A combination of system dynamics and machine learning: Explaining and predicting the progression of patients developing dementia

Aug 3, 2026

Dementia is a progressive and complex disease and one of the largest healthcare challenges worldwide. This study integrates qualitative system dynamics with quantitative machine learning to explain and predict cognitive decline and to inform medical decision-making. Data were provided by the Rush Alzheimer’s Disease Center, with variables selected through a system dynamics model focused on factors linked to decline. Descriptive analyses were followed by unsupervised k-means clustering and supervised models using regression decision trees and random forests. Results show that combining both approaches improves interpretability and prediction.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages117

Reading Time4 h 55 min

AccessFree to download

Introduction

Theoretical background

Dementia patient trajectories as a complex system

Common variables for dementia

Conceptual model

Methods

Sample description

Pre-processing and data selection

Data analysis methods

Ethical considerations

Results

Sample characteristics

Clustering trajectories

Predicting MMSE with decision trees

Predicting MMSE with random forest

Discussion & Conclusion

Discussion

Conclusion

References

Appendix

Master Thesis A combination of system dynamics and machine learning: Explaining and predicting the progression of patients developing dementia Author: B.L. Turan (s1081125) Frist supervisor: H. S.A. Mahmoud, MSc Second supervisor: dr. H.P.L.M. Korzilius Personal information Full name: Berke Levent Turan Student number: s1081125 [E-mail: ](E-mail: berke.turan@ru.nl)[berke.turan@ru.nl](E-mail: berke.turan@ru.nl) [This page is left intentionally blank] Acknowledgements The Rush Alzheimer's Disease Center deserves special acknowledgment for providing the data that made this study possible. I would like to thank the participants and the Rush Alzheimer's Disease Center staff for their contributions to the Memory and Aging Project study. I also want to thank my supervisor Hesham Mahmoud for the guidance during this master's thesis. Your guidance was invaluable for this study and helped me navigate the challenges that arose along the way. Your expertise and support were crucial in successfully completing this project and the personal growth that came with it. I would like to extend my appreciation to dr. Hubert Korzilius for assisting me in finding a suitable topic for my thesis. Furthermore, I would like to thank my family and friends. Their encouragement and belief in my abilities have been the driving force, not only during this master's thesis but also throughout my entire school career. Here is a quote that describes my life recently: He who would accomplish little must sacrifice little; he who would achieve much must sacrifice much; he who would attain highly must sacrifice greatly. James Allen B.L. Turan Nijmegen, 23 June 2023 Abstract Dementia is a progressive and complex disease and one of the largest healthcare problems of this century worldwide. This study examined the integrated performance of qualitative system dynamics and quantitative machine learning for explaining and predicting cognitive decline in dementia to support medical decision-making. Rush Alzheimer's Disease Center provided the data used in this study, and specific variables were selected based on a system dynamics model covering variables contributing to cognitive decline in dementia patients—these variables were then analysed with descriptive analyses followed by unsupervised and supervised machine learning. Unsupervised machine learning was done with the k-Means clustering algorithm for uncovering hidden patterns, and supervised machine learning was done with regression decision trees and the random forest algorithm for predicting. The findings of this study revealed that the combination of system dynamics and machine learning strengthen each other, mitigates the limitations of each approach, and provides a more thorough understanding of the factors that contribute to cognitive decline in dementia patients. The proposed iterative integration process, with the involvement of experts, could further increase the performance of this integration, leading to collective insights that can support medical decision-making. Index 1. Introduction......................................................................................................................8 2. Theoretical background ................................................................................................ 12 2.1 Dementia patient trajectories as a complex system ................................................. 12 2.2 Common variables for dementia .............................................................................. 14 2.3 Conceptual model .................................................................................................... 15 3. Methods........................................................................................................................... 16 3.1 Sample description ................................................................................................... 16 3.2 Pre-processing

FAQs

What is the main goal of the study?

To explain and predict cognitive decline in dementia by integrating qualitative system dynamics with quantitative machine learning for medical decision-making support.

How were variables and data selected?

Variables were chosen based on a system dynamics model covering contributors to cognitive decline, and then analyzed with descriptive methods before applying machine learning.

Which machine learning techniques were used for prediction and pattern discovery?

Unsupervised learning used k-means clustering to uncover hidden patterns, while supervised learning used regression decision trees and random forest to predict MMSE.

Aug 3, 2026

Upload by Theodore