Unsupervised Machine Learning and EMF radiation in schools: a study of 205 schools in Greece

TL;DR

This study uses unsupervised machine learning, specifically hierarchical clustering, to analyze RF-EMF exposure in Greek schools, revealing that most schools have exposure levels significantly below safety limits, regardless of urban population density.

Contribution

The paper introduces the application of hierarchical clustering to RF-EMF exposure data in schools, providing new insights into exposure patterns across a large region.

Findings

97.5% of schools have RF-EMF levels at least 3500 times below safety limits

Exposure levels are not related to urban population density

Most schools' RF-EMF exposure is well below health safety thresholds

Abstract

The expansion of network infrastructure in Greece has raised concerns about the possible negative health effects on sensitive groups, such as children, from exposure to long-term radiofrequency electromagnetic fields (RF-EMFs). The objective of this study is to apply Unsupervised Machine Learning methods such as hierarchical clustering, in order to establish patterns of EMF radiation in schools. To this end we performed measurements in the majority schools units in the region of Thessaly in order to calculate the mean value for RF - EMF exposure in the 27 MHz - 3 GHz range, which covers the whole spectrum of RF - EMF sources. Hierarchical clustering dendrogram analysis shows that population density in urban areas of Thessaly bears no relation to the level of EMF exposure in schools. Furthermore, in $97.5\%$ of schools found in the Thessaly region, the exposure level is at least 3500 times below the Greek exposure limits while in $2.5\%$ it is a little less than 500 times below the limit.