Propensity Score Menggunakan Bootstrap Aggregating Classification and Regression Trees (Bagging CART) pada Kasus Infeksi Oportunistik Pasien HIV/AIDS

Musa, Marsuddin (2019) Propensity Score Menggunakan Bootstrap Aggregating Classification and Regression Trees (Bagging CART) pada Kasus Infeksi Oportunistik Pasien HIV/AIDS. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 06211750012009-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Penelitian observasional bidang kesehatan seringkali tidak melakukan Randomized Controlled Trials (RCT) terhadap subjek penelitian. Hal ini menyebabkan terjadinya ketidakseimbangan kovariat antara kelompok treatment dan kontrol, sehingga dapat diperoleh hasil estimasi yang bias dan kesimpulan yang tidak akurat terhadap efek treatment. Permasalahan tersebut dapat diatasi dengan mereduksi bias pada variabel confounding menggunakan metode Propensity Score Stratification (PSS) dengan Classification and Regression Trees (CART) dan Bootstrap Aggregating (Bagging) CART. Penelitian ini bertujuan untuk mengetahui pengaruh pemberian terapi antiretroviral (ARV) sebagai variabel confounding terhadap infeksi oportunistik pada pasien HIV/AIDS, selain itu akan dilakukan perbandingan hasil ketepatan klasifikasi dan besarnya bias yang dapat direduksi oleh metode PSS CART dan PSS bagging CART. Hasil analisis menunjukkan kedua metode tersebut mampu mereduksi bias dengan menghasilkan 5 strata. Pohon klasifikasi CART menunjukkan variabel yang berpengaruh terhadap faktor confounding (pemberian terapi ARV) berdasarkan strata yang terbentuk yaitu dukungan keluarga, konsep diri, sikap terhadap HIV/AIDS, dan umur pasien. Besar bias yang dapat direduksi oleh metode PSS bagging CART sebesar 89,54%, lebih tinggi dibandingkan metode PSS CART sebesar 61,02%. Hal yang sama ditunjukkan dalam ketepatan klasifikasi yang dihasilkan, metode bagging CART dapat memprediksi data baru dengan ketepatan hasil klasifikasi lebih tinggi yaitu sebesar 63,04% dengan AUC sebesar 54,76%, dibandingkan metode CART sebesar 60,87% dengan AUC sebesar 53,33%. Hasil pengujian Average Treatment Effect (ATE) baik menggunakan metode PSS CART maupun PSS bagging CART menunjukkan kesimpulan yang sama yaitu adanya efek yang signifikan pemberian terapi ARV dan pendampingan terhadap infeksi oportunistik.
====================================================================================================================================
Observational research in the field of health often do not perform Randomized Controlled Trials (RCT) on research subjects. This causes a covariate imbalance between treatment and control groups, so that the biased estimation results and inaccurate conclusions for the treatment effect. These problems can be handled by reducing the bias in the confounding variable using the Propensity Score Stratification (PSS) method with Classification and Regression Trees (CART) and Bootstrap Aggregating (Bagging) CART. This study aims to determine the effect of antiretroviral therapy (ARV) as a confounding variable on opportunistic infections in HIV/AIDS patients, in addition, to the comparison of the results of the classification accuracy and the percent of bias that can be reduced by the PSS CART and PSS CART bagging methods. The results of the analysis showed that both methods were able to reduce bias by resulting 5 strata. The CART classification tree shows variables that influence the confounding factor (giving ARV therapy) based on the strata formed, namely family support, self-concept, attitudes to HIV/AIDS, and the age of the patient. The percent of bias that can be reduced by the PSS bagging CART method is 89,54%, higher than the PSS CART method of 61,02%. The same is shown in the accuracy of the classification produced, the CART bagging method can predict new data with the accuracy of the classification results higher at 63,04% with AUC 54,76%, compared to the CART method of 60,87% with AUC 53,33%. The Average Treatment Effect (ATE) test results using both PSS CART and PSS bagging CART methods showed the same conclusion, namely there is a significant effect of the giving ARV therapy and counseling to opportunistic infections.

Item Type:	Thesis (Masters)
Additional Information:	RTSt 519.536 Mus p-1 2019 3100020083442
Uncontrolled Keywords:	Propensity Score, CART, Bagging CART, HIV/AIDS
Subjects:	Q Science Q Science > QA Mathematics Q Science > QA Mathematics > QA278.2 Regression Analysis. Logistic regression Q Science > QA Mathematics > QA278.55 Cluster analysis Q Science > QA Mathematics > QA76.9 Computer algorithms. Virtual Reality. Computer simulation. R Medicine > R Medicine (General)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis
Depositing User:	Marsuddin Musa
Date Deposited:	08 Dec 2025 08:41
Last Modified:	08 Dec 2025 08:41
URI:	http://repository.its.ac.id/id/eprint/68206

Actions (login required)

View Item