Psychiatry - Research Publications
Permanent URI for this collection
Search Results
1 - 5 of 5
-
ItemS12. A MACHINE LEARNING FRAMEWORK FOR ROBUST AND RELIABLE PREDICTION OF SHORT- AND LONG-TERM CLINICAL RESPONSE IN INITIALLY ANTIPSYCHOTIC-NAïVE SCHIZOPHRENIA PATIENTS BASED ON MULTIMODAL NEUROPSYCHIATRIC DATA(Oxford University Press (OUP), 2020-05-18)Abstract Background The treatment response of patients with schizophrenia is heterogeneous, and markers of clinical response are missing. Studies using machine learning approaches have provided encouraging results regarding prediction of outcomes, but replicability has been challenging. In the present study, we present a novel methodological framework for applying machine learning to clinical data. Herein, algorithm selection and other methodological choices were based on model performance on a simulated dataset, to minimize bias and avoid overfitting. We subsequently applied the best performing machine learning algorithm to a rich, multimodal neuropsychiatric dataset. We aimed to 1) classify patients from controls, 2) predict short- and long-term clinical response in a sample of initially antipsychotic-naïve first-episode schizophrenia patients, and 3) validate our methodological framework. Methods We included data from 138 antipsychotic-naïve, first-episode schizophrenia patients, who had undergone assessments of psychopathology, cognition, electrophysiology, structural magnetic resonance imaging (MRI). Perinatal data and long-term outcome measures were obtained from Danish registers. Baseline diagnostic classification algorithms also included data from 151 matched healthy controls. Short-term treatment response was defined as change in psychopathology after the initial antipsychotic treatment period. Long-term treatment response (4–16 years) was based on data from Danish registers. The simulated dataset was generated to resemble the real data with respect to dimensionality, multimodality, and pattern of missing data. Noise levels were tunable to enable approximation to the signal-to-noise ratio in the real data. Robustness of the results was ensured by running two parallel, fundamentally different machine learning pipelines, a ‘single algorithm approach’ and an ‘ensemble approach’. Both pipelines included nested cross-validation, missing data imputation, and late integration. Results We significantly classified patients from controls with a balanced accuracy of 64.2% (95% CI = [51.7, 76.7]) for the single algorithm approach and 63.1% (95% CI = [50.4, 75.8]) for the ensemble approach. Post hoc analyses showed that the classification primarily was driven by the cognitive data. Neither approach predicted short- and long-term clinical response. To validate our methodological framework based on simulated data, we selected the best, a medium, and the most poorly performing algorithm on the simulated data and applied them to the real data. We found that the ranking of the algorithms was kept in the real data. Discussion Our rigorous modelling framework incorporating simulated data and parallel pipelines discriminated patients from controls, but our extensive, multimodal neuropsychiatric data from antipsychotic-naïve schizophrenia patients were not predictive of the clinical outcome. Nevertheless, our novel approach holds promise as an important step to obtain reliable, unbiased results with modest sample sizes when independent replication samples are not available.
