Rudy Herteno
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Rudy Herteno has a total of 81 co-authors. Here are the top co-authors by number of shared publications:
- Mohammad Reza Faisal
Lambung Mangkurat University
17 shared publications - Friska Abadi
Department of Computer Science, Faculty of Mathematics and Natural Science, Lambung Mangkurat University, Banjarbaru, Indonesia, Indonesia
14 shared publications - Radityo Adi Nugroho
Department of Computer Science, Faculty of Mathematics and Natural Science, Lambung Mangkurat University, Banjarbaru, Indonesia, Indonesia
13 shared publications - Setyo Wahyu Saputro
Department of Computer Science, Faculty of Mathematics and Natural Science, Lambung Mangkurat University, Banjarbaru, Indonesia, Indonesia
10 shared publications - Dodon Turianto Nugrahadi
Lambung Mangkurat University
9 shared publications - Irwan Budiman
Department of Computer Science, Lambung Mangkurat University, Banjarbaru, Indonesia
7 shared publications
Latest Publications
- Open Access
- 1 June 2026
- Published by Poltekkes Kemenkes Surabaya in Indonesian Journal of electronics, electromedical engineering, and medical informatics
- Vol. 8  (3) , 335-351
Depression Level Classification Using Compact Cross-Domain Feature Engineering on Sleep, Physical Activity, and Demographic Data
Depression is a common mental health disorder and a major public health concern, and early identification of depressive symptoms using population survey data can support exploratory risk analysis. However, many previous studies formulated depression prediction as a binary classification task. They used broad predictor sets, while multiclass depression-level classification with compact and interpretable cross-domain features remains less explored. This study developed a compact cross-domain feature engineering approach for classifying depression levels using sleep, physical activity, and demographic data from NHANES 2017–2018. A total of 5,068 respondents were included after preprocessing and PHQ-9 label construction. The target variable was divided into three classes: no-to-minimal depression, mild depression, and depression. Twenty raw predictors were transformed into 15 engineered features representing sleep patterns, sleep-related problems, physical activity, sedentary behavior, and interactions with age and income. Logistic Regression with class_weight = balanced was evaluated using stratified 5-fold cross-validation and compared with several baseline classifiers. The Final 15 FE Only scenario achieved an accuracy of 0.6215 ± 0.0091, macro F1-score of 0.4501 ± 0.0104, balanced accuracy of 0.5146 ± 0.0179, and depression-class recall of 0.6122 ± 0.0622. Compared with Raw Features, depression-class recall increased from 0.5360 ± 0.0541 to 0.6122 ± 0.0622, although the improvement was not statistically significant. These findings indicate that compact cross-domain features can improve sensitivity toward the depression class in an interpretable Logistic Regression setting, but overall predictive gains remain modest. The proposed model is more suitable for exploratory and population-level screening support rather than a stand-alone clinical diagnosis
- 1 June 2026
- Published by Poltekkes Kemenkes Surabaya in Indonesian Journal of electronics, electromedical engineering, and medical informatics
- Vol. 8  (3) , 335-351
Open AccessDepression Level Classification Using Compact Cross-Domain Feature Engineering on Sleep, Physical Activity, and Demographic Data
Depression is a common mental health disorder and a major public health concern, and early identification of depressive symptoms using population survey data can support exploratory risk analysis. However, many previous studies formulated depression prediction as a binary classification task. They used broad predictor sets, while multiclass depression-level classification with compact and interpretable cross-domain features remains less explored. This study developed a compact cross-domain feature engineering approach for classifying depression levels using sleep, physical activity, and demographic data from NHANES 2017–2018. A total of 5,068 respondents were included after preprocessing and PHQ-9 label construction. The target variable was divided into three classes: no-to-minimal depression, mild depression, and depression. Twenty raw predictors were transformed into 15 engineered features representing sleep patterns, sleep-related problems, physical activity, sedentary behavior, and interactions with age and income. Logistic Regression with class_weight = balanced was evaluated using stratified 5-fold cross-validation and compared with several baseline classifiers. The Final 15 FE Only scenario achieved an accuracy of 0.6215 ± 0.0091, macro F1-score of 0.4501 ± 0.0104, balanced accuracy of 0.5146 ± 0.0179, and depression-class recall of 0.6122 ± 0.0622. Compared with Raw Features, depression-class recall increased from 0.5360 ± 0.0541 to 0.6122 ± 0.0622, although the improvement was not statistically significant. These findings indicate that compact cross-domain features can improve sensitivity toward the depression class in an interpretable Logistic Regression setting, but overall predictive gains remain modest. The proposed model is more suitable for exploratory and population-level screening support rather than a stand-alone clinical diagnosis
