This study investigated the statistical insight of big data analytics for efective pandemic analysis. The COVID19 pandemic dataset, containing 5010 negatives and 4990 positives, was obtained from a prominent source known as the Kaggle repository. The dataset was categorized into four classes: COVID-19, Normal, Pneumonia, and Tuberculosis using diverse statistical metrics of image characteristics: mean intensity of a color channel, homogeneity, dissimilarity, correlation, and density. The objective of the statistical analysis was to validate the use of big data analysis for pandemic preparedness, with a working hypothesis that the intensity of the colour channel images would be diferent (p < 0.05) between COVID-19 patients and patients with other health conditions. A statistical package for the Social Sciences version 20 was used for the analysis. The outcomes show that the mean intensity of color channels (Blue, Green, Red) in COVID-19 and Tuberculosis cases was greater in the blue and green channels than in Normal and Pneumonia cases. Pneumonia and normal cases exhibited comparable and reduced mean intensities across all three channels. The Normal condition exhibited the largest mean contrast, whereas Pneumonia displayed the lowest. The mean dissimilarity and homogeneity indicate that tuberculosis demonstrated the highest dissimilarity, signifying greater diversity in pixel intensity. COVID-19 and Pneumonia exhibited comparable homogeneity; however Tuberculosis demonstrated more homogeneity. The mean correlation indicates that Normal images exhibited the highest correlation, whilst Tuberculosis demonstrated the lowest. The mean density study indicates that normal cases demonstrated the highest mean density, whereas tuberculosis exhibited the lowest mean density. In conclusion, our findings established the advantages of statistics for modeling and analyzing big data in the pandemic domain, covers the state-of-the-art for practical analysis.
Big Data Analytics have become common in many areas, from research to practice, and even in everyday
life. It is a prominent tool in pandemic preparedness and control as an aspect of the fourth industrial
revolution (4IR) [
illness surveillance and response. The system helps healthcare professionals track, monitor, and analyze
infectious illness reports via social media to prevent and control child-related diseases. The proposed
technique was validated using use-case scenarios and performance comparisons [
to dependability problems. Based on the 72 reliability-related incidents, despite the ubiquity, it is
also observed that out of those 72 occurrences, 29 incidences resulted in fatalities or serious injuries,
demonstrating that dependability problems may cause significant damage [
comprehend, analyze, and predict infectious disease patterns. AI technology is still growing and has dificulties. Challenges include dependability, safety, durability, and trustworthiness. Dependability is important because AI systems must be trusted. The system may work as intended. Resource allocation and public health crisis response depend on these criteria. Examine reliability, strategy, and failures.
research concerns. The authors discussed AI dependability’s out-of-distribution detection, training set
influence, adversarial attacks, model correctness, and uncertainty quantification. A few examples show
study options. A recent study explored AI reliability evaluation, data gathering, and test preparation,
and how to improve system designs for AI dependability. Since data is vital for reliability assessment
[
of the present investigation. Section 3 delineates the approach, attributes of the dataset, and the research instrument. Section 4 presents a comprehensive analysis of statistical insights on big data analytics for the efective detection of infectious diseases. Section 5 finishes the research and proposes future endeavors for the implementation of efective pandemic control utilizing statistical tools.
Big Data Analytics is crucial for advancing the Fourth Industrial Revolution and tackling the challenges presented by the pandemic. The Fourth Industrial Revolution is a significant transformation characterized by the convergence of physical, digital, and biological technologies. Currently, Big Data propels substantial progress in automation and robotics, enabling these systems to analyze operational data for enhanced eficiency and flexibility. It also enables the IoT, where interconnected devices across diverse platforms generate vast data that, when analyzed, can improve operational eficiencies and predictive maintenance capabilities. Moreover, in smart manufacturing, Big Data enhances the optimization of production processes, minimizes downtime, and enables real-time customization of output. During the
has enabled efective epidemiological surveillance by analyzing data from several sources, including
travel records, social media, and government reports, to track the virus’s spread [
outcomes, and tailored treatment options. Furthermore, the rapid progress and distribution of vaccinations have been expedited by Big Data, which has enabled the analysis of massive clinical trial data. It has also impacted public policy, aiding governments in developing specific lockdown measures and informed reopening strategies based on real-time data concerning infection rates and public sentiment.
