1. Introduction
The coronavirus disease 2019 (COVID-19) has created significant threats to the public health in the world [1]. Despite the remarkable progress in fighting diseases, infectious diseases still have particular importance in epidemiology and public health [2]. The COVID-19 is one of the contagious and infectious diseases of the 21st century, which was first reported in Wuhan, China at the end of December 2019 [3]. The COVID-19 pandemic has affected all aspects of human life [4]. In the beginning, it was assumed that it is a disease like other diseases and can be controlled [5], but it highly affected people’s lives in the world [6]. One of the problems that caused the spread of COVID-19 in different countries was the lack of a proper planning to deal with the disease [7]. Incorrect prediction of its risks caused an increase in the number of casualties and patients [8]. One of the essential factors in controlling infectious diseases is the correct and timely actions of governments against this disease [9]. The decisions of governments can be helpful when they are made based on the correct information [5]. The prediction of disease risks can lead to making more correct decisions and a better understanding of problems when it has high accuracy [8]. The COVID-19 pandemic has disrupted the social and economic life of people [10]; testing and measuring the accuracy of prediction methods can be beneficial to estimate its future trends. Various statistical methods have been used to predict or measure the accuracy of prediction methods [11, 12, 13, 14, 15]. The artificial neural network (ANN) models have been applied for predicting the risks of diseases in other countries [14, 16]. In this study, we aim to use the ANN model to predict the number of COVID-19 cases in Iran and also measure the accuracy of these models, which can be a road map for the use of forecasting methods in future studies. 

2. Materials and Methods
 This is a descriptive, analytical, comparative study. The study population consists of all people infected with COVID-19 in Iran from May 21, 2020 to May 21, 2021. Three input, output, and intermediate layers were used for prediction by the ANN model. A combination of essential computational processing elements connected to each other was used in the form of layers. These processing elements called “neurons” simulate a multi-input nonlinear processor [17]. Connections between neurons have different coefficients. Each neuron receives all inputs with different weights and combines them [18]. The combined inputs are passed through the activation function to produce an output, which can be used as an input to other units [19]. After receiving daily statistics of infected cases, the appropriate ANN model was designed. A genetic algorithm was used for network learning. The ANN was univariate for making predictions. After designing and reaching the optimal ANN model and the optimal number of layers, network training was conducted using Levenberg-Marquardt algorithm (LMA), which finds the minimum of a multivariable nonlinear function. 
Finally, the model output was obtained. The LMA interpolates between the Gauss-Newton algorithm and the gradient descent method. The LMA is more robust than the Gauss-Newton algorithm, which means that it often finds a solution even if it starts far from the final minimum. On the other hand, for well-behaved functions and reasonable starting parameters, LMA is slightly slower than the Gauss-Newton algorithm. The LMA is the most popular method for solving curve-fitting problems. The accuracy of ANN prediction or the mean absolute percentage error (MAPE) was finally measured.

3. Result
The time series of cases infected with COVID-19 in Iran from May 21, 2020 to May 21, 2021 are shown in Figure 1.

The designed ANN had 12 inputs. The ANN divided the data into two categories of testing and training, as shown in Table 1.

Table 2 shows the results of assessing the accuracy of the designed ANN model in forecasting the COVID-19 cases.

The errors of the forecasted values based on the ANN model is plotted in Figure 2.

The trained ANN model was finally used for 30-day out-of-sample forecasting, whose results are shown in Table 3.

Figure 3 shows the mechanism of reaching the best answer.

