Disease-driven domain generalization for neuroimaging-based assessment of Alzheimer's disease.

Development of deep learning models to evaluate structural brain changes caused by cognitive impairment in MRI scans holds significant translational value. The efficacy of these models often encounters challenges due to variabilities arising from different data generation protocols, imaging equipment, radiological artifacts, and shifts in demographic distributions. Domain generalization (DG) techniques show promise in addressing these challenges by enabling the model to learn from one or more source domains and apply this knowledge to new, unseen target domains. Here we present a framework that utilizes model interpretability to enhance the generalizability of classification models across various cohorts. We used MRI scans and clinical diagnoses from four independent cohorts: Alzheimer's Disease Neuroimaging Initiative (ADNI, n = 1821), the Framingham Heart Study (FHS, n = 304), the Australian Imaging Biomarkers & Lifestyle Study of Ageing (AIBL, n = 661), and the National Alzheimer's Coordinating Center (NACC, n = 4647). With this data, we trained a deep neural network to focus on areas of the brain identified as relevant to the disease for model training. Our approach involved training a classifier to differentiate between structural neurodegeneration in individuals with normal cognition (NC), mild cognitive impairment (MCI), and dementia due to Alzheimer's disease (AD). This was achieved by aligning class-wise attention with a unified visual saliency prior, which was computed offline for each class using all the training data. Our method not only competes with state-of-the-art approaches but also shows improved correlation with postmortem histology. This alignment with the gold standard evidence is a significant step towards validating the effectiveness of DG frameworks, paving the way for their broader application in the field.

Cerebral Microbleeds in Different Brain Regions and Their Associations With the Digital Clock-Drawing Test: Secondary Analysis of the Framingham Heart Study.

BACKGROUND: Cerebral microbleeds (CMB) increase the risk for Alzheimer disease. Current neuroimaging methods that are used to detect CMB are costly and not always accessible. OBJECTIVE: This study aimed to explore whether the digital clock-drawing test (DCT) may provide a behavioral indicator of CMB. METHODS: In this study, we analyzed data from participants in the Framingham Heart Study offspring cohort who underwent both brain magnetic resonance imaging scans (Siemens 1.5T, Siemens Healthcare Private Limited; T2*-GRE weighted sequences) for CMB diagnosis and the DCT as a predictor. Additionally, paper-based clock-drawing tests were also collected during the DCT. Individuals with a history of dementia or stroke were excluded. Robust multivariable linear regression models were used to examine the association between DCT facet scores with CMB prevalence, adjusting for relevant covariates. Receiver operating characteristic (ROC) curve analyses were used to evaluate DCT facet scores as predictors of CMB prevalence. Sensitivity analyses were conducted by further including participants with stroke and dementia. RESULTS: The study sample consisted of 1020 (n=585, 57.35% female) individuals aged 45 years and older (mean 72, SD 7.9 years). Among them, 64 (6.27%) participants exhibited CMB, comprising 46 with lobar-only, 11 with deep-only, and 7 with mixed (lobar+deep) CMB. Individuals with CMB tended to be older and had a higher prevalence of mild cognitive impairment and higher white matter hyperintensities compared to those without CMB (P<.05). While CMB were not associated with the paper-based clock-drawing test, participants with CMB had a lower overall DCT score (CMB: mean 68, SD 23 vs non-CMB: mean 76, SD 20; P=.009) in the univariate comparison. In the robust multiple regression model adjusted for covariates, deep CMB were significantly associated with lower scores on the drawing efficiency (ß=-0.65, 95% CI -1.15 to -0.15; P=.01) and simple motor (ß=-0.86, 95% CI -1.43 to -0.30; P=.003) domains of the command DCT. In the ROC curve analysis, DCT facets discriminated between no CMB and the CMB subtypes. The area under the ROC curve was 0.76 (95% CI 0.69-0.83) for lobar CMB, 0.88 (95% CI 0.78-0.98) for deep CMB, and 0.98 (95% CI 0.96-1.00) for mixed CMB, where the area under the ROC curve value nearing 1 indicated an accurate model. CONCLUSIONS: The study indicates a significant association between CMB, especially deep and mixed types, and reduced performance in drawing efficiency and motor skills as assessed by the DCT. This highlights the potential of the DCT for early detection of CMB and their subtypes, providing a reliable alternative for cognitive assessment and making it a valuable tool for primary care screening before neuroimaging referral.