has accelerated the adoption of digital solutions, including telemedicine, remote work, and online education, which are vital to the current digital transformation. This integration highlights the essential role of Big Data in crisis management and as a core component of digital and industrial advancement in the
tonomous systems. The authors introduce the "SMART" statistical framework to evaluate the robustness of AI models used in various domains, including healthcare. The study highlights how AI reliability decreases in out-of-distribution settings and suggests that statistical models outperform deep learning in long-term failure prediction. However, it lacks real-world infectious disease applications and validation on epidemiological datasets.
during COVID-19. It discusses statistical inference challenges in high-dimensional data, highlighting the efectiveness of Bayesian models in long-term trend prediction. However, it has a limited focus on disease classification and weak integration of statistical methods in real-time applications.
Mwamnyange et al. [
tructure for early detection of infectious disease outbreaks. The authors propose an AI-driven data integration framework leveraging social media, wearable technology, and traditional clinical databases.
used for disease outbreak prediction. It evaluates traditional statistical methods such as time-series forecasting and Bayesian inference, alongside AI-based models. The study highlights the importance of integrating epidemiological data with environmental and social media data for robust predictions.
science methodologies such as agent-based modeling and network theory. It provides insights into how big data can enhance disease spread predictions, but lacks emphasis on statistical uncertainty quantification and model generalizability across varying outbreak conditions.
study refines predictive models for disease progression, risk assessment, and individualized treatment using EHRs, medical imaging, genomic data, and wearable sensors. Case studies show it reduces hospital remissions and optimizes healthcare resource management. AI-driven healthcare analytics is lauded, but real-time flexible models and scalability across healthcare systems are not.
Table 1 contains a summary of the review of the related studies on big data analytics for pandemic Summary of the review of the related studies on big data analytics for pandemic classification.
Methodology Dataset Type Results Limitations SMART framework, survival analy- Simulated AI reliability AI reliability decreases in No epidemiological validasis, hazard functions datasets out-of-distribution; statisti- tion, not focused on disease cal models perform better in detection failure prediction AI-based ML (SVM, RF, Neural Net- Syndromic surveillance, 91% sensitivity in outbreak Lacks statistical validation, works), social media and genomic search trends, clinical data prediction poor model interpretability data fusion High-dimensional statistical mod- COVID-19 epidemiological Bayesian models outper- Limited disease classificaels (LASSO, Bayesian Inference), data, EHRs form AI in long-term trend tion focus, weak integration computational epidemiology prediction of statistical methods MapReduce, Hadoop-based data in- Clinical records, social me- Faster epidemiological data Lacks statistical validation, tegration dia, surveillance data processing, early outbreak prioritizes data processing
detection over classification AI-driven data integration, public Social media, wearable tech, Enhanced real-time disease Lacks statistical validahealth surveillance, wearable data clinical databases tracking tion, no adaptive learning fusion methodology Investigation of statistical and AI Epidemiological, environ- Identifies gaps in predictive No standardized framework models (Bayesian inference, time- mental, social media data analytics models for model comparison series, ML) Agent-based modelling, network Global epidemiological and Insights into disease spread Lacks statistical uncertainty theory, computational epidemiol- mobility data via big data quantification, limited ogy model generalizability Investigates an integrated health Empirical study with com- Health data from clinics, AI-based screening, big data big data system for detecting and parative analysis. hospitals, and government analytics, disease detection preventing infectious diseases. records in China. algorithms.
