Modeling Electromagnetic Field Radiation Variability for Minimization of Exposure Rate in Public Health Environments: A Machine Learning Approach (Published)
Public Health awareness of the exposure to radiofrequency electromagnetic fields (Rf-EMF) is necessary for epidemiological studies on possible bio-effects. The Knowledge will promote good health, reduce health risk factors, lower morbidity and mortality rates often associated with public health environment (PHE). Modeling EMF radiation variability and its application is vital given the prospective growth in both the number of mobile devices and equipment radiating electromagnetic fields (EMF) and the increasing concerns in the general public. The main goal of this work is to develop a framework based on Machine Learning (ML) approach to determine the exposure level on time-based variances within and during operational procedures of frequency bands used in mobile telecommunication, digital devices, transmitters, broadcasting terminals etc. The system adopts the measurements and instrumentation methodology, using a handheld Poniie pn8000 multi-field EMF metre and EME Spy exposimeter with a detection limit of 0.0066mW/m (2) with sampling at certain interval in a specific frequency ensemble. The measurements were synthesized as exposure coefficients with assessment based on EMF radiation from emitting devices and matching with International Commission on Non-Ionizing Radiation Protection (ICNIRP) standards endorsed by WHO. Consequently, obtained data was processed and segmented into training and test sets on 70:30 ratios respectively. Computing modeling of the system was performed with R programming language using data obtained from field measurements and records. The prosed system used Ensemble ML-Random Forest Algorithm and Decision Tree algorithm (DTA). Performance evaluation and results obtained clearly demonstrates 87% based on the Confusion Matrix (CM) outcomes for Random Forest and Decision Tree presents 56%. This justifies the capability of ML techniques in accurately classifying EMF radiation to mitigate exposure rate and improve healthcare throughput. The model was transformed and deployed in a production system API environment with Graphical User Interface (GUI) for flexibility. This work revealed relating potential health bio-effects associated with EMF exposure and the different parameters that are currently used for evaluating, limiting, and mitigating the impact of the exposure on the general public.
Keywords: EME Spy Exposimeter, Ensemble ML-Random Forest Algorithm and Decision Tree algorithm (DTA), Multi-Field EMF Metre, Non-Ionizing Radiation Protection (ICNIRP), Public Health Environment (PHE), Radiofrequency Electromagnetic Fields (Rf-EMF)