- Open Access
- 9 May 2026
- Published by Poltekkes Kemenkes Surabaya in Journal of Electronics, Electromedical Engineering, and Medical Informatics
- Vol. 8  (3) , 1017-1032
Wavelength Configuration and Signal Duration for Low-Complexity PPG-Based Anemia Detection: A Preliminary Validation Study
Anemia remains a major global health problem, while standard diagnosis still depends on invasive hemoglobin testing, which may be less practical for repeated and resource-limited screening. Photoplethysmography (PPG) offers a potential non-invasive alternative, but the contribution of different wavelength configurations to anemia classification remains unclear. This preliminary subject-based validation study evaluated the effect of PPG wavelength configuration and recording duration on low-complexity anemia classification. A public dataset containing green, red, and infrared PPG recordings from 52 subjects was used, consisting of 42 normal and 10 anemia subjects. Eight morphological and temporal features were extracted from each wavelength. Seven signal configurations, namely Green, Red, IR, Green+Red, Green+IR, Red+IR, and all channels, were evaluated across 30, 45, 60, and 90 s recording durations. Support Vector Machine, Logistic Regression, Random Forest, and Extra Trees classifiers were trained using class-weighted learning and assessed with 5-fold subject-based cross-validation to reduce subject-level data leakage. The Red+IR configuration with a class-weighted SVM at 90 s achieved the best pooled performance, with a macro F1-score of 0.754, F1-Anemia of 0.588, anemia recall of 0.500, anemia precision of 0.714, accuracy of 0.769, and an error rate of 0.231. Fold-wise analysis showed substantial variability, with a macro F1-score of 0.617 ± 0.251, sensitivity of 0.467 ± 0.506, specificity of 0.846 ± 0.144, ROC-AUC of 0.864 ± 0.150, and PR-AUC of 0.694 ± 0.344. These findings suggest that adding more PPG wavelengths does not necessarily improve classification performance. However, the model still missed 5 of 10 anemia cases, and the limited anemia recall, small minority class, and demographic imbalance indicate that the results should be interpreted as preliminary and require validation on larger, more balanced datasets.
- 9 May 2026
- Published by Poltekkes Kemenkes Surabaya in Journal of Electronics, Electromedical Engineering, and Medical Informatics
- Vol. 8  (3) , 1017-1032
Open AccessWavelength Configuration and Signal Duration for Low-Complexity PPG-Based Anemia Detection: A Preliminary Validation Study
Anemia remains a major global health problem, while standard diagnosis still depends on invasive hemoglobin testing, which may be less practical for repeated and resource-limited screening. Photoplethysmography (PPG) offers a potential non-invasive alternative, but the contribution of different wavelength configurations to anemia classification remains unclear. This preliminary subject-based validation study evaluated the effect of PPG wavelength configuration and recording duration on low-complexity anemia classification. A public dataset containing green, red, and infrared PPG recordings from 52 subjects was used, consisting of 42 normal and 10 anemia subjects. Eight morphological and temporal features were extracted from each wavelength. Seven signal configurations, namely Green, Red, IR, Green+Red, Green+IR, Red+IR, and all channels, were evaluated across 30, 45, 60, and 90 s recording durations. Support Vector Machine, Logistic Regression, Random Forest, and Extra Trees classifiers were trained using class-weighted learning and assessed with 5-fold subject-based cross-validation to reduce subject-level data leakage. The Red+IR configuration with a class-weighted SVM at 90 s achieved the best pooled performance, with a macro F1-score of 0.754, F1-Anemia of 0.588, anemia recall of 0.500, anemia precision of 0.714, accuracy of 0.769, and an error rate of 0.231. Fold-wise analysis showed substantial variability, with a macro F1-score of 0.617 ± 0.251, sensitivity of 0.467 ± 0.506, specificity of 0.846 ± 0.144, ROC-AUC of 0.864 ± 0.150, and PR-AUC of 0.694 ± 0.344. These findings suggest that adding more PPG wavelengths does not necessarily improve classification performance. However, the model still missed 5 of 10 anemia cases, and the limited anemia recall, small minority class, and demographic imbalance indicate that the results should be interpreted as preliminary and require validation on larger, more balanced datasets.
