Article Per Year (5 Year)
p-Index From 2021 - 2026
0.408
P-Index
This Author published in this journals
All Journal Jurnal Teknologi Informasi LOGIC : Jurnal Ilmu Komputer dan Pendidikan
Egi Mutiara Br Sitepu
Institut Teknologi dan Bisnis Indonesia
Humanities Computer Science & IT Control & Systems Engineering Education Other

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

 1 
Prediksi Diagnosa Penyakit Diabetes Menggunakan Algoritma Decision Tree Berdasarkan Indikator Kesehatan Tubuh Egi Mutiara Br Sitepu; Roberto Kaban
LOGIC : Jurnal Ilmu Komputer dan Pendidikan Vol. 4 No. 2 (2026): Logic : Jurnal Ilmu Komputer dan Pendidikan
Publisher : Shofanah Media Berkah

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

Abstract

Meningkatnya jumlah penderita diabetes mellitus secara global telah menimbulkan urgensi yang nyata terhadap ketersediaan sistem deteksi dini yang akurat, terjangkau, dan dapat diinterpretasikan secara klinis oleh tenaga medis. Penelitian ini bertujuan mengeksplorasi penerapan algoritma Decision Tree varian C4.5 dalam pembangunan model prediksi diagnosis diabetes yang bersumber dari data klinis secara otomatis. Sebagai dataset, digunakan Pima Indians Diabetes yang diperoleh dari UCI Machine Learning Repository, berisikan 768 rekaman medis perempuan berketurunan Indian Pima dengan delapan variabel prediktor kesehatan. Seluruh tahapan penelitian meliputi eksplorasi awal data, penanganan nilai tidak valid menggunakan imputasi median, normalisasi Min-Max, konstruksi pohon keputusan, serta evaluasi kinerja model dengan metode 10-fold cross-validation. Pengujian pada data independen menghasilkan akurasi sebesar 77,92%, recall 75,93%, presisi 66,13%, F1-Score 70,69%, dan nilai AUC-ROC 0,823. Variabel glukosa plasma tercatat memberikan kontribusi tertinggi terhadap kepentingan fitur yakni sebesar 38,14%, yang menegaskan posisinya sebagai penanda klinis utama dalam diagnosis diabetes. Temuan ini mengindikasikan bahwa algoritma Decision Tree C4.5 berpotensi dijadikan instrumen skrining awal diabetes pada fasilitas layanan kesehatan tingkat pertama.
Implementasi MobileNetV2 untuk Klasifikasi Jenis Acne Vulgaris pada Platform Website Azwa Guntara; Egi Mutiara Br Sitepu
Jurnal Teknologi Informasi Vol 5, No 2 (2026): Agustus 2026
Publisher : Lembaga Penelitian dan Pengabdian Masyarakat - Universitas Teknologi Digital Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26798/juti.v5i2.2850

Abstract

Acne vulgaris is one of the most common skin diseases affecting adolescents and adults. The identification of acne types is generally performed through visual examination by medical professionals. However, limited access to dermatology services and the high cost of consultation often lead many individuals to perform self-treatment without knowing the type of acne they have. This study aims to develop a web-based Acne Vulgaris classification system using a CNN with the MobileNetV2 architecture. The dataset used in this study was obtained from Kaggle and consists of 4,617 images divided into five classes: Blackheads, Cyst, Papules, Pustules, and Whiteheads. The dataset was split into 60% training data, 20% validation data, and 20% testing data. The preprocessing stage included image resizing to 224×224 pixels, data normalization, and image augmentation. The model was developed using transfer learning and fine-tuning on the last 30 layers of MobileNetV2. The training process employed the Adam optimizer with a learning rate of 0.00001, a batch size of 32, and 50 epochs. The results showed that the fine-tuned MobileNetV2 model achieved an accuracy of 79.96%, outperforming the transfer learning model, which achieved an accuracy of 75.27%. The model also obtained a precision of 81%, a recall of 80%, and an F1-score of 80%. The best-performing model was then implemented into a web-based application using the Flask framework, enabling automatic classification of Acne Vulgaris based on images uploaded by users.
Co-Authors Azwa Guntara Roberto Kaban