Association between cognitive status and structural brain changes in Alzheimer's disease: Clinical implication of lightweight deep learning-aided diagnosis.

Smiling difficulties in Alzheimer's disease linked to reduced nucleus accumbens and pallidum brain volume: Deep learning insights.

Subjective cognitive complaints in cognitively normal patients with Parkinson's disease predict development of dementia: A 5-year longitudinal observation.

Midlife obesity is a known risk factor for cognitive impairment, whereas its association in late life is complex, giving rise to the concept of the 'obesity paradox.' The weight-adjusted waist index (WWI), an indicator reflecting central obesity, has recently emerged. However, evidence regarding the association between WWI and cognitive impairment in Chinese older adults remains scarce. This study explores WWI's association with cognitive decline in older adults, addressing gaps in central obesity's role in neurocognitive health. A total of 5001 older adults aged ≥65 years with normal cognition from the Chinese Longitudinal Healthy Longevity Survey were included in this longitudinal analysis, with a median follow-up duration of 4 years. A time-varying Cox proportional hazards regression model was used to evaluate the association between WWI, waist circumference (WC), body mass index (BMI) and incident cognitive impairment. Nonlinear correlations were investigated using restricted-cubic-spline curves. Subgroup analyses and sensitivity analyses were conducted to enhance the robustness of findings. The incidence of cognitive impairment across the four WWI quartile groups (Q1-Q4) was 6.7%, 7.8%, 9.3% and 13.4%, respectively. WWI was positively associated with incident cognitive impairment, whether treated as a continuous variable (hazard ratio [HR] = 1.14, 95% confidence interval [95% CI] = 1.06-1.23) or a categorised variable (Q4 vs. Q1: HR = 1.70, 95% CI = 1.29-2.24; Q3 vs. Q1: HR = 1.43, 95% CI = 1.08-1.90) in models adjusted for multiple covariates. WC showed a similar trend, while BMI demonstrated no significant association. Associations persisted across subgroups and sensitivity analyses. Elevated WWI and WC, but not BMI, were significantly associated with an increased risk of incident cognitive impairment. The findings suggested that WWI may be a more precise indicator of the association between obesity and cognitive impairment.

Exploring regional BOLD fluctuations to understand the pathophysiology of mild cognitive impairments in individuals with Parkinson's disease.

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder characterized by multiscale structural brain degeneration. Many MRI-based machine learning approaches rely on coarse volumetric measures or black-box models with limited anatomical interpretability. This study aims to localize anatomically meaningful brain regions that discriminate AD from cognitively normal (CN) subjects using a hierarchical tissue-based (HTB) MRI framework. The method models gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF) volumetric changes at lobar, gyral, and 246 fine-grained subregions defined by the Brainnetome atlas. T1-weighted MRI scans from 454 participants (227 AD, 227 CN) obtained from ADNI and MIRIAD were preprocessed using AC-PC alignment, N4 bias correction, skull stripping, and nonlinear registration to MNI space. A total of 561 HTB features were extracted to train Random Forest and XGBoost classifiers using five-fold stratified cross-validation with Bayesian hyperparameter optimization. The XGBoost model achieved the best performance (Accuracy: 79.74%, ROC-AUC: 85.07%), comparable to recent atlas-based MRI classification studies, while providing improved multiscale anatomical interpretability. SHAP analysis revealed consistent hierarchical atrophy patterns in hippocampal subregions, medial amygdala, and areas 35/36 and 28/34, demonstrating that hierarchical anatomical modeling with explainable machine learning enables transparent localization of clinically meaningful AD biomarkers without reliance on black-box architectures.

Associations of plasma GFAP and P-tau217 with imaging ATN markers and cognitive decline across Centiloid scales.

Mild cognitive impairment in Parkinson's disease (PD-MCI) is regarded as a transitory stage between normal cognition and dementia, and patients with PD-MCI have an increased risk of progression into PD dementia (PDD). We aimed to explore whether specific cognitive domain dysfunction is associated with dementia conversion in PD-MCI, and quantitative electroencephalography (qEEG) biomarkers for the domain dysfunction. We retrospectively reviewed 72 patients with PD-MCI. All patients underwent EEG and neuropsychological tests, and their cognitive conversion into PDD was assessed. We assessed five cognitive domains based on neuropsychological tests: memory, attention, language, frontal/executive, and visuospatial domain MCI. The effects of each domain dysfunction on dementia conversion were investigated using Cox regression analysis after controlling for age, sex, and education. We compared the qEEG pattern between patients with specific cognitive domain impairment, which showed a significant association with dementia conversion and its counterpart. The frequency of each cognitive domain impairment was as follows: 51 frontal/executive (70.8%), 42 attention (58.3%), 38 memory (52.8%), 27 visuospatial (37.5%), and 26 language (36.1%) impairment. Among cognitive domains, only language impairment was associated with dementia conversion (p = 0.045), and it was associated with faster conversion among the converters (p = 0.007). Language impairment was associated with higher middle beta power in the bilateral parieto-occipital regions, high beta power in the bilateral parieto-occipital regions and right frontal region, and gamma power in the bilateral occipital regions. These findings suggest that language impairment could be a prognostic marker with distinct qEEG findings in PD-MCI.

