Betül Uzbaş
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Betül Uzbaş has a total of 21 co-authors. Here are the top co-authors by number of shared publications:
- Gizemnur Erol Doğan
Konya Teknik Üniversitesi
5 shared publications Ahmet Arslan
Enelsis Industrial Electronic Systems Research and Development Co. Ltd
1 shared publication- Ahmet Arslan
Department of Computer Engineering, Faculty of Engineering, Selçuk University, 42250, Konya, Turkey
1 shared publication - Ahmet Babalık
Department of Computer Engineering, Division of Computer Sciences, Faculty of Engineering and Natural Sciences, Konya Technical University, Konya, Türkiye
1 shared publication - Ayse Merve Acılar
Necmettin Erbakan Universitesi
1 shared publication - Beyza Kalcı
Konya Teknik Üniversitesi
1 shared publication
Latest Publications
- Open Access
- 9 August 2026
- Published by Pamukkale University in Pamukkale University Journal of Engineering Sciences
- No. Advanced O
Sex determination from cephalometric radiography using data mining
Context—The skull is a crucial reference structure in skeletal analysis due to its hard and durable nature, which enables it to resist environmental degradation and preserve its morphological characteristics over extended periods. Because of these properties, craniofacial structures are widely used in forensic science, anthropology, and medical studies for identification purposes, including sex determination.Objective—This study aims to perform sex determination using cephalometric radiography (CR) and to evaluate the effectiveness of statistical and correlation-based feature selection methods in improving classification performance. Additionally, the study investigates whether a reduced set of informative cephalometric features (CFs) can achieve comparable accuracy to the full feature set, thereby increasing model efficiency and interpretability.Method—A dataset was created by extracting the and coordinates of 19 cephalometric landmarks (CLs) from CR images of 300 patients (150 females and 150 males). Based on these landmarks, 8 Euclidean distance (8-ED), 3 angular (3-A), and 3 frontal sinus (3-FS) measurements were calculated, resulting in a total of 14 CFs. Initially, statistical analyses were conducted to determine the relationship between each feature and the sex variable, identifying the most discriminative parameters. Following this, correlation analysis was performed to examine inter-feature relationships and to detect redundant information. Feature combinations with high correlation and strong association with the target variable were generated using predefined threshold values. To evaluate the effectiveness of these feature sets, comparative classification experiments were carried out using machine learning (ML) algorithms, with a particular focus on performance differences between full and reduced feature subsets.Results—The experimental results showed that using the Support Vector Machine (SVM) classifier with 4 CFs yielded an accuracy of 89.33%. Notably, a comparable level of classification performance was achieved using only 4 CFs selected based on strong statistical significance and high correlation with the sex class. This finding suggests that many of the original CFs contain redundant or less informative data, and that careful feature selection can simplify the model without compromising its predictive capability.Conclusion—The results demonstrate that effective feature selection based on statistical and correlation analyses can significantly reduce the number of CFs required while maintaining high classification performance in sex determination from cephalometric radiographs. This reduction not only improves computational efficiency but also enhances model interpretability and applicability in real-world scenarios. The proposed approach provides a robust framework for optimizing feature sets in similar biomedical classification problems.