4. Discussion
 The out-of-sample forecasting values obtained from the ANN model designed in this study showed that the accuracy of the ANN model in forecasting the COVID-19 cases in Iran was 94%, and the ANN was able to estimate the number of infected cases in Iran. Its error rate was 6%, indicating the very high accuracy of the model. In the study by Omidi et al, who investigated the use of ANN in health systems and forecasting injury risks, the ANN model was highly accurate [17]. In Borghi et al.’s study, the ANN was also highly accurate in forecasting the time series of COVID-19 [14]. Comparing the ANN model with other methods in Maestrini et al.’s study showed that the accuracy of the AANN was higher [20]. The ANN model accuracy in our study was higher than that in the studies by Asgari et al. and Omidi et al. [18, 19].
The proper design of ANN plays a significant role in the accuracy of forecasting . The important factors in the proper design of the ANN are the optimal number of input, output, and intermediate layers, the number of neurons in each layer, and how to train the mdoel. The ANN model in our study had 12 inputs. For network training, the LMA was used. The optimal number of input, intermediate, and output layers was 1, and the optimal number of neurons in the intermediate layer was 7. Asgari et al. and Omidi et al. [18, 19] examined the ANN design accuracy in the forecasting, and showed that the design of ANN had a significant impact on the final accuracy of the ANN model. 

5. Conclusion
The ANN model has high accuracy in forecasting the number of COVID-19 cases in Iran. The outputs of this model can be used as a basis for decisions in controlling the COVID-19.

Ethical Considerations
Compliance with ethical guidelines
All ethical principles are considered in this article. The participants were informed of the purpose of the research and its implementation stages.

Funding
This research did not receive any grant from funding agencies in the public, commercial, or non-profit sectors.

Authors' contributions
Both authors equally contributed to preparing this article.

Conflict of interest
The authors declared no conflict of interest.

