Machine learning with multiple modalities of brain magnetic resonance imaging data to identify the presence of bipolar disorder.

BACKGROUND: Bipolar disorder (BD) is a chronic psychiatric mood disorder that is solely diagnosed based on clinical symptoms. These symptoms often overlap with other psychiatric disorders. Efforts to use machine learning (ML) to create predictive models for BD based on data from brain imaging are expanding but have often been limited using only a single modality and the exclusion of the cerebellum, which may be relevant in BD. METHODS: In this study, we sought to improve ML classification of BD by combining information from structural, functional, and diffusion-weighted imaging. Participants (108 BD I, 78 control) with BD type I and matched controls were recruited into an imaging study. This dataset was randomly divided into training and testing sets. For each of the three modalities, a separate ML model was selected, trained, and then used to generate a prediction of the class of each test subject. Majority voting was used to combine results from the three models to make a final prediction of whether a subject had BD. An independent replication sample was used to evaluate the ability of the ML classification to generalize to data collected at other sites. RESULTS: Combining the three machine learning models through majority voting resulted in an accuracy of 89.5 % for classification of the test subjects as being in the BD or control group. Bootstrapping resulted in a 95 % confidence interval of 78.9 %-97.4 % for test accuracy. Performance was reduced when only using 2 of the 3 modalities. Analysis of feature importance revealed that the cerebellum and nodes of the emotional control network were among the most important regions for classification. The machine learning model performed at chance on the independent replication sample. CONCLUSION: BD I could be identified with high accuracy in our relatively small sample by combining structural, functional, and diffusion-weighted imaging data within a single site but not generalize well to an independent replication sample. Future studies using harmonized imaging protocols may facilitate generalization of ML models.

Cerebellar morphological differences and associations with extrinsic factors in bipolar disorder type I.

BACKGROUND: The neural underpinnings of bipolar disorder (BD) remain poorly understood. The cerebellum is ideally positioned to modulate emotional regulation circuitry yet has been understudied in BD. Literature suggests differences in cerebellar activity and metabolism in BD, however findings on structural differences remain contradictory. Potential reasons include combining BD subtypes, small sample sizes, and potential moderators such as genetics, adverse childhood experiences (ACEs), and pharmacotherapy. METHODS: We collected 3 T MRI scans from participants with (N = 131) and without (N = 81) BD type I, as well as blood and questionnaires. We assessed differences in cerebellar volumes and explored potentially influential factors. RESULTS: The cerebellar cortex was smaller bilaterally in participants with BD. Polygenic propensity score did not predict any cerebellar volumes, suggesting that non-genetic factors may have greater influence on the cerebellar volume difference we observed in BD. Proportionate cerebellar white matter volumes appeared larger with more ACEs, but this may result from reduced ICV. Time from onset and symptom burden were not associated with cerebellar volumes. Finally, taking sedatives was associated with larger cerebellar white matter and non-significantly larger cortical volume. LIMITATIONS: This study was cross-sectional, limiting interpretation of possible mechanisms. Most of our participants were White, which could limit the generalizability. Additionally, we did not account for potential polypharmacy interactions. CONCLUSIONS: These findings suggest that external factors, such as sedatives and childhood experiences, may influence cerebellum structure in BD and may mask underlying differences. Accounting for such variables may be critical for consistent findings in future studies.

Intoxicating effects of alcohol depend on acid-sensing ion channels.

Persons at risk for developing alcohol use disorder (AUD) differ in their sensitivity to acute alcohol intoxication. Alcohol effects are complex and thought to depend on multiple mechanisms. Here, we explored whether acid-sensing ion channels (ASICs) might play a role. We tested ASIC function in transfected CHO cells and amygdala principal neurons, and found alcohol potentiated currents mediated by ASIC1A homomeric channels, but not ASIC1A/2 A heteromeric channels. Supporting a role for ASIC1A in the intoxicating effects of alcohol in vivo, we observed marked alcohol-induced changes on local field potentials in basolateral amygdala, which differed significantly in Asic1a-/- mice, particularly in the gamma, delta, and theta frequency ranges. Altered electrophysiological responses to alcohol in mice lacking ASIC1A, were accompanied by changes in multiple behavioral measures. Alcohol administration during amygdala-dependent fear conditioning dramatically diminished context and cue-evoked memory on subsequent days after the alcohol had cleared. There was a significant alcohol by genotype interaction. Context- and cue-evoked memory were notably worse in Asic1a-/- mice. We further examined acute stimulating and sedating effects of alcohol on locomotor activity, loss of righting reflex, and in an acute intoxication severity scale. We found loss of ASIC1A increased the stimulating effects of alcohol and reduced the sedating effects compared to wild-type mice, despite similar blood alcohol levels. Together these observations suggest a novel role for ASIC1A in the acute intoxicating effects of alcohol in mice. They further suggest that ASICs might contribute to intoxicating effects of alcohol and AUD in humans.