- 9 August 2026
- Published by Pamukkale University in Pamukkale University Journal of Engineering Sciences
- No. Advanced O
Open AccessSex determination from cephalometric radiography using data mining
Context—The skull is a crucial reference structure in skeletal analysis due to its hard and durable nature, which enables it to resist environmental degradation and preserve its morphological characteristics over extended periods. Because of these properties, craniofacial structures are widely used in forensic science, anthropology, and medical studies for identification purposes, including sex determination.Objective—This study aims to perform sex determination using cephalometric radiography (CR) and to evaluate the effectiveness of statistical and correlation-based feature selection methods in improving classification performance. Additionally, the study investigates whether a reduced set of informative cephalometric features (CFs) can achieve comparable accuracy to the full feature set, thereby increasing model efficiency and interpretability.Method—A dataset was created by extracting the and coordinates of 19 cephalometric landmarks (CLs) from CR images of 300 patients (150 females and 150 males). Based on these landmarks, 8 Euclidean distance (8-ED), 3 angular (3-A), and 3 frontal sinus (3-FS) measurements were calculated, resulting in a total of 14 CFs. Initially, statistical analyses were conducted to determine the relationship between each feature and the sex variable, identifying the most discriminative parameters. Following this, correlation analysis was performed to examine inter-feature relationships and to detect redundant information. Feature combinations with high correlation and strong association with the target variable were generated using predefined threshold values. To evaluate the effectiveness of these feature sets, comparative classification experiments were carried out using machine learning (ML) algorithms, with a particular focus on performance differences between full and reduced feature subsets.Results—The experimental results showed that using the Support Vector Machine (SVM) classifier with 4 CFs yielded an accuracy of 89.33%. Notably, a comparable level of classification performance was achieved using only 4 CFs selected based on strong statistical significance and high correlation with the sex class. This finding suggests that many of the original CFs contain redundant or less informative data, and that careful feature selection can simplify the model without compromising its predictive capability.Conclusion—The results demonstrate that effective feature selection based on statistical and correlation analyses can significantly reduce the number of CFs required while maintaining high classification performance in sex determination from cephalometric radiographs. This reduction not only improves computational efficiency but also enhances model interpretability and applicability in real-world scenarios. The proposed approach provides a robust framework for optimizing feature sets in similar biomedical classification problems.
- 6 June 2026
- Published by Mersin University in Bilgisayar Bilimleri ve Teknolojileri Dergisi
- Vol. 6 (2)
A Comparative Analysis of Feature Selection Approaches on Base and Ensemble Machine Learning Models in Breast Cancer Diagnosis
In this study, combinations of machine learning and ensemble models using different feature selection techniques were examined on the Breast Cancer Wisconsin dataset, and the differences in performance of these models across various feature selection techniques were compared. Logistic Regression (LR), Support Vector Machines (SVM), Multilayer Perceptron (MLP), k-Nearest Neighbor (k-NN), Decision Trees (DT), Random Forest (RF), Naive Bayes (NB), XGBoost, LightGBM and ensemble learning methods were used in the study. Two different feature selection methods (Filter and Wrapper) were applied to optimize model performance. To prevent data leakage, feature selection operations were performed only on the training data. To validate the models, 5-fold cross-validation was performed with 10 different seed values. The performance of the predictions was evaluated using Accuracy, Precision, Recall, Specificity, F1 Score, Matthews Correlation Coefficient (MCC) and ROC-AUC. The models with the highest accuracy obtained from various combinations of feature selection and machine learning were analyzed using the nonparametric Friedman test, and the difference between the two most successful models was evaluated using the Nadeau & Bengio-corrected t-test. Finally, the decision-making process of the most successful model was explained using SHAP (Shapley Additive exPlanations) analysis for transparency and interpretability.
- 6 June 2026
- Published by Mersin University in Bilgisayar Bilimleri ve Teknolojileri Dergisi
- Vol. 6 (2)
A Comparative Analysis of Feature Selection Approaches on Base and Ensemble Machine Learning Models in Breast Cancer Diagnosis
In this study, combinations of machine learning and ensemble models using different feature selection techniques were examined on the Breast Cancer Wisconsin dataset, and the differences in performance of these models across various feature selection techniques were compared. Logistic Regression (LR), Support Vector Machines (SVM), Multilayer Perceptron (MLP), k-Nearest Neighbor (k-NN), Decision Trees (DT), Random Forest (RF), Naive Bayes (NB), XGBoost, LightGBM and ensemble learning methods were used in the study. Two different feature selection methods (Filter and Wrapper) were applied to optimize model performance. To prevent data leakage, feature selection operations were performed only on the training data. To validate the models, 5-fold cross-validation was performed with 10 different seed values. The performance of the predictions was evaluated using Accuracy, Precision, Recall, Specificity, F1 Score, Matthews Correlation Coefficient (MCC) and ROC-AUC. The models with the highest accuracy obtained from various combinations of feature selection and machine learning were analyzed using the nonparametric Friedman test, and the difference between the two most successful models was evaluated using the Nadeau & Bengio-corrected t-test. Finally, the decision-making process of the most successful model was explained using SHAP (Shapley Additive exPlanations) analysis for transparency and interpretability.