Investigates the role of data analyt- Case study analysis with EHRs, medical imaging, ge- ML predictive modeling, ics in predictive healthcare, focused machine learning models ap- nomic sequencing, wearable statistical analysis, Apache on risk assessment and treatment plied to patient data. sensor data from City Hos- Hadoop, Spark,Python, personalization. pital. Tableau. classification.
The decision-making process is significantly enhanced by the application of statistical methods, even in
instances involving extensive data sets, sometimes referred to as big data [
The COVID-19 pandemic dataset utilized in this investigation was obtained from a prominent source known as the Kaggle database (https://www.kaggle.com/datasets/rishanmascarenhas/covid19temperatureoxygenpulse-rate). The distribution of the pandemic datasets sample in this analysis which had values ranging from 0 to 9. The details include the ID, Oxygen, Pulse Rate, Temperature, and results (+/−) . In total, the dataset contains 5010 negative and 4990 positive examples that support replicability, making it large enough to be classified as big data. Figure 2 shows the distribution of the datasets by classes and evidence of balanced classes using explorative data analysis. There is no ethical concern regarding this dataset because it was sourced from opensource and has been deanonymized to protect privacy and security.
Framing the research as a classification problem: Let = {(1, 1), (2, 2), ...(, )} constitute
the collection of training cases of size d. = {1, 2..., } be the set of labels where is a feature
with corresponding label, and is a set of features according to classification task definition. The
preliminary phase of this data analysis and categorization commenced with the data preparation;
the dataset sufers from missing data and duplicated data. This was succeeded by a crucial step of
eliminating non-relevant data from the converted raw dataset. Duplication and missing values were
addressed using an imputation technique, wherein the meaning of the respective column was calculated
and used to replace the missing values. The issue was resolved through the identification of essential
aspects, conducted in accordance with [
with Levene’s test implemented in SPSS v 21.0 (IBM Corp., Armonk, NY, USA). This was accomplished using Intel(R) Core (TM) i7-4600U CPU running from 2.10GHz to 2.70GHz on a Windows 10 professional with 8GB RAM. The statistical analysis was to validate the use of big data analysis for pandemic preparedness, with a working hypothesis that the intensity of the colour channel images would be diferent ( < 0.05) between COVID-19 patients and patients with other health conditions.
In statistics, statistical significance denotes the likelihood that the outcomes of a study or experiment are attributable to factors other than random chance. Various metrics and tests are typically employed to ascertain the statistical significance of a result. This research utilized primary criteria for statistical significance contingent upon the specific type of pandemic dataset examined. Statistical significance is a crucial concept; nonetheless, it is essential to recognize that statistical significance does not equate to practical relevance. Always consider the context of the findings and additional metrics, such as efect size, when evaluating results. The equations shown below describe these metrics and express them statistically, using mean intensity of a Color channel, SEM, mean correlation, mean homogeneity, mean dissimilarity, mean correlation, mean correlation and mean density. The detailed description of statistical metrics used, with accompanying formulas are presented in this section. (1) (2) (3) (4) (5) (6) (7) where is the intensity of the pixel in the color channel, and N is the total number of pixels. Equation 2 shows the Standard Error of the Mean (SEM) for pandemic classification. =
1 ∑︁
=1 = √
where is the standard deviation of pixel intensities and is the number of pixels for pandemic
= ∑︁ (, )( − ) 2 , where (, ) is the probability of intensity pairs in the GLCM for pandemic classification.
= ∑︁
(, ) 1 + | − | = ∑︁ (, )| − | , , = ∑︀( − )( − ) (, )| where and are the means and standard deviations of intensity levels for pandemic classification.
