Working Memory Training in Amnestic and Non-amnestic Patients With Mild Cognitive Impairment: Preliminary Findings From Genotype Variants on Training Effects

Machine learning-based quantification for disease uncertainty increases the statistical power of genetic association studies

MOTIVATION

Allowance for increasingly large samples is a key to identify the association of genetic variants with Alzheimer's disease (AD) in genome-wide association studies (GWAS). Accordingly, we aimed to develop a method that incorporates patients with mild cognitive impairment and unknown cognitive status in GWAS using a machine learning-based AD prediction model.

RESULTS

Simulation analyses showed that weighting imputed phenotypes method increased the statistical power compared to ordinary logistic regression using only AD cases and controls. Applied to real-world data, the penalized logistic method had the highest AUC (0.96) for AD prediction and weighting imputed phenotypes method performed well in terms of power. We identified an association (P<5.0×10-8) of AD with several variants in the APOE region and rs143625563 in LMX1A. Our method, which allows the inclusion of individuals with mild cognitive impairment, improves the statistical power of GWAS for AD. We discovered a novel association with LMX1A.

AVAILABILITY AND IMPLEMENTATION

Simulation codes can be accessed at https://github.com/Junkkkk/wGEE_GWAS.

Cognitive training based on functional near-infrared spectroscopy neurofeedback for the elderly with mild cognitive impairment: a preliminary study

Introduction

Mild cognitive impairment (MCI) is often described as an intermediate stage of the normal cognitive decline associated with aging and dementia. There is a growing interest in various non-pharmacological interventions for MCI to delay the onset and inhibit the progressive deterioration of daily life functions. Previous studies suggest that cognitive training (CT) contributes to the restoration of working memory and that the brain-computer-interface technique can be applied to elicit a more effective treatment response. However, these techniques have certain limitations. Thus, in this preliminary study, we applied the neurofeedback paradigm during CT to increase the working memory function of patients with MCI.

Methods

Near-infrared spectroscopy (NIRS) was used to provide neurofeedback by measuring the changes in oxygenated hemoglobin in the prefrontal cortex. Thirteen elderly MCI patients who received CT-neurofeedback sessions four times on the left dorsolateral prefrontal cortex (dlPFC) once a week were recruited as participants.

Results

Compared with pre-intervention, the activity of the targeted brain region increased when the participants first engaged in the training; after 4 weeks of training, oxygen saturation was significantly decreased in the left dlPFC. The participants demonstrated significantly improved working memory compared with pre-intervention and decreased activity significantly correlated with improved cognitive performance.

Conclusion

Our results suggest that the applications for evaluating brain-computer interfaces can aid in elucidation of the subjective mental workload that may create additional or decreased task workloads due to CT.

Social and non-social working memory in neurodegeneration

Although social functioning relies on working memory, whether a social-specific mechanism exists remains unclear. This undermines the characterization of neurodegenerative conditions with both working memory and social deficits. We assessed working memory domain-specificity across behavioral, electrophysiological, and neuroimaging dimensions in 245 participants. A novel working memory task involving social and non-social stimuli with three load levels was assessed across controls and different neurodegenerative conditions with recognized impairments in: working memory and social cognition (behavioral-variant frontotemporal dementia); general cognition (Alzheimer's disease); and unspecific patterns (Parkinson's disease). We also examined resting-state theta oscillations and functional connectivity correlates of working memory domain-specificity. Results in controls and all groups together evidenced increased working memory demands for social stimuli associated with frontocinguloparietal theta oscillations and salience network connectivity. Canonical frontal theta oscillations and executive-default mode network anticorrelation indexed non-social stimuli. Behavioral-variant frontotemporal dementia presented generalized working memory deficits related to posterior theta oscillations, with social stimuli linked to salience network connectivity. In Alzheimer's disease, generalized working memory impairments were related to temporoparietal theta oscillations, with non-social stimuli linked to the executive network. Parkinson's disease showed spared working memory performance and canonical brain correlates. Findings support a social-specific working memory and related disease-selective pathophysiological mechanisms.

Cortical Thickness Changes After Computerized Working Memory Training in Patients With Mild Cognitive Impairment

Background

Adaptive computerized working memory (WM) training has shown favorable effects on cerebral cortical thickness as compared to non-adaptive training in healthy individuals. However, knowledge of WM training-related morphological changes in mild cognitive impairment (MCI) is limited.

