Journal of Current Science and Technology

Accurate Air Quality Index Prediction Using MPRKDNN with Optimized Feature Selection

Jiss Kuruvilla — 2025-12-20

Air quality forecasting is essential for managing environmental and health impacts in rapidly urbanizing regions. The AQI short for Air Quality Index is a standardized measure used to communicate the severity of air pollution based on several pollutant indicators. However, accurately classifying AQI levels remains challenging due to the highly irregular nature of real-world datasets, which often include missing values, noise and redundant variables. Prediction accuracy largely depends on algorithmic complexity and the quality of input data preparation and refinement. To overcome these practical data-related limitations and improve AQI classification, a structured and adaptive model design becomes necessary. This study presents a modular and organized learning framework, Multivariate Piecewise Radial Kernelized Deep Neural Network (MPRKDNN), designed to enhance AQI classification through intelligent preprocessing and targeted feature selection. This process involves estimating missing values using Multivariate Piecewise Constant Interpolation (MPCI) and detecting outliers using the Tietjen-Moore statistical test. Radial Basis Kernelized Quadratic Discriminant Analysis (RBK-QDA) is used to retain the relevant variables and reduce dimensionality. The final output is fed into a deep feed-forward neural network trained using Stochastic Gradient Descent (SGD) for final classification.

The model is evaluated using multicity AQI datasets from India during 2017 to 2023. Comparative studies conducted against baseline deep learning and hybrid models show that MPRKDNN consistently improves classification accuracy, reduces RMSE, and maintains computational efficiency. These results emphasize the importance of integrating structured data preprocessing and kernel-based feature selection to enhance the robustness and interpretability of the AQI prediction system.

An Explainable Approach to Sentiment Analysis of Thai Hotel Reviews Using a Fine-Tuned Language Model and SHAP

Sabsathit Takaew — 2025-12-20

Sentiment analysis plays a pivotal role in the hotel industry, where user-generated reviews significantly influence customer decisions. However, traditional machine learning (ML) methods often struggle with the linguistic nuances of languages such as Thai. This study investigates the effectiveness of fine-tuning WangchanBERTa, a monolingual Thai large language model (LLM), for sentiment classification of hotel reviews from Bangkok. The model's performance was compared with ML algorithms, including extreme gradient boosting (XGBoost), support vector machines (SVM), logistic regression (LR), and multinomial naïve Bayes (MNB). The comparison highlights the advantages of deep contextual understanding enabled by transformer-based architecture. To improve interpretability, Shapley Additive Explanation (SHAP) was applied to the best-performing model to analyze feature importance. The results show that the fine-tuned LLM outperformed all ML models, achieving over 92% across all evaluation metrics (accuracy, precision, recall, and F1-score). SHAP analysis enhanced transparency by revealing sentiment drivers relevant to the hotel domain. This study contributes to the advancement of natural language processing (NLP) for low-resource languages by demonstrating the effectiveness of domain-specific fine-tuning combined with explainable artificial intelligence (XAI) in practical applications.

Effects of Short-term Drought Stress on Chlorophyll Fluorescence and Proline Content of Ficus annulata

Sirilak Nimnuan — 2025-12-20

Drought is still one of the key factors that directly affects the rate of photosynthesis and the reduction of plant growth and yield. This study was carried out to investigate the effect of drought and re-watering trials on the contents of chlorophyll, carotenoids, proline, CO₂ fluxes, and the photosynthetic efficiency of Ficus annulata. Treatments included control (no drought) and drought-stressed plants exposed to 21 days of drought followed by re-watering, with four replications conducted over 56 days. The results showed that drought stress greatly reduced the amounts of chlorophyll and carotenoids, with the highest reduction in relative water content (RWC) observed at 76–79% (p ≤ 0.05). Conversely, proline content significantly increased during drought stress, exhibiting the highest value of 108.16 µg/g FW before re-watering (p ≤ 0.05). A 21-day short-term drought had a statistically significant effect on changes in chlorophyll fluorescence parameters and CO₂ flux (p ≤ 0.05). However, the overall plant response after re-watering showed no significant difference compared with the control (p > 0.05), suggesting recovery of physiological efficiency. Our findings indicated that F. annulata has the capacity to mitigate carbon dioxide emissions. These physiological responses enhance the plant's suitability for drought resistance, and re-watering supports effective survival under drought stress.

