Numerous disorders are characterised by fatigue as a highly disabling symptom. Fatigue plays a particularly important clinical role in multiple sclerosis (MS) where it exerts a profound impact on quality of life. Recent concepts of fatigue grounded in computational theories of brain–body interactions emphasise the role of interoception and metacognition in the pathogenesis of fatigue. So far, however, for MS, empirical data on interoception and metacognition are scarce. This study examined interoception and (exteroceptive) metacognition in a sample of 71 persons with a diagnosis of MS. Interoception was assessed by prespecified subscales of a standard questionnaire (Multidimensional Assessment of Interoceptive Awareness [MAIA]), while metacognition was investigated with computational models of choice and confidence data from a visual discrimination paradigm. Additionally, autonomic function was examined by several physiological measurements. Several hypotheses were tested based on a preregistered analysis plan. In brief, we found the predicted association of interoceptive awareness with fatigue (but not with exteroceptive metacognition) and an association of autonomic function with exteroceptive metacognition (but not with fatigue). Furthermore, machine learning (elastic net regression) showed that individual fatigue scores could be predicted out‐of‐sample from our measurements, with questionnaire‐based measures of interoceptive awareness and sleep quality as key predictors. Our results support theoretical concepts of interoception as an important factor for fatigue and demonstrate the general feasibility of predicting individual levels of fatigue from simple questionnaire‐based measures of interoception and sleep.
After a first episode of major depressive disorder (MDD), there is substantial risk for a long-term remitting-relapsing course. Prevention and early interventions are thus critically important. Various studies have examined the feasibility of detecting at-risk individuals based on out-of-sample predictions about the future occurrence of depression. However, functional magnetic resonance imaging (fMRI) has received very little attention for this purpose so far. Here, we explored the utility of generative models (i.e. different dynamic causal models, DCMs) as well as functional connectivity (FC) for predicting future episodes of depression in never-depressed adults, using a large dataset (N = 906) of task-free ("resting state") fMRI data from the UK Biobank (UKB). Connectivity analyses were conducted using timeseries from pre-computed spatially independent components of different dimensionalities. Over a three-year period, 50% of selected participants showed indications of at least one depressive episode, while the other 50% did not. Using nested cross-validation for training and a held-out test set (80/20 split), we systematically examined the combination of 8 connectivity feature sets and 17 classifiers. We found that a generative embedding procedure based on combining regression DCM (rDCM) with a support vector machine (SVM) enabled the best predictions, both on the training set (0.63 accuracy, 0.66 area under the curve, AUC) and the test set (0.62 accuracy, 0.64 AUC; p < 0.001). However, on the test set, rDCM was only slightly superior to predictions based on FC (0.59 accuracy, 0.61 AUC). Interpreting model predictions based on SHAP (SHapley Additive exPlanations) values suggested that the most predictive connections were widely distributed and not confined to specific networks. Overall, our analyses suggest (i) ways of improving future fMRI-based generative embedding approaches for the early detection of individuals at-risk for depression and that (ii) achieving accuracies of clinical utility may require combination of fMRI with other data modalities.
Numerous disorders are characterised by fatigue as a highly disabling symptom. Fatigue plays a particularly important clinical role in multiple sclerosis (MS) where it exerts a profound impact on quality of life. Recent concepts of fatigue grounded in computational theories of brain-body interactions emphasise the role of interoception and metacognition in the pathogenesis of fatigue. So far, however, for MS, empirical data on interoception and metacognition are scarce. This study examined interoception and (exteroceptive) metacognition in a sample of 71 persons with a diagnosis of MS. Interoception was assessed by pre-specified subscales of a standard questionnaire (MAIA), while metacognition was investigated with computational models of choice and confidence data from a visual discrimination paradigm. Additionally, autonomic function was examined by several physiological measurements. Several hypotheses were tested based on a preregistered analysis plan. In brief, we found the predicted association of interoceptive awareness with fatigue (but not with exteroceptive metacognition) and an association of autonomic function with exteroceptive metacognition (but not with fatigue). Furthermore, machine learning (elastic net regression) showed that individual fatigue scores could be predicted out-of- sample from our measurements, with questionnaire-based measures of interoceptive awareness and sleep quality as key predictors. Our results support theoretical concepts of interoception as an important factor for fatigue and demonstrate the general feasibility of predicting individual levels of fatigue from simple questionnaire- based measures of interoception and sleep.
Chronic fatigue, characterized by persistent physical and/or mental exhaustion, is a frequent and debilitating symptom in medicine. Despite its impact, clinical management remains a challenge. A key problem is the absence of any biomarkers; as a consequence, diagnosis rests entirely on patients’ self-report. This contributes to patient stigmatization and highlights the need for objective diagnostic tools. In this study, we explored the feasibility of constructing computational assays of chronic fatigue, using clinical and functional neuroimaging data from over 2,200 participants in the UK Biobank. Whole-brain analyses of functional and effective connectivity were followed by machine learning, based on a preregistered analysis plan and a strict separation of training data and held-out test data. We found that clinical data, including prior medical diagnoses, cancer history, sleep-related information, and alcohol consumption, enabled a statistically significant prediction of chronic fatigue (61% balanced accuracy, p=0.001). Combining clinical information with brain connectivity data again enabled statistically significant predictions (up to 64% balanced accuracy) but did not consistently outperform the model trained on clinical data only. Across all models, sleep-related information, especially insomnia symptoms, emerged as a particularly important feature for prediction. Our results suggest a high degree of heterogeneity amongst individuals with chronic fatigue. While the predictive performance achieved in this study is not yet sufficient for clinical application, our findings provide a foundation for future developments of objective assays of fatigue. In particular, our results highlight the importance of sleep-related information and suggest new avenues for harnessing neuroimaging information for the prediction of fatigue.
Errors in scientific code pose a significant risk to the accuracy of research. Yet, formal code review remains uncommon in academia. This is a particular challenge for interdisciplinary fields, such as mental health research, which increasingly rely on computational approaches. This paper presents a pragmatic experience-based framework for code review procedures in mental health research. In order to facilitate practical implementation, it includes a structured checklist for identifying common coding issues, from data handling errors to flawed statistical analyses. We discuss barriers to introducing code review and how to overcome them, revisit best practices from software engineering, and highlight the emerging role of large language models in automating aspects of code review. We argue that, despite perceived costs, code review significantly enhances the reliability of research results and fosters a culture of transparency and continuous learning. Our proposal provides an adaptable model for integrating code review into research workflows, helping mitigate errors before publication and strengthening trust in scientific results.