-
Research Article
Comparative Study of INGARCH and SARIMA in Modeling and Forecasting Dengue Cases
Frasiah Wambui Kariuki*,
Anthony Kibira Wanjoya,
Bonface Miya Malenje
Issue:
Volume 12, Issue 6, November 2023
Pages:
150-160
Received:
3 October 2023
Accepted:
26 October 2023
Published:
7 November 2023
Abstract: A crucial focus of public health surveillance systems is to provide reliable forecasts of epidemiological time series. This work utilized data collected through a national public health surveillance system in Thailand to evaluate and compare the performance of a seasonal autoregressive integrated moving average and an Integer generalized autoregressive conditionally heteroscedastic model for modeling and forecasting case occurrence of dengue. The comparison uses weekly reported cases of dengue hemorrhagic fever in Amnat Charoen province Thailand, from January 1st, 2006, to October 7th, 2017 (612 weeks). The results from the in-sample evaluation using the root mean square error and mean absolute error as well as a visual inspection of predicted values show that the two approaches are adequate tools for use in epidemiological surveillance as there is no significant difference in their forecast accuracy for in-sample performance. The incorporation of the weather variables improves the predictive performance of the models and from the model coefficients the study findings reveal that there is a positive relationship between temperature and rainfall and the occurrence of dengue. Overall, the findings in this study support the usefulness of the two approaches as effective tools practitioners can utilize for monitoring and for providing early warning signals of potential outbreaks of epidemics.
Abstract: A crucial focus of public health surveillance systems is to provide reliable forecasts of epidemiological time series. This work utilized data collected through a national public health surveillance system in Thailand to evaluate and compare the performance of a seasonal autoregressive integrated moving average and an Integer generalized autoregres...
Show More
-
Research Article
Early Diagnosis of Pneumonia from Chest X-Rays Using a Capsule Network Model: Enhancing Accuracy and Efficiency in Automated Image Classification
Mbae Karwitha Maureen,
Thomas Mageto,
Anthony Wanjoya
Issue:
Volume 12, Issue 6, November 2023
Pages:
161-173
Received:
2 October 2023
Accepted:
20 October 2023
Published:
9 November 2023
Abstract: Pneumonia is a significant public health concern worldwide, causing substantial morbidity and mortality. Early, accurate diagnosis is vital in ensuring timely treatment and improving patient outcomes. Chest X-ray analysis is the standard procedure used most frequently to diagnose pneumonia, but the accurate and timely interpretation of these images can be complex and time consuming. This research aimed to develop a capsule network (CapsNet) model for image classification, based on the Capsule network model introduced by Sabour and his colleagues in 2017 enabling automated chest X-ray analysis for early detection of pneumonia. Pneumonia impacts diverse populations, with vulnerable groups such as the elderly, young children and immunocompromised individuals at heightened risk. Delayed or missed diagnoses can lead to severe complications and increased healthcare costs. The reliance on human expertise for chest X-ray interpretation introduces the potential or errors, therefore there is a dire need to develop automated and precise diagnostic models and tools which are crucial for facilitating timely interventions. In this study secondary data obtained from Mendeley data was comprehensively pre-processed thoroughly by applying image resizing, standardization and normalization for consistent image quality, followed by a gaussian blur for noise reduction, and histogram equalization for contrast enhancement. The enhanced dataset enabled the main features of the pneumonia-infected images to be captured effectively during model training. The dataset was split into sets for training, testing and validation in an 80%, 10% and 10% ratio. The training set was used to train the CapsNet model which demonstrated a commendable performance with a 96% accuracy, a precision of 96.97% and a recall of 97.42%. The Capsule Network model shows a significant promise as a tool for improving the efficiency and accuracy of pneumonia diagnosis, thus befitting patients and healthcare providers.
Abstract: Pneumonia is a significant public health concern worldwide, causing substantial morbidity and mortality. Early, accurate diagnosis is vital in ensuring timely treatment and improving patient outcomes. Chest X-ray analysis is the standard procedure used most frequently to diagnose pneumonia, but the accurate and timely interpretation of these images...
Show More
-
Research Article
Modelling Stroke Risk Factors Using Classical and Bayesian Quantile Regression Models
Kirui Dennis*,
Charity Wamwea,
Bonface Malenje,
Levi Bor
Issue:
Volume 12, Issue 6, November 2023
Pages:
174-179
Received:
3 October 2023
Accepted:
25 October 2023
Published:
11 November 2023
Abstract: The assessment of stroke risk and mortality, the second leading global cause of death, is of paramount importance. Stroke prediction is a vital pursuit due to its multifactorial nature, involving variables like age, sex, gender, hypertension, BMI and heart disease, which introduce considerable complexity. These diverse factors often lead to substantial uncertainty in stroke prediction models. Our research delves into the evaluation of two distinct methodologies for quantifying this uncertainty: Bayesian and classical quantiles. Bayesian quantiles are calculated from the posterior distribution of a Bayesian logistic regression model, accounting for prior information and spatial correlations. In contrast, classical quantiles are based on the assumption that stroke probabilities conform to a normal distribution. The results reveal that, across all coefficients, the Bayesian model produces narrower intervals compared to the classical model, indicating higher accuracy and confidence. Hence, we conclude that Bayesian quantiles outperform classical quantiles in the context of stroke prediction in Kenya. We recommend their adoption in future research and applications, acknowledging their superior performance and reliability in enhancing stroke prediction models, ultimately contributing to improved public health outcomes. This research represents a significant step towards a better understanding and management of stroke risks and mortality on a global scale.
