The Journal of Prevention of Alzheimer's Disease

INTRODUCTIONThere is need to detect and intervene in pre-clinical phases of Alzheimers disease (AD). Electronic health records (EHRs) may help predict AD using machine learning methods. METHODSWe identified EHRs for 19,473 cases with AD and 111,922 controls. Records spanned 10 or more years prior to AD diagnosis. We trained a random forest model (employing 5-fold cross-validation with 2,499 features) to predict AD 10 years prior to its onset using a 75/25% train/test split and then computed permuted feature importance. RESULTSWe achieved an area under the ROC curve of 0.80. Feature importance identified factors associated with AD, including age, sex, race, ethnicity, BMI, cardiovascular diseases, inflammation, pain, sleep and mood disorders, trauma, other neurodegenerative disorders, diuretics, colon-related disorders and procedures, seizures, and vitamin B12. DISCUSSIONThis is the first EHR-based model to predict AD 10 years prior to onset, which could help predict AD and inform prevention/early intervention.

IntroductionAlzheimers disease is a progressive neurodegenerative disorder where early detection remains difficult. To address this challenge, we analysed a large proteomics dataset from older adults, including individuals diagnosed through clinical and imaging confirmation of brain amyloid deposition. We hypothesized that amyloid positivity could be detected using blood-based proteomic profiles combined with statistical and machine learning methods. MethodsWe applied descriptive and inferential statistical analyses alongside supervised classification approaches, and group comparisons between amyloid positive and amyloid negative individuals were conducted. Classification methods, including random forests, gradient boosting, and neural networks, were used to evaluate prediction of amyloid status. All data were normalized and privacy compliant. ResultsDistinct proteomic signatures were associated with disease status. Significant protein expression differences were observed between amyloid positive (n=337) and amyloid negative (n=651) groups. Classification models reached balanced performance with prediction accuracy up AUC of 0.80. Eight proteins (i.e. SERPINA1, C3, CRP, APOE4, CFH, VTN, C1QTNF5, and PON1) emerged as strong predictors from the best-performing classifiers, representing potential biomarker candidates.: Discussion and ConclusionsCombining statistical and machine learning methods enabled robust identification of patterns distinguishing amyloid profiles. This strategy supports biomarker discovery and development of accessible blood-based diagnostic and therapeutic targets. Significance StatementThis study leverages high-throughput proteomic profiling and machine learning to identify peripheral blood-based protein signatures associated with cerebral amyloid pathology, a hallmark of Alzheimers disease. By integrating clinical data with proteomic biomarkers, we aimed to develop a non-invasive, scalable predictive tool that can support early detection and risk stratification of Alzheimers disease, potentially improving screening efficiency and guiding future therapeutic strategies. Furthermore, this analysis allowed for mechanistic insight into the biology of amyloid in Alzheimers disease. HighlightsO_LIDeveloped a machine learning-based proteomic model to predict amyloid positivity in Alzheimers disease using plasma blood samples. C_LIO_LIIdentified a protein machine learning generated signature linked to central amyloid pathology. C_LIO_LISuggested the potential for scalable, early detection tools to support precise and targeted diagnosis and treatment planning in Alzheimers disease using machine learning. C_LI

Alzheimers disease (AD) patients have varying responses to AD drugs and there may be no single treatment for all AD patients. Trial after trial shows that identifying non-responsive and responsive subgroups and their corresponding moderators will provide better insights into subject selection and interpretation in future clinical trials. We aim to extensively investigate pre-treatment features that moderate treatment effect of Galantamine, Bapineuzumab, and Semagacestat from completed trial data. We obtained individual-level patient data from ten randomized clinical trials. Six Galantamine trials and two Bapineuzumab trials were from Yale University Open Data Access Project and two Semagacestat trials were from the Center for Global Clinical Research Data. We included a total of 10,948 subjects. The trials were conducted worldwide from 2001 to 2012. We estimated treatment effect using causal forest modeling on each trial. Finally, we identified important pre-treatment features that determine treatment efficacy and identified responsive or nonresponsive subgroups. As a result, patients pre-treatment conditions that determined the treatment efficacy of Galantamine differed by dementia stages, but we consistently observed that non-responders in Galantamine trials had lower BMI (25 vs 28, P < .001) and increased ages (74 vs 68, P < .001). Responders in Bapineuzumab and Semagacestat trials had lower A{beta}42levels (6.41 vs 6.53 pg/ml, P < .001) and smaller whole brain volumes (983.13 vs 1052.78 ml, P < .001). 6 positive treatment trials had subsets of patients who had, in fact, not responded. 4 "negative" treatment trials had subsets of patients who had, in fact, responded. This study suggests that analyzing heterogeneity in treatment effects in "positive" or "negative" trials may be a very powerful tool for identifying distinct subgroups that are responsive to treatments, which may significantly benefit future clinical trial design and interpretation.

