Mental disorders pose a worldwide growing public health burden. One of the major challenges for healthcare systems remains to respond to the need of patients with mental disorders for continuous and flexible treatment. The EVA64 study evaluates novel programs of flexible and integrative treatment (FIT) in hospitals. This manuscript presents results from the evaluation of FIT hospitals in comparison to hospitals from regular routine care. In addition to data from adult patients, we also present data from affiliated child and adolescent psychiatric wards employing FIT programs. Using comprehensive claims data, primary outcomes are the utilization of inpatient care and sick leave for a priori defined clusters of mental disorders. We stratify between patients already under treatment (ongoing treatment) and patients with incident treatment cases (initial treatment) at the point of inclusion in the study. In the initial treatment group, we found a significant reduction in the length of inpatient stay of 4.1 days in FIT hospitals compared to routine care. While patients with mood affective disorders (−1.8 days) and patients with neurotic, stress-related, and somatoform disorders (−3.6 days) showed an even stronger effect of the reduction of inpatient lengths of stay, the effect was significantly weaker in patients with mental and behavioral disorders due to use of alcohol (+3.3 days). Regarding the duration of sick leave, we found no significant treatment effect of FIT programs compared to routine care. In the ongoing treatment group of adult patients, we found a significantly lower utilization of inpatient treatment by 1.3 days as well as a shorter duration of sick leave by 4.3 days in FIT hospitals compared to routine care. In the cohort of children and adolescent patients, we also did not observe a significant treatment effect in either the initial treatment group or the ongoing treatment group. Registration: this study was registered in the database “Health Services Research Germany” (trial number: VVfD_EVA64_15_003713).

Altered emotion processing and regulation mechanisms play a key role in eating disorders. We recently reported increased fMRI responses in brain regions involved in emotion processing (amygdala, dorsolateral prefrontal cortex) in acutely underweight anorexia nervosa (AN) patients while passively viewing negatively valenced images. We also showed that patients’ ability to downregulate activity elicited by positively valenced pictures in a brain region involved in reward processing (ventral striatum) was predictive of worse outcomes (increased rumination and negative affect). The current study tries to answer the question of whether these alterations are only state effects associated with undernutrition or whether they constitute a trait characteristic of the disorder that persists after recovery. Forty-one individuals that were weight-recovered from AN (recAN) and 41 age-matched healthy controls (HC) completed an established emotion regulation paradigm using negatively and positively valenced visual stimuli. We assessed behavioral (arousal) and fMRI measures (activity in the amygdala, ventral striatum, and dorsolateral prefrontal cortex) during emotion processing and regulation. Additionally, measures of disorder-relevant rumination and affect were collected several times daily for 2 weeks after scanning via ecological momentary assessment. In contrast to our previous findings in acute AN patients, recAN showed no significant alterations either on a behavioral or neural level. Further, there were no associations between fMRI responses and post-scan momentary measures of rumination and affect. Together, these results suggest that neural responses to emotionally valenced stimuli as well as relationships with everyday rumination and affect likely reflect state-related alterations in AN that improve following successful weight-recovery.

Die vorliegende Habilitationsschrift zu „Computer-based diagnostic and prognostic approaches in medical research using brain MRI“ ist in zwei Abschnitte gegliedert. Konkret wird im ersten Abschnitt eine Übersicht über verschiedene Aspekte des Computer- und MRT-basierten Vorhersageansatzes gegeben. Im zweiten Abschnitt werden die Artikel aus diesem Feld beschrieben, die ich für die Habilitation eingereicht habe. Konkret beginnt der erste Abschnitt der Habilitationsschrift damit, das grundlegende methodische Konzept des Vorhersageansatzes zu beschreiben. Danach werden die drei prozeduralen Stadien beschrieben, die seine Anwendung charakterisieren, d.h. die Phase der Feature-Bestimmung, des Trainings von Regressionsalgorithmen und schließlich des Tests dieser Algorithmen mit Daten unbekannter Genese. Daran schließt sich eine Beschreibung der Entwicklung des Ansatzes in Form von drei Epochen an, die charakterisiert sind durch die Entdeckung diagnostischer Information in Signalen der Magnetresonanz, die erste Nutzung statistischer Regressionsverfahren zu deren Analyse, und die massenhaften Anwendung des Ansatzes. Schließlich werden zum Ende des ersten Abschnittes die Forschungsfragen skizziert, die mit dem Ansatz adressiert werden, d.h. die automatisierte Diagnostik, die Verfeinerung bestehender diagnostischer Richtlinien und die Identifikation neuer Biomarker. Im zweiten Abschnitt beschreibe ich im Detail die Forschungsartikel, die ich im Rahmen der Habilitation eingereicht habe. Über diese Artikel oder Studien hinweg wurden alle oben genannten Forschungsfragen adressiert, die mit dem Verfahren in der Literatur untersucht werden. Darüber hinaus wurden vielfältige technische Herausforderungen des Ansatzes in unterschiedlicher Weise bearbeitet. Zusammenfassend lässt sich daher sagen, dass die vorliegende Habilitationsschrift und die darin beschriebenen Fachartikel einen umfassenden Überblick über die konzeptionelle und methodische Vielfalt des Ansatzes geben. / This habilitation thesis on ‘Computer-based diagnostic and prognostic approaches in medical research using brain MRI’ is divided in two parts – an introductory first part that gives an overview on various aspects of the computer- and MRI-based disease prediction approach and a second part describing the research articles from this field that I submitted for habilitation. In particular, in the first part the habilitation synopsis starts by outlining the basic methodological concept of the disease prediction approach and by describing the three fundamental procedural stages characterizing it, i.e. the feature determination, training and test stages. Then, it continues by delineating the development of the approach in terms of three epochs that are characterized by the discovery of diagnostic information in MR signals, the first use of statistical regression techniques to analyze this information, and the mass use of the approach. Finally, it outlines the research aims pursued with the approach, i.e. automated diagnosis, refinement of diagnostic guidelines, and identification of novel diagnostic biomarkers. In the second part, I describe the peer-reviewed research articles that I submitted for habilitation. Across these articles or studies respectively, all of the three research aims pursued with the approach were addressed. Furthermore, technical challenges connected to the approach were addressed in various different fashions. Thus together, these studies and this habilitation thesis provide a substantial overview on the methodological and conceptual diversity of the field.