Concurrent Detection of Cognitive Impairment and Amyloid Positivity with a Next-Generation Digital Cognitive Assessment

Concurrent Detection of Cognitive Impairment and Amyloid Positivity with a Next-Generation Digital Cognitive Assessment | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Concurrent Detection of Cognitive Impairment and Amyloid Positivity with a Next-Generation Digital Cognitive Assessment Ali Jannati, Karl Thompson, Claudio Toro-Serey, Joyce Gomes-Osman, and 6 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6768373/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 11 Dec, 2025 Read the published version in Alzheimer's Research & Therapy → Version 1 posted 9 You are reading this latest preprint version Abstract Background: Early identification of cognitive impairment and brain pathology associated with Alzheimer’s disease (AD) is essential to maximize benefits from lifestyle interventions and emerging pharmacologic disease-modifying treatments (DMT). Digital cognitive assessments (DCAs) can quickly capture an array of metrics that can be used to train machine learning models to concurrently evaluate different outcomes. DCAs have the potential to optimize clinical workflows and enable efficient assessment of cognitive function and the likelihood of a given underlying pathology. Methods: We assessed the ability of a next-generation DCA, the Digital Clock and Recall (DCR), to concurrently estimate brain amyloid-beta (Aβ) status and detect cognitive impairment, as compared with traditional cognitive assessments, including the MMSE, RAVLT, a DCA, Cognivue®, and blood-based biomarkers in 930 participants from the Bio-Hermes-001 clinical study. Results: Aβ42/40, pTau-181, and pTau-217 poorly classified cognitive impairment (AUCs: 0.63; 0.66; 0.72, respectively), but accurately classified Aβ status (AUCs: 0.81; 0.78; 0.89, respectively). MMSE, RAVLT, and Cognivue poorly classified Aβ status (AUCs: 0.71, 0.72, 0.70, respectively). However, separate multimodal, DCR-based machine-learning classification models, run in parallel, accurately classified both cognitive impairment (AUC=0.85) and Aβ status (AUC=0.83). Conclusions: DCAs that leverage digital technologies to generate advanced metrics, such as the DCR, enable accurate and efficient detection of cognitive impairment associated with AD pathology. They have the potential to empower health systems and primary care providers to help their patients make timely treatment decisions. Alzheimer’s disease mild cognitive impairment dementia digital cognitive assessment Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Dementia is the most feared health condition as we age ( 1 ), and Alzheimer’s disease (AD) is the most common cause of dementia ( 2 ). Nearly one out of four people globally has MCI ( 3 ). Yet, over 90% of individuals with MCI currently remain undiagnosed ( 4 ), and, in the US, less than 30% of annual wellness exams include cognitive assessments ( 5 ). The Food and Drug Administration (FDA) has approved the anti-amyloid antibody therapies lecanemab ( 6 ) and donanemab ( 7 ) for the treatment of mild cognitive impairment (MCI) and mild dementia due to AD. At the same time, significant evidence exists supporting the value of lifestyles that promote brain-healthy behaviors to reduce the risk of progression from MCI to dementia and minimize brain-related disability ( 8 – 13 ). However, for these and future pharmacologic and non-pharmacologic disease-modifying treatments (DMTs) to be truly impactful, healthcare systems must address the challenge of early diagnosis. This is critical for maximizing the benefits of disease-modifying therapies (DMTs) and empowering patients to plan for the future, establish advance care directives, adopt meaningful lifestyle changes, and set achievable personal goals. Primary care providers (PCPs) are strategically positioned to ensure the timely diagnosis of MCI or mild dementia due to AD. Blood-based biomarkers (BBMs) promise to enable early and efficient determination of amyloid brain deposition and AD pathology - indeed, several have received clearance or breakthrough designation from the FDA ( 14 – 18 ). However, they require establishing cognitive impairment, and existing pencil-paper assessments of cognition can be unwieldy, inefficient, and inaccurate. PCPs need more efficient and scalable assessments that integrate into their already time-pressured workflows without increasing provider burden. First-generation digital cognitive assessments (DCAs) were designed to streamline cognitive evaluation and have been proposed as potential solutions to the challenges of conducting cognitive assessments in primary care settings. Compared with traditional paper-pencil tests, DCAs allow for automated scoring, standardize data flow, alleviate subjectivity in scoring, simplify data storage, integrate with existing hospital systems, and ease longitudinal tracking over time. However, they offer limited time saving and little improvement in sensitivity and specificity. Next-generation DCAs capitalize on the power of technology to implement ‘process metrics’ – a framework for neuropsychological assessment that derives insights about behavior, brain, and cognitive function not only from an individual's final test achievement, but also from their process of completing cognitive tasks ( 19 ). It is thus possible to transform the sensor array of a consumer device (e.g., a tablet) into a medical device, and leverage multimodal signal processing and artificial intelligence (AI) models to assess both the output and process of a given task. This promises to enable the development of highly time-efficient assessments of superior sensitivity and specificity, ideally suited for integration into primary care settings ( 20 – 22 ). Furthermore, by ‘multiplexing’ the same output metrics across multiple outcomes, it may be possible to apply distinct AI algorithms to the same process-level data, enabling the concurrent evaluation of different outcomes. This approach could allow a single brief assessment to simultaneously characterize cognitive function and estimate the likelihood of a specific underlying pathology (Fig. 1 ). We hypothesized that the Digital Clock and Recall™ (DCR) ( 23 – 25 ), a next-generation multimodal DCA, would be able to concurrently detect cognitive impairment (CI) and predict the likelihood of a positive amyloid-beta (Aβ) positron emission tomography (PET) of the brain, thus potentially providing an efficient and cost-effective means to detect MCI due to AD, and identify patients for DMTs in primary care. Our objectives were: ( 1 ) assess the ability of the DCR to concurrently predict CI and Aβ status, ( 2 ) compare DCR’s performance to other cognitive assessments and BBMs, and ( 3 ) provide evidence for the additive benefit of combining insights from the DCR with BBMs and genetic testing to improve the detection of CI and Aβ status. Methods Data were obtained from 930 participants with PET, cognitive, genetic, and BBM data in the Bio-Hermes-001 (BH) study (ClinicalTrials.gov Identifier: NCT04733989) ( 26 ). The study was administered by the Global Alzheimer’s Platform Foundation (GAP), and study protocols were administered by clinical trial sites within the community. All subjects provided written informed consent before their participation. Participants and Study Cohort Enrollment criteria are listed in the Supplementary Materials and elsewhere ( 27 ). Participants were > 24% non-White or Latino/Hispanic, and 56.8% female. They were classified a priori by the BH study team into cognitively unimpaired (CU; n = 398), MCI (n = 291), or probable Alzheimer’s dementia (pAD; n = 241). Cohort classification was based on assessment with the Mini-Mental State Examination (MMSE) score, Rey Auditory Verbal Learning Test (RAVLT), and the Functional Activities Questionnaire (FAQ), as well as a clinical interview and review of medical records. Alternatively, a cohort status of MCI or pAD could be assigned if a clinical diagnosis was made within 3 months of Visit 1 (see Supplementary Materials for full criteria). Tables 1 and 2 summarize the participants’ demographics, Aβ status, and cohort classification. Authors were blind to cohort classification and Aβ status until data lock. Table 1 Summary statistics for demographic details by Aβ PET status. Metric Aβ- (n = 604) Aβ+ (n = 326) Statistical Test Age (Mean ± SD) (Range) 70.75 ± 6.66 (59–85) 74.24 ± 6.15 (60–85) T=-7.83, p < 0.001 Gender (%F) 354 (59%) 174 (53%) X 2 = 0.003, p = 0.95 Years of education 15.51 ± 2.61 15.39 ± 2.89 T = 0.7, p = 0.49 Ethnicity (%) X 2 = 4.5, p = 0.11 Not Hispanic or Latino 536 (89%) 288 (88%) - Hispanic or Latino 61 (10%) 28 (9%) - Not Reported 7 (1%) 10 (3%) - Race (%) X 2 = 12.75, p = 0.026 White 507 (84%) 293 (90%) - Black / African American 76 (13%) 28 (9%) - Asian 11 (2%) 5 (1%) - American Indian / Alaskan Native 2 (< 1%) 0 - Native Hawaiian or Other Pacific Islander 1 (< 1%) 0 - Unknown 7 (1%) 0 - Cognitively Unimpaired (%) 314 (52%) 84 (26%) X 2 = 0.09, p = 0.77 SUVR (Mean ± SD) 1.00 ± 0.09 1.42 ± 0.22 T=-39.94, p < 0.001 Statistical tests confirmed the Aβ + group had a significantly greater Aβ load than the Aβ- group (mean SUVR 1.00 vs. 1.42). Aβ + participants tended to be older (70.75 vs. 74.24) but had similar mean years of education. Groups were also not significantly different in ethnicity (Hispanic vs. Not Hispanic). Groups were balanced on gender and the proportion of cognitively unimpaired individuals. Statistically different distributions for race were noted between the groups, due in part to the greater proportion of White and lower proportion of Black / African American participants in the Aβ- group. Cognitive Assessments The MMSE can detect mild to moderate dementia in older adults, but it can take over 10 minutes to complete ( 28 ). The RAVLT is widely used to assess verbal memory but can take over 30 minutes to complete ( 29 ). Cognivue® Clarity is an FDA-listed computerized tool designed to detect early signs of cognitive impairment through a 10-minute assessment administered via proprietary hardware ( 30 ). Table 2 Summary statistics for the cognitive cohorts defined by the organizers of the Bio-Hermes study. Metric CU (n = 398) MCI (n = 291) pAD (n = 241) Statistical Test Age (mean ± SD) (range) 70.35 ± 6.40 (60–85) 72.18 ± 6.84 (60–85) 74.42 ± 6.22 (59–85) F = 33.23, p < 0.001 Gender (%F) 242 ( 61 ) 156 ( 54 ) 130 ( 54 ) X 2 = 0.005, p = 0.99 Years of education 15.77 ± 2.48 15.49 ± 2.76 14.95 ± 2.95 F = 8.78, p < 0.001 Ethnicity (%) X 2 = 13.74, p = 0.008 Not Hispanic or Latino 378 271 213 - Hispanic or Latino 29 35 40 - Not Reported 10 4 4 - Race (%) X 2 = 12.75, p = 0.023 White 345 (87) 253 (87) 202 (84) - Black / African American 42 ( 11 ) 30 ( 10 ) 32 ( 13 ) - Asian 7 ( 2 ) 5 ( 2 ) 4 ( 2 ) - American Indian / Alaskan Native 1 (< 1) 2 (< 1) 0 - Native Hawaiian / Pacific Islander 0 1 (< 1) 1 (< 1) - Unknown 0 0 2 (< 1) - SUVR (Mean ± SD) 1.07 ± 0.19 1.14 ± 0.25 1.28 ± 0.29 F = 59.44, p < 0.001 A one-way analysis of variance (ANOVA) test revealed a significant main effect for age across groups. Post-hoc analyses revealed that the cognitively unimpaired (CU) cohort was significantly younger than the mild cognitive impairment (MCI) cohort, which was in turn significantly younger than the probable Alzheimer’s dementia (pAD) cohort. ANOVA with Tukey’s post-hoc comparisons