Enhancing Cybersecurity in Industrial Internet of Things Systems Using Ensemble Learning Against False Data Injection Attacks

Saiprasad Potharaju — 2025-12-20

False Data Injection Attacks (FDIAs) pose critical threats to Industrial Internet of Things (IIoT) systems by manipulating sensor data to cause operational disruptions and safety hazards. Traditional intrusion detection systems struggle to identify the subtle anomalies characteristic of FDIAs, necessitating advanced machine learning approaches. This study develops a weighted voting ensemble framework integrating Random Forest, XGBoost, Neural Network, and Logistic Regression with F1-score-based dynamic weight assignment for optimized FDIA detection. The proposed ensemble was evaluated on the UKMNCT_IIoT_FDIA dataset containing 15,425 instances across 30 features. Using 70–30 train–test split, model performance was assessed through accuracy, precision, recall, F1-score, and confusion matrix analysis. Results demonstrate exceptional performance: 99.71% accuracy, 99.72% precision, 99.72% recall, and 99.72% F1-score. Confusion matrix analysis revealed only 2 false negatives and 9 false positives across 4,627 test instances, substantially outperforming individual classifiers while maintaining computational efficiency suitable for resource-constrained edge devices.

The weighted voting mechanism successfully leverages algorithmic diversity to achieve superior robustness compared to individual models. Tree-based ensembles (Random Forest: 99.74%, XGBoost: 99.68%) substantially outperformed Neural Network (87.14%) and Logistic Regression (83.32%), confirming the importance of non-linear modeling for complex attack pattern detection. The minimal false negative rate (0.04%) represents critical advancement for critical infrastructure protection where undetected attacks carry severe consequences. This research establishes the efficacy of performance-adaptive ensemble learning for IIoT cybersecurity, providing a practical, scalable solution for safeguarding industrial cyber-physical systems against evolving threats.

Potential Benefits of Boxing Training on Attenuating Arterial Stiffness, Heart Rate Variability and Motor Functioning in Chronic Ischemic Stroke Survivors

Natchaya Chondaen — 2025-12-20

Aerobic exercise has demonstrated benefits in improving arterial stiffness, cardiovascular autonomic function, and motor performance in individuals with stroke. While exergames such as the Wii boxing game have shown potential to enhance cardiovascular fitness, their accessibility may be limited due to technological or equipment constraints. As an alternative, a home-based boxing program may serve as a practical intervention to improve atherosclerosis-related outcomes and autonomic nervous system function in stroke survivors. This study aimed to investigate the effects of moderate-intensity boxing training on arterial stiffness, arterial obstruction, autonomic function, and motor impairments in individuals with chronic ischemic stroke. This study employed a single-cohort feasibility design and was conducted in a community-based setting in Mueang District, Phitsanulok, Thailand. It involved 12 stroke survivors with a mean post-stroke duration of 25.4 months. Participants engaged in 24 one-hour boxing sessions over an 8-week period. Outcome measures included the Cardio-Ankle Vascular Index (CAVI), Ankle-Brachial Index (ABI), Heart Rate Variability (HRV), and Fugl-Meyer Assessment (FMA), evaluated at baseline, 4 weeks, and 8 weeks. Significant improvements were observed in CAVI values on both sides, with reductions from 9.55 to 8.45 (right) and from 9.30 to 8.60 (left) (p = 0.001). HRV analysis showed enhanced autonomic function, with increases in LFnu from 32.75 to 41.96 and LF/HF ratio from 0.51 to 0.99 (p = 0.017). Motor performance, as measured by the FMA, significantly improved from 80 to 96 (p = 0.001), while ABI values remained unchanged. These findings suggest that an accessible, moderate-intensity boxing program may be an effective strategy for supporting vascular health, autonomic regulation, and motor recovery in chronic stroke rehabilitation.

High-Dimensional Quantum Key Distribution for Secure Healthcare Communication Systems: Integrating Internet of Medical Things, Electronic Health Records, and Smart Medical Gate