Objective

The primary objective of this double-blind randomized study was to investigate differences in longitudinal cortical thickness trajectories after adaptive and non-adaptive WM training in patients with MCI. We also investigated the genotype effects on cortical thickness trajectories after WM training combining these two training groups using longitudinal structural magnetic resonance imaging (MRI) analysis in Freesurfer.

Method

Magnetic resonance imaging acquisition at 1.5 T were performed at baseline, and after four- and 16-weeks post training. A total of 81 individuals with MCI accepted invitations to undergo 25 training sessions over 5 weeks. Longitudinal Linear Mixed effect models investigated the effect of adaptive vs. non-adaptive WM training. The LME model was fitted for each location (vertex). On all statistical analyzes, a threshold was applied to yield an expected false discovery rate (FDR) of 5%. A secondary LME model investigated the effects of LMX1A and APOE-ε4 on cortical thickness trajectories after WM training.

Results

A total of 62 participants/patients completed the 25 training sessions. Structural MRI showed no group difference between the two training regimes in our MCI patients, contrary to previous reports in cognitively healthy adults. No significant structural cortical changes were found after training, regardless of training type, across all participants. However, LMX1A-AA carriers displayed increased cortical thickness trajectories or lack of decrease in two regions post-training compared to those with LMX1A-GG/GA. No training or training type effects were found in relation to the APOE-ε4 gene variants.

Conclusion

The MCI patients in our study, did not have improved cortical thickness after WM training with either adaptive or non-adaptive training. These results were derived from a heterogeneous population of MCI participants. The lack of changes in the cortical thickness trajectory after WM training may also suggest the lack of atrophy during this follow-up period. Our promising results of increased cortical thickness trajectory, suggesting greater neuroplasticity, in those with LMX1A-AA genotype need to be validated in future trials.

Predicting MCI to AD Conversation Using Integrated sMRI and rs-fMRI: Machine Learning and Graph Theory Approach

BACKGROUND

Graph theory and machine learning have been shown to be effective ways of classifying different stages of Alzheimer's disease (AD). Most previous studies have only focused on inter-subject classification with single-mode neuroimaging data. However, whether this classification can truly reflect the changes in the structure and function of the brain region in disease progression remains unverified. In the current study, we aimed to evaluate the classification framework, which combines structural Magnetic Resonance Imaging (sMRI) and resting-state functional Magnetic Resonance Imaging (rs-fMRI) metrics, to distinguish mild cognitive impairment non-converters (MCInc)/AD from MCI converters (MCIc) by using graph theory and machine learning.

METHODS

With the intra-subject (MCInc vs. MCIc) and inter-subject (MCIc vs. AD) design, we employed cortical thickness features, structural brain network features, and sub-frequency (full-band, slow-4, slow-5) functional brain network features for classification. Three feature selection methods [random subset feature selection algorithm (RSFS), minimal redundancy maximal relevance (mRMR), and sparse linear regression feature selection algorithm based on stationary selection (SS-LR)] were used respectively to select discriminative features in the iterative combinations of MRI and network measures. Then support vector machine (SVM) classifier with nested cross-validation was employed for classification. We also compared the performance of multiple classifiers (Random Forest, K-nearest neighbor, Adaboost, SVM) and verified the reliability of our results by upsampling.

RESULTS

We found that in the classifications of MCIc vs. MCInc, and MCIc vs. AD, the proposed RSFS algorithm achieved the best accuracies (84.71, 89.80%) than the other algorithms. And the high-sensitivity brain regions found with the two classification groups were inconsistent. Specifically, in MCIc vs. MCInc, the high-sensitivity brain regions associated with both structural and functional features included frontal, temporal, caudate, entorhinal, parahippocampal, and calcarine fissure and surrounding cortex. While in MCIc vs. AD, the high-sensitivity brain regions associated only with functional features included frontal, temporal, thalamus, olfactory, and angular.

CONCLUSIONS

These results suggest that our proposed method could effectively predict the conversion of MCI to AD, and the inconsistency of specific brain regions provides a novel insight for clinical AD diagnosis.