Advances in modern technologies such as telecommunication have widely expanded the applications of wireless systems. Therefore,humans are continuously exposed to electromagnetic fields (EMFs)produced by widely used devices such as mobile and cordless phonesand Wi-Fi routers. According to the World Health Organization, electromagnetic hypersensitivity (EHS) is the medical term for a variety of nonspecific symptoms that afflicted subjects attribute to exposure to different sources of EMFs. About 25% of the general population reports different levels of environmental intolerance to factors such as EMFs, and studies performed in Europe show that about 75% of general practitioners had visited patients complaining
of EHS. In this paper, multilayer perceptron neural network (MLPNN)–based models are proposed to predict the subjective health symptoms in inhabitants living in the vicinity of mobile phone base stations. The classifier uses several parameters such as demographic data, environmental exposure to a mobile phone station, and thehealth conditions of an individual as input to estimate subjective health symptoms. Out of 699 data sets recorded from 363 men and 336 women via questionnaire, 70% were used for training, 15% forvalidation, and the remaining 15% for testing the developed system.
The performance of the developed system (sensitivity and specificity) in predicting the subjective health symptoms is as follows: headache (72%, 91%), fatigue (8%, 98%), sleep disturbance (97%, 93%), dizziness (65%, 85%), vertigo (65%, 84%). These promising results suggest that this system might be useful as a means for predicting the health symptoms in people living in the vicinity of mobile phone base stations, which ultimately enhances the quality
of life of these individuals through providing appropriate medical care and introducing effective methods for reducing the effect of these exposures.
H. Parsaei,1 M. Faraz,1 and S. M. J. Mortazavi 2,3
1Medical Physics and Medical Engineering Department, School
of Medicine, Shiraz University of Medical Sciences, Shiraz, Iran.
2Department of Diagnostic Imaging, Fox Chase Cancer Center,
Philadelphia, Pennsylvania, USA.
3Ionizing and Non-ionizing Radiation Protection Research Center
(INIRPRC), Shiraz University of Medical Sciences, Shiraz, Iran.
Excerpts from the pdf: Link: eco.2017.0011
It has been reported that about 25% of the general population reports different levels of environmental intolerance to factors such as EMFs(Nordin & Nordin, 2016). Furthermore, studies performed in Europe show that about 75% of general practitioners had visited patients complaining of EHS (Slottje et al., 2016). Although the underlying mechanisms are not fully understood, now we know that the health symptoms linked to exposure to EMFs are real and cause functional impairment.
Regarding the challenging issue of EHS, we have previously shown that when the self-reported hypersensitive participants were asked to report their perception about the real and sham exposures, only 25% could discriminate the real exposure/sham exposure phases(this simply could be due to chance). Furthermore, when all these
hypersensitive participants were connected to intensive care unitmonitors and the alterations in their heart rate, respiration, and blood pressure during real and sham exposure phases were recorded,no statistically significant changes between the means of these parameters were detected in real/sham exposures. At that time
(this dates back to 2011), we concluded that psychological factors are possibly involved in EHS (S. M. J. Mortazavi et al., 2011). It is worth noting that this conclusion was flawed due to the limitations
we had in our previous studies, and when we obtained sufficient
data, we realized that EHS was not linked to psychological issues.
Later, we introduced a novel multiphase method for effective screening of the patients diagnosed with EHS (Khademi et al., 2014; S. A. Mortazavi et al., 2014).
In this paper, a system based on a multilayer perceptron neural network (MLPNN) is presented for predicting
subjective health symptoms in people living near mobile phone base stations. The characteristics of this method, its objectives, and how it was developed and evaluated are presented in detail in this paper.
The presented method is to predict the subjective health symptoms in individuals living near mobile base stations. More specifically, the main objective was to determine if these individuals may have health symptoms such as headache, fatigue, sleep disturbance, discomfort depression, loss of memory, dizziness, libido decrease, nervousness,
Table 1. Performance of the Developed MLPNN-Based
System in Predicting Subjective Health Symptoms
for People Living Near Mobile Phone Base Stations
SYMPTOM SENSITIVITY SPECIFICITY ACCURACY
(%) (%) (%)
Headache 71.8 90.9 83.8
Sleep disturbance 82.1 83.3 82.9
Dizziness 65.2 85.4 81.0
Vertigo 65.0 84.7 81.0
Fatigue 8.3 98.9 88.6
Results suggest that people who ‘‘overuse’’ their cell phones preferably should not live near a base station. For the sleep disturbance symptom, the three most effective features were ‘‘daily mobilephone usage,’’ ‘‘duration of exposure to the antenna,’’ and ‘‘cordless phone use.’’ As we can see again, overusing cell phones and cordless phones along with living near base stations may cause a sleep disturbance symptom. The same results are obtained for the dizziness
In terms of this application, this model may help physicians and scientists reduce the health risks of EMFs via predicting the subjective health symptomsf or people currently living or who would like to move to houses in the
vicinity of mobile phone base stations. In other words, this MLPNNbased system can be used to investigate if a person has EHS or not and ultimately can help us predict the health risks of living in the vicinity of mobile base stations.
Accurate prediction of the risk of subjective health symptoms in inhabitants living in the vicinity of mobile phone base stations canenhance the quality of their life through providing appropriate health care and suggesting effective methods for reducing the severity of these symptoms. In this paper, we proposed an MLPNN-based model
for predicting the risk of several symptoms such as headache, fatigue,sleep disturbance, discomfort depression, loss of memory, dizziness,libido decrease, nervousness, and palpitations. Evaluation of the data collected in this survey that was conducted on 699 people living in the vicinity of cellular phone base stations, in Shiraz, Fars, Iran,
reveals that the developed system can successfully predict the risk of subjective health symptoms (for most symptoms) with sensitivities > 65% and specificities > 83%. We hope that the robustness and accuracy of the developed system will help scientists promote the applications of an MLPNN and pattern recognition techniques in
improving the health of individuals living in the vicinity of mobile phone base stations.
Address correspondence to:
S. M. J. Mortazavi
Fox Chase Cancer Center
Doss Lab, R-432
333 Cottman Avenue
Philadelphia, PA 19111
Received: March 19, 2017
Accepted: April 20, 2017