Skripsi

PERANCANGAN PROTOTIPE CLINICAL DECISION SUPPORT SYSTEM UNTUK PREDIKSI RISIKO DIABETES TIPE 2 MENGGUNAKAN XGBOOST DAN EXPLAINABLE AI

---

Type 2 diabetes is a chronic disease that requires early detection to reduce the risk of complications. This study aims to design a web-based prototype Clinical Decision Support System (CDSS) to predict diabetes risk using the XGBoost algorithm and improve interpretability through the SHAP method. The dataset used was obtained from Kaggle and consisted of approximately 100,000 records with eight clinical features. The research employed a comparative experimental method by comparing XGBoost with Logistic Regression, Random Forest, and Decision Tree using accuracy, precision, recall, F1-score, and AUC metrics, with a focus on improving recall. The evaluation results showed that XGBoost achieved a recall of 92.24% and an AUC of 0.9799, making it more suitable for screening contexts that prioritize reducing the risk of false negatives, while SHAP was able to transparently explain feature contributions to the model output. Classification threshold analysis was also conducted to understand the trade-off between recall and precision in the context of medical screening. The developed CDSS successfully integrated prediction and interpretation into a single platform; however, the system remains a prototype/proof-of-concept, is not intended as a clinical diagnostic tool, and requires further clinical validation before being used in real healthcare services.

---

Availability

Detail Information

Series Title

-

Call Number

T2007582026

Publisher

Indralaya : Prodi Sistem Komputer, Fakultas Ilmu Komputer Universitas Sriwijaya., 2026

Collation

xv, 161 hlm.; ilus.; tab.; 29 cm.

Language

Indonesia

ISBN/ISSN

-

Classification

610.285 07

Content Type

Text

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Prodi Sistem Komputer
Prediksi Resiko Diabetes--XGBOOST

Specific Detail Info

-

Statement of Responsibility

MI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment