Propensity score stratification menggunakan Bootstrap Aggregating Multivariate Adaptive Regression Spline pada kasus Diabetes Melitus Tipe 2

Putra, Romy Yunika (2019) Propensity score stratification menggunakan Bootstrap Aggregating Multivariate Adaptive Regression Spline pada kasus Diabetes Melitus Tipe 2. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 06211750012005-Master Tesis.pdf Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Penelitian non-experimental atau penelitian observasional yaitu penelitian empiris dari efek perlakuan ketika eksperimen acak Ramdomized Controlled Trial (RCT) tidak layak dilakukan pengacakan. RCT sangat diperlukan dalam penelitian agar asumsi independensi bisa terpenuhi, sehingga efek bias dapat diminimalisir. Namun, RCT tidak bisa dipraktekkan pada semua bidang terutama pada bidang kesehatan yang menyangkut nyawa manusia. Dalam beberapa kasus, peneliti biasanya menggunakan penelitian non-experimental yang rentan terhadap bias sehingga menyebabkan diragukannya hasil estimasi yang diperoleh. Berdasarkan hal tersebut dibutuhkan suatu metode yang digunakan untuk mengatasi masalah bias akibat observasi yang tidak random. Salah satu metode yang dapat mengatasi bias adalah propensity score yaitu Propensity Score Stratification (PSS). PSS bertujuan untuk mendapatkan kelompok strata yang balance pada setiap kovariat sehingga tidak ada perbedaan mean dan selanjutnya layak dilakukan estimasi ATE. Untuk mengestimasi nilai PSS digunakan Bootstrap Aggregating (Bagging) Multivariate Adaptive Regression Spline (MARS). Studi kasus pada PSS-Bagging MARS ini diterapkan pada komplikasi penyakit Diabetes Melitus (DM) tipe 2. Hasil penelitian menunjukkan bahwa variabel yang menjadi confounding adalah aktivitas olahraga. Estimasi efek perlakuan (ATE) memberikan hasil bahwa variabel aktivitas olahraga merupakan variabel yang berpengaruh terhadap variabel komplikasi penyakit pada pasien DM tipe 2. Dengan menggunakan strata sebanyak 2, bias yang mampu direduksi adalah sebesar 52,580% dengan nilai standar error sebesar 0,031.
=================================================================================================================================
Non-experimental research commonly referred to as observational research, namely empirical research on the effects of treatment when randomized Ramdomized Controlled Trial (RCT) experiments were not feasible for randomization. RCT is needed in research so that the assumption of independence can be fulfilled, so that the effects of bias can be minimized. However, the RCT cannot be practiced in all fields, especially in the health sector which concerns human life. In some cases, researchers usually use non-experimental studies that are susceptible to bias, thus causing doubtful results of the estimates obtained. Based on this, a method is needed to overcome the problem of bias due to non-random observations. One method that can overcome bias is the propensity score, namely Propensity Score Stratification (PSS). PSS aims to get a balanced strata group at each covariate so that there is no mean difference and then it is worth doing the ATE estimation. To estimate the value of PSS, Bootstrap Aggregating (Bagging) Multivariate Adaptive Regression Spline (MARS) is used. Case studies on MARS PSS-Bagging are applied to complications of type 2 diabetes mellitus (DM). The results showed that the variable that became confounding was diabetes gymnastics. The treatment effect estimation (ATE) results that the diabetes gymnastics is a variable that affects the disease complication in type 2 DM patients. Using two strata, the bias that can be reduced is 52.580% with a standard error value of 0.031.

Item Type:	Thesis (Masters)
Additional Information:	RTSt 519.535 Put p-1 2019 3100020084283
Uncontrolled Keywords:	ATE, Bias, Bagging MARS, Diabetes Melitus, Penelitian Non-exsperimental, PSS, RCT
Subjects:	Q Science > QA Mathematics Q Science > QA Mathematics > QA278.2 Regression Analysis. Logistic regression R Medicine > R Medicine (General)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis
Depositing User:	Romy Yunika Putra
Date Deposited:	30 Oct 2025 07:47
Last Modified:	30 Oct 2025 07:47
URI:	http://repository.its.ac.id/id/eprint/67930

Actions (login required)

View Item