- Open Access
- 5 April 2026
- Published by Poltekkes Kemenkes Surabaya in Indonesian Journal of electronics, electromedical engineering, and medical informatics
- Vol. 8  (2) , 248-261
Metaheuristic-Based Hyperparameter Optimization Analysis of Deep Neural Network for Cross-Project Defect Prediction in Mobile Applications
Software Defect Prediction (SDP) plays a strategic role in identifying software defects during the early stages of development, thereby enabling more efficient allocation of testing resources, particularly in the rapidly evolving mobile application domain characterized by fast release cycles. The commonly used Within-Project Defect Prediction (WPDP) approach is often constrained by the limited availability of historical data, especially in projects at early stages of development. As an alternative, Cross-Project Defect Prediction (CPDP) leverages historical data from other projects as training sources. Moreover, the performance of the Deep Neural Network (DNN) used in SDP is highly dependent on accurate hyperparameter configurations, where manual tuning requires substantial time and computational resources without guaranteeing optimal results. To address this issue, this study analyzes and compares the effectiveness of three metaheuristic algorithms, namely Particle Swarm Optimization (PSO), Genetic Algorithm (GA), and Grey Wolf Optimizer (GWO), in optimizing DNN hyperparameters within a CPDP framework. This study utilizes 14 open-source Android mobile application projects and employs the Leave-One-Out Cross-Validation technique. The performance of each combination is evaluated using ROC-AUC as the primary metric. The Wilcoxon Signed-Rank Test with a Bonferroni correction is used to assess the statistical significance of the observed performance differences. The experimental results demonstrate that GWO-DNN achieves the best performance, with an average ROC-AUC of 0.721, and is the only combination that remains statistically significant after Bonferroni correction, with a small effect size based on Cliff’s delta. Overall, the findings of this study indicate that metaheuristic-based hyperparameter tuning is a sufficiently effective approach for improving the capability of DNN in cross-project software defect prediction within the mobile application domain, although the observed improvements remain moderate.
- 5 April 2026
- Published by Poltekkes Kemenkes Surabaya in Indonesian Journal of electronics, electromedical engineering, and medical informatics
- Vol. 8  (2) , 248-261
Open AccessMetaheuristic-Based Hyperparameter Optimization Analysis of Deep Neural Network for Cross-Project Defect Prediction in Mobile Applications
Software Defect Prediction (SDP) plays a strategic role in identifying software defects during the early stages of development, thereby enabling more efficient allocation of testing resources, particularly in the rapidly evolving mobile application domain characterized by fast release cycles. The commonly used Within-Project Defect Prediction (WPDP) approach is often constrained by the limited availability of historical data, especially in projects at early stages of development. As an alternative, Cross-Project Defect Prediction (CPDP) leverages historical data from other projects as training sources. Moreover, the performance of the Deep Neural Network (DNN) used in SDP is highly dependent on accurate hyperparameter configurations, where manual tuning requires substantial time and computational resources without guaranteeing optimal results. To address this issue, this study analyzes and compares the effectiveness of three metaheuristic algorithms, namely Particle Swarm Optimization (PSO), Genetic Algorithm (GA), and Grey Wolf Optimizer (GWO), in optimizing DNN hyperparameters within a CPDP framework. This study utilizes 14 open-source Android mobile application projects and employs the Leave-One-Out Cross-Validation technique. The performance of each combination is evaluated using ROC-AUC as the primary metric. The Wilcoxon Signed-Rank Test with a Bonferroni correction is used to assess the statistical significance of the observed performance differences. The experimental results demonstrate that GWO-DNN achieves the best performance, with an average ROC-AUC of 0.721, and is the only combination that remains statistically significant after Bonferroni correction, with a small effect size based on Cliff’s delta. Overall, the findings of this study indicate that metaheuristic-based hyperparameter tuning is a sufficiently effective approach for improving the capability of DNN in cross-project software defect prediction within the mobile application domain, although the observed improvements remain moderate.