DREAMER: a computational framework to evaluate readiness of datasets for machine learning.

BACKGROUND: Machine learning (ML) has emerged as the predominant computational paradigm for analyzing large-scale datasets across diverse domains. The assessment of dataset quality stands as a pivotal precursor to the successful deployment of ML models. In this study, we introduce DREAMER (Data REAdiness for MachinE learning Research), an algorithmic framework leveraging supervised and unsupervised machine learning techniques to autonomously evaluate the suitability of tabular datasets for ML model development. DREAMER is openly accessible as a tool on GitHub and Docker, facilitating its adoption and further refinement within the research community.. RESULTS: The proposed model in this study was applied to three distinct tabular datasets, resulting in notable enhancements in their quality with respect to readiness for ML tasks, as assessed through established data quality metrics. Our findings demonstrate the efficacy of the framework in substantially augmenting the original dataset quality, achieved through the elimination of extraneous features and rows. This refinement yielded improved accuracy across both supervised and unsupervised learning methodologies. CONCLUSION: Our software presents an automated framework for data readiness, aimed at enhancing the integrity of raw datasets to facilitate robust utilization within ML pipelines. Through our proposed framework, we streamline the original dataset, resulting in enhanced accuracy and efficiency within the associated ML algorithms.

Prediction of Alzheimer's disease progression within 6 years using speech: A novel approach leveraging language models.

INTRODUCTION: Identification of individuals with mild cognitive impairment (MCI) who are at risk of developing Alzheimer's disease (AD) is crucial for early intervention and selection of clinical trials. METHODS: We applied natural language processing techniques along with machine learning methods to develop a method for automated prediction of progression to AD within 6 years using speech. The study design was evaluated on the neuropsychological test interviews of n = 166 participants from the Framingham Heart Study, comprising 90 progressive MCI and 76 stable MCI cases. RESULTS: Our best models, which used features generated from speech data, as well as age, sex, and education level, achieved an accuracy of 78.5% and a sensitivity of 81.1% to predict MCI-to-AD progression within 6 years. DISCUSSION: The proposed method offers a fully automated procedure, providing an opportunity to develop an inexpensive, broadly accessible, and easy-to-administer screening tool for MCI-to-AD progression prediction, facilitating development of remote assessment. HIGHLIGHTS: Voice recordings from neuropsychological exams coupled with basic demographics can lead to strong predictive models of progression to dementia from mild cognitive impairment. The study leveraged AI methods for speech recognition and processed the resulting text using language models. The developed AI-powered pipeline can lead to fully automated assessment that could enable remote and cost-effective screening and prognosis for Alzehimer's disease.

Speech patterns during memory recall relates to early tau burden across adulthood.

INTRODUCTION: Early cognitive decline may manifest in subtle differences in speech. METHODS: We examined 238 cognitively unimpaired adults from the Framingham Heart Study (32-75 years) who completed amyloid and tau PET imaging. Speech patterns during delayed recall of a story memory task were quantified via five speech markers, and their associations with global amyloid status and regional tau signal were examined. RESULTS: Total utterance time, number of between-utterance pauses, speech rate, and percentage of unique words significantly correlated with delayed recall score although the shared variance was low (2%-15%). Delayed recall score was not significantly different between ß-amyoid-positive (Aß+) and -negative (Aß-) groups and was not associated with regional tau signal. However, longer and more between-utterance pauses, and slower speech rate were associated with increased tau signal across medial temporal and early neocortical regions. DISCUSSION: Subtle speech changes during memory recall may reflect cognitive impairment associated with early Alzheimer's disease pathology. HIGHLIGHTS: Speech during delayed memory recall relates to tau PET signal across adulthood. Delayed memory recall score was not associated with tau PET signal. Speech shows greater sensitivity to detecting subtle cognitive changes associated with early tau accumulation. Our cohort spans adulthood, while most PET imaging studies focus on older adults.

Flortaucipir tau PET findings from former professional and college American football players in the DIAGNOSE CTE research project.