To investigate alterations in resting-state electroencephalogram (EEG) microstates across the cognitive spectrum of Parkinson's disease (PD) and to evaluate their utility as electrophysiological biomarkers of cognitive impairment. Resting-state EEG was recorded using a 19-channel system during a 3-min eyes-closed session in 36 healthy controls (HC), 38 PD patients with normal cognition (PDNC), and 39 PD patients with dementia (PDD). Temporal parameters (duration, occurrence, coverage) of six canonical microstates (A-F) were computed and compared across groups. Correlation analyses were conducted between microstate metrics and Montreal Cognitive Assessment (MoCA) scores. Significant group differences were found in microstate dynamics. The PDD group exhibited a longer mean duration of microstates A, C, and E, and a lower occurrence per second of microstates B and C compared to both the PDNC and HC groups (all P < 0.05). Critically, the duration of microstates A and C showed significant negative correlations with MoCA scores (P < 0.05), while the occurrence of microstates B and C demonstrated positive correlations with MoCA scores (P < 0.05). Specific EEG microstate abnormalities are associated with cognitive status in PD. The prolongation of microstates A and C and the reduced occurrence of microstates B and C are stage-sensitive biomarkers that reflect the severity of cognitive decline, providing novel insights into the neural mechanisms of PD-related cognitive dysfunction.

Alzheimer disease (AD) and its related disorders (ADRDs) are characterized by a high frequency of copathologies. We aimed to determine the specificity of plasma pTau217, glial fibrillary acidic protein (GFAP), and neurofilament light chain (NfL) for AD neuropathological change (ADNC) in the presence of common ADRD copathologies. pTau217, GFAP, and NfL were measured using S-PLEX immunoassays from Meso Scale Discovery in banked plasma samples from 2 groups of participants in the Massachusetts Alzheimer's Disease Research Center (MADRC) Longitudinal Cohort study: (1) participants spanning the cognitive spectrum, who underwent brain autopsy, and blood collection within 6 years before death, and (2) participants with normal cognition and no neurologic diagnosis during 5 years of follow-up, but no autopsy data (normal controls [NCs]). Cross-sectional associations between biomarker levels and ADNC, primary neuropathologic diagnosis (NPDx1), and presence of non-AD copathologies were evaluated using linear regression models controlling for age, sex, and time to death. One hundred eighty-seven participants with brain autopsy (NPDx1: AD n = 85; other n = 102; mean age: 74.3 years, 38.5% female; interval blood collection-death [mean ± SD]: 2.8 ± 1.6 years) and 67 NC without brain autopsy (mean age: 66.5 years, 71.6% female) were included. pTau217, but not GFAP, levels increased stepwise with increasing Thal phases (β = 0.61; 95% CI [0.24-0.97] to β = 0.91 [0.55-1.27]) and Braak stages (β = 0.59; [0.16-1.01] to β = 0.74 [0.33-1.15]). Although 23% of individuals with a non-AD NPDx1 had increased pTau217 levels using a cutoff defined by the contrast between ADNC and NC, the majority (62%) had intermediate/high ADNC copathology and the remaining pTau217+ individuals had borderline increased levels. By contrast, 48% of individuals without ADNC had increased GFAP levels. pTau217 and GFAP were not different in the presence or absence of cerebral amyloid angiopathy, α-synuclein or TDP-43 proteinopathies, or primary tauopathies. NfL was not specifically associated with ADNC. Plasma pTau217, but not GFAP or NfL, levels accurately reflect the presence of ADNC in the brain even in individuals with an NPDx1 of a non-AD dementia. Thus, a positive plasma pTau217 test in an individual with a suspected non-AD dementia should not necessarily be considered a misdiagnosis of the presumed non-AD dementia or as a false positive, but rather as evidence of ADNC copathology.