- Open Access
- 27 December 2025
- Published by Poltekkes Kemenkes Surabaya in Indonesian Journal of electronics, electromedical engineering, and medical informatics
- Vol. 8  (1) , 130-140
The Effect of Smote-Tomek on the Classification of Chronic Diseases Based on Health and Lifestyle Data
Machine learning models for chronic disease prediction are often trained on imbalanced healthcare datasets, where non-disease cases dominate. This condition can lead to misleadingly high accuracy while failing to identify patients with chronic diseases, limiting clinical usefulness. This study aims to analyze the impact of class imbalance on model performance and to evaluate the effectiveness of the SMOTE–Tomek resampling technique in improving chronic disease prediction. This research provides empirical evidence that accuracy alone is insufficient for evaluating healthcare models and demonstrates that imbalance-aware preprocessing is essential for valid and reliable chronic disease detection. Five classification models, such as Support Vector Machine, Random Forest, K-Nearest Neighbors, Gradient Boosting, and XGBoost, were evaluated on a lifestyle-based chronic disease dataset under two conditions: without resampling and with SMOTE–Tomek. Model performance was assessed using accuracy, precision, recall, F1-score, and AUC. Without SMOTE–Tomek, all models failed to detect chronic disease cases, producing near-zero recall and F1-scores despite accuracy exceeding 80%. After applying SMOTE–Tomek, substantial improvements were observed across all models, particularly in recall and AUC. Support Vector Machine achieved the best overall performance, with an accuracy of 92.9%, a precision of 92%, a recall of 93.9%, an F1-score of 0.93, and an AUC of 0.98. The findings confirm that handling class imbalance is a prerequisite for meaningful chronic disease prediction. The consistent increase in recall and AUC across all evaluated models confirms that the improvement stems from enhanced class separability rather than metric inflation. The proposed approach supports more reliable early screening and decision-support systems in preventive healthcare
- 27 December 2025
- Published by Poltekkes Kemenkes Surabaya in Indonesian Journal of electronics, electromedical engineering, and medical informatics
- Vol. 8  (1) , 130-140
Open AccessThe Effect of Smote-Tomek on the Classification of Chronic Diseases Based on Health and Lifestyle Data
Machine learning models for chronic disease prediction are often trained on imbalanced healthcare datasets, where non-disease cases dominate. This condition can lead to misleadingly high accuracy while failing to identify patients with chronic diseases, limiting clinical usefulness. This study aims to analyze the impact of class imbalance on model performance and to evaluate the effectiveness of the SMOTE–Tomek resampling technique in improving chronic disease prediction. This research provides empirical evidence that accuracy alone is insufficient for evaluating healthcare models and demonstrates that imbalance-aware preprocessing is essential for valid and reliable chronic disease detection. Five classification models, such as Support Vector Machine, Random Forest, K-Nearest Neighbors, Gradient Boosting, and XGBoost, were evaluated on a lifestyle-based chronic disease dataset under two conditions: without resampling and with SMOTE–Tomek. Model performance was assessed using accuracy, precision, recall, F1-score, and AUC. Without SMOTE–Tomek, all models failed to detect chronic disease cases, producing near-zero recall and F1-scores despite accuracy exceeding 80%. After applying SMOTE–Tomek, substantial improvements were observed across all models, particularly in recall and AUC. Support Vector Machine achieved the best overall performance, with an accuracy of 92.9%, a precision of 92%, a recall of 93.9%, an F1-score of 0.93, and an AUC of 0.98. The findings confirm that handling class imbalance is a prerequisite for meaningful chronic disease prediction. The consistent increase in recall and AUC across all evaluated models confirms that the improvement stems from enhanced class separability rather than metric inflation. The proposed approach supports more reliable early screening and decision-support systems in preventive healthcare
- Open Access
- 27 December 2025
- Published by Poltekkes Kemenkes Surabaya in Indonesian Journal of electronics, electromedical engineering, and medical informatics
- Vol. 8  (1) , 1-16
Comparative Study of Filter, Wrapper, and Hybrid Feature Selection Using Tree-Based Classifiers for Software Defect Prediction
Software defect prediction (SDP) is essential for improving software reliability by enabling the early identification of modules that may contain defects before the release stage. SDP commonly exhibits redundant or non-contributory metrics, underscoring the need for feature selection to derive a more informative subset. To address this problem, the present study investigates and compares the effectiveness of three feature-selection strategies: SelectKBest (SKB), Recursive Feature Elimination (RFE), and the hybrid SKB+RFE, in enhancing the performance of tree-based classifiers on the NASA Metrics Data Program (MDP) data collections. The study utilizes three classification algorithms, namely Random Forest (RF), Extra Trees (ET), and Bagging (Decision Tree), with Area Under the Curve (AUC) serving as the primary metric for assessing model performance. Experimental results reveal that the RFE and Extra Trees combination yields the top performance, producing an average AUC of 0.7855. This is subsequently followed by the SKB+RFE+ET configuration, which achieves an AUC of 0.7809, and SKB+ET at 0.7776. These findings demonstrate that iterative wrapper-based approaches such as RFE can identify more relevant and effective feature subsets than filter or hybrid strategies, with the RFE+Extra Trees configuration yielding the strongest overall predictive performance and wrapper-based methods exhibiting higher stability across heterogeneous datasets. Even without hyperparameter tuning and relying solely on class-weighting rather than explicit resampling techniques, the findings offer empirical insight into the isolated influence of feature selection on predictive performance. Overall, the study confirms that RFE combined with Extra Trees offers the strongest predictive performance on NASA MDP data collections and forms a foundation for developing more adaptive and robust models.
