Article Per Year (5 Year)
p-Index From 2020 - 2025
3.011
P-Index
This Author published in this journals
All Journal Jurnal Sisfokom (Sistem Informasi dan Komputer) MATRIK : Jurnal Manajemen, Teknik Informatika, dan Rekayasa Komputer Buletin Ilmiah Sarjana Teknik Elektro Journal of Robotics and Control (JRC) Journal of Applied Data Sciences International Journal of Robotics and Control Systems Sewagati Jurnal Penyakit Dalam Indonesia Journal of Medicine and Health Technology Bioma : Jurnal Biologi dan Pembelajaran Biologi
Pamungkas, Yuri
Unknown Affiliation
Aerospace Engineering Agriculture, Biological Sciences & Forestry Arts Humanities Biochemistry, Genetics & Molecular Biology Computer Science & IT Control & Systems Engineering Decision Sciences, Operations Research & Management Education Electrical & Electronics Engineering Engineering Health Professions Immunology & microbiology Languange, Linguistic, Communication & Media Mechanical Engineering Medicine & Pharmacology Neuroscience Public Health Social Sciences Other

Published : 21 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 21 Documents
Search

 1   2   3 
Effectiveness of CNN Architectures and SMOTE to Overcome Imbalanced X-Ray Data in Childhood Pneumonia Detection Pamungkas, Yuri; Ramadani, Muhammad Rifqi Nur; Njoto, Edwin Nugroho
Journal of Robotics and Control (JRC) Vol 5, No 3 (2024)
Publisher : Universitas Muhammadiyah Yogyakarta

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.18196/jrc.v5i3.21494

Abstract

Pneumonia is a disease that causes high mortality worldwide in children and adults. Pneumonia is caused by swelling of the lungs, and to ensure that the lungs are swollen, a chest X-ray can be done. The doctor will then analyze the X-ray results. However, doctors sometimes have difficulty confirming pneumonia from the results of chest X-ray observations. Therefore, we propose the combination of SMOTE and several CNN architectures be implemented in a chest X-ray image-based pneumonia detection system to help the process of diagnosing pneumonia quickly and accurately. The chest X-ray data used in this study were obtained from the Kermany dataset (5216 images). Several stages of pre-processing (grayscaling and normalization) and data augmentation (shifting, zooming, and adjusting the brightness) are carried out before deep learning is carried out. It ensures that the input data for deep learning is not mixed with noise and is according to needs. Then, the output data from the augmentation results are used as input for several CNN deep learning architectures. The augmented data will also utilize SMOTE to overcome data class disparities before entering the CNN algorithm. Based on the test results, the VGG16 architecture shows the best level of performance compared to other architectures. In system testing using SMOTE+CNN Architectures (VGG16, VGG19, Xception, Inception-ResNet v2, and DenseNet 201), the optimum accuracy level reached 93.75%, 89.10%, 91.67%, 86.54% and 91.99% respectively. SMOTE provides a performance increase of up to 4% for all CNN architectures used in predicting pneumonia.
Pelatihan Pengembangan Media Ajar Berbasis Tools Artificial Intelligence untuk Guru di SMAN 1 Probolinggo Pamungkas, Yuri; Sain, Anabela Aulia; Putri, Ziyan Nadia; Larasati, Alya Puti; Iqbal, Muhammad; Risald, Randi Achtiar; Kendenan, Valentino; Rachmadiana, Josephine Larissa; Ginting, Tsamarah Amelia Putri; Nur, Rossa Alfi; Balqis, Dayana Satira
Sewagati Vol 8 No 3 (2024)
Publisher : Pusat Publikasi ITS

