Demam Berdarah Dengue (DBD) merupakan salah satu penyakit menular yang masih menjadi masalah kesehatan di Indonesia karena jumlah kasusnya terus meningkat setiap tahun. Proses penegakan diagnosis DBD memerlukan analisis terhadap beberapa parameter laboratorium sehingga dibutuhkan suatu metode yang mampu membantu tenaga kesehatan dalam mengidentifikasi penyakit secara lebih cepat dan objektif. Penelitian ini bertujuan menerapkan metode Random Forest untuk melakukan klasifikasi penyakit DBD berdasarkan data laboratorium pasien di Puskesmas Alang-Alang Lebar. Data yang digunakan merupakan data sekunder rekam medis pasien yang meliputi jenis kelamin, usia, kadar hemoglobin, hematokrit, jumlah leukosit, jumlah trombosit, suhu tubuh, dan hasil diagnosis. Sebelum proses pemodelan, data melalui tahapan data cleaning, transformasi data, normalisasi, dan pemilihan fitur. Model Random Forest kemudian dibangun dan dievaluasi menggunakan metode 10-Fold Cross Validation. Pengukuran kinerja model dilakukan menggunakan Confusion Matrix dengan parameter Accuracy, Precision, Recall, dan F1-Score. Hasil penelitian menunjukkan bahwa metode Random Forest mampu mengklasifikasikan penyakit DBD dengan rata-rata Accuracy sebesar 85,23%, Precision sebesar 85,39%, Recall sebesar 85,23%, dan F1-Score sebesar 85,04%. Analisis feature importance menunjukkan bahwa jumlah trombosit dan suhu tubuh merupakan atribut yang memberikan kontribusi paling besar terhadap proses klasifikasi. Hasil tersebut menunjukkan bahwa metode Random Forest memiliki kemampuan yang baik dalam mengenali pola data laboratorium pasien sehingga berpotensi dimanfaatkan sebagai sistem pendukung keputusan untuk membantu proses identifikasi awal penyakit DBD di fasilitas pelayanan kesehatan.
Kata Kunci: Demam Berdarah Dengue, Random Forest, Machine Learning, Klasifikasi.
Dengue Hemorrhagic Fever (DHF) remains one of the major public health concerns in Indonesia due to the increasing number of reported cases each year. Diagnosing DHF requires the analysis of several laboratory parameters, making it necessary to develop a method that can assist healthcare professionals in identifying the disease more quickly and objectively. This study aims to implement the Random Forest method to classify DHF based on patients' laboratory data collected at the Alang-Alang Lebar Community Health Center. The study employed secondary data obtained from patients' medical records, including gender, age, hemoglobin level, hematocrit, leukocyte count, platelet count, body temperature, and diagnosis results. Before model development, the dataset underwent several preprocessing stages, including data cleaning, data transformation, normalization, and feature selection. The Random Forest model was then trained and evaluated using 10-Fold Cross Validation. Model performance was assessed using a Confusion Matrix with evaluation metrics consisting of Accuracy, Precision, Recall, and F1-Score. The experimental results indicate that the Random Forest method achieved an average Accuracy of 85.23%, Precision of 85.39%, Recall of 85.23%, and F1-Score of 85.04%. Furthermore, feature importance analysis revealed that platelet count and body temperature were the most influential variables in the classification process. These findings demonstrate that the Random Forest algorithm is capable of identifying patterns in patients' laboratory data and has the potential to be utilized as a decision support tool for the early identification of Dengue Hemorrhagic Fever in healthcare facilities.
Keywords: Dengue Hemorrhagic Fever (DHF), Random Forest, Machine Learning, Classification.