Nov 2026· Computer Science and Information Technology· 0 citations· 24 references
TL;DR
A checkpoint-based multimodal transformer-convolutional neural networks framework for early flock-level E. coli infection risk prediction using environmental, production, behavioural, and visual poultry data is proposed, demonstrating the potential of multimodal artificial intelligence (AI) for proactive poultry health monitoring.
Abstract
Poultry farming remains an important contributor to global food security and commercial livestock production. However, infectious diseases such as Escherichia coli (E. coli) cause mortality, poor feed efficiency, reduced growth performance, and economic losses in broiler production systems. This study proposes a checkpoint-based multimodal transformer-convolutional neural networks (CNN) framework for early flock-level E. coli infection risk prediction using environmental, production, behavioural, and visual poultry data. Flock monitoring records collected from 2022 to 2025 were structured across six production checkpoints: day 3, day 7, day 14, day 21, day 28, and day 31. After long-format conversion, approximately 90,000 temporal observations were used for transformer modelling, with 72,000 records for training and 18,000 for testing. The CNN component evaluated 249 poultry images across healthy, low-risk, medium-risk, high-risk, and non-broiler classes. The transformer model achieved 99.96% accuracy, while the CNN model achieved 95.58% accuracy. The integrated dashboard generated flock risk scores, contributing factors, alerts, gradient-weighted class activation mapping (Grad-CAM) explanations, and veterinary advisory recommendations, demonstrating the potential of multimodal artificial intelligence (AI) for proactive poultry health monitoring.
Poultry farming is a critical component of global food security, yet it is highly vulnerable to fast spreading infectious diseases such as coccidiosis, salmonellosis and Newcastle disease. Conventional diagnostic pathways depend on clinical examination and laboratory tests, which are time consuming, expertise dependent and often inaccessible for small and medium scale farmers. This paper presents a combined survey and implementation of PoultryGuard, a smartphone-centric framework for early poultry disease detection from chicken fecal images using convolutional neural networks. First, recent progress in image-based poultry disease detection is reviewed, with emphasis on fecal-image classification and transfer learning strategies. Then, the design and implementation of an EfficientNet-B3 based classifier trained on 8,067 annotated fecal images across four classes (healthy, coccidiosis, salmonella and Newcastle disease) is described. The model is deployed inside a Streamlit application that integrates user authentication, real time image acquisition, a structured veterinary knowledge base, automated diagnostic report generation in PDF format, Telegram based emergency alerts, a veterinary hospital locator and an admin analytics dashboard. Experimental evaluation on a heldout test set shows that the implemented EfficientNet-B3 model achieves 93.73% accuracy, with macro-averaged precision, recall and F1-scores above 0.93, while maintaining low inference latency suitable for smartphone assisted deployment. The study shows that poultry disease management in resource constrained farm environments can be revolutionized by combining deep learning with clinical decision support and communication modules.
S. K, D. S, Deeksha V Panchal et al.· 2026 International Conferenc...· 0 citations
Highly pathogenic avian influenza (HPAI) remains a recurrent threat to poultry production and One-Health surveillance in India. We developed a national relative spatial risk map for India using curated outbreak records spanning January 2006 to April 2024 (predominantly HPAI H5N1 and H5N8), buffered pseudo-absence sampling, H3 resolution-7 hexagons, and 94 environmental, livestock, land-cover, and anthropogenic predictors. Under 5-fold spatial block cross-validation, eight base classifiers were trained and all performed above chance (AUC > 0.76). The Gaussian-process stacked meta-learner achieved the highest AUC (0.853), but the improvement over the strongest individual base learner, Random Forest (AUC 0.851; Brier 0.155; ECE 0.079), was small and statistically non-significant. Its principal added value was a companion uncertainty layer, the GPR posterior standard deviation, which showed internal consistency with ensemble disagreement across base models (r = 0.74, p < 0.001). Feature attribution ranked human population density, extensive chicken density, June precipitation, and seasonal humidity variables among the predictors most associated with model outputs, with interpretation constrained by passive-surveillance bias and multicollinearity. The resulting relative spatial risk surface, prediction-uncertainty surface, and subdistrict risk-uncertainty classification layers identify eastern, northeastern, coastal, and selected southern regions as priorities for targeted surveillance and prospective validation.
Satish Gaikwad, T. R. Arun, Basavaraj Shrinivasa et al.· Frontiers in Veterinary Scie...· 0 citations
The sustainable development of the poultry industry is constrained by frequent disease outbreaks and delayed clinical diagnosis, while conventional disease-control approaches based on manual experience are inadequate for precise management in large-scale farming. This study proposes an intelligent poultry-disease early-warning system based on deep-learning-driven fecal image recognition. Fecal morphological features are used as key indicators for early disease detection, and a dataset containing 10,548 valid samples across five categories—healthy, coccidiosis, Newcastle disease, infectious bursal disease, and salmonellosis—is constructed in collaboration with large-scale poultry farms in Guangdong Province. EfficientNet-B3 is adopted as the backbone network, and a Convolutional Block Attention Module (CBAM) and lightweight Lite-FPN multi-scale feature fusion structure are embedded to enhance fine-grained lesion recognition. The system integrates edge computing, industrial imaging, and wireless sensing for real-time deployment in poultry houses. Results show that warning accuracy reaches 91.6%, system availability reaches 99.94%, and the mortality-and-culling rate decreases by 22.6%, demonstrating the engineering feasibility and practical utility of the proposed system.