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12962/j26139960.v8i3.1022

Abstract

Pemanfaatan kecerdasan buatan (AI) memiliki peran yang sangat signifikan dalam pengembangan media ajar modern. AI memungkinkan media ajar dapat disesuaikan dengan kebutuhan masing-masing siswa. Sistem AI juga dapat memberikan umpan balik secara real-time dan membantu guru memahami kemajuan belajar siswa. Oleh karena itu, dalam kegiatan pengabdian masyarakat ini, tim dosen departemen teknologi kedokteran bermaksud untuk mengadakan pelatihan pengembangan media ajar berbasis tools AI kepada guru di SMAN 1 Probolinggo. Dengan adanya pelatihan ini, para guru diharapkan mampu memanfaatkan media ajar berbasis AI untuk menunjang proses pembelajaran yang interaktif di dalam kelas. Dalam pengabdian masyarakat ini, terdapat beberapa rangkaian kegiatan yang dilakukan mulai dari perencanaan, survei lokasi, penyusunan modul, pelaksanaan pelatihan, evaluasi acara, dan pembuatan laporan. Dalam pelaksanaannya, terdapat dua sesi acara yaitu penyampaian materi/diskusi dan praktik langsung. Pada akhir pelatihan, peserta juga diminta untuk mengisi formulir survei kepuasan untuk menilai keberhasilan pelaksanaan pelatihan. Dari seluruh rangkaian acara pelatihan di SMAN 1 Probolinggo, dapat disimpulkan bahwa proses pelaksanaan pengabdian masyarakat telah berjalan dengan lancar dengan jumlah peserta pelatihan mencapai 30 guru. Antusiasme peserta pelatihan juga tergolong sangat baik. Hal ini terlihat saat proses pre dan post-test berlangsung, serta hasil dari pengukuran tingkat kepuasan peserta yang mencapai 90%.
Leveraging Topic Modelling to Analyze Biomedical Research Trends from the PubMed Database Using LDA Method Pamungkas, Yuri
Jurnal Sisfokom (Sistem Informasi dan Komputer) Vol 13, No 2 (2024): JULY
Publisher : ISB Atma Luhur

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.32736/sisfokom.v13i2.2117

Abstract

Biomedical research has become an essential entity in human life. However, finding trends related to research topics in the health sector contained in the repository is a challenging matter. In this study, we implemented topic modelling to analyze biomedical research trends using the LDA method. Topic modelling was carried out using data from 7000 articles from PubMed, which were processed with text processing such as lowercase, punctuation removal, tokenization, stop-word removal, and lemmatization. For topic modelling, the LDA with corpus conditions varied to 75% and 100% for validation. Alpha and beta parameters are also set with variations between 0.01, 0.31, 0.61, 0.91, symmetry, and asymmetry when the number of the corpus is changed. When the number of the corpus is 75%, the optimal number of topics is 7, with a coherence value of 0.52. Whereas when the number of the corpus is 100%, the optimal number of topics is 10 with a coherence value of 0.51. In addition, based on the results of article topic modelling, several topics are trending, including disease diagnosis, patient care, and genetic or cell research. Based on the classification of biomedical topics into seven categories, the optimal accuracy, precision, and recall values using the Random Forest algorithm were obtained, namely 85.57%, 87.36%, and 87.58%. The results of this study suggest that topic modelling using the LDA can be used to identify trends in biomedical research with high accuracy. This information can help stakeholders make informed decisions about the direction of future research.
Work Fatigue Detection of Search and Rescue Officers Based on Hjorth EEG Parameters Pamungkas, Yuri; Indriani, Ratri Dwi; Crisnapati, Padma Nyoman; Thwe, Yamin
Journal of Robotics and Control (JRC) Vol 5, No 6 (2024)
Publisher : Universitas Muhammadiyah Yogyakarta

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.18196/jrc.v5i6.23511

Abstract

Work fatigue can cause a decrease in cognitive function, such as decreased thinking ability, concentration, and memory. A tired brain cannot work optimally, interfering with a person's ability to perform tasks that require complex thinking. In general, to evaluate work fatigue in a person, self-assessment activities using the Perceived Stress Scale (PSS) are the method most often used by researchers or practitioners. However, this method is prone to bias because sometimes people try to hide or exaggerate their tiredness at work. Therefore, we propose to evaluate people's work fatigue based on their EEG data in this study. A total of 25 participants from SAR officers recorded their EEG data in relaxed conditions (pre-SAR operations) and fatigue conditions (post-SAR operations). Recording was performed on the brain's left (fp1 t7) and right (fp2 t8) hemispheres. The EEG data is then processed by filtering, artifact removal using ICA method, signal decomposition into several frequency bands, and Hjorth feature extraction (activity, mobility, and complexity). The main advantage of Hjorth parameters compared to other EEG features is its ability to provide rich information about the complexity and mobility of the EEG signal in a relatively simple and fast way. Based on the results of activity feature extraction, feature values will tend to increase during the post-SAR operation conditions compared to the pre-operation SAR conditions. In addition, the results of the classification of pre-and post-operative SAR conditions using Bagged Tree algorithm (10-fold cross validation) show that the highest accuracy can be obtained is 94.8%.
Analisis Prediktif Mutasi EGFR pada Adenokarsinoma Paru Menggunakan Pendekatan Pembelajaran Mesin Njoto, Edwin Nugroho; Pamungkas, Yuri; Putri, Atina I.W.; Haykal, Muhammad. Najib; Eljatin, Dwinka Syafira; Djaputra, Edith Maria
Jurnal Penyakit Dalam Indonesia
Publisher : UI Scholars Hub

