Tilburg University

Tilburg University Repository
Not a member yet
    232451 research outputs found

    Essays on Consistency and Randomization in Machine Learning and Fraud Detection

    No full text
    This dissertation addresses statistical questions in the fields of machine learningand insurance fraud detection. Random forests are a popular method used in many fields to make predictions. They are averages of decision trees with two layers of randomization: data sampling and split randomization. The first two chapters separately study the effectiveness of each randomization. In particular, the first chapter establishes pointwise consistency of trees and illustrates that growing large trees in ensembles, often the default choice in practice, is not always a good idea. The second chapter looks at the effectiveness of split randomization for different data characteristics. While prior literature has focused on the amount of noise in the data, this chapter offers a novel perspective on forest performance by showing that randomization is effective in the presence of irrelevant and correlated covariates, possibly opening the way for a better understanding of why random forests work well in many applications. The third chapter, separate from the first two, studies how insurance companies can choose which claims to investigate for fraud. Originated from a collaboration with Achmea, this chapter formalizes selection mechanisms, illustrates that selecting based on prior beliefs can lead to inconsistent learning of fraud characteristics and proposes arandomized strategy conjectured to be consistent

    Cognitive complaints in patients with suspected obstructive sleep apnea are associated with sleepiness, fatigue, and anxiety, not with final diagnosis or objective cognitive impairment

    No full text
    This study examined the nature, severity, and predictors of cognitive complaints in patients referred for suspected obstructive sleep apnea (OSA). The sample included 127 patients classified as no OSA (AHI, apnea/hypopnea index < 5, N = 32), mild OSA (AHI 5-15, N = 46), moderate OSA (AHI 15-30, N = 25), or severe OSA (AHI > 30, N = 24), and 53 healthy controls (HCs), matched for age, sex, education, and IQ. Cognitive complaints were assessed using the Cognitive Failure Questionnaire (CFQ) and the Behavioral Rating Inventory of Executive Functioning Adult Version (BRIEF-A). Regression analyses examined predictors of cognitive complaints including AHI, sleepiness, anxiety, depression, fatigue, and neuropsychological performance. Compared to HCs, those with mild OSA reported more forgetfulness, distractibility, and working memory issues, while those with severe OSA reported more difficulties with initiative, both with large effect sizes. Cognitive complaints were linked to sleepiness, anxiety, and fatigue (ss's 0.29-0.37), but not AHI or cognitive performance. Cognitive complaints were not specific to subjects with OSA but were also common among individuals with sleep complaints suspected for OSA. In conclusion, cognitive complaints were associated with anxiety, fatigue, and sleepiness rather than objective cognitive performance or impairment

    Construction and validation of the implicit theories of sexual offense questionnaire (itsoq) in a general and (sub)clinical population sample

    No full text
    This study developed and assessed the psychometric properties of a questionnaire assessing Implicit Theories (ITs) of sexual offense (Polaschek & Ward, 2002; Ward & Keenan, 1999), named the Implicit Theories of Sexual Offense Questionnaire (ITSOQ). We used existing cognition questionnaires to create a potential item pool, and selected items based on item properties (e.g., mean, SD, range) from three male general population samples (n = 427) and three (sub)clinical population samples (n = 69), i.e., pedophilia-supportive forum users (n = 20), and sexual (n = 28) and violent (n = 21) forensic mental health system clients. A principal component analysis for the general population sample supported a four-component solution for the ITSOQ, including two victim-specific ITs (Factor 1: Children 14-16 years, Factor 2: Women), a sexual social desirability index (SSDI; Factor 3), and a component reflecting the antisocial uncontrollability IT (Factor 4). Analyses indicated measurement invariance, and higher scores for the (sub)clinical population were found for the antisocial uncontrollability and SSDI factors, with low to moderate effect sizes. Additionally, (small) associations between self-reported sexual interest in children and adults and the victim-specific child and women factors were found. Implications and directions for future research are discussed

    Exploring barriers and facilitators to engagement of an online acceptance and commitment therapy intervention for cancer survivors with chronic painful chemotherapy-induced peripheral neuropathy:Qualitative interview study