INTRODUCTION: Tau is a key pathology in chronic traumatic encephalopathy (CTE). Here, we report our findings in tau positron emission tomography (PET) measurements from the DIAGNOSE CTE Research Project. METHOD: We compare flortaucipir PET measures from 104 former professional players (PRO), 58 former college football players (COL), and 56 same-age men without exposure to repetitive head impacts (RHI) or traumatic brain injury (unexposed [UE]); characterize their associations with RHI exposure; and compare players who did or did not meet diagnostic criteria for traumatic encephalopathy syndrome (TES). RESULTS: Significantly elevated flortaucipir uptake was observed in former football players (PRO+COL) in prespecified regions (p < 0.05). Association between regional flortaucipir uptake and estimated cumulative head impact exposure was only observed in the superior frontal region in former players over 60 years old. Flortaucipir PET was not able to differentiate TES groups. DISCUSSION: Additional studies are needed to further understand tau pathology in CTE and other individuals with a history of RHI.

Optimal blood tau species for the detection of Alzheimer's disease neuropathology: an immunoprecipitation mass spectrometry and autopsy study.

Plasma-to-autopsy studies are essential for validation of blood biomarkers and understanding their relation to Alzheimer's disease (AD) pathology. Few such studies have been done on phosphorylated tau (p-tau) and those that exist have made limited or no comparison of the different p-tau variants. This study is the first to use immunoprecipitation mass spectrometry (IP-MS) to compare the accuracy of eight different plasma tau species in predicting autopsy-confirmed AD. The sample included 123 participants (AD = 69, non-AD = 54) from the Boston University Alzheimer's disease Research Center who had an available ante-mortem plasma sample and donated their brain. Plasma samples proximate to death were analyzed by targeted IP-MS for six different tryptic phosphorylated (p-tau-181, 199, 202, 205, 217, 231), and two non-phosphorylated tau (195-205, 212-221) peptides. NIA-Reagan Institute criteria were used for the neuropathological diagnosis of AD. Binary logistic regressions tested the association between each plasma peptide and autopsy-confirmed AD status. Area under the receiver operating curve (AUC) statistics were generated using predicted probabilities from the logistic regression models. Odds Ratio (OR) was used to study associations between the different plasma tau species and CERAD and Braak classifications. All tau species were increased in AD compared to non-AD, but p-tau217, p-tau205 and p-tau231 showed the highest fold-changes. Plasma p-tau217 (AUC = 89.8), p-tau231 (AUC = 83.4), and p-tau205 (AUC = 81.3) all had excellent accuracy in discriminating AD from non-AD brain donors, even among those with CDR < 1). Furthermore, p-tau217, p-tau205 and p-tau231 showed the highest ORs with both CERAD (ORp-tau217 = 15.29, ORp-tau205 = 5.05 and ORp-tau231 = 3.86) and Braak staging (ORp-tau217 = 14.29, ORp-tau205 = 5.27 and ORp-tau231 = 4.02) but presented increased levels at different amyloid and tau stages determined by neuropathological examination. Our findings support plasma p-tau217 as the most promising p-tau species for detecting AD brain pathology. Plasma p-tau231 and p-tau205 may additionally function as markers for different stages of the disease.

Exploring cognitive progression subtypes in the Framingham Heart Study.

INTRODUCTION: Alzheimer's disease (AD) is a heterogeneous disorder characterized by complex underlying neuropathology that is not fully understood. This study aimed to identify cognitive progression subtypes and examine their correlation with clinical outcomes. METHODS: Participants of this study were recruited from the Framingham Heart Study. The Subtype and Stage Inference (SuStaIn) method was used to identify cognitive progression subtypes based on eight cognitive domains. RESULTS: Three cognitive progression subtypes were identified, including verbal learning (Subtype 1), abstract reasoning (Subtype 2), and visual memory (Subtype 3). These subtypes represent different domains of cognitive decline during the progression of AD. Significant differences in age of onset among the different subtypes were also observed. A higher SuStaIn stage was significantly associated with increased mortality risk. DISCUSSION: This study provides a characterization of AD heterogeneity in cognitive progression, emphasizing the importance of developing personalized approaches for risk stratification and intervention. Highlights: We used the Subtype and Stage Inference (SuStaIn) method to identify three cognitive progression subtypes.Different subtypes have significant variations in age of onset.Higher stages of progression are associated with increased mortality risk.