also revealed that the CU cohort had significantly fewer years of education compared to the pAD cohort but was not significantly different from the MCI cohort. Cohorts were significantly different in ethnicity (Hispanic vs. Not Hispanic) and race. Finally, a significant difference in standardized uptake value ratio (SUVR) was found among cohorts, and post-hoc comparisons revealed that the pAD cohort had significantly higher SUVR values than both the MCI and CU cohorts. Similarly, the MCI cohort had significantly greater SUVR values than the CU cohort. The Linus Health DCR ( 23 – 25 ) is an FDA-listed Class II software as a medical device and is a machine learning (ML)-enabled adaptation of the Mini-Cog ( 25 ). The 3-minute DCR is completed with an iPad and Apple Pencil and can be administered by medical or research assistants without specialized training. The DCR includes performance in 3-word immediate and delayed recall tests, intervened by a digital clock drawing test (DCTclock™) ( 31 , 32 ). Drawing strokes and recorded speech are analyzed via automated scoring and ML algorithms to assess verbal memory, executive function, visuospatial reasoning, information processing, and motor function, and to generate both traditional and process metrics ( 23 , 24 , 31 ). Apolipoprotein E (APOE) Genotyping The APOE e4 allele is a major risk factor for AD ( 33 ) and is associated with episodic memory impairment ( 34 ). APOE genotyping was conducted on a blood sample and assayed by C 2 N Diagnostics laboratories ( 27 ). Blood-Based Biomarkers (BBMs) In addition to the APOE assay, C 2 N Diagnostics provided plasma Aβ42 and Aβ40 concentrations measured by mass spectroscopy ( 35 ). C 2 N’s PrecivityAD™ Amyloid Probability Score (APS) combines the Aβ42/40 ratio, APOE genotype, and age to produce a 0–100 score where higher scores represent a higher likelihood of Aβ positivity. pTau-217 and pTau-181 concentrations were measured by Lilly Research Laboratories ( 36 ) and the Simoa HD-1 (Quanterix) ( 37 ), respectively. Positron Emission Tomography Amyloid PET scans were collected using the FDA-approved tracer 18 F-florbetapir ( 38 ) and uploaded to a portal accessible to specialists. Automated preprocessing ( 39 ) and quantification of standardized uptake value ratios (SUVR) ( 40 ) were performed centrally by IXICO Technologies, Inc. (Wilmington, DE, USA). Experts were blind to participants’ clinical information or cognitive and BBM results ( 27 ). This expert determination was used to define a binary Aβ status, which was used as ground truth for all experiments. Amyloid PET analysis indicated that 326 participants were Aβ+ (35%) and 604 were Aβ– (65%). The SUVRs confirmed the expected significant difference between Aβ- and Aβ + groups (1.003 ± 0.1 vs. 1.423 ± 0.2; Cohen’s d = 3.7). Mean (± SD) SUVR was 1.071 (± 0.2), 1.148 (± 0.2), and 1.283 (± 0.3) in the CU, MCI, and pAD cohorts, respectively (see Table S2 for additional statistics). Authors were blind to participant Aβ status until data lock and did not participate in establishing SUVR cutoffs, nor did they participate in the analysis of PET scans. Analytical Approach Separate ML models were trained to predict cognitive impairment (CI; combining MCI and pAD) or PET Aβ status by each cognitive test and BBM (Fig. 1 ). All models were trained and tested using leave-one-out cross-validation. During the inference phase, the training loop model was applied to the left-out subject’s data to produce a probability indicating their CI or Aβ status. This process was repeated for all subjects in the sample. The binary classification cutoff value for Aβ status was determined for each model using the concordance probability method ( 41 ). For each model, we report the mean performance metrics across all iterations. We also report model performance for a 3-class classification paradigm that specifies positive and negative CI or Aβ-status classes, but also an ‘indeterminate’ class. The two thresholds were determined by simultaneously optimizing sensitivity and specificity while minimizing the percentage of subjects in the indeterminate range (see Supplementary Materials). The 3-class classification paradigm was developed to capture the subjects for whom confidence of CI or Aβ status was relatively lower and provide a potential clinical workflow directing these subjects for confirmatory testing. Given the numerous variables derived from the DCR, a random forest classifier was employed, incorporating features from both the Command and Copy Clock tasks, immediate and delayed recall tests, and the participant’s age. Clock features included those extracted by the DCTclock algorithm ( 31 ), newly developed clock-drawing process metric features, and temporal and spatial features extracted from the Apple Pencil (e.g., altitude, azimuth, pressure). Immediate and delayed recall features included the number of words repeated/remembered correctly, and acoustic and linguistic metrics calculated from the speech recordings ( 24 ). Recursive feature elimination ( 42 ) was applied independently to each model training process to identify the subset of DCR features that yielded optimal performance for the CI and Aβ-status classification models, respectively (Fig. 1 ). For other cognitive tests and BBMs, logistic regressions were trained using each test’s most predictive variable combined with the participant’s age and/or APOE status. For other cognitive tests, predictors included the MMSE total score, the RAVLT long-delay score, and the Cognivue Clarity average test score ( 30 ). BBM models used APS, Aβ42/40, pTau-181, and pTau-217 levels as input features. An age-only logistic regression was also trained as a control. Critically, both MMSE and RAVLT were part of the BH a priori criteria for cognitive cohort definition and, therefore, were positively biased in their CI classification performance. These analyses, while circular, were included to provide a hypothetical ‘ceiling’ to assess CI classification performance. Biased results are highlighted in red in the corresponding figures (see figure legends). We performed bootstrap non-inferiority procedures ( 41 ) to compare DCR against the other tests for CI and Aβ-status classification, and to compare unimodal (i.e., BBM or DCA separately) models to ensemble models (i.e., combinations of BBMs and DCAs). First, separately for each cognitive assessment or BBM, we took the mean prediction probability for each participant across 10 cross-validated model training iterations. Then, on each of the 5,000 bootstrapping iterations, we sampled the target-positive and target-negative groups (e.g., with and without CI) with replacement and calculated the difference in the area under the receiver operating characteristic curve (AUC) between the two models. Inferiority, non-inferiority, and superiority were established if the 95% confidence interval of the bootstrapped differences was lower than, within, or higher than the AUC difference plus the margin M, respectively. The choice of M depends on several factors, including a clinician’s willingness to accept a test despite lower efficacy, potentially due to lower risks and costs, easier administration, etc. Currently, BBMs are not widely adopted in clinical practice, and guidelines for their clinical acceptance have not been firmly established. Thus, we used a relatively liberal threshold of M = 0.1 to assess non-inferiority. Finally, we evaluated whether the combination of DCAs and BBMs produced additive improvements for Aβ-status predictions. Leave-one-out cross-validation procedures were used to generate subject-wise test predictions for each test, after which logistic regressions were trained on a given ensemble of DCA and BBM to predict Aβ status. Further analyses examined the effects of adding the APOE genotype to the ensembles. These experiments were conducted to mirror recently suggested clinical workflows for AD diagnosis that recommend CI assessment followed by BBMs and eventual confirmatory PET scans or cerebrospinal fluid evaluation ( 43 – 46 ). Results Cognitive Impairment (CI) Classification Performance Figure 2A displays the mean AUC for each cognitive assessment and BBM for cohort classification (see Supplementary Materials for ROC curves). Cohort classification by MMSE and RAVLT is inherently circular and thus biased, as performance on these assessments was part of the cohort definition criteria (red borders; see Supplementary Materials). Nonetheless, the DCR 3-class model performance [AUC=0.89; confidence interval (CI)=0.88,0.9] was comparable to those of the MMSE (AUC=0.82; CI=0.808,0.824) and RAVLT (AUC=0.89; CI=0.885,0.896) and superior to Cognivue Clarity (AUC=0.74; CI=0.730,0.746). All BBMs performed worse at classifying cognitive cohorts when compared to the DCAs (Figure 2A). Model performance was worse for Aβ42/40 (AUC=0.63; CI=0.620,0.639) and pTau-181 (AUC=0.66; CI=0.651,0.673), while pTau-217 (AUC=0.72; CI=0.706,0.726) approached the classification performance of Cognivue, albeit significantly lower than DCR, MMSE, or RAVLT (see Table S3 for additional statistics). The APS performance (AUC=0.60, CI=0.56,0.61) was slightly better than Aβ42/40 but did not surpass either pTau-217 or pTau-181. An age-only model demonstrated the worst overall performance (AUC=0.63, CI=0.62,0.64). Non-inferiority analyses indicated that for cognitive cohort classification, the DCR was superior to all BBMs, marginally superior to Cognivue, and equivalent to MMSE and RAVLT despite their biased performance (Figure 2B). Aβ Status Classification Performance Figures 3 displays the AUC curves and mean AUC values for Aβ classification by each cognitive assessment and BBM. MMSE (AUC=0.71; CI=0.697,0.717), RAVLT (AUC=0.72; CI=0.715,0.734), and Cognivue (AUC=0.70; CI=0.690,0.711) were overall poor predictors of Aβ status. DCR 3-class model (AUC=0.87; CI=0.844,0.892) outperformed all other cognitive tests. pTau-217 (AUC=0.89; CI=0.881,0.893) outperformed Aβ42/40 (AUC=0.81; CI=0.802,0.821) and pTau-181 (AUC=0.78; CI=0.765,0.785), both of which had lower performance than the DCR. APS (AUC=0.85, CI=0.84,0.86) outperformed Aβ42/40 and pTau-181 but was less performant than pTau-217 or DCR. An age-only model performed the worst (AUC=0.65, CI=0.64,0.66). These findings held when the population was limited to CU or CI (MCI+pAD) cohorts (Figure S2, Table S3). Overall, DCR was non-inferior to all BBMs and substantially outperformed the other cognitive assessments in Aβ classification (Figure 3). Combination of BBMs and Cognitive Assessments Figure 4 shows the benefit of combining each cognitive assessment with each BBM in classifying Aβ status with a secondary logistic regression. All cognitive assessments added value to Aβ42/40 by increasing the AUC, with DCR adding more than double the others (AUC=0.87; CI=0.853,0.892). Combining pTau-181 or pTau-217, which on their own outperformed Aβ42/40 (Figure 2), with MMSE, RAVLT, or Cognivue did not improve their performance. However, combining DCR with any BBM improved their classification performance. Combining the DCR with pTau-217, which by itself had the strongest classification performance (AUC=0.89; CI=0.881,0.893), increased the AUC to 0.91 (CI=0.90,0.911). See Tables S4 and S5 for further statistics. Adding the APOE genotype to the DCR/BBM ensembles produced additional performance gains for classifying Aβ status (Table S6). Figure 5 shows the results from bootstrap non-inferiority analyses for DCR and a combination of all BBMs (base models) compared to these ensembles. DCR combined with a panel of all BBMs and APOE genotype produced the strongest results (AUC=0.94). Despite the significant gains produced by ensemble models, the DCR was equivalent or non-inferior to all other complex models in classifying Aβ-PET. Table S7 lists thresholds for DCR-based CI and Aβ-PET status classification. Discussion The access to pharmacologic and non-pharmacologic DMTs for AD is exciting but highlights the challenge of early diagnosis. A positive amyloid PET scan is a hallmark of AD but can also appear in other conditions ( 47 – 49 ) and normal aging ( 50 ). Further, some individuals with AD-related pathology on PET never develop MCI or have MCI that never progresses to dementia ( 51 ). The presence of biomarkers alone is not sufficient for DMT eligibility or to establish a diagnosis of MCI or mild dementia due to AD ( 52 , 53 ). Thus, recently developed guidelines advocate for establishing CI as the first step ( 45 , 54 ), which is challenging in the time-constrained setting of primary care. DCAs have long been hailed as streamlining cognitive evaluation and offering efficient screening and triage tools in primary-care settings. However, to date, they have not fulfilled expectations. Our findings indicate that simply digitizing a paper-and-pencil test and computationally assessing the results is not enough. ML-enabled analysis of the process through which an assessment is completed ( 19 ) and assessing multiple cognitive domains using graphomotor ( 31 ) and voice/speech metrics ( 24 ) is critical for DCAs to add substantial value, particularly in primary care settings ( 19 – 22 ). One such ‘new generation’ DCA, the Digital Clock and Recall™ (DCR), is superior to the MMSE ( 23 ) and Mini-Cog ( 25 ) for detecting CI despite being completed in a fraction of the time, has great performance in detecting verbal memory impairment on the RAVLT ( 24 ), and can detect functional impairment ( 55 ). Furthermore, by borrowing methodology from biology and ‘multiplexing’ results through simultaneous processing of data from a single source through multiple ML models, it may be possible to deploy ML models based on different subsets of DCR metrics to concurrently evaluate different outcomes and thus concurrently predict CI and Aβ-PET status (Figs. 1 and S1). Confirming our hypothesis, we found that DCR-based models successfully identified both CI and Aβ status. Furthermore, we demonstrated that the DCR was the only cognitive assessment that increased the power of BBMs to predict Aβ-PET status, thus creating a paradigm of additive model improvements as more is learned about the person’s health. The DCR outperformed the MMSE and Cognivue and was non-inferior to the RAVLT for cognitive impairment classification despite its shorter time to complete(~ 3 min) and despite the biased results for MMSE and RAVLT due to their inclusion in cohort classification by study organizers. Moreover, the DCR outperformed Aβ42/40 and pTau-181 and was non-inferior to pTau-217 for Aβ classification. Rentz et al. ( 32 ) previously showed that DCTclock, which is a part of the DCR, successfully discriminates CU individuals from those with MCI or mild dementia (AUC = 0.86) and predicts Aβ and tau PET burden in preclinical adults. Our results confirm and expand those findings with a much larger sample size that also includes CI individuals. Our findings demonstrate the utility of a new generation of DCAs, like the DCR, for addressing the challenges exposed by DMTs and the growing demand for early diagnosis of MCI due to AD. First, they offer PCPs a sensitive tool that can seamlessly integrate into their workflow and enable time-efficient and cost-effective means to triage patients for treatments or confirmatory testing. Recent studies at the University of Massachusetts and Indiana University have demonstrated the feasibility and acceptability of the implementation of the DCR in primary-care settings ( 22 , 56 , 57 ). For the many patients with MCI due to AD who may be ineligible for pharmacologic DMTs ( 58 , 59 ), or those who choose not to take medications, early diagnosis is still essential to enable timely lifestyle interventions, which can decrease dementia cases by ~ 45% ( 13 ). Second, the finding that the DCR performed strongly at both CI and PET Aβ-status classification has important implications for patients and healthcare systems. For a patient, it can take several days to receive the result of a blood test, but an appropriate DCA can assess CI in a few minutes, deliver results immediately at the point of care, and provide an indication for acquiring PET or CSF biomarkers. For healthcare systems, such DCAs enable early patient identification at scale, with low-cost personnel and resources, and without overwhelming laboratory or imaging facilities. It is important to note that many of the recent publications showing the sensitivity and specificity of BBMs for detecting Aβ status were conducted on populations that were already established as cognitively impaired ( 60 – 63 ) (see ( 64 ) for a review). Third, DCAs offer important advantages in lower-income countries, which have the highest projected dementia prevalence. The 3-minute DCR is delivered via a commercially available tablet, is non-invasive, does not require specialized personnel or medical facilities, and a single device can screen numerous patients. Such a solution could increase access to new therapies and help mitigate healthcare disparities due to lower socioeconomic status or living in remote or rural areas. Finally, DCAs like the DCR can streamline recruitment for AD clinical trials, which cost more than other therapeutic areas, compounded by recruitment costs and the time required for participant identification. During 1995–2021, the costs of developing AD drugs were estimated at $ 42.5 billion ( 65 ), with 50–70% of the costs devoted to participant identification and screening ( 66 ). Novel therapeutic pipelines go beyond amyloid and tau as primary targets (20–25% during 2019–2022), enabling evaluation of comparative effectiveness among treatments. DCAs like the DCR can offer time- and cost-efficient solutions to streamline screening but can also enable scalable, objective longitudinal monitoring of patients’ treatment response, potentially capturing not only cognitive but also pathological outcomes. Combining Digital Cognitive Assessments (DCAs) and Blood-Based Biomarkers (BBMs) BBMs are useful for screening, diagnosis, and treatment-response monitoring ( 52 , 53 , 67 ) and represent a major advance in the evaluation of patients with AD ( 68 – 70 ). Several studies have shown associations between biomarker positivity and cognitive decline ( 67 , 71 , 72 ). Ultimately, BBMs can facilitate the identification and monitoring of DMT-eligible patients. However, as noted by Alcolea et al., “...blood biomarkers for neurodegeneration are not anticipated to substitute clinical judgment” ( 73 ) (see also ( 53 )). Accordingly, our results show that relative to cognitive assessments, BBMs are poor predictors of CI and that combining appropriate DCAs with BBMs can help detect CI and improve the prediction of Aβ-PET status over and above BBMs or DCAs alone. These synergistic results are particularly compelling considering both the recently revised AD diagnostic criteria ( 52 ) and the “clinical-biological construct” proposed for AD ( 53 ). Research into racial and ethnic differences in BBMs is only just now being conducted. Next-generation DCAs like the DCR have been shown to be less biased by demographic groups ( 23 ) and may prove to be a strong first-line screening tool for AD pathology. Further research, including the development of normative datasets and cutoffs for various demographic groups, is required for BBMs to reach their full potential and enable understanding of the complex relationship between BBMs, cognition, and pathology. Recent studies have explored the value of combining multiple biomarkers as well as genetic factors to increase the accuracy of AD diagnosis. For example, the addition of pTau-181 and APOE improved the Aβ classification ability of plasma Aβ42/40 ( 72 ). In another study, combining measures of memory and executive function with pTau-217 and APOE improved the ability of pTau-217 to predict dementia due to AD within 4 years (from AUC of 0.83 to 0.91) ( 74 ). Our analyses yielded similar results with AUC = 0.91 for the combination of DCR and pTau-217 without APOE genotyping. This greatly improves upon the reported accuracy of specialists’ clinical diagnosis (AUC = 0.71) ( 74 ). Our results highlight the utility of combining DCAs with a panel of BBMs and genetic profiling to produce a holistic assessment of a person’s likelihood for CI and AD-related pathology. Conclusions At present, as summarized in Fig. 6 , PCPs wait for a patient-initiated complaint to complete a ‘for-cause’ evaluation, which then leads to a specialist referral. At that point, patients tend to be cognitively impaired, resulting in most patients with MCI remaining undiagnosed ( 4 ). Specialist evaluation leads to the diagnosis, but treatment often occurs late and cannot meaningfully prevent or reduce disability. Only about 8% of patients in such a workflow may be eligible for DMTs ( 59 ). New-generation, AI-enabled DCAs, such as the DCR, offer efficient and effective tools for concurrently identifying cognitive impairment and AD-related pathology. They integrate easily into clinical workflows in primary care and will facilitate the early identification of individuals with MCI or early dementia due to AD, empowering PCPs to help their patients and their families make more effective and timely decisions. Abbreviations Aβ Amyloid-beta AD Alzheimer’s Disease APOE Apolipoprotein E APS Amyloid Probability Score AUC Area Under the Receiver Operating Characteristic Curve BBM Blood-Based Biomarker BH Bio-Hermes CSF Cerebrospinal Fluid CU Cognitively Unimpaired DCA Digital Cognitive Assessment DCR Digital Clock and Recall DMT Disease-Modifying Treatment FAQ Functional Activities Questionnaire FDA Food and Drug Administration GAP Global Alzheimer’s Platform MCI Mild Cognitive Impairment ML Machine Learning MMSE Mini-Mental State Examination MRI Magnetic Resonance Imaging pAD Probable Alzheimer’s Dementia PCP Primary Care Provider PET Positron Emission Tomography pTau-181 Phosphorylated Tau-181 pTau-217 Phosphorylated Tau-217 RAVLT Rey Auditory Verbal Learning Test ROC Receiver Operating Characteristic SUVR Standardized Uptake Value Ratio Declarations Ethics approval and consent to participate The study was performed in accordance with the Declaration of Helsinki and its later amendments. The study procedures were explained to participants verbally and through written informed consent that was approved by the local IRB of each site participating in the GAP consortium (see the Bio-Hermes study website(75) for a list of study sites). If, in the opinion of the site principal investigator, the participant did not have the capacity to sign the informed consent form, a legally authorized representative was used to grant consent on behalf of the participant. Consent for publication Not applicable Availability of data and materials The data underlying the findings of this study were collected as part of the Bio-Hermes-001 study (ClinicalTrials.gov Identifier: NCT04733989) and are governed by the Global Alzheimer's Platform (GAP) consortium agreement. Data will be made available via the Alzheimer’s Disease Data Initiative (ADDI) Workbench in the future and at the discretion of GAP. All requests for data access should be made directly to GAP. The code used to calculate the reported results is available from Linus Health, Inc. upon reasonable request and with the permission of Linus Health, Inc. Usage restrictions apply to the availability of this code, which is not immediately publicly available. Competing interests AJ, KT, CTS, CH, JS, and ST are employees of Linus Health and declare ownership of shares or share options in the company. DB is a co-founder of Linus Health and declares ownership of shares or share options in the company. APL is a co-founder of Linus Health and declares ownership of shares or share options in the company. APL serves as a paid member of the scientific advisory boards for Neuroelectrics, Magstim Inc., TetraNeuron, AscenZion, Bitbrain, Skin2Neuron, and MedRhythms. Funding The Bio-Hermes-001 study was organized by the Global Alzheimer's Platform (GAP) and funded by the Alzheimer’s Drug Discovery Foundation (ADDF). Neither GAP nor ADDF had any influence on the analysis, decision to publish, or manuscript preparation. Authors' contributions AJ defined the aims of the analysis, interpreted the data, and drafted and revised the manuscript for content, including medical writing. KT played a major role in the design of the analysis, as well as the analysis and interpretation of data, and revised the manuscript for content. CTS contributed to the analysis and interpretation of data and revised the manuscript for content. JGO revised the manuscript for content. RB revised the manuscript for content. CH analyzed the data and revised the manuscript for content. JS interpreted the data and revised the manuscript for content. DB contributed to the study concept and design, definition of aims of the analysis, interpretation of data, and revision of the manuscript. ST played a major role in the design and interpretation of the analysis, and in drafting and revising the manuscript for content, including medical writing. APL contributed to the study concept and design, played a major role in the conception and definition of aims and interpretation of data, and drafted and revised the manuscript for content, including medical writing. All authors read and approved the final manuscript. Acknowledgements We thank the participants, organizers, and staff of the Bio-Hermes-001 study. References Nichols E, Steinmetz JD, Vollset SE, Fukutaki K, Chalek J, Abd-Allah F, et al. Estimation of the global prevalence of dementia in 2019 and forecasted prevalence in 2050: an analysis for the Global Burden of Disease Study 2019. The Lancet Public Health. 2022 Feb;7(2):e105–25. National Institute on Aging [Internet]. [cited 2023 Feb 23]. Alzheimer’s Disease Fact Sheet. Available from: https://www.nia.nih.gov/health/alzheimers-disease-fact-sheet Salari N, Lotfi F, Abdolmaleki A, Heidarian P, Rasoulpoor S, Fazeli J, et al. The global prevalence of mild cognitive impairment in geriatric population with emphasis on influential factors: a systematic review and meta-analysis. BMC Geriatr. 2025 May 6;25(1):313. Mattke S, Jun H, Chen E, Liu Y, Becker A, Wallick C. Expected and diagnosed rates of mild cognitive impairment and dementia in the U.S. Medicare population: observational analysis. Alz Res Therapy. 2023 Jul 22;15(1):128. Alzheimer’s Association. 2022 Alzheimer’s disease facts and figures. Alzheimers Dement. 2022 Apr;18(4):700–89. Food and Drug Administration (FDA). FDA. FDA; 2023 [cited 2023 Jul 22]. FDA Converts Novel Alzheimer’s Disease Treatment to Traditional Approval. Available from: https://www.fda.gov/news-events/press-announcements/fda-converts-novel-alzheimers-disease-treatment-traditional-approval Center for Drug Evaluation and Research. FDA approves treatment for adults with Alzheimer’s disease. FDA [Internet]. 2024 Jul 2 [cited 2025 Feb 11]; Available from: https://www.fda.gov/drugs/news-events-human-drugs/fda-approves-treatment-adults-alzheimers-disease Kulmala J, Ngandu T, Havulinna S, Levälahti E, Lehtisalo J, Solomon A, et al. The Effect of Multidomain Lifestyle Intervention on Daily Functioning in Older People. J Am Geriatr Soc. 2019 Jun;67(6):1138–44. Chowdhary N, Barbui C, Anstey KJ, Kivipelto M, Barbera M, Peters R, et al. Reducing the Risk of Cognitive Decline and Dementia: WHO Recommendations. Front Neurol. 2021;12:765584. Lehtisalo J, Palmer K, Mangialasche F, Solomon A, Kivipelto M, Ngandu T. Changes in Lifestyle, Behaviors, and Risk Factors for Cognitive Impairment in Older Persons During the First Wave of the Coronavirus Disease 2019 Pandemic in Finland: Results From the FINGER Study. Frontiers in Psychiatry [Internet]. 2021 [cited 2022 Aug 18];12. Available from: https://www.frontiersin.org/articles/10.3389/fpsyt.2021.624125 Solomon A, Handels R, Wimo A, Antikainen R, Laatikainen T, Levälahti E, et al. Effect of a Multidomain Lifestyle Intervention on Estimated Dementia Risk. J Alzheimers Dis. 2021;82(4):1461–6. Ornish D, Madison C, Kivipelto M, Kemp C, McCulloch CE, Galasko D, et al. Effects of intensive lifestyle changes on the progression of mild cognitive impairment or early dementia due to Alzheimer’s disease: a randomized, controlled clinical trial. Alz Res Therapy. 2024 Jun 7;16(1):122. Livingston G, Huntley J, Liu KY, Costafreda SG, Selbæk G, Alladi S, et al. Dementia prevention, intervention, and care: 2024 report of the Lancet standing Commission. The Lancet [Internet]. 2024 Jul 31 [cited 2024 Aug 8];0(0). Available from: https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(24)01296-0/fulltext Commissioner O of the. FDA. FDA; 2025 [cited 2025 May 19]. FDA Clears First Blood Test Used in Diagnosing Alzheimer’s Disease. Available from: https://www.fda.gov/news-events/press-announcements/fda-clears-first-blood-test-used-diagnosing-alzheimers-disease Beckman Coulter Receives FDA Breakthrough Device Designation for Alzheimer’s Disease Blood Test [Internet]. [cited 2025 May 19]. Available from: httpss://www.beckmancoulter.com/about-beckman-coulter/newsroom/press-releases/2025/q1/2025-jan28-bec-receives-fda-breakthrough-device-designation Quanterix. Breaking Ground in Alzheimer’s Diagnosis: Simoa® p-Tau 217 Blood Test Receives FDA Breakthrough Device Designation [Internet]. Quanterix. 2024 [cited 2025 May 19]. Available from: https://www.quanterix.com/breaking-ground-in-alzheimers-diagnosis-simoa-p-tau-217-blood-test-receives-fda-breakthrough-device-designation/ Roche granted FDA Breakthrough Device Designation for blood test to support earlier Alzheimer’s disease diagnosis [Internet]. [cited 2025 May 19]. Available from: https://www.roche.com/media/releases/med-cor-2024-04-11 Spear Bio. FDA breakthrough device designation of novel pTau 217 blood test [Internet]. Spear Bio. 2025 [cited 2025 May 19]. Available from: https://spear.bio/blog/2025/01/13/spear-bio-secures-fda-breakthrough-device-designation-for-its-novel-ptau-217-blood-test-advancing-scalable-solutions-for-early-alzheimers-disease-diagnosis/ Libon DJ, Swenson R, Lamar M, Price CC, Baliga G, Pascual-Leone A, et al. The Boston Process Approach and Digital Neuropsychological Assessment: Past Research and Future Directions. Loewenstein D, editor. JAD. 2022 Jun 14;87(4):1419–32. Libon DJ, Matusz EF, Cosentino S, Price CC, Swenson R, Vermeulen M, et al. Using digital assessment technology to detect neuropsychological problems in primary care settings. Front Psychol. 2023 Nov 17;14:1280593. Libon DJ, Swenson R, Price CC, Lamar M, Cosentino S, Bezdicek O, et al. Digital assessment of cognition in neurodegenerative disease: a data driven approach leveraging artificial intelligence. Front Psychol. 2024 Jul 5;15:1415629. Doerr AJ, Orwig TA, McNulty M, Sison SDM, Paquette DR, Leung R, et al. Digital Assessment of Cognitive Health in Outpatient Primary Care: Usability Study. JMIR Form Res. 2025 Mar 12;9:e66695. Jannati A, Toro-Serey C, Gomes-Osman J, Banks R, Ciesla M, Showalter J, et al. Digital Clock and Recall is superior to the Mini-Mental State Examination for the detection of mild cognitive impairment and mild dementia. Alz Res Therapy. 2024 Jan 2;16(1):2. Banks R, Higgins C, Greene BR, Jannati A, Gomes-Osman J, Tobyne S, et al. Clinical classification of memory and cognitive impairment with multimodal digital biomarkers. Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring. 2024;16(1):e12557. Gomes-Osman J, Borson S, Toro-Serey C, Banks R, Ciesla M, Jannati A, et al. Digital Clock and Recall: a digital, process-driven evolution of the Mini-Cog. Front Hum Neurosci. 2024 Aug 26;18:1337851. Beauregard DW, Mohs R, Dwyer J, Hollingshead S, Smith K, Bork J, et al. Bio-Hermes: A Validation Study to Assess a Meaningful Relationship Between Blood and Digital Biomarkers with Aβ PET Scans for Alzheimer’s Disease. Alzheimer’s & Dementia. 2022;18(S5):e063676. Mohs RC, Beauregard D, Dwyer J, Gaudioso J, Bork J, MaGee‐Rodgers T, et al. The Bio‐Hermes Study: Biomarker database developed to investigate blood‐based and digital biomarkers in community‐based, diverse populations clinically screened for Alzheimer’s disease. Alzheimer’s & Dementia. 2024 Feb 28;alz.13722. Borson S, Scanlan J, Brush M, Vitaliano P, Dokmak A. The mini-cog: a cognitive “vital signs” measure for dementia screening in multi-lingual elderly. Int J Geriatr Psychiatry. 2000 Nov;15(11):1021–7. Schmidt M. Rey auditory verbal learning test: A handbook. Vol. 17. Western Psychological Services Los Angeles, CA; 1996. Cahn-Hidalgo D, Estes PW, Benabou R. Validity, reliability, and psychometric properties of a computerized, cognitive assessment test (Cognivue®). WJP. 2020 Jan 19;10(1):1–11. Souillard-Mandar W, Penney D, Schaible B, Pascual-Leone A, Au R, Davis R. DCTclock: Clinically-Interpretable and Automated Artificial Intelligence Analysis of Drawing Behavior for Capturing Cognition. Frontiers in Digital Health [Internet]. 2021 [cited 2022 Feb 1];3. Available from: https://www.ncbi.nlm.nih.gov/labs/pmc/articles/PMC8553980/ Rentz DM, Papp KV, Mayblyum DV, Sanchez JS, Klein H, Souillard-Mandar W, et al. Association of Digital Clock Drawing With PET Amyloid and Tau Pathology in Normal Older Adults. Neurology. 2021 Apr 6;96(14):e1844–54. Belloy ME, Andrews SJ, Le Guen Y, Cuccaro M, Farrer LA, Napolioni V, et al. APOE Genotype and Alzheimer Disease Risk Across Age, Sex, and Population Ancestry. JAMA Neurology [Internet]. 2023 Nov 6 [cited 2023 Dec 7]; Available from: https://doi.org/10.1001/jamaneurol.2023.3599 Wolk DA, Dickerson BC, the Alzheimer’s Disease Neuroimaging Initiative, Weiner M, Aiello M, Aisen P, et al. Apolipoprotein E (APOE) genotype has dissociable effects on memory and attentional-executive network function in Alzheimer’s disease. Proceedings of the National Academy of Sciences. 2010 Jun 1;107(22):10256–61. Monane M, Johnson KG, Snider BJ, Turner RS, Drake JD, Maraganore DM, et al. A blood biomarker test for brain amyloid impacts the clinical evaluation of cognitive impairment. Ann Clin Transl Neurol. 2023 Oct;10(10):1738–48. Palmqvist S, Janelidze S, Quiroz YT, Zetterberg H, Lopera F, Stomrud E, et al. Discriminative Accuracy of Plasma Phospho-tau217 for Alzheimer Disease vs Other Neurodegenerative Disorders. JAMA. 2020 Aug 25;324(8):772–81. Karikari TK, Pascoal TA, Ashton NJ, Janelidze S, Benedet AL, Rodriguez JL, et al. Blood phosphorylated tau 181 as a biomarker for Alzheimer’s disease: a diagnostic performance and prediction modelling study using data from four prospective cohorts. The Lancet Neurology. 2020;19(5):422–33. Clark CM, Schneider JA, Bedell BJ, Beach TG, Bilker WB, Mintun MA, et al. Use of florbetapir-PET for imaging β-amyloid pathology. Jama. 2011;305(3):275–83. Wolz R, Aljabar P, Hajnal JV, Hammers A, Rueckert D, Alzheimer’s Disease Neuroimaging Initiative. LEAP: learning embeddings for atlas propagation. Neuroimage. 2010 Jan 15;49(2):1316–25. Grecchi E, Foley C, Gispert JD, Wolz R. P3‐434: CENTILOID PET SUVR ANALYSIS USING THE SUPRATENTORIAL WHITE MATTER AS REFERENCE REGION. Alzheimer’s & Dementia [Internet]. 2018 Jul [cited 2023 Aug 26];14(7S_Part_24). Available from: https://alz-journals.onlinelibrary.wiley.com/doi/10.1016/j.jalz.2018.06.1797 Liu X. Classification accuracy and cut point selection. Stat Med. 2012 Oct 15;31(23):2676–86. Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, et al. Scikit-learn: Machine Learning in Python. Journal of Machine Learning Research. 2011;12:2825–30. Hansson O, Edelmayer RM, Boxer AL, Carrillo MC, Mielke MM, Rabinovici GD, et al. The Alzheimer’s Association appropriate use recommendations for blood biomarkers in Alzheimer’s disease. Alzheimer’s & Dementia. 2022;18(12):2669–86. Therriault J, Janelidze S, Benedet AL, Ashton NJ, Arranz Martínez J, Gonzalez-Escalante A, et al. Diagnosis of Alzheimer’s disease using plasma biomarkers adjusted to clinical probability. Nat Aging. 2024 Nov;4(11):1529–37. Udeh‐Momoh CT, Mielke MM, Schindler SE, Hansson O, Khachaturian AS, Weiss J. Bridging the Gap: The Global CEOi Collaborative Workgroup for Adoption of Alzheimer’s disease Blood‐Based Biomarkers in Clinical Practice. Alzheimer’s & Dementia. 2023 Dec;19(S24):e082822. Schindler SE, Galasko D, Pereira AC, Rabinovici GD, Salloway S, Suárez-Calvet M, et al. Acceptable performance of blood biomarker tests of amyloid pathology — recommendations from the Global CEO Initiative on Alzheimer’s Disease. Nat Rev Neurol. 2024 Jun 12;1–14. Diaz-Galvan P, Przybelski SA, Lesnick TG, Schwarz CG, Senjem ML, Gunter JL, et al. β-Amyloid Load on PET Along the Continuum of Dementia With Lewy Bodies. Neurology. 2023 Jul 11;101(2):e178–88. Tan RH, Kril JJ, Yang Y, Tom N, Hodges JR, Villemagne VL, et al. Assessment of amyloid β in pathologically confirmed frontotemporal dementia syndromes. Alzheimers Dement (Amst). 2017 May 29;9:10–20. Gurol ME, Becker JA, Fotiadis P, Riley G, Schwab K, Johnson KA, et al. Florbetapir-PET to diagnose cerebral amyloid angiopathy: A prospective study. Neurology. 2016 Nov 8;87(19):2043–9. Haller S, Montandon ML, Lilja J, Rodriguez C, Garibotto V, Herrmann FR, et al. PET amyloid in normal aging: direct comparison of visual and automatic processing methods. Sci Rep. 2020 Oct 7;10(1):16665. SantaCruz KS, Sonnen JA, Pezhouh MK, Desrosiers MF, Nelson PT, Tyas SL. Alzheimer disease pathology in subjects without dementia in 2 studies of aging: the Nun Study and the Adult Changes in Thought Study. J Neuropathol Exp Neurol. 2011 Oct;70(10):832–40. Jack CR, Andrews JS, Beach TG, Buracchio T, Dunn B, Graf A, et al. Revised criteria for diagnosis and staging of Alzheimer’s disease: Alzheimer’s Association Workgroup. Alzheimers Dement. 2024 Aug;20(8):5143–69. Dubois B, Villain N, Schneider L, Fox N, Campbell N, Galasko D, et al. Alzheimer Disease as a Clinical-Biological Construct—An International Working Group Recommendation. JAMA Neurol. 2024 Dec 1;81(12):1304. Frisoni GB, Festari C, Massa F, Ramusino MC, Orini S, Aarsland D, et al. European intersocietal recommendations for the biomarker-based diagnosis of neurocognitive disorders. The Lancet Neurology. 2024 Mar 1;23(3):302–12. Ciesla M, Toro-Serey C, Jannati A, Banks RE, Gomes-Osman J, Showalter J, et al. Detecting functional impairment with the Digital Clock and Recall. Journal of Alzheimer’s Disease. 2024;102(2):329–37. Fowler NR, Hammers DB, Perkins AJ, Summanwar D, Higbie A, Swartzell K, et al. Feasibility and Acceptability of Implementing a Digital Cognitive Assessment for Alzheimer Disease and Related Dementias in Primary Care. Ann Fam Med. 2025 Apr 29;240293. Summanwar D, Fowler NR, Hammers DB, Perkins AJ, Brosch JR, Willis DR. Agile Implementation of a Digital Cognitive Assessment for Dementia in Primary Care. Ann Fam Med. 2025 Apr 29;240294. Cummings J, Apostolova L, Rabinovici GD, Atri A, Aisen P, Greenberg S, et al. Lecanemab: Appropriate Use Recommendations. J Prev Alz Dis [Internet]. 2023 [cited 2023 Jul 25]; Available from: https://link.springer.com/article/10.14283/jpad.2023.30 Pittock RR, Aakre J, Castillo AM, Ramanan VK, Kremers WK, Jack CR, et al. Eligibility for Anti-Amyloid Treatment in a Population-Based Study of Cognitive Aging. Neurology [Internet]. 2023 Aug 16 [cited 2023 Aug 23]; Available from: https://www.neurology.org/lookup/doi/10.1212/WNL.0000000000207770 Brum WS, Cullen NC, Janelidze S, Ashton NJ, Zimmer ER, Therriault J, et al. A two-step workflow based on plasma p-tau217 to screen for amyloid β positivity with further confirmatory testing only in uncertain cases. Nat Aging. 2023 Aug 31;3(9):1079–90. Manjavong M, Kang JM, Diaz A, Ashford MT, Eichenbaum J, Aaronson A, et al. Performance of Plasma Biomarkers Combined with Structural MRI to Identify Candidate Participants for Alzheimer’s Disease-Modifying Therapy. J Prev Alzheimers Dis. 2024;11(5):1198–205. Cano A, Capdevila M, Puerta R, Arranz J, Montrreal L, de Rojas I, et al. Clinical value of plasma pTau181 to predict Alzheimer’s disease pathology in a large real-world cohort of a memory clinic. EBioMedicine. 2024 Oct;108:105345. Arranz J, Zhu N, Rubio-Guerra S, Rodríguez-Baz Í, Ferrer R, Carmona-Iragui M, et al. Diagnostic performance of plasma pTau217, pTau181, Aβ1-42 and Aβ1-40 in the LUMIPULSE automated platform for the detection of Alzheimer disease. Alzheimers Res Ther. 2024 Jun 26;16(1):139. Garcia-Escobar G, Manero RM, Fernández-Lebrero A, Ois A, Navalpotro-Gómez I, Puente-Periz V, et al. Blood Biomarkers of Alzheimer’s Disease and Cognition: A Literature Review. Biomolecules. 2024 Jan 11;14(1):93. Cummings JL, Goldman DP, Simmons‐Stern NR, Ponton E. The costs of developing treatments for Alzheimer’s disease: A retrospective exploration. Alzheimer’s & Dementia. 2022 Mar;18(3):469–77. Cummings J, Lee G, Nahed P, Kambar MEZN, Zhong K, Fonseca J, et al. Alzheimer’s disease drug development pipeline: 2022. Alzheimer’s & Dementia: Translational Research & Clinical Interventions. 2022;8(1):e12295. Bucci M, Chiotis K, Nordberg A, Alzheimer’s Disease Neuroimaging Initiative. Alzheimer’s disease profiled by fluid and imaging markers: tau PET best predicts cognitive decline. Molecular Psychiatry. 2021;26(10):5888–98. Thijssen EH, La Joie R, Wolf A, Strom A, Wang P, Iaccarino L, et al. Diagnostic value of plasma phosphorylated tau181 in Alzheimer’s disease and frontotemporal lobar degeneration. Nat Med. 2020 Mar;26(3):387–97. Leuzy A, Mattsson‐Carlgren N, Palmqvist S, Janelidze S, Dage JL, Hansson O. Blood‐based biomarkers for Alzheimer’s disease. EMBO Mol Med. 2022 Jan 11;14(1):e14408. Mattsson-Carlgren N, Palmqvist S. The emerging era of staging Alzheimer’s disease pathology using plasma biomarkers. Brain. 2023 May 2;146(5):1740–2. Li RX, Ma YH, Tan L, Yu JT. Prospective biomarkers of Alzheimer’s disease: A systematic review and meta-analysis. Ageing Research Reviews. 2022 Nov 1;81:101699. Palmqvist S, Stomrud E, Cullen N, Janelidze S, Manuilova E, Jethwa A, et al. An accurate fully automated panel of plasma biomarkers for Alzheimer’s disease. Alzheimer’s & Dementia. 2023 Apr;19(4):1204–15. Alcolea D, Beeri MS, Rojas JC, Gardner RC, Lleó A. Blood Biomarkers in Neurodegenerative Diseases: Implications for the Clinical Neurologist. Neurology. 2023 Jul 25;101(4):172–80. Palmqvist S, Tideman P, Cullen N, Zetterberg H, Blennow K, the Alzheimer’s Disease Neuroimaging Initiative, et al. Prediction of future Alzheimer’s disease dementia using plasma phospho-tau combined with other accessible measures. Nat Med. 2021 Jun;27(6):1034–42. Global Alzheimer’s Platform. Bio-Hermes study [Internet]. [cited 2022 May 11]. Available from: https://globalalzplatform.org/biohermesstudy/ Additional Declarations Competing interest reported. AJ, KT, CTS, CH, JS, and ST are employees of Linus Health and declare ownership of shares or share options in the company. DB is a co-founder of Linus Health and declares ownership of shares or share options in the company. APL is a co-founder of Linus Health and declares ownership of shares or share options in the company. APL serves as a paid member of the scientific advisory boards for Neuroelectrics, Magstim Inc., TetraNeuron, AscenZion, Bitbrain, Skin2Neuron, and MedRhythms. Supplementary Files SupplementaryMaterialsSubmissiontoART.docx Cite Share Download PDF Status: Published Journal Publication published 11 Dec, 2025 Read the published version in Alzheimer's Research & Therapy → Version 1 posted Editorial decision: Revision requested 19 Jul, 2025 Reviews received at journal 02 Jul, 2025 Reviews received at journal 24 Jun, 2025 Reviewers agreed at journal 18 Jun, 2025 Reviewers agreed at journal 12 Jun, 2025 Reviewers invited by journal 11 Jun, 2025 Editor assigned by journal 29 May, 2025 Submission checks completed at journal 29 May, 2025 First submitted to journal 28 May, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6768373","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":463315360,"identity":"908dae96-fece-4690-bf3b-da1aafaf9248","order_by":0,"name":"Ali Jannati","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA80lEQVRIiWNgGAWjYFCCA2CSh5+dsQHESGBgYG5gJkKLAY9kM1wLY2MzEVYZMBgchrAIa+FvPPzscUHNHxnjw8xtj3n+MOTxsx9sf1yAR4vEgWPmxjOOGfCYHWZsN+ZtYyiW7ElsbJ6B1ysHzKR52MBa2qR5GxgSNxwAauHBo0P+wPFv0jz/DHiMm4FagA5L3H/+IX4tBgfOmEnzthnwGDCDtLABbZEgYIvhgTNl0rx9xjwSQIdJzm2TSJxx42HjbHxa5G4c3ybN803Onr+9/ZnEmz82if39yQc+49MCDDJULj61UMDfQISiUTAKRsEoGNkAAOzFTB6bc6dpAAAAAElFTkSuQmCC","orcid":"","institution":"Harvard Medical School","correspondingAuthor":true,"prefix":"","firstName":"Ali","middleName":"","lastName":"Jannati","suffix":""},{"id":463315361,"identity":"6b256372-4283-47f4-965f-273e15a329da","order_by":1,"name":"Karl Thompson","email":"","orcid":"","institution":"Linus Health, Inc","correspondingAuthor":false,"prefix":"","firstName":"Karl","middleName":"","lastName":"Thompson","suffix":""},{"id":463315362,"identity":"d6825c89-2b3c-4e7d-8e06-3bc27c860ae1","order_by":2,"name":"Claudio Toro-Serey","email":"","orcid":"","institution":"Linus Health, Inc","correspondingAuthor":false,"prefix":"","firstName":"Claudio","middleName":"","lastName":"Toro-Serey","suffix":""},{"id":463315363,"identity":"a6351155-cf5f-4d5a-8246-ba2c89887e94","order_by":3,"name":"Joyce Gomes-Osman","email":"","orcid":"","institution":"University of Miami Miller School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Joyce","middleName":"","lastName":"Gomes-Osman","suffix":""},{"id":463315364,"identity":"a08b8615-edcc-4b73-b7d9-c9995a2ec43f","order_by":4,"name":"Russell E. Banks","email":"","orcid":"","institution":"Michigan State University","correspondingAuthor":false,"prefix":"","firstName":"Russell","middleName":"E.","lastName":"Banks","suffix":""},{"id":463315365,"identity":"abf1a54e-1f15-410a-9e9b-9a0c0d8720ad","order_by":5,"name":"Connor Higgins","email":"","orcid":"","institution":"Linus Health, Inc","correspondingAuthor":false,"prefix":"","firstName":"Connor","middleName":"","lastName":"Higgins","suffix":""},{"id":463315366,"identity":"c2277c80-74bb-48b7-86ea-4b65dc8904e6","order_by":6,"name":"John Showalter","email":"","orcid":"","institution":"Linus Health, Inc","correspondingAuthor":false,"prefix":"","firstName":"John","middleName":"","lastName":"Showalter","suffix":""},{"id":463315367,"identity":"ad62456c-cd12-4c9e-bff3-a96ede078009","order_by":7,"name":"David Bates","email":"","orcid":"","institution":"Linus Health, Inc","correspondingAuthor":false,"prefix":"","firstName":"David","middleName":"","lastName":"Bates","suffix":""},{"id":463315368,"identity":"49138f08-d8a1-4cb5-b84f-0fa41f676dbc","order_by":8,"name":"Sean Tobyne","email":"","orcid":"","institution":"Linus Health, Inc","correspondingAuthor":false,"prefix":"","firstName":"Sean","middleName":"","lastName":"Tobyne","suffix":""},{"id":463315369,"identity":"5a79cad9-b92c-4749-8715-d89fbb0b8dc0","order_by":9,"name":"Alvaro Pascual-Leone","email":"","orcid":"","institution":"Harvard Medical School","correspondingAuthor":false,"prefix":"","firstName":"Alvaro","middleName":"","lastName":"Pascual-Leone","suffix":""}],"badges":[],"createdAt":"2025-05-28 13:08:26","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6768373/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6768373/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s13195-025-01913-5","type":"published","date":"2025-12-11T15:57:42+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":83682274,"identity":"eb438797-7a25-4c2b-8b0e-ccc4a0e77836","added_by":"auto","created_at":"2025-05-30 16:24:07","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":90231,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMultiplexed machine-learning (ML) pipeline for simultaneous cognitive and pathological assessment using the Digital Clock and Recall (DCR).\u003c/strong\u003e A consumer-grade tablet is used to administer a digital cognitive assessment, in this case the DCR. (\u003cstrong\u003e1. Elicitation Layer\u003c/strong\u003e), capturing rich multimodal data streams during task performance. These data are processed into distinct feature sets that represent thousands of possible metrics, including demographics, traditional accuracy metrics, process metrics, acoustic features, drawing time series, and novel DCTclock metrics (\u003cstrong\u003e2. Data Processing Layer\u003c/strong\u003e). Models training steps are visualized in \u003cstrong\u003e3. Model Training Layer.\u003c/strong\u003e Feature selection techniques such as recursive feature elimination (RFE) are applied to identify the most informative feature sets for each model outcome variable (\u003cstrong\u003e4. Feature Selection\u003c/strong\u003e), which are then input into separate and distinct ML models optimized to predict cognitive impairment status and amyloid beta pathology status (\u003cstrong\u003e5. Model Optimization). \u003c/strong\u003eThus the feature sets for each model can overlap but do include unique patterns of features and interactions. Following model optimizing, the decision threshold(s) are optimized according to how the model outcome will be used in practice (\u003cstrong\u003e6. Thresholding\u003c/strong\u003e). This multiplexed approach enables a single, time-efficient assessment to support the concurrent prediction of distinct clinical outcomes, enhancing sensitivity and specificity in primary care-compatible workflows.\u003c/p\u003e","description":"","filename":"1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6768373/v1/5a456ff2e1c00b12064bd6a4.jpg"},{"id":83682276,"identity":"4b66e6c3-1e4c-4f46-b8a6-f4b0cef12b95","added_by":"auto","created_at":"2025-05-30 16:24:07","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":68024,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eA. \u003c/strong\u003eMean area under the curve (AUC) values (y-axis) averaged across model training iterations for the prediction of cohort classification (cognitively unimpaired [CU] versus MCI or pAD) for each of the cognitive assessments (orange bars) and blood-based biomarkers (BBMs, purple bars). Higher values indicate better performance. Note the AUC values for MMSE and RAVLT (red bordered bars) were part of the criteria for cohort determination and their performance is thus biased. DCR outperformed all other cognitive tests and biomarkers except for the RAVLT, including the MMSE’s biased performance. All BBMs demonstrated poor performance classifying cognitive impairment. \u003cstrong\u003eB. \u003c/strong\u003eResults of the non-inferiority analysis comparing DCR classification performance against other cognitive tests and BBMs. Comparisons are listed on the y-axis while the difference in AUC is plotted on the x-axis. Median AUC values (blue dots) and the 95% confidence interval (blue bars) were calculated from 5000 bootstrap iterations. Vertical dashed lines indicate the margin=0.1 threshold used to establish non-inferiority/equivalence or superiority of one model’s accuracy over another. DCR was statistically superior to all BBMs for the classification of cognitive impairment, and very nearly superior to Cognivue. Despite the biased performance of the MMSE and RAVLT, the DCR had equivalent performance for classifying cognitive impairment.\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6768373/v1/9925562029e9926cb97e8b55.jpg"},{"id":83682577,"identity":"10a8b291-79a1-470d-aa06-61746d0320bc","added_by":"auto","created_at":"2025-05-30 16:32:07","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":88639,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eA. \u003c/strong\u003ePredictive accuracy of blood biomarkers, cognitive assessments, and DCR for PET amyloid status as shown by receiver operating characteristic curves. \u003cstrong\u003eB. \u003c/strong\u003eMean AUC values (y-axis) averaged across model training iterations for the prediction of Aβ status for each of the cognitive assessments and blood-based biomarkers (BBMs). Higher values indicate better performance. DCR outperformed all other cognitive tests, AB 42/40, and pTau-181 in classification performance. Only pTau-217 demonstrated a higher average AUC. \u003cstrong\u003eC. \u003c/strong\u003eResults of the non-inferiority analysis comparing DCR classification performance against other cognitive tests and BBMs. Comparisons are listed on the y-axis while the difference in AUC is plotted on the x-axis. Median AUC values (blue dots) and the 95% confidence interval (blue bars) were calculated from 5000 bootstrap iterations, sampling with replacement from the distributions of predictions made by each model. Vertical dashed lines indicate the margin=0.1. DCR was statistically superior to Cognivue and MMSE for the classification of Aβ status, and nearly superior but at least equivalent to RAVLT. DCR was found to have statistically equivalent performance to all BBMs for classifying Aβ status.\u003cstrong\u003e \u003c/strong\u003eAPS, C2N amyloid probability score; AUC, area under the receiver operating characteristic curve; DCR, Digital Clock and Recall; MMSE, Mini-Mental State Examination; RAVLT, Rey Auditory Verbal Learning Test.\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6768373/v1/65bac2dc927df1c17a97e6f4.jpg"},{"id":83682277,"identity":"5731a717-7a43-42b1-a354-963a1d49ee64","added_by":"auto","created_at":"2025-05-30 16:24:07","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":45745,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eChange in area under the curve (AUC) for prediction of Aβ status on brain PET scan between each biomarker alone versus a combination of that biomarker and each cognitive assessment (MMSE, RAVLT, Cognivue, or DCR).\u003c/strong\u003e Results are expressed in percent change from the AUC for BBM alone. Note that baseline AUC differed between the three BBMs: Aβ42/40 = 0.81; pTau-181 = 0.78; pTau-217 = 0.89; APS = 0.852. Positive change values indicate that the combination of the cognitive assessment with the biomarker produced a higher AUC than that with the biomarker alone. Negative change values indicate that the addition of the cognitive assessment weakened the predictive value of the combination. Aβ, amyloid-beta; AUC, area under the receiver operating characteristic curve; DCR, Digital Clock and Recall; MMSE, mini-mental state examination; pTau, phosphorylated tau; RAVLT, Rey Auditory Verbal Learning Test.\u003c/p\u003e","description":"","filename":"4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6768373/v1/f728a795ac990b8d7a1e9be5.jpg"},{"id":83682279,"identity":"199d0834-b076-459a-aad5-27669f5c89af","added_by":"auto","created_at":"2025-05-30 16:24:07","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":79811,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eNon-inferiority plot showing the bootstrapped 95% confidence interval of the difference in AUC between the base (DCR, All BBMs) and ensemble models.\u003c/strong\u003e Each base model denotes a predictive model using only that feature as a predictor. Ensemble models are denoted by color. DCR by itself is often strong enough when compared to more complex models. All ensemble models are contained within the equivalent zones. Aβ, amyloid-beta; APOE, apolipoprotein E; APS, amyloid probability score; AUC, area under the receiver operating characteristic curve; BBM, blood-based biomarker; DCR, Digital Clock and Recall; MMSE, mini-mental state examination; pTau, phosphorylated tau; RAVLT, Rey Auditory Verbal Learning Test.\u003c/p\u003e","description":"","filename":"5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6768373/v1/052677369409d5ac77a74f20.jpg"},{"id":83682578,"identity":"c9ecbb2e-7626-4d27-82df-5a50aec301cf","added_by":"auto","created_at":"2025-05-30 16:32:07","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":58669,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCurrent clinical workflow compared to the envisioned clinical workflow for diagnosing and treating patients with cognitive impairment using DCAs such as the Digital Clock and Recall. \u003c/strong\u003eAD, Alzheimer’s disease; BBM, blood-based biomarker; CSF, cerebrospinal fluid; DCA, digital cognitive assessment; MRI, magnetic resonance imaging; PCP, primary care provider; PET, positron emission tomography.\u003c/p\u003e","description":"","filename":"6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6768373/v1/c78bafef2672e495e796bb6d.jpg"},{"id":98244048,"identity":"5569588c-8c3a-4817-b22c-3189ef92e742","added_by":"auto","created_at":"2025-12-15 16:12:44","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1621024,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6768373/v1/c80524ad-87b6-4e57-aea6-747fc576592b.pdf"},{"id":83682280,"identity":"5cfe8930-32e3-4a71-abd0-46a683ce4618","added_by":"auto","created_at":"2025-05-30 16:24:07","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":186393,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryMaterialsSubmissiontoART.docx","url":"https://assets-eu.researchsquare.com/files/rs-6768373/v1/146ba8698a72fc76068732d6.docx"}],"financialInterests":"Competing interest reported. AJ, KT, CTS, CH, JS, and ST are employees of Linus Health and declare ownership of shares or share options in the company. DB is a co-founder of Linus Health and declares ownership of shares or share options in the company. APL is a co-founder of Linus Health and declares ownership of shares or share options in the company. APL serves as a paid member of the scientific advisory boards for Neuroelectrics, Magstim Inc., TetraNeuron, AscenZion, Bitbrain, Skin2Neuron, and MedRhythms.","formattedTitle":"Concurrent Detection of Cognitive Impairment and Amyloid Positivity with a Next-Generation Digital Cognitive Assessment","fulltext":[{"header":"Introduction","content":"\u003cp\u003eDementia is the most feared health condition as we age (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e), and Alzheimer\u0026rsquo;s disease (AD) is the most common cause of dementia (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). Nearly one out of four people globally has MCI (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e). Yet, over 90% of individuals with MCI currently remain undiagnosed (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e), and, in the US, less than 30% of annual wellness exams include cognitive assessments (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e). The Food and Drug Administration (FDA) has approved the anti-amyloid antibody therapies lecanemab (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e) and donanemab (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e) for the treatment of mild cognitive impairment (MCI) and mild dementia due to AD. At the same time, significant evidence exists supporting the value of lifestyles that promote brain-healthy behaviors to reduce the risk of progression from MCI to dementia and minimize brain-related disability (\u003cspan additionalcitationids=\"CR9 CR10 CR11 CR12\" citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e). However, for these and future pharmacologic and non-pharmacologic disease-modifying treatments (DMTs) to be truly impactful, healthcare systems must address the challenge of early diagnosis. This is critical for maximizing the benefits of disease-modifying therapies (DMTs) and empowering patients to plan for the future, establish advance care directives, adopt meaningful lifestyle changes, and set achievable personal goals. Primary care providers (PCPs) are strategically positioned to ensure the timely diagnosis of MCI or mild dementia due to AD. Blood-based biomarkers (BBMs) promise to enable early and efficient determination of amyloid brain deposition and AD pathology - indeed, several have received clearance or breakthrough designation from the FDA (\u003cspan additionalcitationids=\"CR15 CR16 CR17\" citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e). However, they require establishing cognitive impairment, and existing pencil-paper assessments of cognition can be unwieldy, inefficient, and inaccurate. PCPs need more efficient and scalable assessments that integrate into their already time-pressured workflows without increasing provider burden.\u003c/p\u003e \u003cp\u003eFirst-generation digital cognitive assessments (DCAs) were designed to streamline cognitive evaluation and have been proposed as potential solutions to the challenges of conducting cognitive assessments in primary care settings. Compared with traditional paper-pencil tests, DCAs allow for automated scoring, standardize data flow, alleviate subjectivity in scoring, simplify data storage, integrate with existing hospital systems, and ease longitudinal tracking over time. However, they offer limited time saving and little improvement in sensitivity and specificity. Next-generation DCAs capitalize on the power of technology to implement \u0026lsquo;process metrics\u0026rsquo; \u0026ndash; a framework for neuropsychological assessment that derives insights about behavior, brain, and cognitive function not only from an individual's final test achievement, but also from their \u003cem\u003eprocess\u003c/em\u003e of completing cognitive tasks (\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e). It is thus possible to transform the sensor array of a consumer device (e.g., a tablet) into a medical device, and leverage multimodal signal processing and artificial intelligence (AI) models to assess both the \u003cem\u003eoutput and process\u003c/em\u003e of a given task. This promises to enable the development of highly time-efficient assessments of superior sensitivity and specificity, ideally suited for integration into primary care settings (\u003cspan additionalcitationids=\"CR21\" citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e). Furthermore, by \u0026lsquo;multiplexing\u0026rsquo; the same output metrics across multiple outcomes, it may be possible to apply distinct AI algorithms to the same process-level data, enabling the concurrent evaluation of different outcomes. This approach could allow a single brief assessment to simultaneously characterize cognitive function and estimate the likelihood of a specific underlying pathology (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eWe hypothesized that the Digital Clock and Recall\u0026trade; (DCR) (\u003cspan additionalcitationids=\"CR24\" citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e), a next-generation multimodal DCA, would be able to concurrently detect cognitive impairment (CI) and predict the likelihood of a positive amyloid-beta (Aβ) positron emission tomography (PET) of the brain, thus potentially providing an efficient and cost-effective means to detect MCI due to AD, and identify patients for DMTs in primary care. Our objectives were: (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) assess the ability of the DCR to concurrently predict CI and Aβ status, (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) compare DCR\u0026rsquo;s performance to other cognitive assessments and BBMs, and (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) provide evidence for the additive benefit of combining insights from the DCR with BBMs and genetic testing to improve the detection of CI and Aβ status.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003eData were obtained from 930 participants with PET, cognitive, genetic, and BBM data in the Bio-Hermes-001 (BH) study (ClinicalTrials.gov Identifier: NCT04733989) (\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e). The study was administered by the Global Alzheimer\u0026rsquo;s Platform Foundation (GAP), and study protocols were administered by clinical trial sites within the community. All subjects provided written informed consent before their participation.\u003c/p\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eParticipants and Study Cohort\u003c/h2\u003e \u003cp\u003eEnrollment criteria are listed in the Supplementary Materials and elsewhere (\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e). Participants were \u0026gt;\u0026thinsp;24% non-White or Latino/Hispanic, and 56.8% female. They were classified \u003cem\u003ea priori\u003c/em\u003e by the BH study team into cognitively unimpaired (CU; n\u0026thinsp;=\u0026thinsp;398), MCI (n\u0026thinsp;=\u0026thinsp;291), or probable Alzheimer\u0026rsquo;s dementia (pAD; n\u0026thinsp;=\u0026thinsp;241). Cohort classification was based on assessment with the Mini-Mental State Examination (MMSE) score, Rey Auditory Verbal Learning Test (RAVLT), and the Functional Activities Questionnaire (FAQ), as well as a clinical interview and review of medical records. Alternatively, a cohort status of MCI or pAD could be assigned if a clinical diagnosis was made within 3 months of Visit 1 (see Supplementary Materials for full criteria). Tables\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and \u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e summarize the participants\u0026rsquo; demographics, Aβ status, and cohort classification. Authors were blind to cohort classification and Aβ status until data lock.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSummary statistics for demographic details by Aβ PET status.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMetric\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAβ- (n\u0026thinsp;=\u0026thinsp;604)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAβ+ (n\u0026thinsp;=\u0026thinsp;326)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eStatistical Test\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge (Mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD) (Range)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e70.75\u0026thinsp;\u0026plusmn;\u0026thinsp;6.66 (59\u0026ndash;85)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e74.24\u0026thinsp;\u0026plusmn;\u0026thinsp;6.15 (60\u0026ndash;85)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eT=-7.83, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGender (%F)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e354 (59%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e174 (53%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eX\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.003, p\u0026thinsp;=\u0026thinsp;0.95\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYears of education\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e15.51\u0026thinsp;\u0026plusmn;\u0026thinsp;2.61\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e15.39\u0026thinsp;\u0026plusmn;\u0026thinsp;2.89\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eT\u0026thinsp;=\u0026thinsp;0.7, p\u0026thinsp;=\u0026thinsp;0.49\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEthnicity (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eX\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;4.5, p\u0026thinsp;=\u0026thinsp;0.11\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNot Hispanic or Latino\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e536 (89%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e288 (88%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHispanic or Latino\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e61 (10%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e28 (9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNot Reported\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7 (1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e10 (3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRace (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eX\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;12.75, p\u0026thinsp;=\u0026thinsp;0.026\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWhite\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e507 (84%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e293 (90%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBlack / African American\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e76 (13%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e28 (9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAsian\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e11 (2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e5 (1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAmerican Indian / Alaskan Native\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2 (\u0026lt;\u0026thinsp;1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNative Hawaiian or Other Pacific Islander\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1 (\u0026lt;\u0026thinsp;1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUnknown\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7 (1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCognitively Unimpaired (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e314 (52%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e84 (26%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eX\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.09, p\u0026thinsp;=\u0026thinsp;0.77\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSUVR (Mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.00\u0026thinsp;\u0026plusmn;\u0026thinsp;0.09\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.42\u0026thinsp;\u0026plusmn;\u0026thinsp;0.22\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eT=-39.94, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eStatistical tests confirmed the Aβ\u0026thinsp;+\u0026thinsp;group had a significantly greater Aβ load than the Aβ- group (mean SUVR 1.00 vs. 1.42). Aβ\u0026thinsp;+\u0026thinsp;participants tended to be older (70.75 vs. 74.24) but had similar mean years of education. Groups were also not significantly different in ethnicity (Hispanic vs. Not Hispanic). Groups were balanced on gender and the proportion of cognitively unimpaired individuals. Statistically different distributions for race were noted between the groups, due in part to the greater proportion of White and lower proportion of Black / African American participants in the Aβ- group.