MUFADHOL, Mufadhol and Wibowo, Mochamad Agung and Jie, Ferry (2026) PENGEMBANGAN MODEL RANTAI PASOK DISTRIBUSI OBAT MENGGUNAKAN PENDEKATAN RULE BASED EXPERT SYSTEM DAN ALGORITMA AUTOMLP. Doctoral thesis, UNIVERSITAS DIPONEGORO.

	Text cover aja.pdf Download (246kB)
	Text COVER LENGKAP.pdf Restricted to Repository staff only Download (3MB)
	Text 5 BAB I.pdf Download (1MB)
	Text 6 BAB II.pdf Download (2MB)
	Text 7 BAB III.pdf Restricted to Repository staff only Download (2MB)
	Text 8 BAB IV.pdf Restricted to Repository staff only Download (4MB)
	Text 9 BAB V.pdf Restricted to Repository staff only Download (920kB)
	Text 10 Daftar Pustaka.pdf Download (1MB)
	Text LAMPIRAN.pdf Restricted to Repository staff only Download (3MB)

Abstract

Informasi merupakan aspek yang sangat penting dalam memprediksi kelayakan rantai pasok distribusi obat karena berperan krusial dalam menjamin ketersediaan obat dalam jumlah dan waktu yang tepat. Proses distribusi obat yang berjalan saat ini masih menggunakan sistem dropping, yaitu distribusi dari penyedia obat kepada pelanggan berdasarkan permintaan dalam rentang waktu tertentu. Namun demikian, terdapat potensi permasalahan pada pencatatan dan pelaporan kebutuhan obat yang belum dilakukan secara sistematis, data permintaan dan penggunaan obat sering kali tidak mutakhir, data perencanaan tidak bersifat real time, serta memungkinkan terjadinya kekurangan jenis obat tertentu pada pusat kesehatan masyarakat yang tidak dapat melakukan permintaan obat secara langsung kepada pusat kesehatan lainnya, sehingga dapat meningkatkan risiko terhadap pasien.
Penelitian ini mengenalkan model baru berupa Supply Chain Expert System (SuCES) untuk mengatasi masalah proses distribusi obat. Dataset diperoleh dari basis data melalui pendekatan rule based expert system yang digunakan sebagai inisialisasi input awal fitur. Model dikembangkan menggunakan algoritma AutoMLP dengan skema pembobotan berbasis standart deviasi. Pada tahap prapengolahan, nilai hilang ditangani menggunakan metode imputasi mean, sedangkan permasalahan ketidakseimbangan kelas diatasi menggunakan teknik Synthetic Minority Over-sampling Technique (SMOTE). Model yang dikembangkan dilatih dan diuji menggunakan fungsi MLPClassifier, dengan proses validasi dilakukan melalui teknik Leave One Out Cross Validation (LOOCV).
Hasil evaluasi model menunjukkan kinerja yang sangat baik dengan nilai akurasi sebesar 98,39% dan F1-score sebesar 98,41%. Model mampu mengenali seluruh kebutuhan distribusi obat dengan nilai recall mencapai 100%, sehingga tidak terdapat kasus kekurangan obat pada unit farmasi atau gudang obat di pusat kesehatan masyarakat. Nilai presisi sebesar 96,88% mengindikasikan adanya sedikit kelebihan distribusi, namun kondisi ini jauh lebih dapat diterima dibandingkan risiko kekurangan pasokan obat. Selain itu, nilai mean squared error (MSE) sebesar 0,0161 mencerminkan tingkat kesalahan prediksi yang rendah, sementara nilai area under the curve (AUC) sebesar 98,39% menunjukkan kemampuan model yang sangat baik dalam membedakan kelas positif dan negatif. Dengan demikian, model ini layak digunakan karena lebih menekankan pada ketersediaan obat yang terjamin di lapangan, meskipun tetap perlu monitoring untuk meminimalkan potensi persediaan obat berlebih.
Kata kunci: SuCES, distribusi obat, rule based, AutoMLP, LOOCV

Information plays a crucial role in predicting the feasibility of pharmaceutical supply chain distribution, as it is essential for ensuring the availability of medicines in the right quantities and at the right time. The current pharmaceutical distribution process still relies on a dropping system, in which medicines are delivered from suppliers to customers based on demand within a specific time period. However, there are potential problems related to the recording and reporting of medicine requirements that are not yet conducted systematically. Data on medicine demand and utilization are often not up to date, planning data are not available in real time, and shortages of certain types of medicines may occur at primary healthcare centers that are unable to directly request supplies from other centers, thereby increasing risks to patients.
This study introduces a novel model, namely the Supply Chain Expert System (SuCES), to address challenges in the pharmaceutical distribution process. The dataset is obtained from a database through a rule-based expert system approach, which is used as an initial input feature initialization. The model is developed using the AutoMLP algorithm with a standard deviation–based weighting scheme. During the preprocessing stage, missing values are handled using mean imputation, while class imbalance is addressed using the Synthetic Minority Over-sampling Technique (SMOTE). The proposed model is trained and tested using the MLPClassifier function, and the validation process is conducted using the Leave-One-Out Cross-Validation (LOOCV) technique.
The evaluation results indicate that the proposed model achieves excellent performance, with an accuracy of 98.39% and an F1-score of 98.41%. The model successfully identifies all pharmaceutical distribution requirements, achieving a recall of 100%, which indicates that no medicine shortages occur in pharmacy units or drug warehouses at primary healthcare centers. A precision value of 96.88% suggests a slight level of over-distribution; however, this condition is considerably more acceptable than the risk of insufficient medicine supply. Furthermore, a mean squared error (MSE) of 0.0161 reflects a low prediction error, while an area under the curve (AUC) of 98.39% demonstrates the model’s strong capability to distinguish between positive and negative classes. Overall, the proposed model is suitable for practical implementation, as it prioritizes assured medicine availability in real-world settings, although continuous monitoring is still required to minimize the potential for excess inventory.
Keyword: SuCES, medicine distribution, rule based, AutoMLP, LOOCV

Actions (login required)

View Item