We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Machine learning–driven parallel XGBoost modeling of astaxanthin production by Haematococcus pluvialis in wastewater contaminated with antibiotics and microplastics
Original title: Machine learning–driven parallel XGBoost modeling of astaxanthin production by Haematococcus pluvialis in wastewater contaminated with antibiotics and microplastics
Summary
Scientists found that a microalga can actually make *more* of a valuable antioxidant called astaxanthin (used in supplements, food, and cosmetics) when grown in wastewater polluted with microplastics or antibiotics — essentially turning pollution stress into a useful product. This is a clever way to potentially clean up contaminated water while producing something valuable, but since the algae are grown in contact with microplastics and antibiotics, more safety testing is needed before this astaxanthin ends up in food or animal feed.
Microalgae can produce high value-added products in wastewater (WW). Haematococcus pluvialis is a microalga and can synthesize high amounts of astaxanthin (AXT) under suitable conditions. In this study, high AXT was produced from H. pluvialis, cultivated in WW contaminated with antibiotics or microplastics (MPs), using an advanced ensemble learning technique called Extreme Gradient Boosting (XGBoost). The modeling framework achieves efficient parallel processing by executing tasks on independent datasets concurrently and without requiring synchronization. Furthermore, an interpretability framework was applied to ensure transparency and reliability of the XGBoost models, using SHAP, permutation importance, and partial dependence plots. SHAP analysis reveals that ciprofloxacin (CIP) and light intensity (LI) systematically increase predictions for AXT-A. Furthermore, it identifies polystyrene (PS) and polyethylene (PE) as the primary drivers of AXT-M. In addition, the activities of antioxidant enzymes, including CAT, SOD, MDA, and APX were investigated to show that microalgae produce AXT when stressed in these conditions. The highest AXT productivity in the MP-contaminated WW was 7.72 mg/gd at 50 mg/L concentrations for four MPs, 100 rpm SS, and 90 μmol m s LI. Moreover, in antibiotic-contaminated WW, the highest AP was 5.46 mg/gd at 75 μg/L concentrations for four antibiotics, 100 rpm SS, and 90 μmol m s LI. The highest antioxidant enzyme activities were found in M-50, and the CAT, SOD, MDA, and APX activities of M-50 were 113 ± 3 U/mg protein, 138 ± 4 U/mg protein, 4 ± 0.1 μmol/g, and 17.8 ± 1.3 U/mg protein, respectively. In conclusion, H. pluvialis cultured in both WWs produced significant AXT productivities, with higher production observed in microplastic-contaminated WW compared to antibiotic-contaminated WW. This highlights the feasibility of using microplastic-contaminated WW for high-level AXT production for applications in the textile, dye, feed, food, and environmental industries, while further safety evaluation is required for potential use in food and feed sectors.