Akademik Veri Yönetim Sistemi
Modeling Earth Systems and Environment, cilt.9, sa.2, ss.2745-2756, 2023 (ESCI)
© 2022, The Author(s), under exclusive licence to Springer Nature Switzerland AG.Aspergillus flavus and Aspergillus parasiticus are among the few aspergilli that produce aflatoxins and cause aflatoxin-related toxicity. To protect against these dangers, it is crucial to produce plant-based solutions that not only protect the balance of the ecosystem but also the health of living beings using natural methods. In this study, the effects of different doses of A. aestivus, B. vulgaris, and M. alba leaf extracts on the development and reproduction of A. flavus and A. parasiticus fungi were investigated and an experimental study was conducted. The inhibition rates of A. aestivus, B. vulgaris, and M. alba extracts were calculated by applying them to petri dishes containing A. flavus and A. parasiticus fungi at concentrations of 1, 5, 10, 15, 20, 25, and 30 mg/ml. As a result of the research, the inhibition rates of A. aestivus, B. vulgaris, and M. alba extracts were found to be between 3.13–89.94%, 2.63–76.3%, and 2.29–73.9%, respectively. In terms of the lethal dose, the plant with the highest amount of development-inhibiting active ingredient per unit was A. aestivus with a concentration of 22 mg/ml and an inhibition rate of 50.8%. B. vulgaris and M. alba also showed similar effects with inhibition rates of 50.33% and 50.14% at concentrations of 26 mg/ml, respectively. An artificial intelligence model has been developed to estimate the antifungal effect on doses that have not been measured. In the development of the model, training was performed using the decision tree regressor (DT), extra trees regressor (ET), random forest regressor (RF), gradient boosting regressor (GBR), light gradient boosting machine (LIGHTGBM), K-nearest neighbors regressor (K-NN), AdaBoost regressor (ADA), ridge regression (RIDGE), least angle regression (LAR), and Bayesian ridge (BR) algorithms. The successful algorithm was evaluated according to the performance criteria of mean square error (MSE), mean absolute error (MAE), root mean square error (RMSE), and coefficient of determination R2, and the K-NN model with an R2 value of 0.96 was found to be the most successful. The K-NN model was used to estimate the inhibition percentage values of the unmeasured leaf extract doses with the least error in the 1–30 mg/ml dose range. With this expert model, the development of organic fungicides with varying dose and leaf types can be performed much faster.