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Introduction. Lung adenocarcinoma is a prevalent form of lung cancer, and mutations in the epidermal growth factor receptor (EGFR) gene are known to play a crucial role in its pathogenesis. This study aimed to develop a machine-learning model to predict EGFR mutations in lung adenocarcinoma patients using clinical and radiological features. Methods. A case-control study was conducted using a dataset comprising 160 patients with lung adenocarcinoma. Several machine learning algorithms, including decision tree, linear regression, Naive Bayes, support vector machine, K-nearest neighbor, and random forest, were employed to predict EGFR mutations based on variables such as smoking status, tumor diameter, tumor location, bubble-like appearance on CT-scan, air-bronchogram on CT-scan, and tumor distribution. Results. Most study subjects were over 50 years old (83.75%) and female (53.13%). The analysis results indicated that the random forest model demonstrated the best performance, achieving an accuracy of 83.33%, precision of 86.96%, recall of 80.00%, and an Area Under the Curve (AUC) of 90.0. The Naive Bayes model also performed well, with an accuracy of 85.42%, precision of 82.61%, recall of 86.36%, and an AUC of 91.0. Conclusions. The study highlights the potential of machine learning techniques, particularly random forest and Naive Bayes, in accurately predicting EGFR mutations in lung adenocarcinoma patients based on readily available clinical and radiological features. These findings could contribute to the development of non-invasive, cost-effective, and efficient tools for EGFR mutation detection, ultimately facilitating personalized treatment approaches for lung adenocarcinoma patients.
Hyperparameter Tuning of EfficientNet Method for Optimization of Malaria Detection System Based on Red Blood Cell Image Pamungkas, Yuri; Eljatin, Dwinka Syafira
Jurnal Sisfokom (Sistem Informasi dan Komputer) Vol. 13 No. 3 (2024): NOVEMBER
Publisher : ISB Atma Luhur

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.32736/sisfokom.v13i3.2257

Abstract

Nowadays, malaria has become an infectious disease with a high mortality rate. One way to detect malaria is through microscopic examination of blood preparations, which is done by experts and often takes a long time. With the development of deep learning technology, the observation of blood cell images infected with malaria can be more easily done. Therefore, this study proposes a red blood cell image-based malaria detection system using the EfficientNet method with hyperparameter tuning. There are three parameters which are learning rate, activation function, and optimiser. The learning rate used is 0.01 and 0.001, while the activation functions used are ReLU and Tanh. In addition, the optimisers used include Adam, SGD, and RMSProp. In the implementation, the cell image dataset from the NIH repository was pre-processed such as resizing, rotating, filtering, and data augmentation. Then the data is trained and tested on several EfficientNet models (B0, B1, B3, B5, and B7) and their performance values are compared. Based on the test results, EfficientNet-B5 and B7 models showed the best performance compared to other EfficientNet models. The most optimal system test results are when the EfficientNet B5 model is used with a learning rate of 0.001, ReLU activation function, and Adam optimiser, with values of 97.69% (accuracy), 98.36% (precision), and 97.03% (recall). This research has proven that proper model selection and hyperparameter tuning can maximise the performance of cell image-based malaria detection system. The development of this EfficientNet-based diagnostic method is more sensitive and specific in malaria detection using RBCs.
Work Fatigue Detection of Search and Rescue Officers Based on Hjorth EEG Parameters Pamungkas, Yuri; Indriani, Ratri Dwi; Crisnapati, Padma Nyoman; Thwe, Yamin
Journal of Robotics and Control (JRC) Vol. 5 No. 6 (2024)
Publisher : Universitas Muhammadiyah Yogyakarta