    No full text
    Background: Online self-management interventions for cancer survivors are increasingly being used, but engagement is often difficult for patients. Given the importance of engagement for intervention effectiveness, identifying patient-reported barriers and facilitators is essential. Objective: The aim of this study was to qualitatively examine barriers and facilitators influencing engagement with an online self-management intervention, offered with or without guidance, for cancer survivors experiencing chronic painful chemotherapy-induced peripheral neuropathy (CIPN). Methods: Patients who took part in the Embrace Pain randomized controlled trial, conducted between December 2021 and July 2024, were invited to participate in this study. Eligible participants were adults with chronic painful CIPN, based on criteria including pain, completion of chemotherapy, and European Organisation for Research and Treatment of Cancer QLQ-CIPN20 Questionnaire (ie, cancer-specific measure of sensory, motor, and autonomic neuropathy). The Embrace Pain randomized controlled trial involved evaluating an online self-management acceptance and commitment therapy intervention for pain interference in daily life, with some participants receiving email guidance and others not. Thereafter, 12 patients experiencing chronic painful CIPN participated in semistructured interviews. Data were analyzed using thematic analysis. An inductive coding approach was applied, and Atlas.ti (Lumivero) was used for coding. Results: In total, 2 themes and 17 codes emerged from the data, namely 7 codes for barriers and 10 codes for facilitators. Barriers included program schedule, burden, lack of guidance, irrelevance, mindfulness exercises, usability, and missing content. Facilitators included usability, recognition, positive self-management, program schedule, symptom management, relevance, guidance, experiential exercises, mindfulness exercises, and value-based living. Program schedule, guidance, mindfulness exercises, and usability proved to be barriers for some, while others indicated that they were facilitators for their use. Conclusions: Participants’ perceptions of the intervention varied, with engagement influenced by individual circumstances. These variations highlight the importance of personal context in shaping both uptake and effectiveness, indicating a need for tailored approaches to address diverse needs and challenges faced by participants. International Registered Report Identifier (IRRID): RR2-10.1186/S13063-022-06592-

    Partial ignorance:Strategies for coping with inconvenient information

    No full text
    Willful ignorance - intentionally avoiding inconvenient information - allows individuals to act selfishly without spoiling their positive image. Yet, full avoidance is not always possible, especially in today's information-rich environments. We propose that when ignorance is not possible, individuals use more subtle information-processing strategies to achieve the same result. These partial ignorance strategies include limiting exposure to inconvenient information, seeking contradictory information, and interpreting information in self-serving ways to maintain a positive image while behaving selfishly. Therefore, to effectively promote prosociality, interventions should go beyond confronting individuals with inconvenient information. Instead, interventions should focus on mitigating the more subtle strategies individuals use. We present potentially promising interventions and raise open research questions that merit further exploration. Understanding these subtle information-processing strategies and developing interventions to tackle them is important to promote prosociality

    Voxel-based lesion-symptom mapping localizes residual visual function in hemianopia

    No full text
    Damage to the primary visual cortex (V1) results in visual field deficits on the contralateral side of the world corresponding to the damaged region. Patients with such loss nonetheless show varying residual vision within this apparently blind region, with the neural mechanisms underlying this ability obscured by small study populations. We identified lesions on structural scans from 39 patients (12 female) with hemianopia and occipital lobe damage. We estimated the proportion of damage throughout the visual cortex and correlated this with performance in up to three different tests of residual vision in the blind field. We then performed voxel-based lesion-symptom mapping (VLSM) to determine the precise neural regions associated with residual vision. V1 damage did not show a relationship with residual vision measured with any task, although the extent of damage to V4 and hMT+/V5 both correlated with the ability to detect contrast-modulated stimuli. Indeed, damage to hMT+/V5 severely impacted performance across all three tasks, particularly motion detection. Although hMT+/V5 appeared necessary for residual vision, some patients with intact hMT+/V5 had no residual vision, indicating this area alone was not sufficient. VLSM revealed that damage to the optic radiation laterally was most highly associated with poor residual vision. Thus, VLSM indicates that hMT+/V5 and the optic radiation specifically are critical for residual vision in hemianopia

    Inflammatory markers in pregnancy:identifying drivers in four large cohorts

    No full text
    INTRODUCTION: Adaptations of the immune system throughout gestation have been proposed as important mechanisms regulating successful pregnancy. Dysregulation of the maternal immune system has been associated with adverse maternal and fetal outcomes. The design and interpretation of human biomarker studies require additional insights in the trajectories and drivers of peripheral immune markers.METHODS: The current study mapped maternal inflammatory markers (C-reactive protein (CRP), interleukin (IL)-1β, IL-6, IL-17A, IL-23, interferon-γ) during pregnancy and investigated the impact of demographic, environmental and genetic drivers on maternal inflammatory marker levels in four multi-ethnic and socio-economically diverse population-based cohorts with more than 12,000 pregnant participants. Additionally, pregnancy inflammatory markers were compared to pre-pregnancy levels.RESULTS: Cytokines showed a high correlation with each other, but not with CRP. Inflammatory marker levels showed high variability between individuals, yet high concordance within an individual over time during and pre-pregnancy. Pre-pregnancy body mass index (BMI) explained ~ 9.6% of the variance in CRP, but less than 1% of the variance in cytokines. The polygenic score of CRP was the best predictor of variance in CRP (14.1%). Gestational age and previously identified inflammation drivers, including tobacco use and parity, explained less than 1% of variance in both cytokines and CRP.DISCUSSION: Our findings corroborate differential underlying regulatory mechanisms of CRP and cytokines and are suggestive of an individual inflammatory marker baseline which is, in part, genetically driven.</p

    Exploring the mechanistic link between the oxytocinergic system and mindfulness training in adults with heightened stress:study protocol for a double-blind, randomized, placebo-controlled trial (MOX-MIND)

    No full text
    BackgroundOxytocin administration is increasingly considered a novel therapeutic support option for alleviating psychological distress in stress-related neuropsychiatric conditions, including anxiety disorders and depression.However, oxytocin as a stand-alone treatment may fail to consistently target the relevant central autonomic circuitry without a complementary supportive therapeutic context that similarly stimulates stress-regulatory states and behaviors.Recent findings suggest that the neurophysiological mechanisms underlying stress regulation induced by mindfulness may rely on an activation of the endogenous oxytocinergic system. Accordingly, combining oxytocin with a mindfulness-based training program may enhance efficacy compared to each intervention alone.MethodsA randomized, double-blind, placebo-controlled clinical trial will be conducted in 120 adults with heightened stress complaints, randomly assigned to one of the following four treatment arms (n = 30 per group): (1) oxytocin + mindfulness, (2) mindfulness + placebo, (3) oxytocin, and (4) placebo (control). The oxytocin or placebo nasal spray will be administered four times a week in the morning before each mindfulness session for six weeks, followed by a six-week post-treatment follow-up session.Primary endpoints include self-reported behavioral measures of stress, depression, and anxiety. Secondary endpoints include self-reported behavioral measures of mood, mindfulness skills, quality of life, sleep quality, and negative thinking. Exploratory measures include (i) electroencephalography (EEG), electrocardiography (ECG), skin conductance, and respiration, measured during rest, meditation, stress induction, and stress recovery; (ii) intervention-induced changes in biological samples, including hormonal levels of oxytocin and cortisol and DNA methylation of the oxytocin receptor gene (OXTR); and (iii) stress reactivity in daily life, assessed through experience sampling and heart rate/sleep monitoring. All outcome measures will be assessed at baseline (T0), immediately post-intervention (T1), and at six-week follow-up (T2).DiscussionThese findings could provide valuable insights into how combining oxytocin and mindfulness-based interventions might enhance stress regulation, particularly in populations with impaired oxytocinergic function

    Moving towards futureproof and appropriate long-term care:A rapid review on the role of technology

    No full text
    Purpose: This paper aims to examine the role of technology in the transition towards appropriate care in long-term care settings for older adults.Design/methodology/approach: A rapid review was conducted based on the Cochrane guidelines on Rapid Reviews. Three databases were searched. Studies conducted after 1999 that examined technology in long-term care settings in relation to the concept of appropriate care (defined as a need for health care transformation due to the increasing demand for long-term care and the shortage of long-term care staff) were included.Findings: Six studies were included and examined a wide range of technologies in long-term care. Findings highlight the importance of technology implementation and stakeholder involvement. Studies used the transition towards appropriate care as justification for research on technology but failed to connect their findings back to the transition towards appropriate care.Originality/value: Technology has not yet in been studied in the context of the transition towards appropriate long-term care for older adults

    52,825

    full texts

    232,451

    metadata records
    Updated in last 30 days.
    Tilburg University Repository
    Access Repository Dashboard
    Do you manage Open Research Online? Become a CORE Member to access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard! 👇