- Open Access
- 19 February 2026
- Published by Elsevier in Egyptian Informatics Journal
- Vol. 33
Fully automated Pell & Gregory classification on panoramic radiographs
- 19 February 2026
- Published by Elsevier in Egyptian Informatics Journal
- Vol. 33
Open AccessFully automated Pell & Gregory classification on panoramic radiographs
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 92456-92465
LDA-Driven Classifiers and Stacking Ensembles for Efficient Human Activity Recognition
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 92456-92465
Open AccessLDA-Driven Classifiers and Stacking Ensembles for Efficient Human Activity Recognition
- 31 December 2025
- Published by Gazi University in Gazi University Journal of Science Part A: Engineering and Innovation
- Vol. 12 (4) , 953-965
Early-Stage Diabetes Detection Using Age-Based Feature Selection and Machine Learning Methods
Diabetes is spreading rapidly around the world, and there is an urgent need for governments to develop comprehensive strategies for diabetes detection. Early detection of diabetes is important for early initiation of treatment. In this paper, Data Mining (DM) and Machine Learning (ML) techniques are used to detect early diabetes by age from survey data. The dataset was divided into 3 groups (young, middle-aged, elderly), and a unique feature selection process was performed by averaging the feature importance obtained in Random Forest (RF), Gradient Boosting (GB), and eXtreme Gradient Boosting (XGBoost) algorithms for each group, and the features that should be considered in diabetes detection according to age groups were determined. Then, the features selected for each age group were classified using different ML methods. Accuracies of 96.77%, 98.10% and 99% were obtained for the young, middle-aged and elderly groups, respectively. The characteristics that should be taken into account in the assessment of diabetes according to age groups were also identified.
- 31 December 2025
- Published by Gazi University in Gazi University Journal of Science Part A: Engineering and Innovation
- Vol. 12 (4) , 953-965
Early-Stage Diabetes Detection Using Age-Based Feature Selection and Machine Learning Methods
Diabetes is spreading rapidly around the world, and there is an urgent need for governments to develop comprehensive strategies for diabetes detection. Early detection of diabetes is important for early initiation of treatment. In this paper, Data Mining (DM) and Machine Learning (ML) techniques are used to detect early diabetes by age from survey data. The dataset was divided into 3 groups (young, middle-aged, elderly), and a unique feature selection process was performed by averaging the feature importance obtained in Random Forest (RF), Gradient Boosting (GB), and eXtreme Gradient Boosting (XGBoost) algorithms for each group, and the features that should be considered in diabetes detection according to age groups were determined. Then, the features selected for each age group were classified using different ML methods. Accuracies of 96.77%, 98.10% and 99% were obtained for the young, middle-aged and elderly groups, respectively. The characteristics that should be taken into account in the assessment of diabetes according to age groups were also identified.
