Please use this identifier to cite or link to this item:
http://repository.ipb.ac.id/handle/123456789/170678| Title: | Pengembangan Model Prediksi Senyawa Antimalaria Menggunakan Pendekatan Machine Learning |
| Other Titles: | Developing a Machine Learning Based Predictive Model for Antimalarial Compounds |
| Authors: | Kusuma, Wisnu Ananta Tedjo, Aryo Prasetya, Fatha Ariya |
| Issue Date: | 2025 |
| Publisher: | IPB University |
| Abstract: | Malaria merupakan salah satu penyakit paling mematikan di kawasan tropis dengan tingkat mortalitas yang tinggi akibat infeksi parasit Plasmodium falciparum. Tantangan utama dalam pengendalian malaria adalah munculnya resistensi terhadap obat-obatan antimalaria yang ada. Studi ini bertujuan untuk mengembangkan metode penapisan maya (virtual screening) senyawa herbal antimalaria berbasis machine learning untuk mengidentifikasi kandidat senyawa potensial. Studi ini melakukan perbandingan terhadap tiga algoritma machine learning, yaitu Support Vector Machine (SVM), Extreme Gradient Boosting (XGB), dan Light Gradient Boosting Machine (LGBM). Model model tersebut dilatih menggunakan kombinasi deskriptor molekuler, yakni PubChem Fingerprint dan ECFP. Untuk meningkatkan kinerja prediktif, dilakukan pemilihan fitur menggunakan metode Lasso. Hasil terbaik diperoleh dari model XGB dengan penerapan seleksi Lasso dengan nilai R² sebesar 0.5599, RMSE sebesar 0.3694, dan MAE sebesar 0.2800. Nilai-nilai tersebut mengindikasikan potensi penggunaannya dalam menyaring senyawa-senyawa herbal untuk tahap awal penemuan obat antimalaria. Malaria is one of the deadliest diseases in tropical regions, with a high mortality rate caused by infection with the Plasmodium falciparum parasite. The main challenge in malaria control is the emergence of resistance to existing antimalarial drugs. This study aims to develop a virtual screening method for antimalarial herbal compounds using machine learning to identify potential candidate compounds. The study compares Support Vector Machine (SVM), Extreme Gradient Boosting (XGB), and Light Gradient Boosting Machine (LGBM) as machine learning algorithms. These models were trained using a combination of molecular descriptors, specifically PubChem Fingerprint and ECFP. To improve predictive performance, feature selection was carried out using the Lasso method. The best results were obtained from the XGB model with Lasso selection, achieving an R² of 0.5599, RMSE of 0.3694, and MAE of 0.2800. These values indicate its potential use in screening herbal compounds during the early stages of antimalarial drug discovery. |
| URI: | http://repository.ipb.ac.id/handle/123456789/170678 |
| Appears in Collections: | UT - Computer Science |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| cover_G6401211078_71d314f73d85471b821967217255fd97.pdf | Cover | 468.37 kB | Adobe PDF | View/Open |
| fulltext_G6401211078_87063b5e93294c14b35c0f8ec0977109.pdf Restricted Access | Fulltext | 1.42 MB | Adobe PDF | View/Open |
| lampiran_G6401211078_0c5b97c6c5724ba5b4784a109df6d0d2.pdf Restricted Access | Lampiran | 270.7 kB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.