References

1. Seery V, Raiden SC, Algieri SC, Grisolía NA, Filippo D, De Carli N, et al. Blood neutrophils from children with covid-19 exhibit both inflammatory and anti-inflammatory markers. EBio Medicine. 2021; 67:103357. [DOI:10.1016/j.ebiom.2021.103357] [PMID] [PMCID]
2. Kırbaş İ, Sözen A, Tuncer AD, Kazancıoğlu FŞ. Comparative analysis and forecasting of covid-19 cases in various European countries with ARIMA, NARNN and LSTM approaches. Chaos, Solitons & Fractals. 2020; 138:110015. [DOI:10.1016/j.chaos.2020.110015] [PMID] [PMCID]
3. Vallejo-Janeta AP, Morales-Jadan D, Freire-Paspuel B, Lozada T, Cherrez-Bohorquez C, Garcia-Bereguiain MA, et al. Covid-19 outbreaks at shelters for women who are victims of gender-based violence from Ecuador. International Journal of Infectious Diseases. 2021; 108:531-6. [DOI:10.1016/j.ijid.2021.06.012] [PMID] [PMCID]
4. Demichev V, Tober-Lau P, Lemke O, Nazarenko T, Thibeault C, Whitwell H, et al. A time-resolved proteomic and prognostic map of covid-19. Cell Systems. 2021; 12(8):780-94.e7. [DOI:10.1016/j.cels.2021.05.005] [PMID] [PCMID]
5. Nasirzadeh M, Akhondi M, Jamalizadehnooq A, Khorramnia S. [A Survey on stress, anxiety, depression and resilience due to the prevalence of covid-19 among Anar City households in 2020: A short report (Persian)]. Journal of Rafsanjan University of Medical Sciences. 2020; 19(8):889-98. [DOI:10.29252/jrums.19.8.889]
6. Martinot M, Eyriey M, Gravier S, Bonijoly T, Kayser D, Ion C, et al. Predictors of mortality, ICU hospitalization, and extrapulmonary complications in covid-19 patients. Infectious Diseases Now. 2021; 51(6):518-25. [DOI:10.1016/j.idnow.2021.07.002] [PMID] [PMCID]
7. Ballı S. Data analysis of covid-19 pandemic and short-term cumulative case forecasting using machine learning time series methods. Chaos, Solitons & Fractals. 2021; 142:110512. [DOI:10.1016/j.chaos.2020.110512] [PMID] [PMCID]
8. Ala’raj M, Majdalawieh M, Nizamuddin N. Modeling and forecasting of covid-19 using a hybrid dynamic model based on SEIRD with ARIMA corrections. Infectious Disease Modelling. 2021; 6:98-111. [DOI:10.1016/j.idm.2020.11.007] [PMID] [PMCID]
9. Chaubey A, Sahoo CK. Assimilation of business intelligence: The effect of external pressures and top leaders commitment during pandemic crisis. International Journal of Information Management. 2021; 59:102344. [DOI:10.1016/j.ijinfomgt.2021.102344]
10. Garcia MA, Barreras PV, Lewis A, Pinilla G, Sokoll LJ, Kickler T, et al. Cerebrospinal fluid in covid-19 neurological complications: Neuroaxonal damage, anti-SARS-Cov2 antibodies but no evidence of cytokine storm. Journal of the Neurological Sciences. 2021; 427:117517. [DOI:10.1016/j.jns.2021.117517] [PMID] [PMCID]
11. ArunKumar KE, Kalaga DV, Kumar CMS, Kawaji M, Brenza TM. Forecasting of covid-19 using deep layer recurrent neural networks (RNNs) with gated recurrent units (GRUs) and long short-term memory (LSTM) cells. Chaos, Solitons & Fractals. 2021; 146:110861. [DOI:10.1016/j.chaos.2021.110861] [PMID] [PMCID]
12. ArunKumar KE, Kalaga DV, Sai Kumar CM, Chilkoor G, Kawaji M, Brenza TM. Forecasting the dynamics of cumulative covid-19 cases (confirmed, recovered and deaths) for top-16 countries using statistical machine learning models: Auto-regressive integrated moving average (ARIMA) and seasonal auto-regressive integrated moving average (SARIMA). Applied Soft Computing. 2021; 103:107161. [DOI:10.1016/j.asoc.2021.107161] [PMID] [PMCID]
13. Guleryuz D. Forecasting outbreak of covid-19 in Turkey; Comparison of Box-Jenkins, Brown’s exponential smoothing and long short-term memory models. Process Safety and Environmental Protection. 2021; 149:927-35. [DOI:10.1016/j.psep.2021.03.032] [PMID] [PMCID]
14. Borghi PH, Zakordonets O, Teixeira JP. A covid-19 time series forecasting model based on MLP ANN. Procedia Computer Science. 2021; 181:940-7. [DOI:10.1016/j.procs.2021.01.250] [PMID] [PMCID]
15. Al-qaness MAA, Saba AI, Elsheikh AH, Elaziz MA, Ibrahim RA, Lu S, et al. Efficient artificial intelligence forecasting models for covid-19 outbreak in Russia and Brazil. Process Safety and Environmental Protection. 2021; 149:399-409. [DOI:10.1016/j.psep.2020.11.007] [PMID] [PMCID]
16. Aljaaf AJ, Mohsin TM, Al-Jumeily D, Alloghani M. A fusion of data science and feed-forward neural network-based modelling of covid-19 outbreak forecasting in IRAQ. Journal of Biomedical Informatics. 2021; 118:103766. [DOI:10.1016/j.jbi.2021.103766] [PMID] [PMCID]
17. Omidi MR, Nabi O. Forecasting the number of injured in traffic accidents referred to forensic medicine in Hamadan province using multi-layered artificial neural network. Safety Promotion and Injury Prevention. 2020; 8(1):24-9. [DOI:10.22037/meipm.v8i1.30132]
18. Asgari H, Omidi M R, Omidi N. [Prediction of casualties of traffic accidents in Khuzestan province using dynamic neural network (Persian)]. Traffic Management Studies. 2018; 13(49):23-34. [Link]
19. Omidi MR, Jafari Eskandari M, Raissi S, Shojaei AA. Providing an appropriate prediction model for traffic accidents: A case study on Accidents in Golestan, Mazandaran, Guilan, and Ardebil Provinces. Health in Emergencies and Disasters Quarterly. 2019; 4(3):165-72. https://www.sid.ir/paper/346097/en
20. Maestrini V, Birtolo LI, Francone M, Galardo G, Galea N, Severino P, et al. Cardiac involvement in consecutive unselected hospitalized covid-19 population: In-hospital evaluation and one-year follow-up. International Journal of Cardiology. 2021; 339:235-42. [DOI:10.1016/j.ijcard.2021.06.056] [PMID] [PMCID]