Dior Masrane Reoukadji — 2025-12-20

The emergence and advancement of technologies such as the Internet of Medical Things (IoMT), Electronic Health Records (EHR), and Smart Medical Gate (SMG) have remarkably changed patient care practices. With the digitization of healthcare services, concerns regarding data security have increased. These systems face increasing risks due to cyber threats and the advances of quantum computing technology. For instance, Peter Shor’s quantum algorithms are predicted to affect the integrity and confidentiality of sensitive medical data. This puts classical (non-quantum) cryptographic systems such as RSA (Ron Rivest, Adi Shamir, and Leonard Adleman) and ECC (Elliptic Curve Cryptography) at risk. This work proposes an integrated high-dimensional quantum key distribution (HD-QKD) infrastructure for secure medical data transmission across IoMT, EHR, and SMG ecosystems. It introduces a cloud-based Central EHR/Cloud Server for key management, along with edge Quantum Security Gateways. The system employs qudit encoding (d > 2) over a 50 km optical fiber link with 12-dB attenuation. Its edge-centric design ensures noise resilience and delivers a high information rate per photon. It also provides low-latency security against quantum threats while maintaining compatibility with existing fiber networks through wavelength division multiplexing. Simulations validate the system's potential, achieving secure key rates of 2.5 megabits per second between medical structures-double the rate of prior qubit-based Quantum Key Distribution (QKD) protocols-demonstrating superior scalability and performance for real-time healthcare applications.

In Situ Green Production of Silver Nanoparticles Utilizing Purple Corn Silk Extract for Multifunctional Healthcare Hemp Textiles

Pisutsaran Chitichotpanya — 2025-12-20

This study aimed to create multifunctional healthcare hemp fabrics employing a facile and cost-effective method. Multifunctional hemp was manufactured through in situ green synthesis of silver nanoparticles (AgNPs) employing an anthocyanin extract as both a reducing agent and functional colorant, due to the numerous health benefits linked to anthocyanins derived from purple corn silk (PCS), an agricultural byproduct. XRD and SEM-EDS analyses confirmed AgNP formation and uniform distribution on hemp fibers. The results demonstrated that dyebath pH significantly affected the perceived color, color strength (K/S), UV protection, and antioxidant and antibacterial activities. In an alkaline dyebath, more AgNPs were produced, improving K/S values, UV protection (UPF rating of 50+), and antibacterial efficiency against S. aureus and E. coli, with E. coli exhibiting better efficacy. However, an increase in AgNPs reduced the antioxidant capabilities of the treated fabrics. Overall, this study successfully demonstrated an economical and straightforward method for finishing hemp fabrics for multifunction healthcare textiles. PCS also contains a higher concentration of anthocyanins compared to other natural sources, rendering it an economical anthocyanin resource for textile businesses.

Occupational Stress and Associated Factors among Sugarcane Farmers in Sa Kaeo Province, Thailand

Sootthikarn Mungkhunthod — 2025-12-20

Sugarcane farming is physically demanding and exposes farmers to multiple stressors, increasing their risk of mental health problems. This study aimed to examine stress levels and identify the factors related to stress among sugarcane farmers in Sa Kaeo Province, Thailand. A cross-sectional study was conducted among 276 registered sugarcane growers in Sa Kaeo Province for at least one year using a multi-stage sampling method. Data were gathered through questionnaires between February and April 2025. Data were analyzed using descriptive statistics and multiple logistic regression analysis. The results indicated that most participants experienced high stress (49.3%), followed by moderate stress (25.0%) and severe stress (22.5%). Factors significantly associated with high stress among sugarcane farmers included experienced changes in climate variability, which showed the strongest association, followed by experiencing sugarcane product prices lower than last year, having moderate social support, incurring costs for cultivation of more than 10,000 Baht, lacking access to agricultural water sources, sleeping less than six hours, and sleeping six to seven hours (p < 0.05). These findings highlight the need for targeted interventions such as improved mental health services, agricultural subsidies, and reliable water access. Compared with previous Thai studies on farmer stress, this research provides new evidence by focusing on sugarcane farmers in Sa Kaeo Province, thereby contributing to more context-specific strategies for supporting farmer well-being.

Comparison of Phagocytic Activity of Phagocytes in Whole Blood against Live and Heat-treated at 100°C Saccharomyces cerevisiae

