This Ph.D. project explores how to improve the effectiveness and efficiency ofknowledge discovery in databases (KDD) in medical research through the effectiveintegration of domain expertise. The aim was to develop and evaluate a KDDframework that improves the efficiency and accuracy of knowledge discoveryfrom real-world health data.Knowledge discovery from electronic health records (EHRs) is complex due todata inconsistencies. Collaborative feature engineering, termed knowledge-drivenfeature engineering (KDFE), is crucial. During KDFE, new variables, referred to asfeatures, are generated through collaboration between domain experts, computerscientists, and medical researchers.A case study, involving two medical projects, demonstrated the significantimpact of manual KDFE (mKDFE) on classification performance, measured bythe area under the receiver operating characteristic curve (AUROC). Compared to abaseline, mKDFE increased the average AUROC from 0.62 to 0.82 in Project 1 andfrom 0.61 to 0.89 in Project 2 (p < 0.001).To optimise KDD, an automated KDFE (aKDFE) framework was developed.This framework supports automated feature engineering, constructing informativefeatures from EHR data. The framework effectively collects and aggregatesdomain knowledge to generate features that are more informative than thosedirectly recorded in EHRs or manually engineered (mKDFE), as is common inmany medical research projects today. aKDFE outperforms mKDFE by automatingmanual processes and enhancing predictive power.Clinical decision support systems (CDSSs), like Janusmed Riskprofile, containvaluable domain knowledge in the form of risk scores. Studies were conducted toexplore CDSS risk scores and their impact on aKDFE effectiveness. These findingshighlight the potential of aKDFE to streamline medical research by leveragingboth automated feature engineering and expert knowledge.aKDFE offers several advantages over mKDFE: (i) increased efficiency throughautomated knowledge discovery and feature engineering (FE) processes; (ii)enhanced effectiveness due to superior predictive power; and (iii) explicit andtransparent operation sequences for data pivoting and feature generation fromEHR features.The long-term objective is to equip medical researchers with augmentedcomputational expertise, minimising dependence on data scientists. Futureimprovements may include: (i) assessing advanced event-based models; (ii)leveraging large language models (LLMs) to capture and structure domainknowledge; and (iii) exploring multi-agent knowledge discovery.

The aim of this thesis is to deepen the understanding of sports schools, particularly the methods and decision-making processes underlying student admissions to these schools, as well as the significance of the schools in shaping athletic careers.

The thesis comprises four studies. The first is a scoping study that maps peer-reviewed research on sports schools. It shows that research on sports schools has expanded significantly, particularly since the 2010s, underscoring the schools’ dual role as both sites for individual development and as integral parts of elite talent systems. The second study, based on the concept of communities of practice, is a case study of admission processes at sports schools. Using six focus group interviews with school sport teachers (n = 18), the study shows how admissions take place within sport-specific communities, shaped by sports cultures and assumptions about talent. Despite the subjective nature of athlete assessments, collective engagement among teachers plays a central role in their joint decision regarding which students to admit. The third study, a retrospective cohort study, explores the use of physical tests in admission processes. Analysing test results from cross-country skiers (n = 193), the study finds no or weak correlation between admission test performance and later sporting success, raising questions about the use of such tests in talent identification processes, such as admission to sports schools. The fourth study, informed by the concept of horizons for action, is a case study based on semi-structured interviews, exploring senior-year student athletes’ (n = 10) career development. The study shows how social and cultural contexts shape career development and career decisions, such as choosing a sport, changing clubs and applying to sport schools. While all student athletes aspired to elite careers, their dreams have shifted into pragmatically rational intentions within bounded opportunities. The findings highlight the segmented nature of career horizons among student athletes and their limited awareness of alternative pathways.

Together, these studies provide new insights into the functioning of sports schools, as well as into admission processes and career development within them. They offer valuable perspectives for refining admission procedures and career guidance.

This dissertation investigates how metascientific approaches can enhance the quality and reliability of evidence syntheses in educational psychology. Prompted by the replication crisis, widespread questionable research practices, and the growing dependence on systematic reviews and meta-analyses in education, this work critically examines current research standards and advances innovative solutions rooted in open science.

Study I evaluates the methodological validity and reproducibility of the influential research synthesis Visible Learning by John Hattie. The study reveals several methodological flaws that contest the assumptions of the findings and the failure of being able to reproduce the statistics serves as a warning example of the presence of the replication crisis. 

Study II evaluates the risk of bias and transparency in systematic reviews conducted in educational psychology. Alarmingly, most included systematic reviews were judged as high risk of bias and across the entire sample, there was a lack of data sharing, preregistered protocols, and reproducible primary research data. 

Study III is a proof of concept of a registered report in educational psychology, the study aims to investigate the evidence of a writing intervention by conducting a systematic review. By adhering to the state-of-the-art conducting standards in systematic reviews, this protocol covers all aspects needed to produce reliable evidence as well as being reproducible. 

In Study IV, an innovative open-source Community-Augmented Meta-Analysis combined with a database is developed. The study presents solutions to several well-known problems in systematic reviews by allowing the research community to update, store, calculate, and share educational interventional data in a convenient way.

The findings of the included studies highlight significant gaps in research rigor and transparency, underscoring the necessity of fundamental change to adhere to current standards and modern research practices. 

By incorporating methodological tools such as preregistration, open science, risk of bias assessments and FAIR data principles, this dissertation calls for a paradigm shift in the synthesis and application of evidence in educational psychology. Ultimately, it seeks to promote more trustworthy, transparent, and impactful research to better inform educational policy and practice.