Diabetes-associated cognitive dysfunction (DACD) is characterized by crosstalk between metabolic disturbances and neural dysfunction. Prolactin (PRL), a pituitary hormone involved in glucose regulation and neuroinflammation, may contribute to DACD development. This study aims to investigate the relationships and underlying mechanisms between serum PRL and DACD in patients with type 2 diabetes (T2DM) via neuroimaging. 522 with normal cognition and 418 with mild cognitive impairment (MCI) were included after age- and gender-matching among 1167participants with T2DM. 247 individuals underwent functional magnetic resonance imaging scans to assess brain structure and function. Structural equation modeling (SEM) was used to explore the mechanism linking serum PRL to cognition. Relative weight analysis (RWA) indicated that PRL and LH accounted for 8.14% and 7.53% of MCI risk in T2DM across 1167 participants. However, multivariate logistic regression analysis showed that PRL, rather than LH, was an independent risk factor for MCI in T2DM (OR = 0.907, 95% CI: 0.849-0.968, P = 0.003), with the lowest PRL tertile group showing significant cognitive deficits and reduced hippocampal-amygdaloid volumes. Moreover, SEM indicated that PRL positively influenced cognition and exerted an indirect effect via limbic structures, contributing 14.07% to the total effect. Additionally, the lowest PRL tertile group exhibited a significant decreased functional connectivity between the right amygdala and right frontal lobe, partially mediating the link between serum PRL and MoCA scores. Decreased serum PRL is associated with reduced hippocampal and amygdaloid volumes, impaired amygdala functional connectivity, and diminished cognitive performance in T2DM.

Pathological and neuroimaging changes in the cerebellum of Alzheimer's disease (AD) patients have been well documented. However, the changes in cerebellar amyloid plaque deposition connectivity networks during AD progression based on positron emission tomography (PET) imaging remain unclear. We selected 18F-florbetapir PET (18F-AV45 PET) imaging data from the Alzheimer's disease neuroimaging initiative (ADNI) dataset (n = 612) and employed graph theoretical analysis to examine amyloid plaque deposition connectivity, comparing the connectivity differences across cognitively normal (CN), early mild cognitive impairment (EMCI), late mild cognitive impairment (LMCI), and AD groups. In addition, we combined graph theoretical features with the standardized uptake value ratio (SUVR) of regions of interest and applied them to machine learning models for the early diagnosis of AD. As cognitive decline progressed, significant changes in cerebellar network connectivity were observed across groups. Regarding local connectivity, changes in betweenness centrality were evident in multiple cerebellar regions at different cognitive stages. Cerebellar amyloid networks revealed early changes in amyloid plaque deposition connectivity. The machine learning model achieved an area under the curve (AUC) of 0.950 for distinguishing AD from CN, 0.995 for CN vs. EMCI, 0.964 for EMCI vs. LMCI and 0.632 for LMCI vs. AD. These findings provide new insights into the cerebellar pathological features of AD and highlight the potential of this approach for early identification and prediction of AD progression.

Alzheimer's disease (AD) is a degenerative neurological disorder marked by cognitive decline and functional disability. Despite the extensive use of magnetic resonance imaging (MRI) in machine learning (ML)-based AD studies, the relative and combined contributions of MRI-derived morphometric (MO), microstructural (MS), and graph-theoretical (GT) features are still not well explored in a unified, comparative framework. It remains unclear whether adding multimodal MRI-derived features consistently improves the predictive performance of ML-based approaches for AD diagnosis and cognitive decline. Addressing this gap, this study systematically analyzed the individual (MO, MS, GT) and combined (MO+MS, MO+GT, MS+GT, MO+MS+GT) utility of MRI-based feature sets. We developed an ensemble-based ML framework with a nested cross-validation module for two key tasks: (i) Alzheimer's disease cognitive stage classification (DSC) and (ii) longitudinal cognitive decline prediction (LCDP) in terms of mini-mental state examination (MMSE) score. In this study, we conducted feature ablation and statistical analysis to evaluate performance improvements resulting from the incremental addition of feature sets. The results of the study indicated that the proposed ensemble-based ML approach achieved the best predictive performance (balanced accuracy [BACC]: 0.898 ± 0.051) using a combination of MO and MS feature sets for cognitively normal (CN) vs. AD dementia (CN-ADD). In contrast, the best results for mild cognitive impairment (MCI) vs. ADD (MCI-ADD) and CN-MCI were achieved using the MO feature set alone, with BACC of 0.769 ± 0.116 and 0.652 ± 0.044, respectively. Likewise, for the LCDP task, the MO-based ensemble learner achieved an R2 of 0.212 ± 0.177. These results demonstrate that MO features capture the most robust disease-related information, while multimodal integration offers task-specific and limited benefits. In addition, these findings demonstrate the potential of integrated MRI-derived features in ML frameworks for enhancing ADD diagnosis and cognitive decline prediction and underscore the importance of feature selection based on task complexity.