Precision livestock farming represents a key technological alternative to address the health, production, and environmental challenges of the beef cattle sector. This systematic review analyzes the use of emerging technologies, both hardware and software, focused on cattle health management. The PRISMA methodology and the Parsifal tool were applied to identify, select and evaluate studies published between 2020 and 2025, related to sensors, communication systems, artificial intelligence algorithms and digital platforms. The results show that technologies such as biometric sensors, GPS devices, thermal cameras and IoT communication networks have improved early disease detection, behavioral monitoring and livestock welfare. Likewise, the high performance of artificial intelligence models such as CNN, LSTM, SVM and Random Forest in the prediction of physiological events is highlighted, with efficiencies above 95%. Although significant benefits in efficiency and sustainability are observed, barriers related to technological adoption, connectivity and costs are identified. Finally, the development of scalable and adaptable solutions to diverse rural contexts is proposed as a future line, thus consolidating precision livestock farming as a key model for the sustainability of the livestock sector.
Miller Ruales Luna, John Barco Jiménez, Camilo Eraso Guerrero et al.· Archivos Latinoamericanos de...· 0 citations
The occurrence of foodborne diseases is a considerable public health issue, especially in areas that are quickly becoming urbanized with intricate food delivery systems. In this paper, we present a machine learning-based model for predicting outbreaks, explainability, and spatial risk propagation, validated through a multiyear data set of an epidemiological nature from 12 cities in the Eastern Province of Saudi Arabia (2021–2025). The final data set includes 61 cases and 13 engineered features. In the current research, the proposed architecture uses XGBoost to predict outbreaks, alongside using the random forest for predicting severity and support vector machine (SVM) for comparisons. The XGBoost classifier demonstrates an evenly balanced performance (accuracy = 0.85, precision = 0.78, recall = 0.78) on the testing set. Due to the size of the dataset, the results are provided with the estimation of uncertainty (rather than the exact numbers). Using leakage-safe repeated stratified cross-validation, the mean AUC equals 0.64 [95% interval = (0.20, 1.00)], and the leave-one-year-out validation method is not stable (mean AUC 0.47). Differences between the models (e.g., better single-split cross-validation AUC for SVM) are within confidence intervals. Interpretability is improved by using the SHAP framework to measure feature importance, which shows that the main factors are hospitalization and the severity of symptoms. The graph module also helps in understanding the propagation of disease risk between cities, highlighting the importance of well-connected metropolitan areas as disease hubs. Moreover, the use of a locally deployed Mistral LLM makes the generated explanations more readable. The findings show that our approach presents an appropriate balance of predictiveness, interpretability, and spatial knowledge. With only 61 data points and 11 outbreaks reported, this research is clearly not meant to be an early warning system, but rather a proof-of-concept on how one might be designed. In order to ensure reproducibility, the preprocessing pipeline and synthetic dataset generator have been made available to the community.
N. F. Saleem ALAnsary, Mahmood Berekaa, Raghad Alhotheyfa et al.· Frontiers in Public Health· 0 citations
Aquaculture is a vital element of global food security and economic sustainability. The changing environment from time to time and the rapid spread of diseases in aquaculture leading to huge economic losses, still pose a major challenge in the maintenance of healthy fish. Timely intervention for rapid and accurate fish health forecasting and diagnosis will decrease the fish mortality and increase the efficiency of fish production. In the current paper, a stacking model using heterogenous ensemble learning will be developed using three machine learning algorithms including the Random Forest algorithm, Support Vector Machine (SVM) and XGBoost together with Logistic Regression as the meta-classifier to predict fish health. Data preprocessing stage should be conducted prior to developing a model which will include missing value treatment, feature scaling, feature encoding and irrelevant feature elimination. The efficiency and generality of the proposed model will be estimated via using stratified 10-fold cross-validation technique. The quality of the model will be estimated by using the criteria including accuracy, precision, recall, F1 Score, confusion matrix and ROC curve analysis. In addition, SHapley Additive exPlanations will be applied to increase the explainability of the proposed model by showing the contribution of each environmental and biological factor into predicting fish health. The experimental results reveal that the heterogeneous stacking model can outperform the individual base learners and provide a transparent decision explanation. The framework is proposed to be utilized for intelligent fish health monitoring and contributing to sustainable aquaculture through predicting diseases in time and managing the aquaculture farms effectively, interpretable and scalable way.
Ranjan Kumar, Savita Choudhary· Journal of Intelligent Decis...· 0 citations
