Analysis of coronavirus active cases for Italy
As demonstrated in Fig. 7, the proposed model’s fitting impact outperforms other models used in Italy. As per our forecast, the number of coronavirus positive patients on 29 January 2022 is 10,821,375, while the proposed model predicts 10,819,855.
Analysis of coronavirus active cases for Brazil
As demonstrated in Fig. 8, the proposed model’s fitting impact outperforms other models used in Brazil. As per our forecast, the number of coronavirus positive patients on 29 January 2022 is 25,256,198, while the proposed model predicts 25,254,306.
From 16 January 2022 to 29 January 2022, a forecast vs. actual plot for Brazil’s positive coronavirus patients is shown. According to the proposed model’s predictions, Brazil has 25,254,306 COVID-19 cases on 29 January 2022.
Analysis of coronavirus active cases for India
As demonstrated in Fig. 9, the proposed model’s fitting impact outperforms other models used in India. As per our forecast, the number of coronavirus positive patients on 29 January 2022 is 41,092,522, while the proposed model predicts 41,090,386.
From 16 January 2022 to 29 January 2022, a forecast vs. actual plot for India’s positive coronavirus patients is shown. According to the proposed model’s predictions, India has 41,090,386 COVID-19 cases on 29 January 2022.
Analysis of coronavirus active cases for Germany
As demonstrated in Fig. 10, the proposed model’s fitting impact outperforms other models used in Germany. As per our forecast, the number of coronavirus positive patients on 29 January 2022 is 9,618,245, while the proposed model predicts 9,617,381.
From 16 January 2022 to 29 January 2022, a forecast vs. actual plot for Germany’s positive coronavirus patients is shown. According to the proposed model’s predictions, Germany has 9,617,381 COVID-19 cases on 29 January 2022.
Analysis of coronavirus active cases for Pakistan
As demonstrated in Fig. 11, the proposed model’s fitting impact outperforms other models used in Pakistan. As per our forecast, the number of coronavirus positive patients on 29 January 2022 is 1,417,991, while the proposed model predicts 1,417,481.
From 16 January 2022 to 29 January 2022, a forecast vs. actual plot for Pakistan’s positive coronavirus patients is shown. According to the proposed model’s predictions, Pakistan has 1,417,481 COVID-19 cases on 29 January 2022.
Analysis of coronavirus active cases for Myanmar
As demonstrated in Fig. 12, the proposed model’s fitting impact outperforms other models used in Myanmar. As per our forecast, the number of coronavirus positive patients on 29 January 2022 is 535,080, while the proposed model predicts 534,898.
From 16 January 2022 to 29 January 2022, a forecast vs. actual plot for Myanmar’s positive coronavirus patients is shown. According to the proposed model’s predictions, Myanmar has 534,898 COVID-19 cases on 29 January 2022.
Performance analysis
In comparison to previous models, the proposed model has the highest \({R}^{2}\) values, 0.9992 in phase 1 and 0.9784 in phase 2. The recommended \({R}^{2}\) value is closer to 1, which denotes that the forecast is accurate for the COVID-19 positive patients. The logistic growth model has a higher \({R}^{2}\) than the FTS model.
Discussion
Strengths and weaknesses of the proposed model
The proposed hybrid model combining the FTS technique with the nonlinear logistic growth model stands out for its capabilities of making accurate predictions of the active cases of COVID-19. However, the proposed model is highly efficient in terms of predictive capacities because of its ability to identify linear and nonlinear trends in the data. The model’s R-scores of 0.9992 in phase-1 and 0.9784 in phase-2 underscore its robustness and reliability. Additionally, the model’s adaptability to different countries with varying epidemic dynamics showcases its flexibility and generalizability. The FTS component, while adept at handling nonlinearity, can be sensitive to the choice of intervals and the fuzzification process, potentially impacting prediction accuracy. Moreover, the logistic growth model assumes a saturation point which might not be applicable in scenarios with fluctuating infection rates due to external interventions like lockdowns or mass vaccination drives.
Limitations and differences in prediction results
Further research should be focused on the reasons for variation in predictions for different countries and different time intervals. The differences in the prediction accuracy can be attributed to several factors:
Government interventions
The measures like the implementation of lock down, social distancing measures, and vaccination programs differ greatly in terms of stringency and timing across nations. These interventions can cause sudden changes in infection trends, which can be difficult for the model to follow.
Healthcare Infrastructure
Differences in the health care systems and their capacity around the world help infect the disease and affects the ability to treat it thereby affecting the ability of the model to predict the disease. Countries with well-established health care systems may therefore have different epidemic trends than those with constrained health care systems.
Cultural and behavioral factors
Literature also indicates that people’s adherence to advised health measures, cultural practices, and social norms also influence disease transmission dynamics and thus vary the prediction results.
Phase-specific factors
The two phases analyzed include January 28, 2020, to June 5, 2020 (phase-1), and October 10, 2021 to January 15, 2022 (phase-2), which are two distinct eras of the pandemic. Factors affecting the model during the early phase include high transmission rate and few treatments, while factors during the late phase include vaccination and improved management techniques.
To overcome these limitations and improve the reliability of our model in the future, more variables including population mobility, social distance, and vaccination data, humidity, and average temperature which are known to affect virus spread will be included. We also envisaged the use of more complex algorithms such as neuro-fuzzy systems and weighted sum algorithms in order to increase prediction precision.
Discover more from PressNewsAgency
Subscribe to get the latest posts sent to your email.