-
ItemT18. EFFECTS OF COGNITIVE REMEDIATION ON WHITE MATTER IN INDIVIDUALS AT ULTRA-HIGH RISK FOR PSYCHOSIS – A RANDOMIZED, CONTROLLED CLINICAL TRIAL(Oxford University Press (OUP), 2020-05-18)Abstract Background Individuals at ultra-high risk for psychosis (UHR) present with subtle white matter alterations, which have been associated with clinical and functional outcome. The effect of cognitive remediation on white matter (WM) in UHR-individuals has not been investigated. Methods In a randomized, clinical intervention-trial (FOCUS), UHR-individuals aged 18–40 years were assigned to treatment as usual (TAU) or TAU plus cognitive remediation (CR) for 20 weeks. CR comprised 20 x 2-hour sessions of neurocognitive and social-cognitive training (SCIT). Primary outcome was whole brain fractional anisotropy (FA) derived from diffusion weighted imaging. Secondary outcomes pertained to regions of interest analyses. Planned post-hoc analyses explored dose-response effects of CR on WM. Main analyses of treatment effect of CR on primary and secondary outcomes were conducted using linear mixed models, assessing the interaction of timepoint by group (CR and TAU). Analyses were conducted according to the intention-to-treat principle. Results 111 UHR-individuals and 59 healthy controls were included. Attrition-rate was 30% at 6 months post-treatment follow-up. The CR group completed a mean of 12 hours of neurocognitive training. We found no effect of CR on whole-brain or regional FA. Planned post-hoc analyses revealed significant time*group (high- and low-attendance to CR) interactions in left superior corona radiata (p<0.01), left cingulum cingulate gyrus (P=0.03), and right superior longitudinal fasciculus (P<0,01), corrected. Specifically, when compared to UHR-individuals with high attendance (UHR-high >12 hours), those with low attendance (UHR-low <12 hours) had more co-morbid diagnoses, larger recreational smoking (nicotine and cannabis), more depressive and negative symptoms, and had significantly lower global FA at baseline, and showed a significant increase in FA after treatment. Furthermore, UHR-low displayed large effect-size (ES) improvements on depressive and negative symptoms, and moderate to large ES improvements in several cognitive functions (verbal fluency, verbal working memory, and processing speed). In contrast, UHR-high displayed large ES improvements in UHR-symptoms, and moderate ES improvement on social and occupational functioning. Discussion Contradicting our main hypothesis, we found no effect of CR on whole-brain or regional FA after six months. This may be explained by both the low number of neurocognitive training sessions and the attrition rate. The average of 12 hours of neurocognitive training is considerably lower than the recommended dosage of 25–30 hours necessary for cognitive improvements. The continuous need to develop feasible interventions and enhance adherence is stressed. Nevertheless, non-specific treatment may improve WM-integrity in UHR-individuals with lower global baseline FA in those with more severe psychopathology. The UHR-low subgroup exhibited improvements with large ES in levels of depressive and negative symptoms, as well as cognitive functions. We speculate, whether our results reflect that UHR-individuals with higher baseline FA (approaching the healthy controls), present with a preserved structural capacity for increased demands and new learning, while UHR-individuals characterized by lower FA at baseline may be more amendable to neuroplastic treatment-effects. The results support the value of subgrouping in a clinically heterogenous UHR-population, which also applies to examining WM integrity.
-
ItemHeritability of Memory Functions and Related Brain Volumes: A Schizophrenia Spectrum Study of 214 Twins(Oxford University Press (OUP), 2020-01-01)Abstract Background Memory performance is heritable and shares partial genetic etiology with schizophrenia. How the genetic overlap between memory and schizophrenia is related to intelligence (IQ) and brain volumes has not been formally tested using twin modeling. Methods A total of 214 twins were recruited nationwide by utilization of the Danish registers, including monozygotic and dizygotic twin pairs concordant or discordant for a schizophrenia spectrum disorder and healthy control pairs. Memory/IQ assessments and MRI scans were performed and structural equation modeling was applied to examine the genetic and environmental effects and to quantify associations with schizophrenia liability. Results Significant heritability estimates were found for verbal, visual and working memory. Verbal and visual memory were associated with schizophrenia, and for visual memory the association was due to overlapping genetics. IQ was highly heritable, but only performance IQ was associated with schizophrenia. Genetic factors also contributed to total brain, right superior frontal, left rostral middle frontal and hippocampal volumes. Smaller total brain and hippocampal volumes were associated with schizophrenia, and for the left hippocampus this association was due to overlapping genetic factors. All 3 memory measures were associated with IQ, but only visual memory was associated with total brain and hippocampal volumes. Discussion Specific memory measures and brain volumes were moderately heritable and showed overlap with schizophrenia liability, suggesting partially shared etiological influences. Our findings further suggest that factors impacting IQ also influence memory, whereas memory impairments and brain volume abnormalities appear to represent separate pathological processes in the pathway to schizophrenia.
