Optimization of the Biochemical Diagnosis of Pheochromocytomas and Paragangliomas: A Dual Approach to Reducing False-Positive Results and Improving Clinical Interpretation through Preanalytical Standardization and the Use of Artificial Intelligence

Die Dissertation verfolgt einen dualen Ansatz mit dem Ziel einer verbesserten Diagnostik von Phäochromozytomen und Paragangliomen (PPGL). Die Präanalytik kann durch Standardisierung und Optimierung verbessert werden. Die Interpretation und Stratefizierung kann durch künstliche Intelligenz verbessert werden.:Table of Contents LIST OF ABBREVIATIONS VI INTRODUCTION 1 BACKGROUND AND CLINICAL SIGNIFICANCE OF PHEOCHROMOCYTOMAS AND PARAGANGLIOMAS 1 GENETIC BASIS AND MOLECULAR PATHOPHYSIOLOGY 1 BIOCHEMICAL DIAGNOSTICS: FUNDAMENTALS AND CHALLENGES 3 THE ROLE OF IMAGING IN THE DIAGNOSTIC PATHWAY 4 THE CHALLENGE OF CLINICAL INTERPRETATION DESPITE EXCELLENT TEST PERFORMANCE 5 MANAGEMENT AND THERAPEUTIC STRATEGIES FOR PPGL 6 PREANALYTICAL FACTORS AS CONFOUNDING VARIABLES IN DIAGNOSTICS 7 METHODOLOGICAL APPROACHES TO IMPROVING DIAGNOSTIC ACCURACY 8 MACHINE LEARNING IN MEDICAL DIAGNOSTICS: CURRENT LANDSCAPE AND CHALLENGES 9 THE PARADIGM OF SUPERVISED LEARNING IN PRECISION MEDICINE 10 Logistic Regression (LR) 11 Naïve Bayes (NB) 11 Support Vector Machines (SVM) 12 Random Forest (RF) 12 Artificial Neural Networks (ANN) 12 CURRENT LIMITATIONS AND KNOWLEDGE GAPS 13 RESEARCH OBJECTIVES AND APPROACH 14 MATERIALS AND METHODS 16 STUDY DESIGN AND PATIENT POPULATIONS 16 PATIENT COHORT FOR PREANALYTICAL INVESTIGATIONS (STUDY I) 16 PATIENT COHORT FOR MACHINE LEARNING MODELING (STUDY II) 17 DIAGNOSTIC CRITERIA AND REFERENCE STANDARDS 17 BIOCHEMICAL MEASUREMENTS 18 METHODOLOGY FOR PREANALYTICAL INVESTIGATIONS (STUDY I) 18 Categorization of Preanalytical Variables 18 Meteorological Data Retrieval 19 METHODOLOGY FOR MACHINE LEARNING MODELING (STUDY II) 19 Feature Engineering and Data Preparation 19 Computational Implementation with Python 20 Algorithm Development 20 Validation Framework 21 Clinical Utility Study 21 STATISTICAL ANALYSIS 22 ARCHITECTURE OF THE CLINICAL DECISION SUPPORT SYSTEM (PATENT APPLICATION) 22 AUTHOR’S CONTRIBUTION 24 AUTHOR’S CONTRIBUTION TO THE PUBLICATION “PREANALYTICAL CONSIDERATIONS AND OUTPATIENT VERSUS INPATIENT TESTS OF PLASMA METANEPHRINES TO DIAGNOSE PHEOCHROMOCYTOMA” 24 AUTHOR’S CONTRIBUTION TO THE PUBLICATION “UTILITY OF DISEASE PROBABILITY SCORES TO GUIDE DECISION-MAKING DURING SCREENING FOR PHAEOCHROMOCYTOMA AND PARAGANGLIOMA: A MACHINE LEARNING MODELLING CROSS SECTIONAL STUDY” 24 AUTHOR’S CONTRIBUTION TO THE PATENT APPLICATION “ VERFAHREN ZUR VORHERSAGE EINES NEBENNIERENTUMORS SOWIE EINES METASTASERISIKOS MITHILFE KLINISCH RELEVANTER PARAMETER ” (DE 10 2022 114 246 A1) 25 PUBLICATIONS 26 PAPER “PREANALYTICAL CONSIDERATIONS AND OUTPATIENT VERSUS INPATIENT TESTS OF PLASMA METANEPHRINES TO DIAGNOSE PHEOCHROMOCYTOMA” 26 PAPER “UTILITY OF DISEASE PROBABILITY SCORES TO GUIDE DECISION-MAKING DURING SCREENING FOR PHAEOCHROMOCYTOMA AND PARAGANGLIOMA: A MACHINE LEARNING MODELLING CROSS SECTIONAL STUDY” 37 PUBLISHED PATENT APPLICATION “VERFAHREN ZUR VORHERSAGE EINES NEBENNIERENTUMORS SOWIE EINES METASTASERISIKOS MITHILFE KLINISCH RELEVANTER PARAMETER“ 48 DISCUSSION 111 INTEGRATION OF PREANALYTICAL OPTIMIZATION AND MACHINE LEARNING APPROACHES 111 NOVEL METHODOLOGICAL CONTRIBUTIONS 111 Comprehensive Preanalytical Characterization 111 Advanced Machine Learning Validation Methodology 112 Clinical Decision Support Enhancement 113 CLINICAL IMPLICATIONS AND PRACTICE RECOMMENDATIONS 115 Preanalytical Best Practices 115 Machine Learning Implementation Framework 115 BROADER IMPLICATIONS FOR DIAGNOSTIC MEDICINE 117 Paradigm Shift in Diagnostic Optimization 117 Implications for Rare Disease Diagnostics 117 Future Directions in Precision Diagnostics 118 DISCUSSION OF THE PATENT APPLICATION AND CLINICAL TRANSLATION 118 LIMITATIONS AND FUTURE RESEARCH DIRECTIONS 119 Study Limitations 119 Future Research Priorities 120 CONCLUSIONS 121 SUMMARY 123 BACKGROUND 123 RESEARCH QUESTION AND HYPOTHESIS 123 MATERIALS AND METHODS 124 Preanalytical Investigation: 124 Machine Learning Investigation: 124 RESULTS 125 Preanalytical Findings: 125 Machine Learning…