Wimol Chobchuenchom — 2025-12-20

Phagocytosis by neutrophils, eosinophils, and monocytes plays a critical role in inflammation through pathogen clearance and cytokine secretion. Saccharomyces cerevisiae is extensively used in food and beverage fermentation and typically undergoes heat treatment before consumption. While previous studies have examined phagocytic responses to yeast at baking temperatures (190°C), the effects of moderate heat treatment at 100°C commonly used in steaming and boiling remain poorly characterized. This study aimed to compare the phagocytic activity of neutrophils, eosinophils, and monocytes against live (RT) and heat-treated (ST; 100°C for 20 minutes) S. cerevisiae in human whole blood. Blood samples from 30 healthy volunteers (aged 19–24 years) with normal white blood cell counts were incubated with yeast suspensions (6.30 × 10⁴ cells/µL) for up to 30 minutes. Viability assessment confirmed 96% viability for RT and 0% for ST. All phagocyte types engulfed both RT and ST yeast, indicating recognition of both live and heat-killed cells. At 30 minutes, the phagocytic percentages for neutrophils, eosinophils, and monocytes against RT were 88.76%, 74.11%, and 64.85%, respectively, compared to 83.77%, 53.57%, and 29.08% against ST. Notably, neutrophils against ST showed significantly higher phagocytic indices than against RT at 20–30 minutes, suggesting enhanced ingestion efficiency per cell despite reduced overall activation. Heat treatment significantly decreased phagocytic activity in eosinophils and monocytes. Neutrophils demonstrated superior phagocytic activity compared to eosinophils and monocytes against both RT and ST at all time points. These findings suggest that heat treatment at 100°C alters S. cerevisiae cell wall integrity, differentially affecting phagocyte responses and potentially reducing the immunogenic potential of heat-processed yeast products.

Musculoskeletal Disorders (MSDs) and Ergonomic Assessment of Working Posture in Wood-Cutting Operators

Agung Kristanto — 2025-12-20

The industrial sector, particularly at the furniture SME center, relied heavily on human labor but lacked adequate infrastructure, leading to high rates of musculoskeletal disorders (MSDs) among workers. A preliminary study showed that 93% of workers experienced back pain, while significant percentages reported shoulder pain (67%), lower back pain (47%), leg joint pain (40%), and overall body aches (40%). These preliminary findings underscored the urgent need to assess the ergonomic risks specifically faced by wood-cutting workers. This study aimed to assess ergonomic risks for wood-cutting workers using the Quick Exposure Check (QEC) method and to examine the correlation between self-reported MSDs and ergonomic risk levels. A total of 60 workers from the wood-cutting workstation were surveyed. The independent variable was the workers' pain levels, measured using a modified SNQ-VAS questionnaire, while the dependent variable was the exposure level to injury risks in specific body parts, assessed using the QEC questionnaire. Chi-Square and Fisher's exact analyses were employed to explore the relationship between these variables. Results indicated that 88.33% of workers experienced ergonomic risks at action level 4, necessitating urgent intervention. Significant correlations were found, with MSDs in the upper and lower backs showing strong associations with QEC exposure levels (χ² = 3.965, p = 0.047 for the upper back; χ² = 4.044, p = 0.044 for the lower back). Similarly, correlations were significant for right shoulder and wrist MSDs in relation to QEC levels (χ² = 4.127, p = 0.042; χ² = 3.860, p = 0.049, respectively), as well as for neck MSDs correlated with neck exposure levels (χ² = 29.492, p = 0.033). These findings called for immediate ergonomic improvements to enhance the health and safety of workers at the SME Center.

Health Promotion Research at Rangsit University: A Scoping Review Using the HURS Framework for Institutional Policy Development

Duangnapha Bunsong — 2025-12-20

Universities play a critical role in advancing public health, as behavioral risk factors adopted by young adults, such as smoking, alcohol consumption, physical inactivity, and unhealthy diets contribute to the development of non-communicable diseases (NCDs) later in life. The Healthy University Rating System (HURS), developed by the ASEAN University Network–Health Promotion Network (AUN-HPN), provides a structured framework to evaluate and strengthen health promotion in higher education. This study aimed to systematically map health promotion research at Rangsit University using the HURS framework to identify strengths, gaps, and opportunities for institutional policy development.

A scoping review was conducted following Arksey and O’Malley’s methodology, with refinements by Levac and Peters. Publications in English and Thai (2015–2024) were retrieved from Scopus, ScienceDirect, CINAHL Plus, Thai Citation Index, ThaiLIS, and the Rangsit University Institutional Repository. Eligible studies addressed one or more of the 22 HURS domains. Twenty-seven studies were included, most of which were descriptive or exploratory. Research was concentrated in mental health, digital health, physical activity, and nutrition, while substance use prevention, sexual health, and environmental health were underrepresented. Limited policy translation was observed.

This review offers the first synthesis of health promotion research at Rangsit University aligned with the HURS framework. Findings highlight gaps and policy-relevant opportunities but should be interpreted cautiously, as this single-site case study has limited generalizability.

