INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue IV, April 2026

Development of a Predictive Model for Fowl-Cholera Infection Status

in Poultry Using Advanced Data Mining Analysis Techniques and

Logistic Regression Modeling.

DOI:

Fowl cholera, caused by Pasteurella multocida, remains one of the most economically devastating poultry

diseases worldwide. Rapid and accurate diagnosis is critical for effective intervention, yet traditional methods

often fall short in speed and predictive accuracy. This study presents a Big Data-driven data mining approach to

diagnose fowl cholera in poultry, leveraging a dataset of 500 samples characterized by variables such as bird

age, vaccination history, environmental conditions, clinical symptoms, and mortality rates. Machine learning

algorithms including Logistic Regression, Random Forest, and Gradient Boosting were deployed to model

Fowl cholera, a highly contagious bacterial disease caused by Pasteurella multocida, continues to impose

substantial economic losses on the global poultry industry [1]. In Nigeria, the disease disproportionately affects

small and medium-scale poultry farmers, where limited access to rapid diagnostic tools often delays

interventions and exacerbates mortality rates [2]. Traditional diagnostic approaches, although effective, typically

require well-equipped laboratory infrastructure and extended turnaround times [3].

Advances in big data analytics and machine learning now offer transformative opportunities for proactive disease

surveillance in veterinary practice. By integrating heterogeneous data sources including environmental

monitoring, clinical observations, and vaccination histories predictive models can deliver timely, accurate

diagnostic insights, enabling targeted interventions and reducing losses [4].

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue IV, April 2026

Epidemiology and Burden of Fowl Cholera

Fowl cholera, triggered by Pasteurella multocida, is a persistent poultry disease with both acute and chronic

manifestations. Birds suffering from the acute form often die suddenly, while the chronic form results in

localized lesions and reduced productivity. Outbreaks cause high mortality and economic loss across both

commercial and smallholder systems, especially in regions where poultry contributes meaningfully to household

nutrition and income [5]; [6].

include PCR and serotyping. These approaches are reliable in controlled lab settings, but they are slow and

resource-intensive, limiting utility in urgent field-based intervention contexts [5]. More advanced tools like

whole-genome sequencing can identify strain variability and inform vaccine matching, but are still costly and

infrequently deployed in low-resource settings [5]. A recent Ethiopian study emphasized that without linked,

proactive surveillance, diagnostic gains alone don’t translate to rapid responses on the ground [9].

Predictive Modeling for Disease Surveillance

In recent years, veterinary epidemiology has embraced data-driven approaches, using large, varied data sources

from farm records to sensor outputs to detect disease early and target interventions. For instance, a machine

learning model achieved around 91% accuracy in predicting Newcastle disease outbreaks across over a thousand

farms [10]. Similar approaches are being used to prioritize inspections and forecast risk [11]. Reviews show that

Purely data-driven models often face criticism for being opaque or overly tailored to one dataset. A more

promising direction is to integrate statistical/ML models with epidemiological theory such as compartmental

disease spread patterns, age-based susceptibility, and seasonal trends. This hybrid approach can enhance

transparency, scalability, and actionable relevance [14]; [15]. In the context of fowl cholera, models can be

improved by treating vaccination history and local strain ecology as structured, interpretable inputs, rather than

as anonymous features [5].

Gaps and Opportunities

Despite the rise of predictive analytics, few studies address fowl cholera specifically. Many target other poultry

diseases or general management issues, with very little prospective validation in real farms [10]; [11].

Furthermore, while environmental and management risk factors are recognized, they are seldom combined with

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue IV, April 2026

Dataset

The study utilized a dataset comprising 500 poultry health records from commercial and backyard farms in Osun

State, Nigeria. Key variables included: Age of birds (in weeks), Vaccination history (binary:

complete/incomplete), Environmental factors (temperature, humidity, sanitation score), Clinical symptoms

(respiratory distress, diarrhea, lethargy), Mortality rates (% within last 14 days).

Table 1. Performance Metrics of Different Machine Learning Algorithms

Table 2. Relative Feature Importance for Fowl Cholera Prediction (Random Forest Model)

Fig. 1. Confusion Matrix for Random Forest Classifier

The confusion matrix in Figure 1 illustrates the classification performance of the Random Forest model,

highlighting its ability to correctly identify both infected and non-infected cases with minimal mis-classifications.

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue IV, April 2026

Fig. 2. ROC Curves Comparing All Models

The ROC curves comparing the three classifiers (Logistic Regression, Gradient Boosting, and Random Forest)

are presented in Figure 2. The Random Forest model achieved the highest area under the curve, indicating

superior discriminative ability across different classification thresholds.

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue IV, April 2026

as the most influential features, underscoring the importance of integrating both clinical and environmental data

predictive accuracy for fowl cholera infection status. It achieved an accuracy of 94.6%, with a precision of 93.8%,

a recall of 95.2%, and an impressive ROC-AUC score of 0.97. Gradient Boosting performed slightly lower,

attaining an accuracy of 93.2%, while Logistic Regression recorded the lowest accuracy at 89.5%. These results

indicate that ensemble-based approaches, particularly Random Forest, were more effective in capturing the

complex patterns in the dataset compared to linear models.

Analysis of feature importance from the Random Forest model revealed that environmental sanitation score was

the most influential predictor of fowl cholera infection status. This was followed by vaccination history and the

age of birds. Additional key predictors included mortality rate trends and respiratory symptoms, both of which

contributed significantly to the classification performance. These findings highlight the role of farm hygiene

practices, preventive vaccination programs, and observable clinical signs in predicting disease occurrence.

potential for integration into smart farm monitoring systems, allowing near real-time alerts to farmers. However,

State reduce generalizability. The lack of an independent test set also limits external validation. Future

expansions across diverse ecological and management contexts are necessary to ensure broader applicability.

Moreover, while three algorithms were tested, including XGBoost or deep learning architectures could enhance

feature learning and model interpretability in future research. A clearer articulation of preprocessing and feature

selection steps, as demonstrated in this revised methodology, contributes to methodological transparency and

reproducibility.

This study demonstrates that advanced data mining techniques can significantly improve diagnostic accuracy for

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,

MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue IV, April 2026

13. 13 T. Zheng, et al., “Development and validation of machine-learning models for behavior detection

in poultry,” Poultry Science, vol. 103, no. 10, Art. no. 103256, 2024. doi: 10.1016/j.psj.2024.103256.

14. 14 B. Martínez-López, J. M. Díaz-Cao, M. J. Clavijo, C. González-Crespo, and X. Liu, “Toward

precision veterinary epidemiology: Applications, challenges, and opportunities of digitalization and

the big-data revolution in livestock health,” Journal of the American Veterinary Medical Association,