- 17 November 2025
- Published by Springer Nature in Signal, Image and Video Processing
- Vol. 19 (16) , 1367
Arrhythmia detection from 12-Lead ECG with 2-phase feature extraction: by presenting the evaluation of atrial fibrillation
- 17 November 2025
- Published by Springer Nature in Signal, Image and Video Processing
- Vol. 19 (16) , 1367
Arrhythmia detection from 12-Lead ECG with 2-phase feature extraction: by presenting the evaluation of atrial fibrillation
- Open Access
- 8 August 2025
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 65 (16) , 8486-8496
A Machine Learning-Based Modeling Approach for Dye Removal Using Modified Natural Adsorbents
This study used machine learning models to investigate the potential of biosorbents derived from natural fruit seed waste (apricot, almond, and walnut) for removing a cationic dye. Levulinic acid (LA)-modified powders of almond shell (ASh), apricot kernel shell (APKSh), and walnut shell (WSh) were used to remove methylene blue (MB) from an aqueous solution, producing 105 experimental data points under various circumstances. Attributes included pH (3-5), adsorbent dose (0.4-6.0 g/L), concentration (10-500 mg/L), time (30-600 min), and temperature (25-55 °C). Species information was incorporated into the data set using the One-Hot Encoding method. The data were normalized using the min-max method, and due to the non-normal distribution of the data, Spearman correlation analysis was employed to rank the importance of the attributes. Gradient Boosting (GB), Multilayer Perceptron (MLP), XGBoost (XGB), and Random Forest (RF) algorithms were applied for regression estimation. Based on 5-fold cross-validation results, the GB model achieved the highest performance, with R2 values of 0.8858 for removal percentage and 0.9532 for adsorption capacity.
- 8 August 2025
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 65 (16) , 8486-8496
Open AccessA Machine Learning-Based Modeling Approach for Dye Removal Using Modified Natural Adsorbents
This study used machine learning models to investigate the potential of biosorbents derived from natural fruit seed waste (apricot, almond, and walnut) for removing a cationic dye. Levulinic acid (LA)-modified powders of almond shell (ASh), apricot kernel shell (APKSh), and walnut shell (WSh) were used to remove methylene blue (MB) from an aqueous solution, producing 105 experimental data points under various circumstances. Attributes included pH (3-5), adsorbent dose (0.4-6.0 g/L), concentration (10-500 mg/L), time (30-600 min), and temperature (25-55 °C). Species information was incorporated into the data set using the One-Hot Encoding method. The data were normalized using the min-max method, and due to the non-normal distribution of the data, Spearman correlation analysis was employed to rank the importance of the attributes. Gradient Boosting (GB), Multilayer Perceptron (MLP), XGBoost (XGB), and Random Forest (RF) algorithms were applied for regression estimation. Based on 5-fold cross-validation results, the GB model achieved the highest performance, with R2 values of 0.8858 for removal percentage and 0.9532 for adsorption capacity.
- Open Access
- 31 December 2024
- Published by Adiyaman University in Adıyaman Üniversitesi Mühendislik Bilimleri Dergisi
- Vol. 11 (24) , 475-488
VERİ ÖN İŞLEME TEKNİKLERİNİN SAĞLIK VERİLERİNİN SINIFLANDIRMA BAŞARISINA ETKİSİNİN İNCELENMESİ
Veri madenciliği sürecinin en temel adımlarından biri olan veri ön işleme teknikleri, literatürde sıklıkla başvurulan bir süreçtir. Bu çalışmada Hepatit hastalığına ait veri kümesi üzerinde sağlık alanında sık kullanılan veri ön işleme tekniklerinin etkinliği incelenmiştir. Sırasıyla eksik veri, dengesiz veri kümesi, aykırı veri, normalizasyon ve özellik seçimi işlemleri uygulanmıştır. Veri kümesinin her adımda elde edilen yeni versiyonu için literatürde sıklıkla kullanılan beş makine öğrenmesi yöntemi (KNN, LR, RF, SVM, ANN) ile sınıflandırma yapılmıştır. Elde edilen sonuçlar, doğru ve gerekli veri ön işleme tekniklerinin seçimi ile model başarısına olumlu katkısını desteklemektedir. Tüm aşama sonunda elde edilen model performansları %85 ve üzerinde olup, tüm performans belirleme ölçütleri bazında tutarlı sonuçlar göstermektedir. Her bir veri ön işleme model performansına kademeli olarak katkıda bulunmuş, en yüksek katkı ise son aşamada uygulanan özellik seçimi ile sağlanmıştır. Özellik seçimi, modelin performansını belirgin şekilde iyileştirerek sınıflandırma başarısına önemli ölçüde katkı sağlamıştır.