Abstract: The assessment of stroke risk and mortality, the second leading global cause of death, is of paramount importance. Stroke prediction is a vital pursuit due to its multifactorial nature, involving variables like age, sex, gender, hypertension, BMI and heart disease, which introduce considerable complexity. These diverse factors often lead to substan...
Show More
-
Research Article
Estimating the Determinants of Firm Innovation Inefficiency Through the Conditional Mean of Innovation Inefficiency Given a Composite Error
Issue:
Volume 12, Issue 6, November 2023
Pages:
180-186
Received:
17 October 2023
Accepted:
3 November 2023
Published:
11 November 2023
Abstract: The present paper demonstrates that the estimations of the determinants of firm innovation inefficiency can be obtained through the conditional mean of innovation inefficiency given a composite error. We extract the estimations of the determinants of firm innovation inefficiency by replacing the true parameters in the equation of the conditional mean of innovation inefficiency given a composite error with Maximum Likelihood estimations from the Stochastic Frontier Approach. This is an alternative method for the estimation of the determinants of firm inefficiency besides those which are existent in the relevant literature. Based on statistical theory and algebra, we first present the case where innovation inefficiency is assumed to be distributed as a truncated normal with a nonzero constant mean. Second, we focus on the case where innovation inefficiency is assumed to be distributed as a truncated normal with a mean that varies across firms. There, we show that all the change in the error term of the Stochastic Frontier Knowledge Production Function originates from innovation inefficiency. The latter is modelled as having two components: a) a function of some firm-specific characteristics (variables) and b) random component. Then, we advance to the estimations of the determinants of firm innovation inefficiency via a generalized Stochastic Frontier Approach (generalized production frontier approach). Finally, we replace the true parameters in the equation of the conditional mean of innovation inefficiency given a composite error with Maximum Likelihood estimations from the generalized production frontier approach.
Abstract: The present paper demonstrates that the estimations of the determinants of firm innovation inefficiency can be obtained through the conditional mean of innovation inefficiency given a composite error. We extract the estimations of the determinants of firm innovation inefficiency by replacing the true parameters in the equation of the conditional me...
Show More
-
Research Article
Determinants of Climate Smart Agriculture Technology Practices in Ghana: Application of Multinomial Logistic Regression
Ibrahim Hashim,
Abukari Alhassan,
Richard Puurbalanta*,
Edward Akurugu,
Yahaya Iddrisu,
Salifu Hussein
Issue:
Volume 12, Issue 6, November 2023
Pages:
187-194
Received:
19 November 2023
Accepted:
9 December 2023
Published:
22 December 2023
Abstract: The study primarily examined the determinants of Climate Smart Agriculture technology practices on maize production. Data on socio-demographic and farming characteristics were obtained from the Climate Change, Agriculture and Food Security Partnership for Up Scaling the project’s targeted communities (Bompari, Dazuuri and Toto) in the Lawra municipality of the Upper West Region of Ghana. A total of 300 peasant farmers completed the questionnaire. Results from the model building confirmed models 1 and 2 to have strong explanatory power. Notwithstanding that, further evaluation with the adoption of Likelihood Ratio and log-likelihood favoured model 1 Furthermore, the post estimation results (Average Marginal Effects) from model 1 revealed that farming experience and household head status have no significant impact on predicting Climate Smart Agriculture technology practices. The results also confirmed that farmers who have practiced Climate Smart Agriculture technology for 6 to 10 years were found to be accompanied by a low probability (15.47%) of using improved variety/treated seeds as compared to those farmers who have practiced the technology for a period of 1–5 years. Also, tied ridges as Climate Smart Agriculture technology practiced by farmers resulted in a high probability of 11.44% for high yields relative to low yields. We recommend the need for further study to investigate the underlying reasons, if any, based on the non-significant relationship established at the 5% level between the determinants of mineral chemical fertiliser and monoculture respectively.
Abstract: The study primarily examined the determinants of Climate Smart Agriculture technology practices on maize production. Data on socio-demographic and farming characteristics were obtained from the Climate Change, Agriculture and Food Security Partnership for Up Scaling the project’s targeted communities (Bompari, Dazuuri and Toto) in the Lawra municip...
Show More