will leverage ML algorithms to classify the data and develop software to automate this process, ensuring accuracy and efficiency.
This study aims to fill the gaps in current literature regarding the combination of wearable technology data (specifically accelerometers and temperature) with official meteorological data. By doing so, it will enhance our understanding of how these technologies can be utilized to accurately determine location context. This information will then be used in a machine learning model to generate more reliable predictions about diseases. The results are anticipated to make valuable contributions not just to scholarly research but also to practical implementations in healthcare and environmental monitoring.
