Supervised machine learning for diagnostic classification from large-scale neuroimaging datasets

Brain Imaging and Behavior - There are growing concerns about the generalizability of machine learning classifiers in neuroimaging. In order to evaluate this aspect across relatively large...     

A trait of mind: stability and robustness of sleep across sleep opportunity manipulations during simulated military operational stress

Study ObjectivesWithin-subject stability of certain sleep features across multiple nights is thought to reflect the trait-like behavior of sleep. However, to be considered a trait, a parameter must be both stable and robust. Here, we examined the stability (i.e. across the same sleep opportunity periods) and robustness (i.e. across sleep opportunity periods that varied in duration and timing) of different sleep parameters.MethodsSixty-eight military personnel (14 W) spent 5 nights in the sleep laboratory during a simulated military operational stress protocol. After an adaptation night, participants had an 8-hour sleep opportunity (23:00–07:00) followed by 2 consecutive nights of sleep restriction and disruption which included two 2-hour sleep opportunities (01:00–03:00; 05:00–07:00) and, lastly, another 8-hour sleep opportunity (23:00–07:00). Intra-class correlation coefficients were calculated to examine differences in stability and robustness across different sleep parameters.ResultsSleep architecture parameters were less stable and robust than absolute and relative spectral activity parameters. Further, relative spectral activity parameters were less robust than absolute spectral activity. Absolute alpha and sigma activity demonstrated the highest levels of stability that were also robust across sleep opportunities of varying duration and timing.ConclusionsStability and robustness varied across different sleep parameters, but absolute NREM alpha and sigma activity demonstrated robust trait-like behavior across variable sleep opportunities. Reduced stability of other sleep architecture and spectral parameters during shorter sleep episodes as well as across different sleep opportunities has important implications for study design and interpretation.     

Identifying disease foci from static and dynamic effective connectivity networks: Illustration in soldiers with trauma

Brain connectivity studies report group differences in pairwise connection strengths. While informative, such results are difficult to interpret since our understanding of the brain relies on region‐based properties, rather than on connection information. Given that large disruptions in the brain are often caused by a few pivotal sources, we propose a novel framework to identify the sources of functional disruption from effective connectivity networks. Our approach integrates static and time‐varying effective connectivity modeling in a probabilistic framework, to identify aberrant foci and the corresponding aberrant connectomics network. Using resting‐state fMRI, we illustrate the utility of this novel approach in U.S. Army soldiers (N = 87) with posttraumatic stress disorder (PTSD), mild traumatic brain injury (mTBI) and combat controls. Additionally, we employed machine‐learning classification to identify those significant connectivity features that possessed high predictive ability. We identified three disrupted foci (middle frontal gyrus [MFG], insula, hippocampus), and an aberrant prefrontal‐subcortical‐parietal network of information flow. We found the MFG to be the pivotal focus of network disruption, with aberrant strength and temporal‐variability of effective connectivity to the insula, amygdala and hippocampus. These connectivities also possessed high predictive ability (giving a classification accuracy of 81%); and they exhibited significant associations with symptom severity and neurocognitive functioning. In summary, dysregulation originating in the MFG caused elevated and temporally less‐variable connectivity in subcortical regions, followed by a similar effect on parietal memory‐related regions. This mechanism likely contributes to the reduced control over traumatic memories leading to re‐experiencing, hyperarousal and flashbacks observed in soldiers with PTSD and mTBI. Hum Brain Mapp 39:264–287, 2018. © 2017 Wiley Periodicals, Inc.     

Sleep disturbance mediates the association of adverse childhood experiences with mental health symptoms and functional impairment in US soldiers

Adverse childhood experiences (ACEs) can have long‐term impacts on a person's mental health, which extend into adulthood. There is a high prevalence of ACEs among service members. Further, service members also report frequently experiencing disrupted sleep. We hypothesized that disrupted sleep may serve a mechanistic function connecting ACEs to functional impairment and poorer mental health. In a cross‐sectional sample (n = 759), we found evidence for an indirect effect of ACEs on mental health outcomes through disrupted sleep. In a different sample using two time‐points (n = 410), we found evidence for an indirect effect of ACEs on changes in mental health outcomes and functional impairment during a reset period, through changes in disrupted sleep during the same period. Implications, limitations and future research directions are discussed.     

The ANAM lacks utility as a diagnostic or screening tool for concussion more than 10 days following injury

Congress has mandated that the Department of Defense perform screening for concussion, or mild traumatic brain injury, on all service members redeploying from Iraq and Afghanistan. However, the retrospective diagnosis of concussion is complicated by the subjective nature of the complaints, overlap of symptoms with other conditions, and the normally rapid recovery of neurocognitive function following a concussive event. One diagnostic and screening test in current use by the Department of Defense is the Automated Neuropsychological Assessment Metrics (ANAM). A team of researchers deployed to Iraq between January and April 2009 to test the validity of the ANAM for the diagnosis of concussion in the combat environment. Performance by concussed participants on all six ANAM subtests was compared with that of controls. The ANAM appears to have no utility as an individual diagnostic or population screening tool for the detection of neurocognitive dysfunction from a single, uncomplicated concussion when administered 10 or more days following injury. Further studies are required to determine the modalities providing optimal sensitivity and specificity for use as diagnostic or screening tests beyond the first 72-hour acute postinjury period.     

Assessment of acute concussion in the combat environment

Despite the prevalence of concussion in soldiers deployed to Iraq and Afghanistan, neuropsychological tests used to assist in concussion management have not been validated on the battlefield. This study evaluated the validity of the Automated Neuropsychological Assessment Metrics (ANAM) in the combat environment. Cases meeting criteria for concussion, healthy controls, and injured controls were assessed. Soldiers were administered the ANAM, traditional neuropsychological tests, and a background questionnaire. Cases were enrolled within 72 h of concussion. Cases exhibited poorer performance than controls on all ANAM subtests, with significant differences on simple reaction time (SRT), procedural reaction time (PRT), code substitution, and matching to sample (p<.001). Discriminant ability of scores on SRT and PRT subtests was 71%, which improved to 76% when pre-deployment baseline scores were available. An exploratory clinical decision tool incorporating ANAM scores and symptoms improved discriminant ability to 81%. Results provide initial validation of the ANAM for detecting acute effects of battlefield concussion.