- 27 December 2025
- Published by Poltekkes Kemenkes Surabaya in Indonesian Journal of electronics, electromedical engineering, and medical informatics
- Vol. 8  (1) , 1-16
Open AccessComparative Study of Filter, Wrapper, and Hybrid Feature Selection Using Tree-Based Classifiers for Software Defect Prediction
Software defect prediction (SDP) is essential for improving software reliability by enabling the early identification of modules that may contain defects before the release stage. SDP commonly exhibits redundant or non-contributory metrics, underscoring the need for feature selection to derive a more informative subset. To address this problem, the present study investigates and compares the effectiveness of three feature-selection strategies: SelectKBest (SKB), Recursive Feature Elimination (RFE), and the hybrid SKB+RFE, in enhancing the performance of tree-based classifiers on the NASA Metrics Data Program (MDP) data collections. The study utilizes three classification algorithms, namely Random Forest (RF), Extra Trees (ET), and Bagging (Decision Tree), with Area Under the Curve (AUC) serving as the primary metric for assessing model performance. Experimental results reveal that the RFE and Extra Trees combination yields the top performance, producing an average AUC of 0.7855. This is subsequently followed by the SKB+RFE+ET configuration, which achieves an AUC of 0.7809, and SKB+ET at 0.7776. These findings demonstrate that iterative wrapper-based approaches such as RFE can identify more relevant and effective feature subsets than filter or hybrid strategies, with the RFE+Extra Trees configuration yielding the strongest overall predictive performance and wrapper-based methods exhibiting higher stability across heterogeneous datasets. Even without hyperparameter tuning and relying solely on class-weighting rather than explicit resampling techniques, the findings offer empirical insight into the isolated influence of feature selection on predictive performance. Overall, the study confirms that RFE combined with Extra Trees offers the strongest predictive performance on NASA MDP data collections and forms a foundation for developing more adaptive and robust models.
- Open Access
- 2 December 2025
- Published by Institute of Advanced Engineering and Science in Computer Science and Information Technologies
- Vol. 6  (3) , 283-293
Characteristics ransomware stop/djvu remk and erqw variants with static-dinamic analysis
Ransomware has developed into various new variants every year. One type of ransomware is STOP/DJVU, containing more than 240+ variants. This research to determine changes in differences characteristics and impact between ransomware variants STOP/DJVU remk, which is a variant from 2020, and the erqw variant from 2023, through a mixed-method research approach. Observation, simulation using mixing static and dynamic malware analysis methods. Both variants are from the Malware Bazaar site. The total characteristics based on dynamic analysis, the remk variant has 177, and the erqw variant has 190, which increased by 1.8%. The total characteristics based on static analysis, the remk variants have 586, and the erqw variants have 736, which increased by 5.7%. All characteristics from remk to erqw increasing in dynamic analysis, except the number of payloads that decreased about 20%. In static analysis, all characteristics from remk to erqw increase except the number of sections decreased about 1.5%. It can be the affected CPU performance, because the remk variant affects performance by increasing CPU work by 3.74%, while the erqw variant affects performance by reducing CPU work by 1.18%, both compared with normal CPU. which will affect the ransomware's destructive work and require changes in its handling.
- 2 December 2025
- Published by Institute of Advanced Engineering and Science in Computer Science and Information Technologies
- Vol. 6  (3) , 283-293
Open AccessCharacteristics ransomware stop/djvu remk and erqw variants with static-dinamic analysis
Ransomware has developed into various new variants every year. One type of ransomware is STOP/DJVU, containing more than 240+ variants. This research to determine changes in differences characteristics and impact between ransomware variants STOP/DJVU remk, which is a variant from 2020, and the erqw variant from 2023, through a mixed-method research approach. Observation, simulation using mixing static and dynamic malware analysis methods. Both variants are from the Malware Bazaar site. The total characteristics based on dynamic analysis, the remk variant has 177, and the erqw variant has 190, which increased by 1.8%. The total characteristics based on static analysis, the remk variants have 586, and the erqw variants have 736, which increased by 5.7%. All characteristics from remk to erqw increasing in dynamic analysis, except the number of payloads that decreased about 20%. In static analysis, all characteristics from remk to erqw increase except the number of sections decreased about 1.5%. It can be the affected CPU performance, because the remk variant affects performance by increasing CPU work by 3.74%, while the erqw variant affects performance by reducing CPU work by 1.18%, both compared with normal CPU. which will affect the ransomware's destructive work and require changes in its handling.
