Journal of Emerging Technology and Digital Transformation

Synergistic Intelligence: A Stacking Ensemble Approach for Accurate and Scalable Diabetes Prediction

2026-01-06T05:26:35+00:00

Diabetes mellitus poses a rapidly escalating global health crisis, currently affecting over 537 million adults and demanding scalable, automated diagnostic solutions. However, current machine learning interventions often face critical bottlenecks, particularly model overfitting and poor generalization due to severe class imbalance in clinical datasets. To overcome these limitations, this study engineers a robust, clinically applicable Stacking Ensemble framework validated on the Pima Indians Diabetes Dataset. We employed a rigorous data preprocessing pipeline that utilizes the Synthetic Minority Oversampling Technique (SMOTE) to rectify class distribution, ensuring unbiased decision boundaries. By strategically integrating the complementary strengths of Logistic Regression, Support Vector Machines, K-Nearest Neighbors, and Naive Bayes via a meta-learning architecture, our approach successfully mitigates the individual weaknesses of single classifiers. The proposed ensemble demonstrated superior performance, achieving an accuracy of 81.5% and a critical recall rate of 84.0%, significantly reducing the risk of missed diagnoses compared to baseline models. Crucially, the system maintains exceptional computational efficiency with an inference latency of only 27.43 ms, confirming its viability for real-time deployment in resource-constrained medical environments. This research bridges the gap between algorithmic complexity and practical utility, offering a scalable, interpretable solution for early diabetes detection.

Accident Hotspot Prediction and Prevention Using Machine Learning

2026-01-23T07:37:18+00:00

In cities the rising rate of road accidents is one of the major threats to not only the safety of the citizens but also the overall effectiveness of the traffic system. Conventional reactive measures like post-incident review and blanket safety initiatives tend to be deficient in keeping accidents at bay prior to the crash. In an attempt to fill this gap, the present research offers a proactive data-driven approach built upon machine learning (ML) methods that aim at predicting the sites of accidents (the so-called hotspots). Historical crash reports, environmental factors (e.g., weather, lighting), time series (e.g., peak times, time of the year), and structural characteristics of road networks will be measured expressly to analyze their relations with the destination of the crashes. Traffic incident logs, weather archives, and geospatial road data will all be publicly available datasets which will be used to train and validate such ML models as Random Forests, Support Vector Machines (SVM), and Neural Networks. The evaluation of these will be on the measures of their prediction of high-risk areas in terms of their accuracy, precision and recall. The system can therefore facilitate the smooth running of traffic by allowing prompt responses before the incidence of accidents occur to assist the traffic authorities to maximize the patrol units, enact local safety measures, upgrade urban infrastructure, and eventually, minimize accidents in the metropolitan road systems.

A Scalable Intelligent Traffic Signal Framework Using IoT Sensing and Predictive ML Models

2026-01-16T03:57:13+00:00

Traffic congestion in the urban setting is one of the biggest remaining challenges of modern cities. Rapid increase in population, increasing number of vehicles and insufficient development of road infrastructure as a whole has worsened congestion problems, leading to increase in travel time, high fuel consumption and high emission level and decline in overall urban living standards. Traditional fixed-time traffic signal controllers are based on predetermined schedules and are unable to adjust to real-time fluctuations in traffic flow and pedestrian behavior and unforeseen events such as accidents or road maintenance. These limitations make them less effective and contribute to inefficient traffic management. To overcome these problems, a Smart Traffic Light Control System with the integration of Internet of Things (IoT) sensor networks and advanced Machine Learning (ML) algorithms for dynamic, data-driven signal optimization is proposed in this study. The system continuously gathers and processes the real-time traffic data from the IoT devices such as Vehicle Count sensors, RFID units, surveillance cameras, environment sensors. These different data streams enable an accurate and comprehensive picture of the traffic behavior through multiple junctions to be created. ML models intend to take both the past and recent real-time traffic datasets and learn to forecast the level of congestion, approximate queue lengths and pinpoint the peak-hour travel flow patterns. Supervised learning methods (such as Random Forest, Gradient Boosting, LSTM network) are employed for traffic forecast, and the reinforcement learning algorithms are used for dynamic adjustment of signal time and phase time. This hybrid method of ML allows the system to self-adapt to ever-changing road conditions and distribute green light intervals in a more efficient way. The proposed model will strive to minimize traffic delays and increase the vehicle throughput, as well as road safety and the impact on the environment. It has a scalable architecture that enables it to interface with the current Intelligent Transportation Systems (ITS) and be extended to cover large urban networks. Experiments using simulation, backed up by actual traffic data, indicate significant betterment in average waiting time and signal responsiveness and movement of the whole traffic as compared to conventional fixed-time and actuated signal systems. These findings validate the hypothesis that the traffic signal management system based on the IoT and ML can introduce a ground-breaking solution to the optimization of urban mobility and congestion and facilitate the evolution and actualization of smarter and more sustainable cities.

AI-Driven Recruitment and Hiring Automation through CV Scanning and AI Agents

2026-01-06T05:32:40+00:00

The presented work is dedicated to the development
of an AI agent that will perform automated CV sorting and
evaluation to assist current hiring processes with a supervised
machine learning model. A Random Forest Regressor containing
300 trees was trained on structured features based on the
unstructured CV information including years of experience,
number of related skills, education level, overlap of skills with the
job description, and coverage ratio. The 50 actual CVs dataset
was divided to 80 percent training set and 20 percent testing
set and the ground truth target used in supervised learning
was HR inspired rule based scoring. The model realized an
impressive predictive power, as seen with the R² of 0.94 and Mean
Absolute Error of 4.46 which signifies a significant agreement
with human based ratings. The trained model is integrated to
an AI agent that automates the ranking of the candidates, short
listing and scheduling of interviews. The system is particularly
effective at unloading HR staff and increasing the efficiency of
the recruitment process; nevertheless, some of the key issues are
the bias in the training data, the tendency of the system to treat
different candidates with unequal consideration, and the limited
interpretability of the model. The study highlights the importance
of introducing feedback loops, reducing bias, and establishing
transparency to ensure that AI-based recruitment systems were
ethically, reliably, and humanly deployed.

Energy-Efficient Resource Management in IoT Networks through Federated Learning

2026-01-23T07:50:01+00:00

The widespread expansion of Internet of Things networks poses inconvenient power consumption challenges and demands better resource utilization since the regular IoT devices have low power supplies and processing power. The traditional systems are not effective in a centralized system as they bring about a lot of communication issues as well as the threat to privacy and less scalability. One of the suggested approaches to enhancing the efficiency of the IoT network in terms of energy consumption and appropriate performance results is a FL-based optimization framework. The suggested solution applies edge computer systems and local data processing that reduces information transfer costs as well as amounts of power consumption. The study is aimed at creating an energy-efficient predictive resource management model that integrates FL lightweight algorithms and optimization techniques to optimize the factors of performance of the IoT network in real-time, such as power consumption and command speed and rational processing. The study will be done by simulation analysis using real world data and federated architectural systems to ensure its performance standards. The proposed achievement in the form of an intelligent sustainable IoT ecosystem that does not require the human touch to function efficiently is there.

A Machine Learning–Driven Framework for Early Detection and Classification of Tomato Leaf Diseases to Enhance Agricultural Productivity and Crop Health

2026-01-23T07:26:53+00:00

Tomato is a global important horticultural crop whose yield and quality are severely affected by bacterial, fungal and viral pathogens that incite foliar diseases. As such, early and accurate diagnosis of such ills is critical for crop management. Traditional methods for diagnosis fell to manual inspection and laboratory analysis are burdensome and time-consuming and impractical in large-scale and resource-constrained agricultural environments. Although recent efforts in deep learning and computer vision have led to automated diagnosis of plant disease, a lot of current approaches are based on laboratory-curated datasets and lack robustness, interpretability or deploy ability under real state conditions. This manuscript proposes a complete framework based on deep learning for the early detection and classification of tomato leaf diseases which simultaneously addresses the problem of accuracy, generalization, explainability, and deployment feasibility. The system exploits transfer learning using state-of-the-art convolutional neural network architectures such as EfficientNetB4, ResNet50, InceptionV3 and MobileNetV3 refined with a combination of laboratory and acquired image datasets collected in field. To counter-act the class imbalance and environmental variability we use plenty of data augmentation, normalization and regularization protocols. The models are evaluated based on a set of stringent performance results such as accuracy, precision, recall, F1-score and AUC. Experimental results show that our model which is effectively based on the EfficientNetB4 model outperforms the competing models with an accuracy of classification ranging from 96 percent to 99 percent, an eurointiention range of almost 0.99, while at the same time ensuring a robust generalization of the results under field-like conditions. Lightweight architectures like MobileNetV3 also help in enabling real-time inference on edge devices making the system practical. In sum, the proposed framework presents a solution that is scalable and interpretable and which can be easily deployed to serve as a solution for precision agriculture in favor of improved disease management, crop resilience fortification and sustainable tomato production.

Smart ATS: An AI-Driven Multi-Stage Resume Scoring and Recruitment Automation System

2026-01-06T05:40:39+00:00

An artificial intelligence-powered Applicant Track-
ing System (ATS) that uses a multi-step algorithmic pipeline
to handle candidate scoring, skill finding, experience analysis,
and resume extraction. The Sentence-BERT model (allMiniLM-
L6-v2) for job-description similarity, RapidFuzz for fuzzy skill
matching, canonical skill-mapping algorithms, and a determin-
istic experience-scoring model power the system’s hybrid scor-
ing architecture. Using weighted evaluation characteristics such
as skill relevance, experience alignment, LLM-based semantic
matching, and penalty adjustments for underqualification or
overqualification, the proposed ATS calculates a normalised 0–10
score. Experimental review on a dataset of over 40 resumes
demonstrates a screening accuracy improvement of over 88%
when compared to manual evaluation methodologies, significantly
reducing HR workload and producing consistent and intelligible
applicant rankings.