Automated composting assisted by machine learning: Substrate optimization, rapid stabilization, and enhanced bioactivity

Composting is a sustainable strategy for recycling organic waste, improving soil fertility, and suppressing plant pathogens. This study focuses on bio-oxidation phase of the compost prepared with an automated machine using a substrate composition defined by multiple linear regression (MLR) from previous data: 49% of olive mill wastewater (OMW), 19.5% of exhausted olive cake (EOC), 15.5% of poultry manure (PM), and 16% green waste (GW). After 24 h, the compost showed 55% organic matter (OM) and a C/N ratio of 33. The phytotoxicity of compost tea was evaluated through germination tests on cress, tomato, fenugreek, sorghum, and maize. Cress achieved the highest germination index (205%), while tomato reached 66%. The best germination was observed at 25% compost tea concentration for all seeds. Antifungal assays against Fusarium oxysporum at 20% concentration revealed strong inhibition when compost tea was vacuum-filtered. Furthermore, predictive modeling of C/N ratio evolution was performed to estimate the time required to reach compost maturity (C/N = 15). The LITE-RNN model showed the highest accuracy (R² = 0.96), predicting a maturation time of 44 h. These findings demonstrate the potential of combining automated composting with artificial intelligence to accelerate compost production, enhance biological activity, and improve disease suppression.