From Limited Data to Meaningful Insights - Two studies on chronic stress prediction using machine learning - Doctoral thesis in medicine (PhD)

Aug 3, 2026

Upload by Violet

This doctoral thesis evaluates how machine learning can enable chronic stress prediction when data are limited, emphasizing sample-size, data quality, and interpretability. Two studies examine chronic stress measurement and modeling in distinct populations, using multiple algorithms in Study 1 and XGBoost with SHAP-based feature explanations in Study 2. The work analyzes predictive performance, variable rankings, and methodological constraints of small datasets, concluding with implications, strengths, and limitations for future applications in healthcare research.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages103

Reading Time4 h 20 min

AccessFree to download

Introduction

Overview on Machine Learning

Chronic stress and effects on various diseases

Material and methods

Study 1: Chronic stress in general practice assistants

Study 2: Chronic stress in the German population

Results

Study 1: Chronic stress in general practice assistants

Study 2: Chronic stress in the German population

Discussion

Main findings

Methodological considerations on analyzing chronic stress data with ML

Conclusions and perspectives

From Limited Data to Meaningful Insights Two studies on chronic stress prediction using machine learning Doctoral thesis to obtain a doctorate (PhD) from the Faculty of Medicine of the University of Bonn Arezoo Bozorgmehr from Shahreza, Iran 2024 Written with authorization of the Faculty of Medicine of the University of Bonn First reviewer: [Prof. Dr. med. Birgitta Weltermann MPH](Prof. Dr. med. Birgitta Weltermann MPH)(USA) Second reviewer: Dr. rer. nat. Javad Ghofrani, University of Luebeck (Germany) Day of oral examination: 06/11/2023 From the Institute of General Practice and Family Medicine Director: [Prof. Dr. med. Birgitta Weltermann MPH](Prof. Dr. med. Birgitta Weltermann MPH)(USA) Table of Contents List of abbreviations ...................................................................................................... 5 1 Introduction.............................................................................................................. 7 1.1 Overview on Machine Learning .............................................................................. 7 1.1.1 Application of machine learning techniques in public health ............................... 8 1.1.2 Impact of sample-size and quality of data ........................................................... 9 1.1.3 Challenges of small datasets ............................................................................ 11 1.1.4 Machine learning model interpretation .............................................................. 13 1.2 Chronic stress and effects on various diseases ................................................... 14 1.2.1 Self-assessment questionnaires for chronic stress ........................................... 16 1.2.2 Chronic stress in healthcare professionals........................................................ 17 1.2.3 Chronic stress in the German population .......................................................... 18 1.2.4 Prediction of chronic stress using machine learning approaches...................... 19 2 Material and methods............................................................................................ 20 2.1 Study 1: Chronic stress in general practice assistants ......................................... 20 2.1.1 Dataset.............................................................................................................. 20 2.1.2 Primary outcome ............................................................................................... 22 2.1.3 Comparison of four machine learning and logistic regression models .............. 22 2.1.4 Model interpretation: Variable rankings in machine learning models ................ 25 2.1.5 Evaluation of the models' performance ............................................................. 27 2.2 Study 2: Chronic stress in the German population ............................................... 27 2.2.1 Dataset.............................................................................................................. 27 2.2.2 Primary outcome ............................................................................................... 29 2.2.3 eXtreme Gradient Boosting (XGBoost) ............................................................. 29 2.2.4 Model interpretation: SHapley Additive exPlanations (SHAP) ........................... 31 2.2.5 Evaluation of the model's performance ............................................................. 33 3 Results.................................................................................................................... 35 3.1 Study 1: Chronic stress in general practice assistants ......................................... 35 3.1.1 Descriptive results............................................................................................. 35 3.1.2 Performance of the machine learning algorithms ..........................................

FAQs

What is the main goal of the thesis?

The thesis investigates how to derive meaningful insights for chronic stress prediction from limited data using machine learning, while addressing model interpretability and data constraints.

How do the two studies differ in approach and dataset?

Study 1 focuses on chronic stress in general practice assistants and compares four machine learning models with logistic regression, while Study 2 models chronic stress in the German population using XGBoost and SHAP for interpretation.

Which methods are used to interpret model behavior?

Study 1 uses variable rankings to interpret machine learning models, and Study 2 applies SHAP (SHapley Additive exPlanations) to explain feature contributions.

Aug 3, 2026

Upload by Violet