-
ItemNo Effects of Cognitive Remediation on Cerebral White Matter in Individuals at Ultra-High Risk for Psychosis-A Randomized Clinical Trial(FRONTIERS MEDIA SA, 2020-08-28)BACKGROUND: Individuals at ultra-high risk for psychosis (UHR) present with subtle alterations in cerebral white matter (WM), which appear to be associated with clinical and functional outcome. The effect of cognitive remediation on WM organization in UHR individuals has not been investigated previously. METHODS: In a randomized, clinical trial, UHR individuals aged 18 to 40 years were assigned to treatment as usual (TAU) or TAU plus cognitive remediation for 20 weeks. Cognitive remediation comprised 20 x 2-h sessions of neurocognitive and social-cognitive training. Primary outcome was whole brain fractional anisotropy derived from diffusion weighted imaging, statistically tested as an interaction between timepoint and treatment group. Secondary outcomes were restricted to five predefined region of interest (ROI) analyses on fractional anisotropy, axial diffusivity, radial diffusivity and mean diffusivity. For significant timepoint and treatment group interactions within these five ROIs, we explored associations between longitudinal changes in WM and cognitive functions/clinical symptoms. Finally, we explored dose-response effects of cognitive remediation on WM. RESULTS: A total of 111 UHR individuals were included. Attrition-rate was 26%. The cognitive remediation group completed on average 12 h of neurocognitive training, which was considerably lower than per protocol. We found no effect of cognitive remediation on whole-brain FA when compared to treatment as usual. Secondary ROI analyses revealed a nominal significant interaction between timepoint*treatment of AD in left medial lemniscus (P=0.016) which did not survive control for multiple comparisons. The exploratory test showed that this change in AD correlated to improvements of mental flexibility in the cognitive remediation group (p=0.001). We found no dose-response effect of neurocognitive training on WM. CONCLUSIONS: Cognitive remediation comprising 12 h of neurocognitive training on average did not improve global or regional WM organization in UHR individuals. Further investigations of duration and intensity of cognitive training as necessary prerequisites of neuroplasticity-based changes are warranted. CLINICAL TRIAL REGISTRATION: ClinicalTrials.gov, identifier NCT02098408.
-
ItemA machine-learning framework for robust and reliable prediction of short- and long-term treatment response in initially antipsychotic-naive schizophrenia patients based on multimodal neuropsychiatric data(NATURE PUBLISHING GROUP, 2020-08-10)The reproducibility of machine-learning analyses in computational psychiatry is a growing concern. In a multimodal neuropsychiatric dataset of antipsychotic-naïve, first-episode schizophrenia patients, we discuss a workflow aimed at reducing bias and overfitting by invoking simulated data in the design process and analysis in two independent machine-learning approaches, one based on a single algorithm and the other incorporating an ensemble of algorithms. We aimed to (1) classify patients from controls to establish the framework, (2) predict short- and long-term treatment response, and (3) validate the methodological framework. We included 138 antipsychotic-naïve, first-episode schizophrenia patients with data on psychopathology, cognition, electrophysiology, and structural magnetic resonance imaging (MRI). Perinatal data and long-term outcome measures were obtained from Danish registers. Short-term treatment response was defined as change in Positive And Negative Syndrome Score (PANSS) after the initial antipsychotic treatment period. Baseline diagnostic classification algorithms also included data from 151 matched controls. Both approaches significantly classified patients from healthy controls with a balanced accuracy of 63.8% and 64.2%, respectively. Post-hoc analyses showed that the classification primarily was driven by the cognitive data. Neither approach predicted short- nor long-term treatment response. Validation of the framework showed that choice of algorithm and parameter settings in the real data was successfully guided by results from the simulated data. In conclusion, this novel approach holds promise as an important step to minimize bias and obtain reliable results with modest sample sizes when independent replication samples are not available.