چكيده فارسي :
در چند سال اخير و پس از شروع همه‌گيري كرونا، چالش اصلي بين ساكنين كره‌ي خاكي، بيماري كوويد-19 بوده است. كرونا ويروس‌ها براي اولين بار در سال 1965 كشف شدند و تا كنون بيش از هفت نوع كرونا ويروس انساني يافت شده است. به‌منظور بررسي داده‌هاي مذكور، ابتدا پيش‌پردازش روي داده‌ها شامل حذف داده‌هاي غيرعددي و داده‌هايي كه ويژگي مطلوبي براي ارزيابي ويروس كوويد-19 نبودند، اجرا شد. در ادامه براي يكسان‌سازي داده‌هاي هدف و داده‌هاي مستقل، كل مجموعه نرمال‌سازي شد. در نهايت يك شبكه عصبي پيشخور چندلايه با الگوريتم پس‌انتشار خطا (MFNN)، طراحي شد تا پس از آموزش، مقادير هدف مورد نظر را پيش‌بيني كنند. در پايان نيز نتايج حاصل از كار و همچنين پيشنهادات لازم براي انجام بهتر كار ارائه شده است.
چكيده لاتين :
Considering the high importance of examining the data obtained from the patients of the Covid-19 pandemic disease, this research aims to predict the parameters related to this disease (including the percentage of infection and the recovery rate) according to the diet of the patients؛ an approach is proposed based on artificial neural networks. For this purpose, a reliable global data set has been extracted from Kaggle, which includes 4 general groups, including the amount of fat in the diet, the amount of energy in kilocalories, the amount of protein in the diet and the amount of food consumed in kilograms. It is to be noted that diet information has been collected from 170 countries from different parts of the world. In order to check the mentioned data, firstly, pre-processing has been implemented on the data with the aim of removing non-numerical data and the data that were not a desirable feature for the evaluation of the Covid-19 virus. Next, to make the target data and the independent data equal, the whole set is normalized. Finally, a multilayer feedforward neural network with error backpropagation algorithm (MFNN) is designed to predict the desired target values after training. Different configurations were designed for the neural network, and the results showed that the neural network with an input layer, a hidden layer, and an output layer along with the Lunberg-Marquardt algorithm achieves the best accuracy. Finally, the accuracy of the algorithm was checked with different parameters, and in general, the results showed that the proposed algorithm can predict the percentage of recovery, the percentage of infection and the rate of death in all diets, with an accuracy of more than 99%.
