A hybrid computational intelligence approach for bioremediation of amoxicillin based on fungus activities from soil resources and aflatoxin B1 controls.

ABSTRACT

Nowadays, releasing the Emerging Pollutants (EPs) in the nature is one of the main reasons for many health and environmental disasters. Amoxicillin as an antibiotic is one of the EPs and categorized as the Endocrine Disrupting Compounds (EDCs) in hazardous materials. Accumulation of amoxicillin in the soil bulk increases the cancer risk, drug resistances and other epidemiological diseases. Hence, the soil bioremediation of antibiotics can be a solution for this problem which is more environmental-friendly system. This study technically creates a bio-engine setup in soil bulk for remediation of amoxicillin based on Aspergillus Flavus (AF) activities and Removal Percentage (RP) of amoxicillin with Aflatoxin B1 Generation (AG) controls. The main novelty is to propose a hybrid computational intelligence approach to do optimization for mechanical and biological aspects and to predict the behavior of bio-engine's effective mechanical and biological features in an intelligent way. The optimization model is formulated by the Central Composite Design (CCD) which is set by the Response Surface Methodology (RSM). The prediction model is formulated by the Random Forest (RF), Adaptive Neuro Fuzzy Inference System (ANFIS) and Random Tree (RT) algorithms. According to the experimental practices from real soil samples in different times and places, concentration of amoxicillin and Aflatoxin B1 are set equal to 25 mg/L (ppm) and 15 µg/L (ppb). Likewise, the outcomes of experiments in CCD-RSM computations are evaluated by curve fitting comparisons between linear, 2FI, quadratic and cubic polynomial equations with considering to regression coefficient and predicted regression coefficient values, ANOVA and optimization by sequential differentiation. Based on the results of CCD-RSM, the RP performance in the optimum conditions is measured around 86% and in 25 days after runtime, the RP and AG are balanced in the safe mode. The proposed hybrid model achieves the 0.99 accuracy. The applicability of the research is done using real field evaluations from drug industrial park in Mashhad city in Iran. Finally, a broad analysis is done and managerial insights are concluded. The main findings of the present research are (I) with application of bioremediation from fungus activities, amoxicillin amounts can be control in soil resources with minimum AG, (II) ANFIS model has the best accuracy for smart monitoring of amoxicillin bioremediation in soil environments and (III) based on the statistical assessments Aeration Intensity and AF/Biological Waste ratio are most effective on the amoxicillin removal percentage.