Olfactory impairment and abnormal frontal EEG oscillations are recognized as early markers of Alzheimer’s disease (AD). Using a publicly available olfactory EEG dataset of 35 subjects spanning normal cognition, amnestic mild cognitive impairment (aMCI), and AD, each with MMSE scores and demographics, stimulus-locked epochs from four electrodes (Fp1, Fz, Cz, Pz) were processed with wavelet-based time–frequency analysis. Band-limited power ratios (delta, theta, alpha, beta) were computed as log-transformed post-odor/baseline values and aggregated to subject-level features. Statistical analyses revealed graded attenuation of odor-evoked frontal (Fp1) band-power ratios across groups, with significant differences in several band–odor combinations. PCA of Fp1 features showed partial separation of diagnostic categories, while multi-channel features offered weaker discrimination. Random forest classifiers trained on Fp1-only features achieved 66.7% test accuracy, outperforming the four-channel model (55.6%), with moderate sensitivity, specificity, and precision. These findings highlight that compact frontal wavelet-derived band-power ratios during olfactory stimulation carry diagnostically relevant information for distinguishing Normal, aMCI, and AD. The transparent pipeline, combining time–frequency processing, subject-level aggregation, and multiclass classification, offers a scalable framework that can be extended to larger cohorts or integrated with multimodal biomarkers.

Alzheimer's disease (AD) classification using machine learning has increasingly relied on multimodal inputs such as Magnetic Resonance Imaging (MRI), cognitive assessments, and biological markers. This study evaluates whether integrating these sources enhances predictive performance compared to using them independently. Neural networks were trained on data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) to classify subjects into Cognitively Normal (CN), Mild Cognitive Impairment (MCI), and AD categories using unimodal, bimodal, and trimodal input configurations. Contrary to expectations, multimodal models did not consistently outperform unimodal ones. The highest test accuracy (81%) was achieved by both the cognitive-only and trimodal models, with the former also demonstrating superior class-wise performance. These findings suggest that neuropsychological features may carry greater diagnostic value than imaging or fluid biomarkers, underscoring the importance of more targeted data fusion strategies. Furthermore, the inclusion of biological markers did not significantly improve early MCI detection, likely due to their limited dimensionality and the model's constrained ability to extract meaningful patterns from such inputs.

To identify mild cognitive impairment (MCI) in Parkinson's disease (PD) using two brief tests the Amnesia Light and Brief Asssment (ALBA) and the door Picture Naming and Immediate Recall (dPICNIR) in 6-8 minutes. The ALBA, the dPICNIR and the third version of the Addenbrooke's Cognitive Examination III (ACE-III) were administered to 124 participants, equally divided into PD patients and socio-demographically matched normal controls (NC). The PD group was divided into those with normal cognitive functions (PD-CN) and with MCI (PD-MCI) using neuropsychological tests. Cognitive impairment in the PD group was mild, with significantly lower ACE-III scores than in NC (91 vs. 96 points). Despite these subtle deficits, gesture recall in the ALBA was significantly lower even in the PD-CN group compared to the NC. PD-MCI patients had other significant deficits in the ALBA and PICNIR tests. In the PD group, the gesture recall in the ALBA and correctly recalled pictures in the dPICNIR correlated with the results of verbal fluency and trail making tests, followed by memory tests and all ACE-III scores except visuospatial one. In contrast, correctly recalled sentence words in the ALBA correlated with the memory and language scores in the ACE-III test and memory test scores. Subtle cognitive changes in PD can be detected through gesture recall test, even in those with normal cognition. The ALBA and PICNIR tests are effective in identifying MCI in PD and provide a brief and valid assessment of cognitive functions. They are freely available at www.abadeco.cz.

