An Explainable Heterogeneous Stacking Ensemble Framework for Fish Health Prediction in Intelligent Aquaculture

Explainable machine learning framework for foodborne disease outbreak prediction in Eastern Province, Saudi Arabia: case study

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).

A Deep Learning Architecture for Smart Fish Farm Management and Early Mortality Prediction

The rapid growth of aquaculture has intensified the need for intelligent, data-driven systems capable of improving fish farm productivity, sustainability, and operational efficiency. Traditional manual management approaches remain limited by labor intensity, inconsistent monitoring, and delayed decision-making, especially in dynamic aquatic environments. This study presents a Deep Learning–based Fish Farming Management System that integrates Internet of Things (IoT) sensing technologies with a Bidirectional Long Short-Term Memory (BiLSTM) network to provide real-time monitoring, predictive analytics, and intelligent decision support. IoT-enabled sensors continuously capture water quality and behavioral parameters, which are transmitted to a centralized data-processing architecture for storage, analysis, and automated alerts.  The BiLSTM model primarily focuses on fish mortality classification, where it analyzes temporal patterns in environmental and behavioral data to predict mortality risk with high reliability. In addition, the system supports auxiliary predictive and recommendation functions, including anomaly detection and rule-assisted recommendations for feeding, aeration, and environmental adjustments, which are derived from learned temporal trends and domain constraints rather than independently optimized predictive models. Experimental results for the mortality classification task demonstrate outstanding performance, with the model achieving 96.55% accuracy, 96.97% precision, 96.67% recall, a Cohen’s Kappa score of 0.9482, and an average AUC of 0.9947, significantly outperforming benchmark models.

An optimized stacking ensemble framework with multi-layer explainable AI for soil heavy metal contamination classification

Multi-Modal Soil Health and Crop Productivity Prediction using Hybrid Deep Learning

The current study describes an advanced hybrid multi-modal approach that will be used in predicting soil health and crop productivity through the application of several types of deep learning models and machine learning approaches to enhance predictions. In the current work, a hybrid model is considered, and three kinds of pre-trained CNN, known as ResNet50, VGG16, and Mobile Net, are used to extract the features of soil. The pre-fetched features are subsequently processed through multiple fully connected layers with SoftMax activation functions for multi-class types of soil classification, namely clay, sandy, and loam. To enhance the generalization and robustness of the models, data augmentation and normalization techniques are applied to the data. The models are optimized with Adam and SGD optimizers using categorical cross-entropy parameters for the loss function. The results of the experimentation show high accuracy and robust performance for the precision-recall and F1-score metrics. The model would highly benefit precision agriculture applications to facilitate autonomous soil type detection, and to enable farmers and growers to better inform crop selection and nutrient management for sustainable yield optimization models in smart farming.

Multi-Plant Disease Classification using ResNet50-based Deep Ensemble Learning Framework

Agricultural productivity and food security are heavily impacted by plant diseases, and thus there is a high demand for accurate and automated plant disease detection that can be achieved by applying deep learning techniques. This research proposes a Multi-Model Ensemble Method Based on Deep Learning for multi-plant disease detection using ResNet50 to improve classification performance across multiple crop varieties. The proposed framework takes five important categories of plants into consideration including tomato, potato, grape, apple and maize, and 10 classes of healthy and diseased plants are generated from the PlantSeg dataset. The Anaconda platform was used along with Python to create a development environment that allows data preprocessing, augmentation, training and testing to be implemented efficiently. The proposed ensemble framework combines the feature extraction power of ResNet50 with several deep learning classifiers so as to obtain a good identification performance at different resolutions and environments. The proposed model performance is tested with the following metrics Accuracy, Precision, Inference Time, and Resolution quality and compared with MobileNetV2, YOLOv8 and the baseline CNN models. Experimental results show that the proposed ensemble ResNet50 framework achieves an accuracy of 98.7% and precision of 98.3%, which is about 6.4%, 4.8%, and 9.2% higher than that of MobileNetV2, YOLOv8, and CNN respectively. Moreover, the proposed method achieves high resolution disease localization capability with an inference time improvement of almost 18% compared with YOLOv8. The proposed system greatly improves the detection accuracy of the early stage and the calculation speed of the system, which is very suitable for smart agriculture applications and real-time monitoring of the health status of crops.

A smart water quality companion for farmers: predicting canal suitability for aquaculture and livestock using validated machine learning models and web based tool

In recent times, global food production has transitioned towards aquaculture and livestock farming owing to the contamination of agricultural crops with harmful fertilisers and pesticides. Quality of intake water plays a crucial role in the growth and wellbeing of the aquatic creatures and livestock. This study aims to develop and validate a machine-learning-driven framework and an open access web tool that integrates a diurnal water quality index with optimised predictive models for aquaculture and livestock applications in the deltaic region of western Godavari, India. Critical parameters were analysed for diurnal samples over two seasons on a semi-perennial canal. From the water quality data using supervised machine learning techniques like linear regression (LR), random forest (RF), decision tree (DT), long short-term memory (LSTM), XGBoost, support vector regression (SVR), and artificial neural networks (ANN), water quality prediction models were developed and compared for model accuracy and feature parameter influence on aquaculture water quality index (AWQI) and livestock water quality index (LWQI). The results indicated that the regression and classification models of LR and SVR showed the highest accuracy when applied to aquaculture ( R 2 : 0.999, MSE: 0.0004 & R 2 : 0.999, MSE: 0.006) and livestock ( R 2 : 0.999, MSE: 0.00001 & R 2 : 0.999, MSE: 0.0043) water quality datasets. This study offers more practical benefits to aquatic and livestock farmers, field officers and experts by enabling dual water quality analysis and prediction through an open-access web tool for instantaneous, quick decision-making on water quality for their intended use.