- 31 December 2024
- Published by Adiyaman University in Adıyaman Üniversitesi Mühendislik Bilimleri Dergisi
- Vol. 11 (24) , 475-488
Open AccessVERİ ÖN İŞLEME TEKNİKLERİNİN SAĞLIK VERİLERİNİN SINIFLANDIRMA BAŞARISINA ETKİSİNİN İNCELENMESİ
Veri madenciliği sürecinin en temel adımlarından biri olan veri ön işleme teknikleri, literatürde sıklıkla başvurulan bir süreçtir. Bu çalışmada Hepatit hastalığına ait veri kümesi üzerinde sağlık alanında sık kullanılan veri ön işleme tekniklerinin etkinliği incelenmiştir. Sırasıyla eksik veri, dengesiz veri kümesi, aykırı veri, normalizasyon ve özellik seçimi işlemleri uygulanmıştır. Veri kümesinin her adımda elde edilen yeni versiyonu için literatürde sıklıkla kullanılan beş makine öğrenmesi yöntemi (KNN, LR, RF, SVM, ANN) ile sınıflandırma yapılmıştır. Elde edilen sonuçlar, doğru ve gerekli veri ön işleme tekniklerinin seçimi ile model başarısına olumlu katkısını desteklemektedir. Tüm aşama sonunda elde edilen model performansları %85 ve üzerinde olup, tüm performans belirleme ölçütleri bazında tutarlı sonuçlar göstermektedir. Her bir veri ön işleme model performansına kademeli olarak katkıda bulunmuş, en yüksek katkı ise son aşamada uygulanan özellik seçimi ile sağlanmıştır. Özellik seçimi, modelin performansını belirgin şekilde iyileştirerek sınıflandırma başarısına önemli ölçüde katkı sağlamıştır.
- 29 December 2023
- Published by Bulent Evcevit University in Karaelmas Science and Engineering Journal
- Vol. 13 (2) , 316-328
Detection of COVID-19 Severity and Mortality from Blood Parameters by Ensemble Learning Methods
COVID-19 is a pandemic that causes a high rate of spread and Acute Respiratory Distress Syndrome (ARDS). Severe pneumonia in infected individuals has resulted in too many patients being admitted to the Intensive Care Unit (ICU). This has placed unprecedented pressure on health systems by exceeding capacities. It is essential to detect the prognosis of this disease so that the health systems can remain active and the conditions of the patients who need to be hospitalized in the ICU do not become critical. In this study, COVID-19 prognosis was detected by using ICU admission (COVID-19 SEVERITY) and COVID-19 related death (COVID- 19 MORTALITY) datasets with Machine Learning (ML) methods. The missing data of the datasets were filled with K-Nearest Neighbor (KNN), and Min-Max normalization was performed. Datasets were divided three times into training and test sets, and the data were balanced with the Synthetic Minority Oversampling Technique (SMOTE). Then, classification was carried out using Ensemble Learning (EL) methods. For COVID-19 SEVERITY and COVID-19 MORTALITY, 89.54% and 97.25% accuracy were achieved with the Adaboost classifier, respectively. Successful and rapid COVID-19 prognosis detection with ML methods will help to use the ICU more efficiently and relieve the pressure on health systems.