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.18196/jrc.v5i6.23511

Abstract

Work fatigue can cause a decrease in cognitive function, such as decreased thinking ability, concentration, and memory. A tired brain cannot work optimally, interfering with a person's ability to perform tasks that require complex thinking. In general, to evaluate work fatigue in a person, self-assessment activities using the Perceived Stress Scale (PSS) are the method most often used by researchers or practitioners. However, this method is prone to bias because sometimes people try to hide or exaggerate their tiredness at work. Therefore, we propose to evaluate people's work fatigue based on their EEG data in this study. A total of 25 participants from SAR officers recorded their EEG data in relaxed conditions (pre-SAR operations) and fatigue conditions (post-SAR operations). Recording was performed on the brain's left (fp1 & t7) and right (fp2 & t8) hemispheres. The EEG data is then processed by filtering, artifact removal using ICA method, signal decomposition into several frequency bands, and Hjorth feature extraction (activity, mobility, and complexity). The main advantage of Hjorth parameters compared to other EEG features is its ability to provide rich information about the complexity and mobility of the EEG signal in a relatively simple and fast way. Based on the results of activity feature extraction, feature values will tend to increase during the post-SAR operation conditions compared to the pre-operation SAR conditions. In addition, the results of the classification of pre-and post-operative SAR conditions using Bagged Tree algorithm (10-fold cross validation) show that the highest accuracy can be obtained is 94.8%.
Impact of Hyperparameter Tuning on ResNet-UNet Models for Enhanced Brain Tumor Segmentation in MRI Scans Pamungkas, Yuri; Triandini, Evi; Yunanto, Wawan; Thwe, Yamin
International Journal of Robotics and Control Systems Vol 5, No 2 (2025)
Publisher : Association for Scientific Computing Electronics and Engineering (ASCEE)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31763/ijrcs.v5i2.1802

Abstract

Brain tumor segmentation in MRI scans is a crucial task in medical imaging, enabling early diagnosis and treatment planning. However, accurately segmenting tumors remains a challenge due to variations in tumor shape, size, and intensity. This study proposes a ResNet-UNet-based segmentation model using LGG dataset (from 110 patients), optimized through hyperparameter tuning to enhance segmentation performance and computational efficiency. The proposed model integrates different ResNet architectures (ResNet18, ResNet34, ResNet50, ResNet101, and ResNet152) with UNet, evaluating their performance under various learning rates (0.01, 0.001, 0.0001), optimizer types (Adam, SGD, RMSProp), and activation functions (Sigmoid). The methodology involves training and evaluating each model using Loss Function, Mean Intersection over Union (mIoU), Dice Similarity Coefficient (DSC), and Iterations per Second as performance metrics. Experiments were conducted on MRI brain tumor datasets to assess the impact of hyperparameter tuning on model performance. Results show that lower learning rates (0.0001 and 0.001) improve segmentation accuracy, while Adam and RMSProp outperform SGD in minimizing segmentation errors. Deeper models (ResNet50, ResNet101, and ResNet152) achieve the highest mIoU (up to 0.902) and DSC (up to 0.928), but at the cost of slower inference speeds. ResNet50 and ResNet34 with RMSProp or Adam provide an optimal trade-off between accuracy and computational efficiency. In conclusion, hyperparameter tuning significantly impacts MRI segmentation performance, and selecting an appropriate learning rate, optimizer, and model depth is crucial for achieving high segmentation accuracy with minimal computational cost.
Enhancing Diabetic Retinopathy Classification in Fundus Images using CNN Architectures and Oversampling Technique Pamungkas, Yuri; Triandini, Evi; Yunanto, Wawan; Thwe, Yamin
Journal of Robotics and Control (JRC) Vol. 6 No. 1 (2025)
Publisher : Universitas Muhammadiyah Yogyakarta

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.18196/jrc.v6i1.25331

Abstract

Diabetic Retinopathy (DR) is a severe complication of diabetes mellitus that affects the retinal blood vessels and is a leading cause of blindness in productive-age individuals. The global increase in diabetes prevalence requires an effective DR classification system for early detection. This study aims to develop a DR classification system using several CNN architectures, such as EfficientNet-B4, ResNet-50, DenseNet-201, Xception, and Inception-ResNet-v2, with the application of the SMOTE oversampling technique to address data class imbalance. The dataset used is APTOS 2019, which has an unbalanced class distribution. Two scenarios were tested, the first without data balancing and the second with SMOTE implementation. The test results show that in the first scenario, Xception achieved the highest accuracy at 80.61%, but model performance was still limited due to majority class dominance. The application of SMOTE in the second scenario significantly improved model accuracy, with EfficientNet-B4 achieving the highest accuracy of 97.78%. Additionally, precision and recall increased dramatically in the second scenario, demonstrating SMOTE's effectiveness in enhancing the model's ability to detect minority classes and reduce prediction errors. DenseNet-201 achieved the highest precision at 99.28%, while Inception-ResNet-v2 recorded the highest recall at 98.57%. Overall, this study proves that the SMOTE method effectively addresses class imbalance in the fundus dataset and significantly improves CNN model performance. Although data balancing can help improve model quality by dealing with data imbalances, it comes at a higher computational cost. Using data balancing techniques with SMOTE significantly increased the iteration time per round on all tested CNN architectures.
Application of 3D Printing Technology for Medical Purposes: A State of the Art Pamungkas, Yuri; Kuswanto, Djoko; Syafira Eljatin, Dwinka; Nugroho Njoto, Edwin
Journal of Medicine and Health Technology Vol. 2 No. 1 (2025)
Publisher : Direktorat Riset dan Pengabdian Kepada Masyarakat, Institut Teknologi Sepuluh Nopember