Weather Radar Utilization for Monitoring Volcanic Activity to Support Flight Safety: A Case Study of Mount Marapi Eruption, West Sumatra, 22 December 2023

Hesti Heningtiyas — 2025-12-20

Volcanic ash can cause damage to aircraft engines and endanger flight safety. Weather radar has the potential to detect the height and direction of eruption cloud distribution, as well as the type of volcanic material. This research aims to close the technological gap in volcanic activity observation and eruption-height detection of volcanic eruptions using weather radar, based on a case study of the Mount Marapi eruption in West Sumatra on 22 December 2023. The method used in this research is to process radar data to produce weather radar products, specifically the CMAX (Column Maximum) product to determine the pattern of eruption activity and multi-VCUT (Vertical Cut) product to describe the eruption intensity, pattern characteristics, height, and distribution direction. The data used is the BMKG weather radar data from the Minangkabau Meteorological Station, West Sumatra. Before data processing, Clutter Identification and Radar Data Quality Control were carried out to reduce observation bias caused by ground-echo clutter.

The analysis results of the weather radar data show multiple episodes of continuous and sporadic eruption activity from Mount Marapi during a single day of observation. These results are more detailed than the VONA (Volcano Observatory Notice for Aviation) reports based on visual observations. This provides an opportunity to develop a volcanic ash early-warning system that improves the accuracy and effectiveness of volcanic activity observations and enhances flight safety.

Nitrogen Use Efficiency of Maize Hybrids under Contrasting Nitrogen Levels in Post-Rice Field Conditions

Sirigul Bunphok — 2025-12-20

Growing maize under both high and low nitrogen conditions in post-rice fields is essential for understanding the factors influencing nitrogen fertilizer efficiency. This study aimed to identify maize hybrids with high nitrogen use efficiency (NUE) and stable performance across different nitrogen levels for potential use in low-input systems and breeding programs. A field experiment was conducted during the dry season of 2023/24 to assess six maize hybrid varieties. The experiment followed a randomized complete block design with two factors, including maize variety and nitrogen fertilizer level, with three replications. Data were collected on yield, yield components, and agronomic traits to assess the effects of nitrogen fertilizer, variety, and their interaction. Among the tested hybrids, Pac789, CP639, P4163, and DK9979C recorded the highest average yields. Under nitrogen-deficient conditions, P4546 exhibited strong tolerance, maintaining stable yield levels with minimal reduction. Under sufficient nitrogen conditions, CP639, P4163, Pac789, and DK9979C demonstrated high NUE values. Across nitrogen treatments, Pac789 displayed both a high shelling percentage and dark green foliage, indicating broad adaptability. Yield under high nitrogen conditions was positively correlated with NUE, while under low nitrogen conditions, yield was closely associated with the low nitrogen tolerance index and the nitrogen deficiency index. P4546 stood out as the most tolerant hybrid under nitrogen-limited conditions, showing only a slight yield reduction. These findings provide valuable insights for farmers selecting hybrids suited to varying soil fertility and nitrogen availability. Moreover, the results offer practical guidance for breeding programs aiming to develop maize varieties with improved NUE, contributing to more productive and sustainable maize cultivation in post-rice agroecosystems.

Encapsulation of Cinnamic Acid in Cellulose Acetate Hybrid Membranes via Electrospinning and Electrospraying: A Preliminary Study Toward Wound Dressing Applications

Supphakrit Noipan — 2025-12-20

Cinnamic acid (CN) was encapsulated in electrospun cellulose acetate (CA) nanofibers and the electrosprayed CA microparticles. The hybrid membrane (HM) was fabricated as a three-layer sandwich structure composed of electrospun CN-loaded CA nanofibers as the outer layers and the electrosprayed CN-loaded CA microparticles as the inner layer. The effects of CA concentration, type of solvent, and addition of CN on the morphology and sizes of either the electrospun nanofibers or the electrosprayed microparticles were investigated. The preliminary potential for using HM as wound dressings was investigated by comparing it with the electrospun nanofibers membrane (FM) and the cast films (CF). The release characteristics of CN from each type of membrane were investigated through total immersion and diffusion using Franz cell methods. For both methods, FM and HM exhibited greater CN released than CF. However, HM allowed more convenient CN release than FM, which contained only nanofibers. The mechanical properties in terms of tensile strength, Young’s modulus, and elongation at break of these membranes were investigated. The antioxidant activities of HM, as determined by the 1,1-diphenyl-2-picrylhydrazyl (DPPH) assay, was the highest among all membrane types. All membranes exhibited antibacterial activitie against Staphylococcus aureus (S. aureus), but not against Escherichia coli (E. coli). Interestingly, HM showed the highest antibacterial activity against S. aureus. FM and especially HM may be promising for future wound dressing applications.