- 29 December 2023
- Published by Bulent Evcevit University in Karaelmas Science and Engineering Journal
- Vol. 13 (2) , 316-328
Detection of COVID-19 Severity and Mortality from Blood Parameters by Ensemble Learning Methods
COVID-19 is a pandemic that causes a high rate of spread and Acute Respiratory Distress Syndrome (ARDS). Severe pneumonia in infected individuals has resulted in too many patients being admitted to the Intensive Care Unit (ICU). This has placed unprecedented pressure on health systems by exceeding capacities. It is essential to detect the prognosis of this disease so that the health systems can remain active and the conditions of the patients who need to be hospitalized in the ICU do not become critical. In this study, COVID-19 prognosis was detected by using ICU admission (COVID-19 SEVERITY) and COVID-19 related death (COVID- 19 MORTALITY) datasets with Machine Learning (ML) methods. The missing data of the datasets were filled with K-Nearest Neighbor (KNN), and Min-Max normalization was performed. Datasets were divided three times into training and test sets, and the data were balanced with the Synthetic Minority Oversampling Technique (SMOTE). Then, classification was carried out using Ensemble Learning (EL) methods. For COVID-19 SEVERITY and COVID-19 MORTALITY, 89.54% and 97.25% accuracy were achieved with the Adaboost classifier, respectively. Successful and rapid COVID-19 prognosis detection with ML methods will help to use the ICU more efficiently and relieve the pressure on health systems.
- Open Access
- 28 May 2023
- Published by Springer Nature in Soft Computing
- Vol. 27 (15) , 10555-10570
Diagnosis of COVID-19 from blood parameters using convolutional neural network
Asymptomatically presenting COVID-19 complicates the detection of infected individuals. Additionally, the virus changes too many genomic variants, which increases the virus’s ability to spread. Because there isn’t a specific treatment for COVID-19 in a short time, the essential goal is to reduce the virulence of the disease. Blood parameters, which contain essential clinical information about infectious diseases and are easy to access, have an important place in COVID-19 detection. The convolutional neural network (CNN) architecture, which is popular in image processing, produces highly successful results for COVID-19 detection models. When the literature is examined, it is seen that COVID-19 studies with CNN are generally done using lung images. In this study, one-dimensional (1D) blood parameters data were converted into two-dimensional (2D) image data after preprocessing, and COVID-19 detection was made with CNN. The t-distributed stochastic neighbor embedding method was applied to transfer the feature vectors to the 2D plane. All data were framed with convex hull and minimum bounding rectangle algorithms to obtain image data. The image data obtained by pixel mapping was presented to the developed 3-line CNN architecture. This study proposes an effective and successful model by providing a combination of low-cost and rapidly-accessible blood parameters and CNN architecture making image data processing highly successful for COVID-19 detection. Ultimately, COVID-19 detection was made with a success rate of 94.85%. This study has brought a new perspective to COVID-19 detection studies by obtaining 2D image data from 1D COVID-19 blood parameters and using CNN.
- 28 May 2023
- Published by Springer Nature in Soft Computing
- Vol. 27 (15) , 10555-10570
Open AccessDiagnosis of COVID-19 from blood parameters using convolutional neural network
Asymptomatically presenting COVID-19 complicates the detection of infected individuals. Additionally, the virus changes too many genomic variants, which increases the virus’s ability to spread. Because there isn’t a specific treatment for COVID-19 in a short time, the essential goal is to reduce the virulence of the disease. Blood parameters, which contain essential clinical information about infectious diseases and are easy to access, have an important place in COVID-19 detection. The convolutional neural network (CNN) architecture, which is popular in image processing, produces highly successful results for COVID-19 detection models. When the literature is examined, it is seen that COVID-19 studies with CNN are generally done using lung images. In this study, one-dimensional (1D) blood parameters data were converted into two-dimensional (2D) image data after preprocessing, and COVID-19 detection was made with CNN. The t-distributed stochastic neighbor embedding method was applied to transfer the feature vectors to the 2D plane. All data were framed with convex hull and minimum bounding rectangle algorithms to obtain image data. The image data obtained by pixel mapping was presented to the developed 3-line CNN architecture. This study proposes an effective and successful model by providing a combination of low-cost and rapidly-accessible blood parameters and CNN architecture making image data processing highly successful for COVID-19 detection. Ultimately, COVID-19 detection was made with a success rate of 94.85%. This study has brought a new perspective to COVID-19 detection studies by obtaining 2D image data from 1D COVID-19 blood parameters and using CNN.