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.12962/j30466865.v2i1.2255

Abstract

Introduction: The application of 3D printing technology in healthcare has revolutionized various medical practices, allowing for personalized solutions tailored to individual patient needs. This study explores the current state of 3D printing in medical applications, highlighting its benefits and challenges. Method: A comprehensive review of existing literature was conducted, focusing on the utilization of 3D printing for creating anatomical models, prosthetics, and bioprinting of living tissues. The analysis included a survey of various additive manufacturing techniques, such as Selective Laser Sintering (SLS) and Stereolithography (SLA), to assess their effectiveness in medical contexts. Results: The findings indicate that 3D printing enhances surgical planning by providing accurate anatomical models, thereby improving surgical outcomes. Additionally, the technology facilitates the production of custom implants and prosthetics, leading to better integration with patients' anatomy. Bioprinting has shown promise in developing artificial organs and regenerative therapies, significantly impacting transplant medicine. Discussion: While 3D printing offers substantial advantages, challenges such as regulatory hurdles, ethical considerations, and the need for standardized practices remain. The technology's potential in personalized medicine is vast, suggesting a future where individualized treatments are commonplace. Continued research and collaboration among medical professionals, technologists, and regulatory bodies are essential to address these challenges and optimize the benefits of 3D printing in healthcare. This study underscores the transformative impact of 3D printing in medicine and its potential to enhance patient care and outcomes across various applications.
Co-Authors Abdul Karim Abdurahman Abdurrahman Achmad Syaifudin Adhi Dharma Wibawa, Adhi Dharma Alifah Putri, Athirah Hersyadea Aung, Myo Min Balqis, Dayana Satira Cahya, Meiliana Dwi Chusak, Thassaporn Derek, Natan Dian Puspita Hapsari Djaputra, Edith Maria Dwinka Syafira Eljatin Elfrida Ratnawati Evi Triandini Fadli, Sonny Faiqoh, Elok Nur Fitriani, Fatimah Nur Forca, Adrian Jaleco Ginting, Tsamarah Amelia Putri Hashim, Uda Haykal, Muhammad Nazhif Haykal, Muhammad. Najib Hedianto, Tri Hidayah, Rizka Nurul Hisbiyah, Yuni Imam Susilo Indriani, Ratri Dwi Indriastuti, Endah Jafari, Nadya Paramitha Karimah, Rumman Kendenan, Valentino Kuswanto, Djoko Larasati, Alya Puti Made Krisnanda Muhammad Faizi, Muhammad Nakkliang, Kanittha Njoto, Edwin Nugroho Nugroho Njoto, Edwin Nur Rochmah, Nur Nur, Rossa Alfi Padma Nyoman Crisnapati Parlindungan, Putra Gelar Perwitasari, Rayi Kurnia Pratasik, Stralen Putra, Gumilar Fardhani Ami Putri Alief Siswanto Putri, Atina I.W. Putri, Atina Irani Wira Putri, Ziyan Nadia Putu Adi Guna Permana Rachmadiana, Josephine Larissa Radiansyah, Riva Satya Ramadani, Muhammad Rifqi Nur Rangkuti, Rahmah Yasinta Ridhoi, Ahmad Risald, Randi Achtiar Sain, Anabela Aulia Sakina Sangsawang, Thosporn Shofwan Hanief Syafira Eljatin, Dwinka Syulthoni, Zain Budi Tedy Apriawan Thwe, Yamin Uda, Muhammad Nur Afnan Wawan Yunanto Yulan, Gao Zulaikha, Ellya