- Open Access
- 26 November 2025
- Published by Poltekkes Kemenkes Surabaya in Journal of Electronics, Electromedical Engineering, and Medical Informatics
- Vol. 8  (1) , 84-104
Comparative Analysis of YOLO11 and Mask R-CNN for Automated Glaucoma Detection
Glaucoma is a progressive optic neuropathy and a major cause of irreversible blindness. Early detection is crucial, yet current practice depends on manual estimation of the vertical Cup-to-Disc Ratio (vCDR), which is subjective and inefficient. Automated fundus image analysis provides scalable solutions but is challenged by low optic cup contrast, dataset variability, and the need for clinically interpretable outcomes. This study aimed to develop and evaluate an automated glaucoma screening pipeline based on optic disc (OD) and optic cup (OC) segmentation, comparing a single-stage model (YOLO11-Segmentation) with a two-stage model (Mask R-CNN with ResNet50-FPN), and validating it using vCDR at a threshold of 0.7. The contributions are fourfold: establishing a benchmark comparison of YOLO11 and Mask R-CNN across three datasets (REFUGE, ORIGA, G1020); linking segmentation accuracy to vCDR-based screening; analyzing precision–recall trade-offs between the models; and providing a reproducible baseline for future studies. The pipeline employed standardized preprocessing (optic nerve head cropping, resizing to 1024×1024, conservative augmentation). YOLO11 was trained for 200 epochs, and Mask R-CNN for 75 epochs. Evaluation metrics included Dice, Intersection over Union (IoU), mean absolute error (MAE), correlation, and classification performance. Results showed that Mask R-CNN achieved higher disc Dice (0.947 in G1020, 0.938 in REFUGE) and recall (0.880 in REFUGE), while YOLO11 attained stronger vCDR correlation (r = 0.900 in ORIGA) and perfect precision (1.000 in G1020). Overall accuracy exceeded 0.92 in REFUGE and G1020. In conclusion, YOLO11 favored conservative screening with fewer false positives, while Mask R-CNN improved sensitivity. These complementary strengths highlight the importance of model selection by screening context and suggest future research on hybrid frameworks and multimodal integration
- 26 November 2025
- Published by Poltekkes Kemenkes Surabaya in Journal of Electronics, Electromedical Engineering, and Medical Informatics
- Vol. 8  (1) , 84-104
Open AccessComparative Analysis of YOLO11 and Mask R-CNN for Automated Glaucoma Detection
Glaucoma is a progressive optic neuropathy and a major cause of irreversible blindness. Early detection is crucial, yet current practice depends on manual estimation of the vertical Cup-to-Disc Ratio (vCDR), which is subjective and inefficient. Automated fundus image analysis provides scalable solutions but is challenged by low optic cup contrast, dataset variability, and the need for clinically interpretable outcomes. This study aimed to develop and evaluate an automated glaucoma screening pipeline based on optic disc (OD) and optic cup (OC) segmentation, comparing a single-stage model (YOLO11-Segmentation) with a two-stage model (Mask R-CNN with ResNet50-FPN), and validating it using vCDR at a threshold of 0.7. The contributions are fourfold: establishing a benchmark comparison of YOLO11 and Mask R-CNN across three datasets (REFUGE, ORIGA, G1020); linking segmentation accuracy to vCDR-based screening; analyzing precision–recall trade-offs between the models; and providing a reproducible baseline for future studies. The pipeline employed standardized preprocessing (optic nerve head cropping, resizing to 1024×1024, conservative augmentation). YOLO11 was trained for 200 epochs, and Mask R-CNN for 75 epochs. Evaluation metrics included Dice, Intersection over Union (IoU), mean absolute error (MAE), correlation, and classification performance. Results showed that Mask R-CNN achieved higher disc Dice (0.947 in G1020, 0.938 in REFUGE) and recall (0.880 in REFUGE), while YOLO11 attained stronger vCDR correlation (r = 0.900 in ORIGA) and perfect precision (1.000 in G1020). Overall accuracy exceeded 0.92 in REFUGE and G1020. In conclusion, YOLO11 favored conservative screening with fewer false positives, while Mask R-CNN improved sensitivity. These complementary strengths highlight the importance of model selection by screening context and suggest future research on hybrid frameworks and multimodal integration
- Open Access
- 1 November 2025
- Published by Institute of Advanced Engineering and Science in Computer Science and Information Technologies
- Vol. 6  (3) , 283-293
Characteristics ransomware stop/djvu remk and erqw variants with static-dinamic analysis
Ransomware has developed into various new variants every year. One type of ransomware is STOP/DJVU, containing more than 240+ variants. This research to determine changes in differences characteristics and impact between ransomware variants STOP/DJVU remk, which is a variant from 2020, and the erqw variant from 2023, through a mixed-method research approach. Observation, simulation using mixing static and dynamic malware analysis methods. Both variants are from the Malware Bazaar site. The total characteristics based on dynamic analysis, the remk variant has 177, and the erqw variant has 190, which increased by 1.8%. The total characteristics based on static analysis, the remk variants have 586, and the erqw variants have 736, which increased by 5.7%. All characteristics from remk to erqw increasing in dynamic analysis, except the number of payloads that decreased about 20%. In static analysis, all characteristics from remk to erqw increase except the number of sections decreased about 1.5%. It can be the affected CPU performance, because the remk variant affects performance by increasing CPU work by 3.74%, while the erqw variant affects performance by reducing CPU work by 1.18%, both compared with normal CPU. which will affect the ransomware's destructive work and require changes in its handling.