Multi-Factor Model for Renal Risk Prediction in HIV Patients Initiating Dolutegravir: A Thai Secondary Hospital Cohort Study

Siriyaporn Wanitchakorn — 2025-12-20

Effective renal risk stratification for patients initiating dolutegravir (DTG) is crucial, as standard estimated glomerular filtration rate (eGFR) monitoring may be insufficient to prevent long-term kidney disease. This study aimed to develop and validate a multi-factor risk model to identify patients susceptible to a significant eGFR reduction (≥25%), thereby enabling targeted monitoring and preventive care. A retrospective cohort study at a Thai secondary care hospital analyzed 1,100 people with HIV who initiating DTG-based regimens between 2021 and 2023. The cohort was randomly divided into a training (n = 880) and a validation (n = 220) dataset. A multi-factor logistic regression model was developed, and its performance was compared against a model using only baseline eGFR. A significant eGFR reduction occurred in 17.82% of the cohort. Five independent risk factors were identified: age >40 years, BMI >23 kg/m², CD4 count <400 cells/µL, baseline eGFR <90 mL/min/1.73 m², and elevated alanine aminotransferase (ALT >40 U/L). The comprehensive 5-factor model demonstrated significantly better predictive performance (AUC 0.780; 95% CI 0.716 – 0.844) than the model using eGFR alone (AUC 0.651; 95% CI 0.571 – 0.730). Identifying at-risk patients is a critical first step in preventing long-term renal disease. This study provides a practical, novel renal risk prediction tool for DTG users, highlighting its clinical utility for proactive care. Future work should focus on prospective validation and integration into clinical workflows.

Microbial Ozone Decontamination of N95 Respirators: Efficacy and Material Preservation

Tanit Boonsiri — 2025-12-20

Ozone gas is a promising method for decontaminating personal protective equipment (PPE), providing broad antimicrobial activity with minimal residue effects. However, its effects on the structural integrity and filtration performance of N95 respirators are not well established. This study evaluated the antimicrobial efficacy of ozone treatment against Pseudomonas aeruginosa, Staphylococcus aureus, and Candida albicans on culture media and N95 respirators, and assessed whether fiber integrity and filtration efficiency were preserved using the GermZero3 prototype sterilizer developed with the National Science and Technology Development Agency (NSTDA). Microbial suspensions (10⁴ CFU/mL in TSB broth) were inoculated onto agar plates and respirator sections, and then exposed to 25–50 ppm ozone for 15–60 min. Viability was assessed by culture, while fiber integrity and filtration efficiency were evaluated by scanning electron microscopy and a NaCl aerosol test. Complete eradication of P. aeruginosa was achieved after 15 min and S. aureus within 45 min. C. albicans showed 99.90–99.98% reduction by 45–60 min, with no statistically significant difference from full clearance. When applied to contaminated respirators, ozone treatment eliminated all three pathogens after 60 min. Fiber morphology remained intact, and filtration efficiency was preserved at 99.99%, exceeding the ≥95% N95 standard. These findings support ozone treatment with the GermZero3 sterilizer as a safe and effective method for extending N95 respirator use during shortages.

An Ensemble Machine Learning Strategy for Accurate and Timely Prediction of Heart Disease

Tassaneeporn Tuion — 2025-12-20

Heart disease is a complicated disorder that is becoming increasingly common. Finding the best treatment plan for any heart disease patient requires early detection. The main goal of this study is to improve the prediction of heart disease by implementing ensemble learning techniques with machine learning (ML) models. The dataset used in this study, comprising 319,755 records from the Centers for Disease Control and Prevention, was downloaded from the Kaggle website. The SMOTE-ENN hybrid approach was used to address class imbalance. To increase the consistency and distribution of numerical variables, data preprocessing entailed standardization and the creation of dummy variables. The machine learning (ML) models Random Forest, Logistic Regression, Support Vector Machine, and Extra Trees were applied to the processed dataset without and with bagging, boosting, and stacking ensemble methods. Stacking, which combined SVM and Logistic Regression, outperformed the baseline models and other ensemble techniques, returning the highest recall score of 0.837731. This study underlines the significance of data balancing and ensemble learning for accurate forecasts based on medical datasets, which are typically large. The findings highlight how ML can enhance early diagnosis and intervention in the treatment of cardiac disease.