Detection of Mild Cognitive Impairment From Non-Semantic, Acoustic Voice Features: The Framingham Heart Study.

BACKGROUND: With the aging global population and the rising burden of Alzheimer disease and related dementias (ADRDs), there is a growing focus on identifying mild cognitive impairment (MCI) to enable timely interventions that could potentially slow down the onset of clinical dementia. The production of speech by an individual is a cognitively complex task that engages various cognitive domains. The ease of audio data collection highlights the potential cost-effectiveness and noninvasive nature of using human speech as a tool for cognitive assessment. OBJECTIVE: This study aimed to construct a machine learning pipeline that incorporates speaker diarization, feature extraction, feature selection, and classification to identify a set of acoustic features derived from voice recordings that exhibit strong MCI detection capability. METHODS: The study included 100 MCI cases and 100 cognitively normal controls matched for age, sex, and education from the Framingham Heart Study. Participants' spoken responses on neuropsychological tests were recorded, and the recorded audio was processed to identify segments of each participant's voice from recordings that included voices of both testers and participants. A comprehensive set of 6385 acoustic features was then extracted from these voice segments using OpenSMILE and Praat software. Subsequently, a random forest model was constructed to classify cognitive status using the features that exhibited significant differences between the MCI and cognitively normal groups. The MCI detection performance of various audio lengths was further examined. RESULTS: An optimal subset of 29 features was identified that resulted in an area under the receiver operating characteristic curve of 0.87, with a 95% CI of 0.81-0.94. The most important acoustic feature for MCI classification was the number of filled pauses (importance score=0.09, P=3.10E-08). There was no substantial difference in the performance of the model trained on the acoustic features derived from different lengths of voice recordings. CONCLUSIONS: This study showcases the potential of monitoring changes to nonsemantic and acoustic features of speech as a way of early ADRD detection and motivates future opportunities for using human speech as a measure of brain health.

Adversarial Learning for MRI Reconstruction and Classification of Cognitively Impaired Individuals.

Game theory-inspired deep learning using a generative adversarial network provides an environment to competitively interact and accomplish a goal. In the context of medical imaging, most work has focused on achieving single tasks such as improving image resolution, segmenting images, and correcting motion artifacts. We developed a dual-objective adversarial learning framework that simultaneously 1) reconstructs higher quality brain magnetic resonance images (MRIs) that 2) retain disease-specific imaging features critical for predicting progression from mild cognitive impairment (MCI) to Alzheimer's disease (AD). We obtained 3-Tesla, T1-weighted brain MRIs of participants from the Alzheimer's Disease Neuroimaging Initiative (ADNI, N=342) and the National Alzheimer's Coordinating Center (NACC, N = 190) datasets. We simulated MRIs with missing data by removing 50% of sagittal slices from the original scans (i.e., diced scans). The generator was trained to reconstruct brain MRIs using the diced scans as input. We introduced a classifier into the GAN architecture to discriminate between stable (i.e., sMCI) and progressive MCI (i.e., pMCI) based on the generated images to facilitate encoding of disease-related information during reconstruction. The framework was trained using ADNI data and externally validated on NACC data. In the NACC cohort, generated images had better image quality than the diced scans (Structural similarity (SSIM) index: 0.553 ± 0.116 versus 0.348 ± 0.108). Furthermore, a classifier utilizing the generated images distinguished pMCI from sMCI more accurately than with the diced scans (F1-score: 0.634 ± 0.019 versus 0.573 ± 0.028). Competitive deep learning has potential to facilitate disease-oriented image reconstruction in those at risk of developing Alzheimer's disease.

APOE genotype-specific methylation patterns are linked to Alzheimer disease pathology and estrogen response.

The joint effects of APOE genotype and DNA methylation on Alzheimer disease (AD) risk is relatively unknown. We conducted genome-wide methylation analyses using 2,021 samples in blood (91 AD cases, 329 mild cognitive impairment, 1,391 controls) and 697 samples in brain (417 AD cases, 280 controls). We identified differentially methylated levels in AD compared to controls in an APOE genotype-specific manner at 25 cytosine-phosphate-guanine (CpG) sites in brain and 36 CpG sites in blood. Additionally, we identified seven CpG sites in the APOE region containing TOMM40, APOE, and APOC1 genes with P < 5 × 10-8 between APOE ε4 carriers and non-carriers in brain or blood. In brain, the most significant CpG site hypomethylated in ε4 carriers compared to non-carriers was from the TOMM40 in the total sample, while most of the evidence was derived from AD cases. However, the CpG site was not significantly modulating expression of these three genes in brain. Three CpG sites from the APOE were hypermethylated in APOE ε4 carriers in brain or blood compared in ε4 non-carriers and nominally significant with APOE expression in brain. Three CpG sites from the APOC1 were hypermethylated in blood, which one of the 3 CpG sites significantly lowered APOC1 expression in blood using all subjects or ε4 non-carriers. Co-methylation network analysis in blood and brain detected eight methylation networks associated with AD and APOE ε4 status. Five of the eight networks included genes containing network CpGs that were significantly enriched for estradiol perturbation, where four of the five networks were enriched for the estrogen response pathway. Our findings provide further evidence of the role of APOE genotype on methylation levels associated with AD, especially linked to estrogen response pathway.

Sex-specific blood biomarkers linked to memory changes in middle-aged adults: The Framingham Heart Study.

The relationship between sex-specific blood biomarkers and memory changes in middle-aged adults remains unclear. We aimed to investigate this relationship using the data from the Framingham Heart Study (FHS). We conducted association analysis, partial correlation analysis, and causal dose-response curves using blood biomarkers and other data from 793 middle-aged participants (≤ 60 years) from the FHS Offspring Cohort. The results revealed associations of adiponectin and fasting blood glucose with midlife memory change, along with a U-shaped relationship of high-density lipoprotein cholesterol with memory change. No significant associations were found for the other blood biomarkers (e.g., amyloid beta protein 42) with memory change. To our knowledge, this is the first sex-specific network analysis of blood biomarkers related to midlife memory change in a prospective cohort study. Our findings highlight the importance of targeting cardiometabolic risks and the need to validate midlife-specific biomarkers that can accelerate the development of primary preventive strategies.

Comparison of Commonly Measured Plasma and Cerebrospinal Fluid Proteins and Their Significance for the Characterization of Cognitive Impairment Status.

BACKGROUND: Although cerebrospinal fluid (CSF) amyloid-ß42 peptide (Aß42) and phosphorylated tau (p-tau) and blood p-tau are valuable for differential diagnosis of Alzheimer's disease (AD) from cognitively normal (CN) there is a lack of validated biomarkers for mild cognitive impairment (MCI). OBJECTIVE: This study sought to determine how plasma and CSF protein markers compared in the characterization of MCI and AD status. METHODS: This cohort study included Alzheimer's Disease Neuroimaging Initiative (ADNI) participants who had baseline levels of 75 proteins measured commonly in plasma and CSF (257 total, 46 CN, 143 MCI, and 68 AD). Logistic regression, least absolute shrinkage and selection operator (LASSO) and Random Forest (RF) methods were used to identify the protein candidates for the disease classification. RESULTS: We observed that six plasma proteins panel (APOE, AMBP, C3, IL16, IGFBP2, APOD) outperformed the seven CSF proteins panel (VEGFA, HGF, PRL, FABP3, FGF4, CD40, RETN) as well as AD markers (CSF p-tau and Aß42) to distinguish the MCI from AD [area under the curve (AUC) = 0.75 (plasma proteins), AUC = 0.60 (CSF proteins) and AUC = 0.56 (CSF p-tau and Aß42)]. Also, these six plasma proteins performed better than the CSF proteins and were in line with CSF p-tau and Aß42 in differentiating CN versus MCI subjects [AUC = 0.89 (plasma proteins), AUC = 0.85 (CSF proteins) and AUC = 0.89 (CSF p-tau and Aß42)]. These results were adjusted for age, sex, education, and APOEϵ4 genotype. CONCLUSIONS: This study suggests that the combination of 6 plasma proteins can serve as an effective marker for differentiating MCI from AD and CN.

Design and Feasibility Analysis of a Smartphone-Based Digital Cognitive Assessment Study in the Framingham Heart Study.

BACKGROUND: Smartphone-based digital technology is increasingly being recognized as a cost-effective, scalable, and noninvasive method of collecting longitudinal cognitive and behavioral data. Accordingly, a state-of-the-art 3-year longitudinal project focused on collecting multimodal digital data for early detection of cognitive impairment was developed. METHODS AND RESULTS: A smartphone application collected 2 modalities of cognitive data, digital voice and screen-based behaviors, from the FHS (Framingham Heart Study) multigenerational Generation 2 (Gen 2) and Generation 3 (Gen 3) cohorts. To understand the feasibility of conducting a smartphone-based study, participants completed a series of questions about their smartphone and app use, as well as sensory and environmental factors that they encountered while completing the tasks on the app. Baseline data collected to date were from 537 participants (mean age=66.6 years, SD=7.0; 58.47% female). Across the younger participants from the Gen 3 cohort (n=455; mean age=60.8 years, SD=8.2; 59.12% female) and older participants from the Gen 2 cohort (n=82; mean age=74.2 years, SD=5.8; 54.88% female), an average of 76% participants agreed or strongly agreed that they felt confident about using the app, 77% on average agreed or strongly agreed that they were able to use the app on their own, and 81% on average rated the app as easy to use. CONCLUSIONS: Based on participant ratings, the study findings are promising. At baseline, the majority of participants are able to complete the app-related tasks, follow the instructions, and encounter minimal barriers to completing the tasks independently. These data provide evidence that designing and collecting smartphone application data in an unsupervised, remote, and naturalistic setting in a large, community-based population is feasible.

Shifting From Active to Passive Monitoring of Alzheimer Disease: The State of the Research.

Most research using digital technologies builds on existing methods for staff-administered evaluation, requiring a large investment of time, effort, and resources. Widespread use of personal mobile devices provides opportunities for continuous health monitoring without active participant engagement. Home-based sensors show promise in evaluating behavioral features in near real time. Digital technologies across these methodologies can detect precise measures of cognition, mood, sleep, gait, speech, motor activity, behavior patterns, and additional features relevant to health. As a neurodegenerative condition with insidious onset, Alzheimer disease and other dementias (AD/D) represent a key target for advances in monitoring disease symptoms. Studies to date evaluating the predictive power of digital measures use inconsistent approaches to characterize these measures. Comparison between different digital collection methods supports the use of passive collection methods in settings in which active participant engagement approaches are not feasible. Additional studies that analyze how digital measures across multiple data streams can together improve prediction of cognitive impairment and early-stage AD are needed. Given the long timeline of progression from normal to diagnosis, digital monitoring will more easily make extended longitudinal follow-up possible. Through the American Heart Association-funded Strategically Focused Research Network, the Boston University investigative team deployed a platform involving a wide range of technologies to address these gaps in research practice. Much more research is needed to thoroughly evaluate limitations of passive monitoring. Multidisciplinary collaborations are needed to establish legal and ethical frameworks for ensuring passive monitoring can be conducted at scale while protecting privacy and security, especially in vulnerable populations.

A Network Analysis of Digital Clock Drawing for Command and Copy Conditions.

Graphomotor and time-based variables from the digital Clock Drawing Test (dCDT) characterize cognitive functions. However, no prior publications have quantified the strength of the associations between digital clock variables as they are produced. We hypothesized that analysis of the production of clock features and their interrelationships, as suggested, will differ between the command and copy test conditions. Older adults aged 65+ completed a digital clock drawing to command and copy conditions. Using a Bayesian hill-climbing algorithm and bootstrapping (10,000 samples), we derived directed acyclic graphs (DAGs) to examine network structure for command and copy dCDT variables. Although the command condition showed moderate associations between variables (µ|ßz|= 0.34) relative to the copy condition (µ|ßz| = 0.25), the copy condition network had more connections (18/18 versus 15/18 command). Network connectivity across command and copy was most influenced by five of the 18 variables. The direction of dependencies followed the order of instructions better in the command condition network. Digitally acquired clock variables relate to one another but differ in network structure when derived from command or copy conditions. Continued analyses of clock drawing production should improve understanding of quintessential normal features to aid in early neurodegenerative disease detection.

Corrigendum: Digital neuropsychological measures by defense automated neurocognitive assessment: reference values and clinical correlates.

[This corrects the article DOI: 10.3389/fneur.2024.1340710.].

Digital neuropsychological measures by defense automated neurocognitive assessment: reference values and clinical correlates.

Introduction: Although the growth of digital tools for cognitive health assessment, there's a lack of known reference values and clinical implications for these digital methods. This study aims to establish reference values for digital neuropsychological measures obtained through the smartphone-based cognitive assessment application, Defense Automated Neurocognitive Assessment (DANA), and to identify clinical risk factors associated with these measures. Methods: The sample included 932 cognitively intact participants from the Framingham Heart Study, who completed at least one DANA task. Participants were stratified into subgroups based on sex and three age groups. Reference values were established for digital cognitive assessments within each age group, divided by sex, at the 2.5th, 25th, 50th, 75th, and 97.5th percentile thresholds. To validate these values, 57 cognitively intact participants from Boston University Alzheimer's Disease Research Center were included. Associations between 19 clinical risk factors and these digital neuropsychological measures were examined by a backward elimination strategy. Results: Age- and sex-specific reference values were generated for three DANA tasks. Participants below 60 had median response times for the Go-No-Go task of 796 ms (men) and 823 ms (women), with age-related increases in both sexes. Validation cohort results mostly aligned with these references. Different tasks showed unique clinical correlations. For instance, response time in the Code Substitution task correlated positively with total cholesterol and diabetes, but negatively with high-density lipoprotein and low-density lipoprotein cholesterol levels, and triglycerides. Discussion: This study established and validated reference values for digital neuropsychological measures of DANA in cognitively intact white participants, potentially improving their use in future clinical studies and practice.

MedPodGPT: A multilingual audio-augmented large language model for medical research and education.

The proliferation of medical podcasts has generated an extensive repository of audio content, rich in specialized terminology, diverse medical topics, and expert dialogues. Here we introduce a computational framework designed to enhance large language models (LLMs) by leveraging the informational content of publicly accessible medical podcast data. This dataset, comprising over 4, 300 hours of audio content, was transcribed to generate over 39 million text tokens. Our model, MedPodGPT, integrates the varied di-alogue found in medical podcasts to improve understanding of natural language nuances, cultural contexts, and medical knowledge. Evaluated across multiple benchmarks, MedPodGPT demonstrated an average improvement of 2.31% over standard open-source benchmarks and showcased an improvement of 2.58% in its zero-shot multilingual transfer ability, effectively generalizing to different linguistic contexts. By harnessing the untapped potential of podcast content, MedPodGPT advances natural language processing, offering enhanced capabilities for various applications in medical research and education.

Associations Between Glucose Metabolism Measures and Amyloid-ß and Tau Load on PET 14 Years Later: Findings From the Framingham Heart Study.

OBJECTIVE: Type 2 diabetes and glucose metabolism have previously been linked to Alzheimer disease (AD). Yet, findings on the relation of glucose metabolism with amyloid-ß and tau pathology later in life remain unclear. RESEARCH DESIGN AND METHODS: We included 288 participants (mean age = 43.1 years, SD = 10.7, range 20-70 years) without dementia, from the Framingham Heart Study, who had available measures of glucose metabolism (i.e., one-time fasting plasma glucose and insulin) and positron emission tomography (PET) measures of amyloid-ß and/or tau 14 years later. We performed linear regression analyses to test associations of plasma glucose (continuously and categorically; elevated defined as >100 mg/dL), plasma insulin, homeostatic model assessment for insulin resistance (HOMA-IR) with amyloid-ß or tau load on PET. When significant, we explored whether age, sex, and APOE ε4 allele carriership (AD genetic risk) modified these associations. RESULTS: Our findings indicated that elevated plasma glucose was associated with greater tau load 14 years later (B [95% CI] = 0.03 [0.01-0.05], P = 0.024 after false discovery rate [FDR] correction) but not amyloid-ß. APOE ε4 carriership modified this association (B [95% CI] = -0.08 [-0.12 to -0.03], P = 0.001), indicating that the association was only present in APOE ε4 noncarriers (n = 225). Plasma insulin and HOMA-IR were not associated with amyloid-ß or τ load 14 years later after FDR correction. CONCLUSIONS: Our findings suggest that glucose metabolism is associated with increased future tau but not amyloid-ß load. This provides relevant knowledge for prevention strategies and prognostics to improve health care.