Deficit irrigation and organic amendments can reduce dietary arsenic risk from rice : introducing machine learning-based prediction models from field data
| Autor: | Parijat Bhattacharya, Sudip Sengupta, Jajati Mandal, Arnab Pari, Kallol Bhattacharyya, Sanjay Halder |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2021 |
| Předmět: |
0106 biological sciences
Field experiment Deficit irrigation Amendment engineering.material Machine learning computer.software_genre 010603 evolutionary biology 01 natural sciences Irrigation management Mathematics Ecology business.industry Dietary Arsenic food and beverages 04 agricultural and veterinary sciences Manure 040103 agronomy & agriculture engineering 0401 agriculture forestry and fisheries Animal Science and Zoology Artificial intelligence business Agronomy and Crop Science Vermicompost computer Predictive modelling |
| ISSN: | 0167-8809 |
| Popis: | Dietary rice consumption can assume a significant pathway of the carcinogenic arsenic (As) in the human system. In search of a viable mitigation strategy, a field experiment was conducted with rice (cv. IET-4786) at geogenically arsenic-contaminated areas (West Bengal, India) for two consecutive years. The research aimed to explore irrigation management (saturation and alternate wetting and drying), and organic amendments (vermicompost, farmyard manure, and mustard cake) efficiencies in reducing As load in the whole soil-plant system. A thrice replicated strip plot design was employed and As content in the soil, plant parts, and the associated soil physicochemical properties were determined through a standard protocol. Results revealed that the most negligible As accumulation in the edible grains was accomplished by vermicompost amendment along with alternate wetting and drying (0.318 mg kg−1) over farmer’s practice of continuous submergence with no manure situation (0.895 mg kg−1). Interestingly, an increase in the grain yield by 25% was also observed. The risk of dietary exposure to As through rice was assessed by target cancer risk (TCR) and severity adjusted margin of exposure (SAMOE) mediated risk thermometer. The adopted strategy made all the risk factors somewhat benign to ensure a better standard of health. The Machine Learning algorithm revealed that Random Forest performed better in predicting grain As concentration than k-Nearest Neighbour and Generalized Regression Model. Hence, if properly calibrated and validated, the former can represent an effective tool for predicting grain As concentration in rice. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect