Pengembangan Sistem Prediksi Risiko Readmission Pasien Gagal Jantung Berbasis Algoritma Xgboost Menggunakan Data Klinis

Umair, Muhammad Mushab (2025) Pengembangan Sistem Prediksi Risiko Readmission Pasien Gagal Jantung Berbasis Algoritma Xgboost Menggunakan Data Klinis. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5026211078-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (8MB) | Request a copy

Abstract

Gagal jantung merupakan salah satu penyakit dengan angka kejadian dan tingkat kematian yang tinggi. Berdasarkan data BPJS Kesehatan tahun 2021, biaya penanganan penyakit jantung juga merupakan yang terbesar, mencapai hingga Rp 7,7 triliun. Salah satu tantangan utama dalam penanganan pasien gagal jantung adalah risiko readmission yang tinggi, terutama pada 30 hari pertama setelah keluar dari rumah sakit. Risiko ini disebabkan oleh faktor-faktor seperti usia lanjut, komorbiditas, dan kondisi klinis tertentu. Penelitian ini bertujuan untuk mengembangkan sistem prediksi risiko readmission pasien gagal jantung berbasis algoritma XGBoost dengan menggunakan data klinis pasien. Metode yang digunakan melibatkan beberapa tahapan, mulai dari pengumpulan dan pra-pemrosesan data, penyeimbangan data menggunakan teknik SMOTE-ENN, hingga pengembangan model prediksi berbasis XGBoost. Selain itu, dilakukan hyperparameter tuning serta interpretasi hasil model menggunakan teknik SHAP (Shapley Additive Explanations). Model terbaik XGBoost dikembangkan dan dioptimalkan melalui hyperparameter tuning menggunakan metode Random Search yang menghasilkan skor akurasi sebesar 88%, skor F1 sebesar 91%, dan skor ROC AUC sebesar 86%. Sistem juga divalidasi oleh dokter spesialis jantung melalui demonstrasi dashboard interaktif yang menampilkan hasil prediksi dan interpretasinya. Hasil validasi menunjukkan sistem memiliki potensi untuk digunakan sebagai alat bantu klinis yang interpretatif dan akurat dalam mendukung pengambilan keputusan medis, dimana mendapat nilai rata-rata 3,78 dari 5. Penelitian ini memberikan kontribusi dalam pengembangan sistem cerdas berbasis data klinis yang transparan dan dapat diandalkan untuk menekan angka readmission pasien gagal jantung di Indonesia.
==================================================================================================================================================
Heart failure is a disease with a high incidence and mortality rate. According to data from BPJS Kesehatan in 2021, the cost of treating heart-related diseases ranked the highest, reaching up to IDR 7.7 trillion. One of the major challenges in managing heart failure patients is the high risk of readmission, particularly within the first 30 days after hospital discharge. This risk is influenced by factors such as advanced age, comorbidities, and specific clinical conditions. This study aims to develop a readmission risk prediction system for heart failure patients using the XGBoost algorithm based on clinical data. The methodology involves several stages, including data collection and preprocessing, class balancing using the SMOTE-ENN technique, and the development of a predictive model using XGBoost. Additionally, hyperparameter tuning and model interpretation were carried out using SHAP (Shapley Additive Explanations). The best XGBoost model was developed and optimized using the Random Search method, achieving an accuracy score of 88%, an F1-score of 91%, and an ROC AUC score of 86%. The system was also validated by a cardiologist through a demonstration of an interactive dashboard that displays prediction results and their interpretations. Validation results showed that the system has potential to be used as a clinically interpretable and accurate decision-support tool, receiving an average score of 3.78 out of 5. This research contributes to the development of a transparent and reliable intelligent system based on clinical data to reduce heart failure readmission rates in Indonesia.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	gagal jantung, klinis, readmission, XGBoost clinical, heart failure, readmission, XGBoost
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. R Medicine > RA Public aspects of medicine > RA971 Health services administration.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis
Depositing User:	Muhammad Mushab Umair
Date Deposited:	25 Jul 2025 06:57
Last Modified:	25 Jul 2025 06:57
URI:	http://repository.its.ac.id/id/eprint/121806

Actions (login required)

View Item