- 1 November 2025
- Published by Institute of Advanced Engineering and Science in Computer Science and Information Technologies
- Vol. 6  (3) , 283-293
Open AccessCharacteristics ransomware stop/djvu remk and erqw variants with static-dinamic analysis
Ransomware has developed into various new variants every year. One type of ransomware is STOP/DJVU, containing more than 240+ variants. This research to determine changes in differences characteristics and impact between ransomware variants STOP/DJVU remk, which is a variant from 2020, and the erqw variant from 2023, through a mixed-method research approach. Observation, simulation using mixing static and dynamic malware analysis methods. Both variants are from the Malware Bazaar site. The total characteristics based on dynamic analysis, the remk variant has 177, and the erqw variant has 190, which increased by 1.8%. The total characteristics based on static analysis, the remk variants have 586, and the erqw variants have 736, which increased by 5.7%. All characteristics from remk to erqw increasing in dynamic analysis, except the number of payloads that decreased about 20%. In static analysis, all characteristics from remk to erqw increase except the number of sections decreased about 1.5%. It can be the affected CPU performance, because the remk variant affects performance by increasing CPU work by 3.74%, while the erqw variant affects performance by reducing CPU work by 1.18%, both compared with normal CPU. which will affect the ransomware's destructive work and require changes in its handling.
- Open Access
- 4 August 2025
- Published by Poltekkes Kemenkes Surabaya in Indonesian Journal of electronics, electromedical engineering, and medical informatics
- Vol. 7  (3) , 514-525
An Empirical Study of Cross-Project and Within-Project Performance in Software Defect Prediction Models Using Tree-Based and Boosting Classifiers
Software Defect Prediction (SDP) is a vital process in modern software engineering aimed at identifying faulty components in the early stages of development. In this study, we conducted a comprehensive evaluation of two widely employed SDP approaches, Within-Project Software Defect Prediction (WP-SDP) and Cross-Project Software Defect Prediction (CP-SDP), using identical preprocessing steps to ensure an objective comparison. We utilized the NASA MDP dataset, where each project was split into 70% training and 30% testing data, and applied three distinct resampling strategies—no sampling, oversampling, and undersampling—to address the challenge of class imbalance. Five classification algorithms were examined, including Support Vector Machine (SVM), Random Forest (RF), Gradient Boosting (GB), XGBoost (XGB), and LightGBM (LGBM). Performance was measured primarily using Accuracy and Area Under the Curve (AUC) metrics, resulting in 360 experimental outcomes. Our findings revealed that WP-SDP, combined with oversampling and Random Forest, demonstrated superior predictive capability on most projects, achieving an Accuracy of 89.92% and an AUC of 0.931 on PC4. Nonetheless, CP-SDP excelled in certain small-scale projects (e.g., MW1), underscoring its potential when local historical data is scarce but inter-project characteristics remain sufficiently similar. This study’s results underscore the importance of selecting a prediction scheme tailored to specific project attributes, class imbalance levels, and available historical data. By establishing a standardized methodological framework, our work contributes to a clearer understanding of the strengths and limitations of WP-SDP and CP-SDP, paving the way for more effective defect detection strategies and improved software quality.
- 4 August 2025
- Published by Poltekkes Kemenkes Surabaya in Indonesian Journal of electronics, electromedical engineering, and medical informatics
- Vol. 7  (3) , 514-525
Open AccessAn Empirical Study of Cross-Project and Within-Project Performance in Software Defect Prediction Models Using Tree-Based and Boosting Classifiers
Software Defect Prediction (SDP) is a vital process in modern software engineering aimed at identifying faulty components in the early stages of development. In this study, we conducted a comprehensive evaluation of two widely employed SDP approaches, Within-Project Software Defect Prediction (WP-SDP) and Cross-Project Software Defect Prediction (CP-SDP), using identical preprocessing steps to ensure an objective comparison. We utilized the NASA MDP dataset, where each project was split into 70% training and 30% testing data, and applied three distinct resampling strategies—no sampling, oversampling, and undersampling—to address the challenge of class imbalance. Five classification algorithms were examined, including Support Vector Machine (SVM), Random Forest (RF), Gradient Boosting (GB), XGBoost (XGB), and LightGBM (LGBM). Performance was measured primarily using Accuracy and Area Under the Curve (AUC) metrics, resulting in 360 experimental outcomes. Our findings revealed that WP-SDP, combined with oversampling and Random Forest, demonstrated superior predictive capability on most projects, achieving an Accuracy of 89.92% and an AUC of 0.931 on PC4. Nonetheless, CP-SDP excelled in certain small-scale projects (e.g., MW1), underscoring its potential when local historical data is scarce but inter-project characteristics remain sufficiently similar. This study’s results underscore the importance of selecting a prediction scheme tailored to specific project attributes, class imbalance levels, and available historical data. By establishing a standardized methodological framework, our work contributes to a clearer understanding of the strengths and limitations of WP-SDP and CP-SDP, paving the way for more effective defect detection strategies and improved software quality.
- Open Access
- 23 April 2025
- Published by Poltekkes Kemenkes Surabaya in Indonesian Journal of electronics, electromedical engineering, and medical informatics
- Vol. 7  (2) , 344-357
Analysis of the Effect of Feature Extraction on Sentiment Analysis using BiLSTM: Monkeypox Case Study on X/Twitter
The monkeypox outbreak has again become a global concern due to its widespread spread in various countries. Information related to the disease is widely shared through social media, especially Twitter which is a major source of public opinion. However, the complexity of language and the diverse viewpoints of users often pose challenges in accurately analyzing sentiment. Therefore, sentiment analysis of tweets about monkeypox is important to understand public perception and its impact on the dissemination of health information. This research contributes to identifying the most effective word embedding-based feature extraction method for sentiment analysis of health issues on social media. The purpose of this study is to compare the performance of word embedding methods namely Word2Vec, GloVe, and FastText in sentiment analysis of tweets about monkeypox using the BiLSTM model. Data totaling 1511 tweets were collected through a crawling process using the Twitter API. After the data is collected, manual labeling is done into three sentiment categories, namely positive, negative, and neutral. Furthermore, the data is processed through a preprocessing stage which includes data cleaning, case folding, tokenization, stopword removal, and stemming. The evaluation results show that FastText with BiLSTM produces the highest accuracy of 90%, followed by Word2Vec at 89%, and GloVe at 87%. FastText proved to be more effective in reducing classification errors, especially in distinguishing between negative and positive sentiments due to its ability to capture subword information and broader context. These findings suggest that the use of FastText can improve the accuracy of sentiment analysis, especially on health issues that develop on social media, so that it can support data-driven decision making by relevant parties in handling information dissemination.
- 23 April 2025
- Published by Poltekkes Kemenkes Surabaya in Indonesian Journal of electronics, electromedical engineering, and medical informatics
- Vol. 7  (2) , 344-357
Open AccessAnalysis of the Effect of Feature Extraction on Sentiment Analysis using BiLSTM: Monkeypox Case Study on X/Twitter
The monkeypox outbreak has again become a global concern due to its widespread spread in various countries. Information related to the disease is widely shared through social media, especially Twitter which is a major source of public opinion. However, the complexity of language and the diverse viewpoints of users often pose challenges in accurately analyzing sentiment. Therefore, sentiment analysis of tweets about monkeypox is important to understand public perception and its impact on the dissemination of health information. This research contributes to identifying the most effective word embedding-based feature extraction method for sentiment analysis of health issues on social media. The purpose of this study is to compare the performance of word embedding methods namely Word2Vec, GloVe, and FastText in sentiment analysis of tweets about monkeypox using the BiLSTM model. Data totaling 1511 tweets were collected through a crawling process using the Twitter API. After the data is collected, manual labeling is done into three sentiment categories, namely positive, negative, and neutral. Furthermore, the data is processed through a preprocessing stage which includes data cleaning, case folding, tokenization, stopword removal, and stemming. The evaluation results show that FastText with BiLSTM produces the highest accuracy of 90%, followed by Word2Vec at 89%, and GloVe at 87%. FastText proved to be more effective in reducing classification errors, especially in distinguishing between negative and positive sentiments due to its ability to capture subword information and broader context. These findings suggest that the use of FastText can improve the accuracy of sentiment analysis, especially on health issues that develop on social media, so that it can support data-driven decision making by relevant parties in handling information dissemination.