INTRODUCTIONWe evaluated the performance of plasma brain-derived (BD)- as well as total-p-Tau181, p-Tau217 and p-Tau231 in detecting beta-amyloid positivity (A{beta}+) and cognitive decline in a Singapore-based cohort of older people with cerebrovascular disease. METHODSBrain amyloid status (A{beta}- [n = 139] vs A{beta}+ [n = 74]) was determined by positron emission tomography (PET) scans. Plasma BD and total p-Tau were measured using NUcleic acid Linked Immuno-Sandwich Assay multiplexing platform (NULISAseq). RESULTSBD-p-Tau217 (area under the curve [AUC] = 0.965) outperformed other BD and total-p-Tau species in detecting PET A{beta}+ (AUC = 0.823-0.937; all p [&le;] 0.008). Using three-range or binary references, BD-p-Tau217 demonstrated high sensitivity (>90%), specificity (>90%), positive (>85%) and negative (>95%) predictive values. BD-p-Tau217-derived High-risk group exhibited faster cognitive decline than the Low-risk group. DISCUSSIONRisk stratification for PET A{beta}+ based on plasma BD-p-Tau217 suggests superior diagnostic and prognostic utility, warranting further assessment.

IntroductionThe study aimed to identify strategies to increase older Black adults participation in Alzheimers disease (AD) biomarker research studies. Methods399 community-dwelling Black and White older adults (age [&ge;] 55) who had never participated in AD research completed a survey about their perceptions of AD research involving blood draw, MRI, and PET. ResultsAlthough most participants expressed interest in AD biomarker research (Black participants: 63.0%, White participants: 80.6%), Black participants were significantly more hesitant than White participants (28.9% vs 15.1%), were more concerned about study risks, (30.8% vs. 11.1%) and perceived multiple barriers to participating in brain scans. Lack of information was perceived as a barrier to participation across groups (45.8%) and return of study results was perceived as a participation incentive (78.9-85.7%) (Ps < .05). DiscussionStrategies to increase Black older adult participation in AD research may include disseminating additional study information and return of results. Declaration of InterestNone of the investigators have a conflict of interest. JE receives support from VA IK HX002283, NIA P30AG072976, and NIA P30AG010133. AJP receives support from NIA (NIA U01 AG057195) and Alzheimers Association (LDRFP-21-818464). SW receives support from multiple NIA grants (K23AG062555, P30AG072976, P30AG010133, and R21AG074179) and the VA for clinical services. She receives book royalties from APPI and DSMB consultant fees (total less than $2000/year). AJS receives support from multiple NIH grants (P30 AG010133, P30 AG072976, R01 AG019771, R01 AG057739, U01 AG024904, R01 LM013463, R01 AG068193, T32 AG071444, and U01 AG068057 and U01 AG072177). He has also received support from Avid Radiopharmaceuticals, a subsidiary of Eli Lilly (in kind contribution of PET tracer precursor); Bayer Oncology (Scientific Advisory Board); Eisai (Scientific Advisory Board); Siemens Medical Solutions USA, Inc. (Dementia Advisory Board); Springer-Nature Publishing (Editorial Office Support as Editor-in-Chief, Brain Imaging and Behavior).

Background: Alzheimers disease (AD) is a multifactorial neurodegenerative disorder in which copper dyshomeostasis, mitochondrial stress, oxidative injury and immune dysregulation may contribute to pathogenesis. Cuproptosis, a copper-triggered regulated cell death pathway, has emerged as a potential mechanistic link to AD, but its therapeutic and biomarker implications remain incompletely defined. Methods: We integrated transcriptomic, machine learning, immune infiltration, QSFR, molecular docking, docking validation and ADME analyses using GEO blood- and brain-based AD cohorts. Differentially expressed genes were intersected with curated cuproptosis-related genes, followed by pathway enrichment, construction and validation of a hybrid ensemble classifier, CIBERSORT-based immune correlation analysis, QSFR-driven target prioritization, ligand docking, consensus docking validation and SwissADME profiling. Results: The transcriptomic analyses revealed reproducible AD associated signatures enriched in neurodegenerative, oxidative stress, mitochondrial and inflammatory pathways. Across multiple machine learning models, FDX1, PDHB, PDHA1, DLAT and DLD consistently emerged as the most important cuproptosis-related genes, with the hybrid ensemble achieving the best diagnostic performance. Immune profiling suggested that these genes are linked to distinct immune infiltration patterns. QSFR and docking prioritized FDX1 as a key target and Clioquinol, PBT2 and Ebselen showed the strongest and most consistent binding behavior. Docking validation confirmed reliable pose reproduction and enrichment over decoys, while ADME analysis supported Clioquinol, PBT2 and Ebselen as the most balanced candidates for further consideration. Conclusion: This integrated workflow identifies a cuproptosis-centered mitochondrial gene module as a robust AD signature and highlights Clioquinol, PBT2 and Ebselen as promising repurposing candidates. The findings provide a prioritized computational framework for future experimental validation of copper-linked therapeutic strategies in AD.

BACKGROUNDEffective treatments for neurodegenerative diseases remain elusive, underscoring the importance of preventive strategies. Probucol, a cholesterol lowering and antioxidant drug with established cardiovascular use, has shown neuroprotective effects in preclinical models of dementia by modulating peripheral lipoprotein amyloid metabolism and preserving capillary integrity. However, no large-scale human studies have examined its association with dementia risk. OBJECTIVETo examine the association between probucol use and incident dementia in older adults. DESIGN, SETTING, AND PARTICIPANTSThis retrospective cohort study used the Japan Medical Data Centre claims database from 2014 to 2023. Adults aged 50 years or older prescribed probucol or statins were included, excluding those with prior dementia or recent drug exposure. Participants were categorized as probucol monotherapy users, statin monotherapy users, or combination users ([&ge;]2 prescriptions). Propensity score matching was used to balance baseline comorbidities. EXPOSURESProbucol or statin therapy. MAIN OUTCOMES AND MEASURESThe primary outcome was incident all cause dementia. Secondary outcomes included Alzheimers disease and mixed Alzheimers and vascular dementia. Odds ratios (ORs) with 95% CIs were calculated using logistic regression adjusted for age and sex. RESULTSAmong 57 231 individuals (52.6% female) followed for up to 10 years (median, 3 years), 7 387 (12.9%) developed dementia. The cohort included 2 896 probucol users (5.1%) and 54 335 statin users (94.9%). Dementia incidence was higher among females (14.7%) than males (10.9%). Dementia incidence was lower in probucol users (5.6%, 162/2 896) than in statin users overall (13.2%, 8 248/62 519), with individual statins ranging from 11.4% (fluvastatin) to 16.7% (pravastatin). Probucol use was associated with a 62% lower adjusted risk of dementia compared with all statin users combined (adjusted OR, 0.38; 95% CI, 0.37-0.38). Protective associations were consistent across individual statin comparisons, with adjusted ORs ranging from 0.30 (pitavastatin) to 0.57 (fluvastatin). CONCLUSIONS AND RELEVANCEIn this large national cohort of Japanese adults, probucol use was associated with a substantially lower risk of incident dementia compared with statins. These findings provide the first large-scale human evidence linking probucol exposure with reduced dementia risk, supporting its further evaluation as a preventive therapy in prospective clinical trials. Key PointsO_ST_ABSQuestionC_ST_ABSIs use of the lipid-lowering and antioxidant agent probucol associated with a reduced risk of incident dementia compared with statin therapy in older adults? FindingsIn this nationwide cohort study of 57 231 Japanese adults aged 50 years and older, dementia incidence was nearly halved in probucol users (5.6%) compared with statin users overall (13.2%), corresponding to 62% lower adjusted odds of dementia. Protective associations were consistent across dementia subtypes and individual statins. MeaningThese findings suggest that probucol, a long-standing and well-tolerated cardiovascular drug, may offer a mechanistically distinct and potentially scalable strategy for dementia prevention, warranting confirmation in prospective intervention trials.

Background: International Classification of Diseases (ICD) codes are often used in epidemiological studies to track disease rates over time. Objective: This evaluation of ICD-code-based algorithms for electronic medical record (EMR) studies of Alzheimers disease (AD) and related dementias (ADRD) examines the impact of incorporating Centers for Medicare and Medicaid (CMS) data as an additional source of diagnostic and treatment information in Department of Veterans Affairs (VA) EMR studies. Methods: We performed a chart review of 100 VA Million Veteran Program (MVP) participants to evaluate algorithm performance. We also assessed genetic associations across algorithms in a large MVP cohort (n=396k). Results: Adding CMS data increased the number of detected cases, sensitivity, and positive predictive value, but decreased specificity and negative predictive value. Genetic analyses showed that broader (ADRD/dementia) algorithms with just VA data performed similarly to narrow (AD-focused) algorithms incorporating both VA and CMS ICD codes. Additionally, narrow AD algorithms based solely on VA data yielded the highest ORs, indicating the largest proportion of late-onset AD cases. Conclusions: We recommend using a broad (ADRD) algorithm without CMS or medication data, particularly for epidemiological studies or a strict AD algorithm including CMS and medication cases for genetic discovery of late-onset AD associations in VA EMR, and a strict AD algorithm without CMS data for applications focused solely on AD and sensitive to misspecification. Careful evaluation of algorithm performance is warranted in different EMR systems, as ICD coding practices vary by institution, as demonstrated by this comparison of VA EMR and CMS data.

Structured AbstractO_ST_ABSINTRODUCTIONC_ST_ABSIf mild cognitive impairment (MCI) is diagnosed after dementia, it suggests either the dementia diagnosis was premature, or the MCI diagnosis is incorrect. We investigated the prevalence and predictors of such "diagnostic reversion"--MCI diagnosis following dementia diagnosis--in a large academic health system. METHODSAmong 5,965 patients aged 50+ with incident dementia in UCSF Health electronic health records, we identified "reverters" with a subsequent MCI diagnosis. We used Group LASSO-regularized logistic regression and random forest models to identify predictors. RESULTSReversion occurred in 13.7% of patients. Lower odds were observed among older adults (OR=0.95/year; 95% CI: 0.92-0.98), while higher odds were found among Spanish speakers (OR=2.26; 95% CI: 1.28-4.00), those with greater cardiovascular risk (OR=1.16; 95% CI: 1.01-1.33), and higher Charlson comorbidity burden (OR=1.09; 95% CI: 1.05-1.14). DISCUSSIONDiagnostic reversion is common and socially patterned, suggesting contributions from misdiagnosis, clinical uncertainty, or variability in clinical presentation and care setting.

BackgroundThere are many myths regarding Alzheimers disease (AD) that have been circulated on the Internet, each exhibiting varying degrees of accuracy, inaccuracy, and misinformation. Large language models such as ChatGPT, may be a useful tool to help assess these myths for veracity and inaccuracy. However, they can induce misinformation as well. The objective of this study is to assess ChatGPTs ability to identify and address AD myths with reliable information. MethodsWe conducted a cross-sectional study of clinicians evaluation of ChatGPT (GPT 4.0)s responses to 20 selected AD myths. We prompted ChatGPT to express its opinion on each myth and then requested it to rephrase its explanation using a simplified language that could be more readily understood by individuals with a middle school education. We implemented a survey using Redcap to determine the degree to which clinicians agreed with the accuracy of each ChatGPTs explanation and the degree to which the simplified rewriting was readable and retained the message of the original. We also collected their explanation on any disagreement with ChatGPTs responses. We used five Likert-type scale with a score ranging from -2 to 2 to quantify clinicians agreement in each aspect of the evaluation. ResultsThe clinicians (n=11) were generally satisfied with ChatGPTs explanations, with a mean (SD) score of 1.0({+/-}0.3) across the 20 myths. While ChatGPT correctly identified that all the 20 myths were inaccurate, some clinicians disagreed with its explanations on 7 of the myths. Overall, 9 of the 11 professionals either agreed or strongly agreed that ChatGPT has the potential to provide meaningful explanations of certain myths. ConclusionsThe majority of surveyed healthcare professionals acknowledged the potential value of ChatGPT in mitigating AD misinformation. However, the need for more refined and detailed explanations of the diseases mechanisms and treatments was highlighted. Impact StatementThere are many statements regarding Alzheimers disease (AD) diagnosis, management, and treatment circulating on the Internet, each exhibiting varying degrees of accuracy, inaccuracy, and misinformation. Large language models are a popular topic currently, and many patients and caregivers may turn to LLMs such as ChatGPT to learn more about the disease. This study aims to assess ChatGPTs ability to identify and address AD myths with reliable information. We certify that this work is novel. Key Points- Geriatricians acknowledged the potential value of ChatGPT in mitigating misinformation in Alzheimers Disease - There remain nuanced cases where ChatGPT explanations are not as refined or appropriate. - Why does this matter? Large language models such as ChatGPT are very popular nowadays and patients and caregivers often may use them to learn about their disease. The paper seeks to determine whether ChatGPT does an appropriate job in moderating understanding of Alzheimers Disease myths.

BackgroundCognitively unimpaired (CU) adults vary substantially in their risk of developing mild cognitive impairment (MCI), yet most subtyping approaches focus on downstream neurobiological or cognitive markers rather than upstream, modifiable risk factors. We aimed to identify clinically meaningful subgroups of CU adults defined by integrated comorbid, behavioral, and social risk profiles, and to evaluate heterogeneity in both incident MCI risk and cardiometabolic treatment effects. MethodsWe conducted a prospective cohort study of 121,322 CU adults aged [&ge;]50 years from the All of Us Research Program. Baseline comorbidities, lifestyle behaviors, and social determinants of health were jointly modeled using the Bayesian Mixed Integrative Data Subtyping framework, which integrates binary and continuous modalities via modality-specific likelihoods and shared latent constructs. Subtype-specific risk of incident MCI was assessed using multivariable Cox proportional hazards models adjusting for demographics and baseline medication use. A double/debiased machine learning interactive regression model with inverse probability of censoring weights to mitigate bias from informative censoring was implemented to estimate the average treatment effects of antihypertensive agents, Glucagon-Like Peptide (GLP) receptor agonists, and non-GLP antidiabetic medications on time to MCI. ResultsFour distinct subtypes were identified: I low-risk healthy aging, II behavioral/social vulnerability, III cardiometabolic-depressive multimorbidity, and IV mixed social-medical vulnerability profiles. Compared with Subtype I, Subtype III demonstrated the highest risk of incident MCI (HR: 3.69, 95% CI: 3.14-4.33), followed by Subtype IV and Subtype II. In treatment effect analyses, antihypertensive use was associated with a modest prolongation of MCI-free survival overall (time ratio:1.04, 95% CI: 1.03-1.06), with the largest benefit observed in Subtype III (time ratio: 1.14, 95% CI: 1.09-1.19). Non-GLP antidiabetic therapies were similarly associated with modest overall delay, with significant benefits in Subtypes I and III. GLP-class therapies were not associated with overall delay but showed a significant association in Subtype III. ConclusionsIntegrative subtyping based on comorbid, behavioral, and social risk factors reveals clinically meaningful heterogeneity in both cognitive risk and treatment response. Aligning dementia prevention strategies with dominant vulnerability pathways may enhance the effectiveness and equity of population-level precision prevention.

INTRODUCTIONThere is a lack of cognitive tools to predict disease progression in mild cognitive impairment (MCI) and Alzheimers disease (AD). METHODSWe assessed patients with MCI, AD, and cognitively healthy controls (cHC) using NIH toolbox assessments for attention/concentration and executive functioning and overall cognitive decline by the Alzheimers Disease Assessment Scale-Cognitive (ADAS-Cog). RESULTSAmong 183 participants over a median follow-up of 540 days, both between- and within-subjects variance in NIH toolbox and ADAS-Cog assessments increased from cHC to MCI to AD patients. Among patients with AD, pattern comparison processing speed (PCPS) and dimensional change card sort tests (DCCS) declined at 3 and 6 months prior to global cognitive decline (p=0.008 & 0.0012). A 5-point decrease in either PCPS or DCCS increased risk of global cognitive decline (HR 1.32 (1.08-1.60) and 1.62 (1.16-2.26)). DISCUSSIONTesting for cognitive domains of attention/concentration and executive functioning may predict subsequent global cognitive, and functional decline.

BACKGROUNDAlzheimers Disease (AD) is a complex clinical phenotype with unprecedented social and economic tolls in an aging global population. Real World Data (RWD) from electronic health records (EHRs) offer opportunities to accelerate precision drug development and scale epidemiological research on AD. A precise characterization of AD cohorts is needed to address the noise abundant in RWD. METHODSWe conducted a retrospective cohort study to develop and test computational models for AD cohort identification using clinical data from 8 Massachusetts healthcare systems. We mined temporal representations from EHR data using a novel transitive sequential pattern mining algorithm (tSPM) to train and validate our models. We then tested our models against a held-out test set from a review of medical records to adjudicate the presence of AD. We trained two classes of models using Gradient Boosting Machine (GBM) to compare the utility of AD diagnosis records versus the tSPM temporal representations (comprising sequences of diagnosis and medication observations) from electronic medical records for characterizing AD cohorts. RESULTSIn a group of 4,985 patients, we identified 219 sequences of medication-diagnosis records for constructing the best classification models. The models with the sequential features improved AD classification by a magnitude of up to 16 percent (over the use of AD diagnosis codes). Six groups of sequences, which we refer to as temporal digital markers, were identified for characterizing the AD cohorts, including sequences that involved (1) a symptom or (2) a risk factor in the past, followed by an AD diagnosis, (3) AD medications, (4) indirect risk factors, symptom management, and potential side effects, (5) comorbidities with possible shared roots or side effects, and (6) plural encounters with of AD diagnosis codes. Discussions of how the identified sequential patterns can be interpreted are provided. CONCLUSIONSWe present sequential patterns of diagnosis and medication codes from electronic medical records, as digital markers of Alzheimers Disease. Classification algorithms developed on the sequential patterns can replace standard features from EHRs to enrich phenotype modeling.

AO_SCPLOWBSTRACTC_SCPLOWO_ST_ABSBackground and PurposeC_ST_ABSAlzheimers Disease (AD) is a complex neurodegenerative disease that has been becoming increasingly prevalent in recent decades. Efforts to identify predictive biomarkers of the disease have proven difficult. Advances in the collection of multi-omic data and deep learning algorithms have opened the possibility of integrating these various data together to identify robust biomarkers for predicting the onset of the disease prior to the onset of symptoms. This study performs a systematic review of recent methods used to predict AD using multi-omic and multi-modal data. MethodsWe systematically reviewed studies from Google Scholar, Pubmed, and Semantic Scholar published after 2018 in relation to predicting AD using multi-omic data. Three reviewers independently identified eligible articles and came to a consensus of papers to review. The Quality in Prognosis Studies (QUIP) tool was used for the risk of bias assessment. Results22 studies which use multi-omic data to either predict AD or develop AD biomarkers were identified. Those studies which aimed to directly classify AD or predict the progression of AD achieved area under the receiver operating characteristic curve (AUC) between .70 - .98 using varying types of patient data, most commonly extracted from blood. Hundreds of new genes, single nucleotide polymorphisms (SNPs), RNA molecules, DNA methylation sites, proteins, metabolites, lipids, imaging features, and clinical data have been identified as successful biomarkers of AD. The most successful techniques to predict AD have integrated multi-omic data together in a single analysis. ConclusionThis review has identified many successful biomarkers and biosignatures that are less invasive than cerebral spinal fluid. Together with the appropriate prediction models, highly accurate classifications and prognostications can be made for those who are at risk of developing AD. These early detection of risk factors may help prevent the further development of cognitive impairment and improve patient outcomes.

INTRODUCTIONDementia is a disabling condition that progressively impairs daily function. Timely identification of older adults at high risk for dementia or cognitive impairment (a potential precursor) is critical to maximizing opportunities for intervention. METHODSUtilizing structured electronic health record data from 122,633 patients aged 55-80 years, we leveraged demographics, encounter diagnoses, and patient problem lists to develop and prospectively validate a ML model. RESULTSThe ML model achieved a C-statistic of 0.811 (95% confidence interval: 0.810, 0.812) with adequate calibration overall and in subgroups based on race and sex. Recommending screening for patients with 3-year predicted risk > 5%, the ML model obtained satisfactory fairness across race and sex subgroups, with a net benefit of 18 true positive MCI/dementia diagnoses per 1,000 patients. DISCUSSIONThe ML model developed in this study can effectively identify individuals at high risk for a future diagnosis of MCI/dementia, potentially facilitating earlier screening and intervention to reduce the burden of cognition-related disability.

Alzheimer's disease is a progressive neurodegenerative disorder that poses a growing global public health challenge. Early and accurate diagnosis is critical for effective treatment, clinical trial participation, and disease management. This systematic review and meta-analysis evaluates the diagnostic performance of machine learning (ML) and deep learning (DL) algorithms for detecting Alzheimer's disease (AD) and mild cognitive impairment (MCI) using neuroimaging and clinical data. Relevant studies were identified from PubMed, IEEE Xplore, and arXiv (2015 to 2025). Random-effects models were applied to estimate pooled performance metrics (AUC, sensitivity, specificity, and F1-score), and subgroup analyses compared results by model type, imaging modality, and validation strategy. Thirty studies met inclusion criteria. The pooled AUC was 0.962, indicating high overall discriminative accuracy. However, studies relying solely on internal validation or with smaller datasets often reported inflated metrics, suggesting potential overfitting and optimism bias. ML and DL methods demonstrate strong potential for early AD detection, but standardized evaluation protocols and external validation are necessary for clinical translation.

Background and ObjectivesProgression from mild cognitive impairment (MCI) to Alzheimers Disease and Related Dementias (AD/ADRD) varies widely across individuals, yet the mechanisms underlying this heterogeneity remain unclear. Identifying clinical and social determinants influencing this transition could enable earlier intervention. While cardiovascular and social risk factors are established contributors to dementia incidence, their role in progression from MCI to dementia may differ. Few studies using real world clinical data have evaluated these potential determinants of MCI progression. MethodsUsing electronic health records (EHR) from patients with incident MCI at UCSF Health (2010-2024), we evaluated cardiovascular (blood pressure [BP], body mass index [BMI], and type II diabetes) and social (marital status, language preference, race/ethnicity, and neighborhood disadvantage) risk factors for rate of progression from MCI to AD/ADRD. Covariate-adjusted Cox proportional hazards models estimated hazard ratios for incident AD/ADRD, with evaluation of interactions by sex. ResultsAmong 6,529 patients, higher systolic BP was associated with AD/ADRD incidence (HR per 10 mmHg: 1.09, 95% CI: 1.05-1.14). BMI was inversely associated with incidence in both males (HR: 0.94; 95% CI: 0.92-0.97) and females (HR:0.98; 95% CI: 0.96-0.99). Compared to married individuals, widowed patients had a higher hazard of progression (HR: 1.15; 95% CI: 1.00-1.32). Spanish-speaking (HR: 1.38; 95% CI: 1.04-1.81), Chinese-speaking (HR: 1.19; 95% CI: 1.00-1.42), and "Other non-English" speaking patients (HR:1.24; 95% CI: 1.03-1.51) had a higher hazard of progression compared to English speakers. Latinx (HR:1.22; 95% CI: 1.01-1.48) and Asian patients (HR:1.14, 95% CI: 1.00-1.30; p=0.04) also had higher hazards of progression compared to White patients. Neighborhood disadvantage was not significantly associated with disease progression. DiscussionCardiovascular and social factors independently influence dementia progression, with some sex-specific patterns. Integrating clinical and social indicators highlights the potential of EHR data to identify high-risk patients earlier in the care continuum and support equitable dementia prevention.

INTRODUCTIONDelaying cognitive decline in Alzheimers disease can significantly impact both function and quality of life. METHODSLongitudinal analysis of National Alzheimers Coordinating Center (NACC) dataset of 7,653 mild dementia CDR-SB AD participants at baseline with prescriptions for diabetes (DBMD), lipid-lowering (LIPL), anti-hypertensive (AHTN), and non-steroidal anti-inflammatory (NSD) medications over 10 years was evaluated for change in cognitive function relative to non-treated stratified by sex and APOE genotype. RESULTSCombination therapy of DBMD+LIPL+AHTN+NSD resulted in a 44% / 35% (MMSE/CDR-SB) delay in cognitive decline at 5 years and 47% / 35% (MMSE/CDR-SB) delay at 10 years. Females and APOE4 carriers exhibited greatest cognitive benefit of combination therapy. DISCUSSIONCombination therapies significantly delayed cognitive decline in NACC AD participants at a magnitude comparable to or greater than beta-amyloid immunomodulator interventions. These data support combination precision medicine targeting AD risk factors to alter the course of the disease that persists for a decade.

ImportanceInterpretable scoring system can contribute to bridge the gap between the timeliness and complexity of diagnosing Alzheimers disease (AD) and promote early intervention at non-specialist settings. ObjectiveTo develop a risk score to predict the likelihood of AD with interpretable machine learning using variables that are obtainable at integrated primary care settings. DesignA secondary data analysis including cohort studies from the Alzheimers Disease Neuroimaging Initiative (ADNI) and the National Alzheimers Coordinating Center (NACC) extracted in August 2023 and March 2024. SettingThe ADNI and NACC are multi-site cohort studies in North America. ParticipantsParticipants with normal cognition or mild cognitive impairment at baseline visit were identified. Participants with the same diagnosis overtime were assigned to the stable group, and those converted to AD were placed in the progressive group. Main Outcome(s) and Measure(s)Cognitive tests and daily functioning measured with Functional Assessment Questionnaire (FAQ) at baseline visit. ResultsA total of 676 participants from ADNI and 4592 participants from NACC were identified. After removing incomplete data, 665 ADNI (mean age [SD]: 73.44 [6.90]; 293 [44.1%] female; 374 stable and 291 progressive) and 3657 NACC participants (mean age [SD]: 70.96 [10.03]; 2405 [65.8%] female; 2445 stable and 1212 progressive) remained. Combinations of 4 measures were selected to generate 10 scorecards using FasterRisk algorithm, showing strong performance (area under the curve [AUC] = 0.868-0.892) in ADNI and remaining robust when validated in NACC (AUC = 0.795). The features were Category Animal [&le;] 20 (2 points), Trail Making Test B [&le;] 143 (-3 points), Logical Memory Delayed [&le;] 3 (4 points), Logical Memory Delayed [&le;] 8 (3 points), and FAQ [&le;] 2 (-5 points). The probable AD risk corresponded to total points: 7.4% (-8), 25.3% (-4), 50% (-1), 74.7% (2), and > 90% ([&ge;] 6). We refer to this model as the (F)unctioning, (LA)nguage, (M)emory, and (E)xecutive functioning or FLAME scorecard. Conclusions and RelevanceOur findings highlight the potential to predict AD development using obtainable information, allowing for applicability at integrated primary care. While our scope centers on AD, this foundation paves the way for other dementia types Key PointsO_ST_ABSQuestionC_ST_ABSCan accessible information, such as demographics, cognitive tests, and functioning questionnaire, yield in reliable results for predicting Alzheimers disease development using interpretable machine learning? FindingsThe results of 665 participants from the Alzheimers Disease Neuroimaging Initiative demonstrated robust performance of determining Alzheimers disease development using four separate measures of (F)unctioning, (LA)nguage, (M)emory, and (E)xecutive functioning or the FLAME scorecard. It remains reliable when externally validated with a separate dataset of 3657 participants from the National Alzheimers Coordinating Center. MeaningThe FLAME scorecard shows potential to be implemented in integrated primary care settings to promote early detection and intervention of cognitive decline due to Alzheimers disease.

ObjectivesTo test the psychometric properties of several dementia care-related scales among Latinos in the US. DesignWe leveraged secondary baseline data from a one-arm mHealth trial on dementia caregiver support. We included 100 responses for caregiver-focused scales and 88 responses for care recipient-focused scales. Scales included the Neuropsychiatric Inventory Questionnaire Severity and Distress scales, six-item Zarit Burden Inventory, Ten-item Center for Epidemiologic Studies Depression Scale, Geriatric Depression Inventory, Quality of Life in Alzheimers Disease, and Single-item Satisfaction With Life Scale. We calculated concurrent validity using Pearson and Spearman correlations and expected correlations amongst all variables in line with the Stress Process Framework. We calculated internal consistency reliability using Cronbachs alpha. ResultsAll concurrent validity correlations followed the expected directionality, with 19/21 inter-scale correlations in the total sample reaching statistical significance (p<0.05), and 17/21 reaching at least a low correlation (0.3). Cronbachs alpha ranged from 0.832 to 0.879 in all scales in the total sample. ConclusionThe English and Spanish caregiver-administered scales tested in this manuscript have good psychometric properties. Clinical ImplicationsThe dementia care-related scales are now appropriately available for use among US Latinos in research and clinical contexts.