Segmentasi Partial Least Squares Structural Equation Modeling Berdasarkan Score Factor Variabel Laten Menggunakan Self-Organizing Maps Pada Pasien Alzheimer

Wahyuni, Wahyuni (2026) Segmentasi Partial Least Squares Structural Equation Modeling Berdasarkan Score Factor Variabel Laten Menggunakan Self-Organizing Maps Pada Pasien Alzheimer. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6003241005_MASTER_THESIS.pdf - Accepted Version Restricted to Repository staff only Download (1MB) | Request a copy

Abstract

Penyakit Alzheimer merupakan gangguan neurodegeneratif progresif yang menjadi penyebab utama demensia global, namun memiliki tantangan besar berupa heterogenitas klinis yang luas. Sifat heterogen ini menyebabkan pendekatan pengobatan yang seragam sering kali tidak efektif, sehingga identifikasi subkelompok pasien menjadi prioritas utama. Penelitian terdahulu telah banyak berupaya memetakan heterogenitas ini menggunakan metode Self-Organizing Maps (SOM), namun pendekatan tersebut umumnya bersifat deskriptif dan memiliki keterbatasan dalam menjelaskan mekanisme hubungan kausal yang kompleks antar variabel pembentuk penyakit. Oleh karena itu, penelitian ini mengusulkan pendekatan integratif Partial Least Squares Structural Equation Modeling (PLS-SEM) dan SOM menggunakan data National Alzheimer’s Coordinating Center (NACC). Metode hibrida ini dirancang untuk memodelkan hubungan kausal antar variabel laten terlebih dahulu, kemudian melakukan segmentasi pasien secara topologis berdasarkan skor variabel laten tersebut. Penelitian ini diharapkan dapat memberikan pemetaan subkelompok pasien yang lebih valid secara teoretis sebagai landasan bagi pengembangan strategi penanganan medis yang lebih presisi. Hasil penelitian menunjukkan bahwa SOM berhasil mengidentifikasi dua subkelompok pasien, yaitu Cluster 1 (95,1%) sebagai fenotipe progresi tipikal dan Cluster 2 (4,9%) sebagai fenotipe berisiko tinggi dengan dominasi faktor vaskular. Analisis MGA menunjukkan adanya perbedaan besaran hubungan struktural yang signifikan antar cluster , di mana koefisien pengaruh (magnitude) faktor risiko vaskular terhadap penurunan fungsi kognitif-sosial dan peningkatan keparahan demensia terlihat jauh lebih kuat dan mengancam pada cluster 2 dibandingkan dengan cluster 1 yang pengaruhnya hampir tidak ada, meskipun secara statistik lokal pengujian di cluster 2 terbatasi oleh kecilnya ukuran sampel. Temuan ini menunjukkan bahwa pendekatan PLS-SEM dan SOM mampu mengungkap heterogenitas pasien Alzheimer dan mendukung pengembangan strategi penanganan penyakit Alzheimer yang lebih personal dan presisi.
=======================================================================================================================================
Alzheimer’s disease is a progressive neurodegenerative disorder that is the leading cause of global dementia, but it presents a major challenge in the form of extensive clinical heterogeneity. This heterogeneous nature causes uniform treatment approaches to often be ineffective, making the identification of patient subgroups a top priority. Previous studies have made significant efforts to map this heterogeneity using Self-Organizing Maps (SOM), but such approaches are generally descriptive and limited in explaining the complex causal mechanisms among disease-forming variables. Therefore, this study proposes an integrative approach combining Partial Least Squares Structural Equation Modeling (PLS-SEM) and SOM using data from the National Alzheimer’s Coordinating Center (NACC). This hybrid method is designed to first model the causal relationships among latent variables, and then topologically segment patients based on the scores of those latent variables. This study is expected to provide a theoretically more valid mapping of patient subgroups as a foundation for the development of more precise medical management strategies. The results showed that SOM successfully identified two patient subgroups, namely cluster 1 (95.1%) as the typical progression phenotype and cluster 2 (4.9%) as the high-risk phenotype dominated by vascular factors. MGA analysis revealed significant differences in the magnitude of structural relationships between cluster s, wherein the impact coefficient (magnitude) of vascular risk factors on the decline of cognitive-social function and increased dementia severity appears much stronger and more threatening in cluster 2 compared to cluster 1 where the effect is almost non-existent, even though the local statistical testing in cluster 2 was limited by its small sample size. These findings indicate that the PLS-SEM and SOM approach is capable of uncovering the heterogeneity of Alzheimer's patients and supporting the development of more personalized and precise management strategies for Alzheimer's disease

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Penyakit Alzheimer, PLS-SEM, Self-Organizing Maps (SOM), Segmentasi Pasien, Heterogenitas Klinis, Alzheimer’s Disease, Patient Segmentation, Clinical Heterogeneity.
Subjects:	Q Science Q Science > QA Mathematics > QA278.3 Structural equation modeling. Q Science > QA Mathematics > QA278.55 Cluster analysis R Medicine > RC Internal medicine > RC0321 Neuroscience. Biological psychiatry. Neuropsychiatry
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis
Depositing User:	Wahyuni Wahyuni
Date Deposited:	29 Jul 2026 05:07
Last Modified:	29 Jul 2026 05:07
URI:	http://repository.its.ac.id/id/eprint/139579

Actions (login required)

View Item