= ∑︀
where represents pixel values in a binary image for pandemic analysis.
task is conceptualized based on the statistical tools called for a detailed analysis of the COVID-19 pandemic dataset. A statistical analysis was conducted to categorize infectious diseases, including
preparation, analysis via SPSS, and essential statistical measures (intensity, SEM, contrast, homogeneity, dissimilarity, correlation, and density). The findings are articulated through sophisticated statistical metrics, including the mean intensity of a color channel, the standard error of the mean (SEM), mean correlation, mean homogeneity, mean dissimilarity, repeated mean correlation, and mean density. 4.2. Analysis of big data analytics for pandemic classification.
statistical analysis of four pandemic data scenarios/conditions. Data is divided into COVID-19, Normal, Pneumonia, and Tuberculosis. For each occurrence, it includes statistical measurements of image attributes, including the mean and standard error. It is a clear presentation of results with mean values and standard errors; Table 2 is the comparison of the blue channels in images across diferent health conditions, and Table 3 presents the comparison of green channels in the image obtained for diferent health conditions. While Table 4 compares the red channel in the image obtained for diferent health conditions, Table 5 compares the contrast of the image obtained for diferent health conditions. The focus of Table 6 is the comparison of dissimilarity of the image obtained for diferent health conditions, and Table 7 shows the comparison of homogeneity of the image obtained for diferent health conditions. Table 8 compares the correlation of the images obtained for diferent health conditions, and finally, Table 9 compares the correlation of the images obtained for diferent health conditions. In COVID-19 and TB cases, blue and green channels have a higher mean intensity than Normal and Pneumonia channels. All three channels show similar and reduced mean intensities for pneumonia and normal patients.
Case Sample size Mean intensity ± SEM p-value COVID-19 554 134.87 ± 0.94 Normal 1779 122.15 ± 0.33 < 0.001 Pneumonia 4061 122.91 ± 1.05 Tuberculosis 729 133.77 ± 1.05 a,b: indicate significant diference (p<0.05); CI: confidence interval.
95% CI 124.29-125.24
Case Sample size Mean intensity ± SEM p-value COVID-19 554 134.59 ± 0.94 Normal 1779 122.15 ± 0.33 < 0.001 Pneumonia 4061 122.91 ± 0.31 Tuberculosis 729 129.22 ± 1.02 a,b,c: indicate significant diference (p<0.05); CI: confidence interval.
95% CI 123.80-124.74
the highest intensity, compared to patients with diferent health conditions as indicated in Table 2, 3 and 4, respectively. Mean contrast of the image from Covid-19 patients (134.36 ± 4.37) was significantly lowered ( < 0.001) compared with mean contrast of the image from the normal patients (163.81 ± 1.08) in Table 5.
demonstrate the highest dissimilarity (0.98 ± 0.01) and homogeneity (0.38 ± 0.00) , respectively, signifying greater diversity in pixel intensity as shown in Table 6 and 7. COVID-19 and Pneumonia exhibit comparable homogeneity (0.26 − 0.27 ; Table 7). The mean correlation indicates that Normal images exhibit the highest correlation (12.06 ± 0.01) as shown in Table 8. The mean density study indicates that normal cases demonstrate the highest mean density (19.20 ± 0.12 ; Table 9). Meanwhile, the mean contrast, dissimilarities, homogeneity, correlation and density were lower ( < 0.05) for
Pattern identification and analysis of infectious and pandemic illness epidemics have underutilized statistical approaches. The global pandemic data study lacks statistical analysis. For pandemic analysis, state-of-the-art statistical methods like mean intensity of a color channel, SEM, mean correlation, homogeneity, dissimilarity, correlation, and density provide depth of finding for informed decisionmaking. This research shows that pandemics are straightforward to control and minimize mortality if statistical methods are used to analyze the pandemic dataset to advise doctors, policymakers, and other stakeholders on disease spread. This study examines statistics, data science, machine learning, and AI in relation to environmental science, natural sciences, medicine, and technology. It shows how statistics may model and analyze huge data in pandemics, covers the state-of-the-art for practical analysis, and shows how to implement methodologies. The paper makes a more substantial contribution to the ifeld of pandemic analytics. Future research may compare results using theoretical machine learning techniques, ML modeling, and validation, to validate the result of this investigation. Multiple data sources enable a more complete picture of disease spread. Maps of COVID-19 hotspots were created using social media, contact tracing apps, and real-time hospital data.