Rahmani S, Ansarihadipour H, Bayatiani MR, Khosrowbeygi A, Babaei S, Rasmi Y.
Exposure to Electromagnetic fields (EMF) has been increasingly associated with cellular redox perturbations through the generation of reactive oxygen species, particularly under conditions of extremely low-frequency electromagnetic fields (ELF-EMF), where biological systems exhibit altered susceptibility to oxidative stress. In this investigation, we evaluated β-thalassemia major samples alongside matched controls, focusing on erythrocytes and plasma oxidative status under controlled exposure to 0.5 mT and 1 mT fields at 50 Hz frequency, thereby simulating clinically relevant electromagnetic conditions. Concomitantly, alterations in oxyhemoglobin concentration were quantified using integrated biochemical and computational approaches, while artificial neural networks (ANN) and specifically a multilayer perceptron (MLP) architecture were developed as a feed forward neural network model to predict hemoglobin functionality. Experimental findings demonstrated significant perturbations in optical density of hemoglobin, alongside elevated methemoglobin and hemichrome formation, indicating structural destabilization within the oxygen-carrying system. Furthermore, Native PAGE and SDS-PAGE analyses revealed pronounced protein aggregation, including formation of high molecular weight protein aggregates affecting plasma proteins, consistent with oxidative destabilization patterns. These molecular alterations were corroborated by changes in antioxidant defense system capacity and broader oxidative biomarkers, as assessed through spectrophotometric analysis. Importantly, computational modeling using a regression model demonstrated strong predictive performance with R2=0.942, supported by minimized sum of squares errors (SSR) and reduced relative errors (RE), highlighting robust agreement between observed and predicted outputs. Additionally, exposure-induced hemoglobin structural changes were significantly more pronounced in β-thalassemia patients, suggesting heightened vulnerability compared with healthy donors. Collectively, these findings integrate electromagnetic bioeffects with hematological dysfunction and demonstrate that ELF-EMF exposure modulates redox balance, protein integrity, and hemoglobin stability, while machine learning frameworks such as ANN provide high-accuracy predictive modeling of oxyhemoglobin concentration under both physiological and pathological conditions.
Reference:
