The Journal of Prevention of Alzheimer's Disease最新文献

Alzheimer's disease (AD) clinical trials continue to face major hurdles in patient identification, resulting in delayed timelines, underpowered studies, and escalating costs. This perspective explores these challenges through the lens of a memory clinic, where hundreds of cases often translate into only a handful of enrollments. We highlight the potential of artificial intelligence (AI) to address this gap by powering chatbots for awareness and pre-screening, decision support tools for case identification, and algorithms for matching patients to trial-specific criteria, automating and streamlining the recruitment process. We also examine critical considerations in developing such AI-driven tools, including data standardization, privacy protections, and ethical safeguards. With thoughtful implementation, these innovations could accelerate more inclusive and efficient AD trials, ultimately bringing therapies to patients faster.

Artificial intelligence (AI), often seen as a harbinger of future innovation, also presents a dilemma: it can perpetuate existing human biases. However, this issue is not novel or unique to AI. Humans have long been the progenitors of biases, and AI, as a product of human creation, often mirrors these inherent tendencies. Here, we present a perspective on the development and use of AI, recognizing it as a tool influenced by human input and societal norms, rather than an autonomous entity. Modern efforts to technologically enabled data collection approaches and model development, particularly in the context of Alzheimer's disease and related dementias, can potentially reduce bias in AI. We also highlight the importance of data sharing from existing legacy cohorts to help accelerate ongoing AI model development efforts for greater scientific good and clinical care.

Background: Harmonizing diverse healthcare datasets is a challenging task due to inconsistent naming conventions. Manual harmonization is time- and resource-intensive, limiting scalability for multi-cohort Alzheimer's Disease research. Large Language Models, or specifically text-embedding models, offer a promising solution, but their rapid development necessitates continuous, domain-specific benchmarking, especially since general established benchmarks lack clinical data harmonization use cases.

Objectives: To evaluate how different text-embedding models perform for the harmonization of clinical variables.

Design and setting: We created a novel benchmark to assess how well different Language Model embeddings can be used to harmonize cohort study metadata with an in-house Common Data Model that includes cohort-to-cohort mappings for a wide range of Alzheimer's Disease cohorts. We evaluated five different state-of-the-art text embedding models for seven different data sets in the context of Alzheimer's disease.

Participants: No patient data were utilized for any of the analyses, as the evaluation was based on semantic harmonization of cohort metadata only.

Measurements: Text descriptions of variables from different modalities were included for the analyses, namely clinical, lifestyle, demographics, and imaging.

Results: Our benchmark results favored different models compared to general-purpose benchmarks. This suggests that models fine-tuned for generic tasks may not translate well to real-world data harmonization, particularly in Alzheimer's disease. We propose guidelines to format metadata to facilitate manual or model-assisted data harmonization. We introduce an open-source library (https://github.com/SCAI-BIO/ADHTEB) and an interactive leaderboard (https://adhteb.scai.fraunhofer.de) to aid future model benchmarking.

Conclusions: Our findings highlight the importance of domain-specific benchmarks for clinical data harmonization in the field of Alzheimer's disease and motivate standards for naming conventions that may support semi-automated mapping applications in the future.

Background: The relationship between alcohol consumption and cerebral small vessel disease (CSVD) remains uncertain, particularly regarding drinking patterns and beverage types. We investigated how total alcohol intake, drinking frequency, and beverage-specific consumption are associated with CSVD burden using cross-sectional data.

Methods: We included 27,326 UK Biobank (UKB) participants with MRI data, among whom 21,130 were current drinkers with full alcohol intake data. Alcohol consumption (frequency and beverage type) was self-reported. CSVD burden was measured via normalized white matter hyperintensity volume (WMHV) on T2-FLAIR MRI. Multivariable linear regression models adjusted for demographics, lifestyle, and vascular risk factors were used to examine associations.

Results: Compared with non-drinkers, alcohol consumers had greater CSVD burden (Beta = 0.07; 95 % CI, 0.00-0.15). Among them, higher drinking frequency (≥5 times/week) was associated with increased CSVD burden (Beta = 0.10; 95 % CI, 0.07-0.13). High consumption of red wine, white wine/champagne, and spirits (≥7 servings/week) correlated positively with CSVD burden. In contrast, low-to-moderate beer/cider intake (≤3 servings/week) was inversely associated with burden. A dose-response relationship between total ethanol intake and CSVD burden was observed, with minimal intake (<1.97 g/day) showing a mild negative association, and higher levels increasing risk.

Conclusion: Greater frequency and volume of alcohol intake, especially from wine and spirits, are linked with higher CSVD burden. Conversely, low beer/cider consumption may be inversely associated with CSVD burden. These findings underscore the importance of moderating alcohol consumption to maintain cerebrovascular health.

The development of effective therapeutics for Alzheimer's Disease and related dementias (ADRD) has been hindered by patient heterogeneity and the limitations of current diagnostic tools. New treatments have no chance of working if given to patients who cannot benefit from them. This perspective explores how advances in Artificial Intelligence (AI), particularly multimodal machine learning, can solve the 'Goldilocks problem' of identifying patients for inclusion in clinical trials and support precision treatment in real-world healthcare settings. We examine the challenges of patient stratification, grounded by a conceptual framework of identifying each person's stage and subtype of dementia. We review data from several clinical trials of Alzheimer's disease therapeutics, to explore how AI-guided patient stratification can improve trial outcomes, reduce costs and improve recruitment. Further, we discuss the integration of AI into clinical workflows, the importance of model interpretability and generalizability, and ethical imperative to address algorithmic bias. By combining AI with scientific insight, clinical expertise, and patient experience, we argue that intelligent analytics can accelerate the discovery and delivery of new diagnostics and therapeutics, ultimately transforming dementia care and improving outcomes for patients around the globe.

Until recently, accurate early detection of clinical symptoms associated with Alzheimer's disease (AD) and related dementias (ADRD) has been difficult. Digital technologies have created new opportunities to capture cognitive and other AD/ADRD related behaviors with greater sensitivity and specificity. Speech captured through digital recordings has shown recent promise at feasible levels of scalability because of the widespread penetration of smartphones. One such study is described in detail to illustrate the depth in which artificial intelligence (AI) analytic approaches can be used to amplify the value of audio recordings. Another modality that has also attracted research interest are ocular scans that have near term potential for validation as a digital biomarker and a point of entry for clinical care workflows. Single modality measures, however, are rapidly giving way to multi-modality sensors that are embedded in all smartphones and other internet-of-things connected devices. Artificial intelligence (AI) driven analytic approaches are able to divine clinical signals from these high dimensional digital data streams. These data driven findings are setting the stage for a future state in which AD/ADRD detection will be possible at the earliest possible stage of the neurodegenerative process and enable interventions that would significantly attenuate or alter the trajectory, preventing disease from reaching the clinical diagnosis threshold.

To effectively combat dementia onset and progression, lifestyle-based interventions targeting multiple risk factors and disease mechanisms through a multidomain approach - tailored and implemented early in the disease process - have emerged as promising. Electronic databases and relevant websites (clinicaltrials.gov, euclinicaltrials.eu, PubMed and EMBASE) were systematically searched for randomized controlled trials (RCTs) testing the combination of multidomain lifestyle and pharmacological interventions. Studies were included if 1) lifestyle intervention was multimodal (≥2 domains); 2) it was combined with drugs, supplements, or medical food; 3) the study population was within the Alzheimer's disease (AD) and related dementias continuum, including cognitively normal individuals at-risk for dementia, people with subjective cognitive decline (SCD), mild cognitive impairment (MCI), or prodromal AD; 4) outcomes included cognitive or dementia-related measure(s), and 5) intervention lasted at least 6 months. Twelve combination RCTs were identified, incorporating 2 to 7 lifestyle domains (physical exercise, cognitive training, dietary guidance, social activities, sleep hygiene, cardiovascular/metabolic risk management, psychoeducation or stress management), combined with pharmacological components (e.g., Omega-3, Tramiprosate, vitamin D, BBH-1001, epigallocatechin gallate, Souvenaid, and metformin). Seven RCTs targeted participants with prodromal AD, MCI or early dementia, five focused on at risk individuals or SCD. Additionally, 2 studies adopted a precision medicine approach by enriching populations with APOE-ε4 carriers. Findings suggest that well-designed interventions - tailored to the right individuals, implemented at the optimal time - may effectively improve cognition. However, further refinement of the RCT methodology is warranted, for better alignment with the multifaceted nature of dementia prevention and management.

Background: Elevated pulse pressure (PP), indicative of arterial stiffness, has been implicated in cognitive impairment and Alzheimer's disease (AD) pathology. However, its role in preclinical AD and interactions with genetic risk factors like apolipoprotein E ε4 (APOE4) remain unclear.

Objectives: To investigate the association between baseline PP and AD biomarkers (amyloid-beta (Aβ) and tau) and cognitive decline, and to determine whether APOE4 carrier status moderates these relationships.

Design: Prospective cohort study and secondary analysis of the Anti-Amyloid Treatment in Asymptomatic Alzheimer's (A4) randomized clinical trial SETTING: Multicenter observational cohort and randomized clinical trial conducted at 67 sites across the United States, Canada, Australia, and Japan.

Participants: This study included 1690 cognitively unimpaired older adults from the A4 and Longitudinal Evaluation of Amyloid Risk and Neurodegeneration (LEARN) studies. Participants underwent baseline PP assessment, Aβ and tau PET imaging, and cognitive testing with longitudinal follow-up over 240 weeks.

Measurements: Blood pressure was measured at baseline, with PP calculated as the difference between systolic and diastolic pressures. AD pathologies were assessed through Aβ PET imaging using 18F-Florbetapir, and regional tau deposition in inferior temporal and meta-temporal regions using 18F-Flortaucipir PET imaging. Cognitive performance was measured using the Preclinical Alzheimer Cognitive Composite (PACC).

Results: Higher baseline PP was significantly associated with increased Aβ (β = 0.078; p = 0.001), inferior temporal tau (β = 0.110; p = 0.032), and meta-temporal tau deposition (β = 0.116; p = 0.022). In longitudinal analyses, elevated PP predicted greater decline in PACC scores (β = -0.020; p < 0.001). APOE4 status moderated these associations, with significant effects of PP on tau deposition and cognitive decline observed exclusively among APOE4 carriers. Mediation analysis indicated that tau deposition significantly mediated the association between PP and cognitive decline (indirect effect β = -0.068; 95 % CI [-0.126, -0.011]).

Conclusions: Elevated PP is associated with increased AD biomarker burden and accelerated cognitive decline in cognitively unimpaired older adults, particularly among APOE4 carriers. Our study suggests that arterial stiffness may contribute to AD pathogenesis and progression via tau pathology. These results highlight the potential of vascular health management as an early intervention target for AD prevention, especially in genetically at-risk populations.