Predicting long-term outcome in anorexia nervosa: a machine learning analysis of brain structure at different stages of weight recovery

• DOI: 10.1017/s0033291723001861
• Publication Date: 2023-08-09T07:00:46Z
• Authors: Dominic Arold; Fabio Bernardoni; Daniel Geisler; Arne Doose; Volkan Uen; Ilka Boehm; Veit Roessner; Joseph A. King; Stefan Ehrlich
• Keywords: Underweight; Anorexia nervosa; Univariate analysis; Orbitofrontal cortex

Abstract Background Anorexia nervosa (AN) is characterized by sizable, widespread gray matter (GM) reductions in the acutely underweight state. However, evidence for persistent alterations after weight-restoration has been surprisingly scarce despite high relapse rates, frequent transitions to other psychiatric disorders, and generally unfavorable outcome. While most studies investigated brain regions separately (univariate analysis), disorders can be conceptualized as network multivariate with only subtle local effects. We tested structural weight-restored individuals a history of AN, their putative biological substrate relation 1-year treatment Methods trained machine learning models on regional GM measures classify healthy controls (HC) ( N = 289) from at three stages AN: patients starting intensive 165, used baseline), partial 115), former stable full 89). Alterations were related outcome both anatomically functionally. Results Patients could classified HC when (ROC-AUC 0.90) but also 0.64). more pronounced worse not detected long-term weight-recovered individuals, i.e. those favorable These greater functional connectivity, merely explained body mass index, even increases cortical thickness observed (insula, lateral orbitofrontal, temporal pole). Conclusions Analyzing might help develop personalized interventions discharge inpatient treatment.