Machine Learning Application to Optimize Spent Mushroom Compost (SMC) Biochar as Filter Media for Heavy Metal Adsorption in Abandoned Mine Water

Abstract The surging global population strains resources, escalating pollution, and exacerbating water scarcity. Sustainable water management necessitates alternative sources such as abandoned mine water. However, these sources often contain hazardous heavy metals, like lead, copper, iron, and manganese, posing grave health and environmental risks. Conventional methods struggle to effectively treat these heavy metals in abandoned mining ponds, urging the search for cost‐efficient and sustainable solutions. Biochar, particularly from spent mushroom compost (SMC), emerges as a potent adsorbent due to its high surface area and binding groups. Yet, the variability in its efficiency remains a challenge. Conventional empirical models fail to capture the dynamic nature of adsorption processes accurately. Adopting machine learning, specifically an adaptive neuro‐fuzzy inference system (ANFIS), shows potential in predicting adsorption efficiency. This study aims to employ ANFIS to forecast SMC biochar's performance in a lab‐scale metal retention pond, providing design charts for diverse initial metal concentrations and pH levels. Validation through real cases aims to enhance accuracy and establish a framework for future heavy metal adsorption capacities. This research offers a sustainable approach to removing heavy metals from abandoned mining ponds whereas the computational modeling in optimizing SMC biochar introduces a novel approach for practical applications.