Objective This study aimed to assess the validity and reliability of the Turkish version of the Brief Cognitive Assessment Tool (BCAT) in differentiating between older adults with major Neurocognitive Disorder (MNCD) and mild neurocognitive disorder (MiNCD), and cognitively normal (CN) individuals. Method Participants were categorized into MNCD (n = 152), MiNCD (n = 73), and CN (n = 53) groups by Diagnostic and Statistical Manual of Mental Disorders-5 criteria. The BCAT was translated and culturally adapted into Turkish. All participants completed the BCAT, Mini Mental State Examination, Montreal Cognitive Assessment (MoCA), Clock Drawing Test, and Öktem’s Auditory Verbal Learning Test. Internal consistency, test–retest reliability, and concurrent validity with MoCA were assessed. Receiver operating characteristic analysis was employed to determine a cutoff point for BCAT. Results BCAT scores showed high internal consistency and test–retest reliability. BCAT and MoCA scores correlated strongly, confirming convergent validity. MNCD group had lower BCAT scores than MiNCD and CN across multiple cognitive domains. At a cutoff score of 33, BCAT distinguished MNCD from MiNCD and CN. Conclusions Our findings suggest that the Turkish version of the BCAT is a valid and reliable screening tool for cognitive impairment in older adults. It effectively differentiates between MNCD, MiNCD, and CN individuals, supporting its clinical utility for early detection of neurocognitive disorders.

Falls are the leading cause of disability, premature institutionalization, and mortality in the aging population. Older adults with cognitive impairment are more susceptible to falls and have a significantly higher risk of falls compared to those with normal cognition. Physical activity (PA) is a key fall prevention strategy; however, evidence on the dose-response relationship between PA and falls among older adults with impaired cognition is limited. We investigated associations of physical activity and health symptoms with falls in cognitively diverse older adults to identify differential effects. We conducted a cross-sectional secondary analysis of 6,781 adults aged ≥ 65 who participated in both 2020 and 2022 Health and Retirement Study (HRS) interview years. Weighted survey logistic regression models estimated associations between PA levels and self-reported falls, defined as reporting at least one fall in the previous two years. Cognitive status was assessed using the 27-point HRS cognition scale. Models sequentially adjusted for demographics and health symptoms, including pain, dyspnea, obesity, and depressive symptoms. In fully adjusted models, high-intensity PA was significantly associated with lower odds of falling (OR = 0.76, 95% CI: 0.60-0.98). Associations observed for moderate PA in partially adjusted models were attenuated and no longer statistically significant after full adjustment (OR = 0.83, 95% CI: 0.66-1.05). Mild PA was not significantly associated with falls across models. Stratified analyses showed a consistent pattern of lower odds of falls with increasing PA intensity in both cognition groups; however, significant associations for moderate and high PA were lost after adjusting for health symptoms in the normal cognition group. Among those with impaired cognition, only high-level PA demonstrated a similar attenuated trend, while mild and moderate PA remained non-significant across models. Chronic pain, dyspnea, and depressive symptoms were significantly associated with increased fall risk in both subgroups, while obesity was only significantly associated with falls in the normal cognition group. Moderate and high-intensity PA were associated with lower odds of falls, with important effect modification by cognitive status and health symptoms. These findings suggest that fall prevention strategies should account for cognitive function and symptom burden when promoting physical activity in older adults.

ABSTRACTBackgroundPost‐stroke cognitive impairment (PSCI) is a common complication following a stroke. Recent findings highlight the role of miR‐502‐3p in both vascular and neurodegenerative diseases. However, the role in PSCI remains uncovered.ObjectiveThis study emphasized the differential expression of miR‐502‐3p and subsequently evaluated the predictive value of miR‐502‐3p expression levels for PSCI.Materials and methodsThe study subjects included 112 patients with PSCI and 161 individuals with post‐stroke cognitive normality. The relative expression of miR‐502‐3p was calculated by qPCR, while its predictive value for PSCI was assessed via ROC curve. Pearson's correlation coefficient was utilized to analyze the correlation between serum miR‐502p‐3p levels and PSCI. Multivariate logistic regression was used to identify risk factors of PSCI.ResultsSerum miR‐502‐3p was identified as significantly elevated in the PSCI group. The area under the ROC curve was 0.850, with a sensitivity of 76.79% and a specificity of 77.64%. The miR‐502‐3p level was positively correlated with both NIHSS and mRS scores. In addition, a negative correlation was observed between the miR‐502‐3p level and MoCA score. Elevated miR‐502‐3p and hypertension were identified as independent risk factors for PSCI.ConclusionSignificantly elevated serum miR‐502‐3p was a promising biomarker for the onset of PSCI. Elevated miR‐502‐3p and hypertension were independent risk factors for PSCI.

Predictive factors of global cognitive impairment in de novo Parkinson disease with normal cognition at baseline: a 10-year cohort study.