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eCognitive Assessments\u003c/h3\u003e\n\u003cp\u003eThe MMSE can detect mild to moderate dementia in older adults, but it can take over 10 minutes to complete (\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e). The RAVLT is widely used to assess verbal memory but can take over 30 minutes to complete (\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e). Cognivue\u0026reg; Clarity is an FDA-listed computerized tool designed to detect early signs of cognitive impairment through a 10-minute assessment administered via proprietary hardware (\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSummary statistics for the cognitive cohorts defined by the organizers of the Bio-Hermes study.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMetric\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCU (n\u0026thinsp;=\u0026thinsp;398)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMCI (n\u0026thinsp;=\u0026thinsp;291)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003epAD (n\u0026thinsp;=\u0026thinsp;241)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eStatistical Test\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge (mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD) (range)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e70.35\u0026thinsp;\u0026plusmn;\u0026thinsp;6.40 (60\u0026ndash;85)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e72.18\u0026thinsp;\u0026plusmn;\u0026thinsp;6.84 (60\u0026ndash;85)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e74.42\u0026thinsp;\u0026plusmn;\u0026thinsp;6.22 (59\u0026ndash;85)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eF\u0026thinsp;=\u0026thinsp;33.23, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGender (%F)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e242 (\u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e61\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e156 (\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e130 (\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eX\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.005, p\u0026thinsp;=\u0026thinsp;0.99\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYears of education\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e15.77\u0026thinsp;\u0026plusmn;\u0026thinsp;2.48\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e15.49\u0026thinsp;\u0026plusmn;\u0026thinsp;2.76\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e14.95\u0026thinsp;\u0026plusmn;\u0026thinsp;2.95\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eF\u0026thinsp;=\u0026thinsp;8.78, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEthnicity (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eX\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;13.74, p\u0026thinsp;=\u0026thinsp;0.008\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNot Hispanic or Latino\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e378\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e271\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e213\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHispanic or Latino\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e29\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e35\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNot Reported\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRace (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eX\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;12.75, p\u0026thinsp;=\u0026thinsp;0.023\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWhite\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e345 (87)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e253 (87)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e202 (84)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBlack / African American\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e42 (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e30 (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e32 (\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAsian\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7 (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e5 (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4 (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAmerican Indian / Alaskan Native\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1 (\u0026lt;\u0026thinsp;1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2 (\u0026lt;\u0026thinsp;1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNative Hawaiian / Pacific Islander\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1 (\u0026lt;\u0026thinsp;1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1 (\u0026lt;\u0026thinsp;1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUnknown\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2 (\u0026lt;\u0026thinsp;1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSUVR (Mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.07\u0026thinsp;\u0026plusmn;\u0026thinsp;0.19\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.14\u0026thinsp;\u0026plusmn;\u0026thinsp;0.25\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.28\u0026thinsp;\u0026plusmn;\u0026thinsp;0.29\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eF\u0026thinsp;=\u0026thinsp;59.44, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eA one-way analysis of variance (ANOVA) test revealed a significant main effect for age across groups. Post-hoc analyses revealed that the cognitively unimpaired (CU) cohort was significantly younger than the mild cognitive impairment (MCI) cohort, which was in turn significantly younger than the probable Alzheimer\u0026rsquo;s dementia (pAD) cohort. ANOVA with Tukey\u0026rsquo;s post-hoc comparisons also revealed that the CU cohort had significantly fewer years of education compared to the pAD cohort but was not significantly different from the MCI cohort. Cohorts were significantly different in ethnicity (Hispanic vs. Not Hispanic) and race. Finally, a significant difference in standardized uptake value ratio (SUVR) was found among cohorts, and post-hoc comparisons revealed that the pAD cohort had significantly higher SUVR values than both the MCI and CU cohorts. Similarly, the MCI cohort had significantly greater SUVR values than the CU cohort.\u003c/p\u003e \u003cp\u003eThe Linus Health DCR (\u003cspan additionalcitationids=\"CR24\" citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e) is an FDA-listed Class II software as a medical device and is a machine learning (ML)-enabled adaptation of the Mini-Cog (\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e). The 3-minute DCR is completed with an iPad and Apple Pencil and can be administered by medical or research assistants without specialized training. The DCR includes performance in 3-word immediate and delayed recall tests, intervened by a digital clock drawing test (DCTclock\u0026trade;) (\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e). Drawing strokes and recorded speech are analyzed via automated scoring and ML algorithms to assess verbal memory, executive function, visuospatial reasoning, information processing, and motor function, and to generate both traditional and process metrics (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e).\u003c/p\u003e\n\u003ch3\u003eApolipoprotein E (APOE) Genotyping\u003c/h3\u003e\n\u003cp\u003eThe APOE e4 allele is a major risk factor for AD (\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e) and is associated with episodic memory impairment (\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e). APOE genotyping was conducted on a blood sample and assayed by C\u003csub\u003e2\u003c/sub\u003eN Diagnostics laboratories (\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e).\u003c/p\u003e\n\u003ch3\u003eBlood-Based Biomarkers (BBMs)\u003c/h3\u003e\n\u003cp\u003eIn addition to the APOE assay, C\u003csub\u003e2\u003c/sub\u003eN Diagnostics provided plasma Aβ42 and Aβ40 concentrations measured by mass spectroscopy (\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e). C\u003csub\u003e2\u003c/sub\u003eN\u0026rsquo;s PrecivityAD\u0026trade; Amyloid Probability Score (APS) combines the Aβ42/40 ratio, APOE genotype, and age to produce a 0\u0026ndash;100 score where higher scores represent a higher likelihood of Aβ positivity. pTau-217 and pTau-181 concentrations were measured by Lilly Research Laboratories (\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e) and the Simoa HD-1 (Quanterix) (\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e), respectively.\u003c/p\u003e\n\u003ch3\u003ePositron Emission Tomography\u003c/h3\u003e\n\u003cp\u003eAmyloid PET scans were collected using the FDA-approved tracer \u003csup\u003e18\u003c/sup\u003eF-florbetapir (\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e) and uploaded to a portal accessible to specialists. Automated preprocessing (\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e) and quantification of standardized uptake value ratios (SUVR) (\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e) were performed centrally by IXICO Technologies, Inc. (Wilmington, DE, USA). Experts were blind to participants\u0026rsquo; clinical information or cognitive and BBM results (\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e). This expert determination was used to define a binary Aβ status, which was used as ground truth for all experiments. Amyloid PET analysis indicated that 326 participants were Aβ+ (35%) and 604 were Aβ\u0026ndash; (65%). The SUVRs confirmed the expected significant difference between Aβ- and Aβ\u0026thinsp;+\u0026thinsp;groups (1.003\u0026thinsp;\u0026plusmn;\u0026thinsp;0.1 vs. 1.423\u0026thinsp;\u0026plusmn;\u0026thinsp;0.2; Cohen\u0026rsquo;s \u003cem\u003ed\u003c/em\u003e\u0026thinsp;=\u0026thinsp;3.7). Mean (\u0026plusmn;\u0026thinsp;SD) SUVR was 1.071 (\u0026plusmn;\u0026thinsp;0.2), 1.148 (\u0026plusmn;\u0026thinsp;0.2), and 1.283 (\u0026plusmn;\u0026thinsp;0.3) in the CU, MCI, and pAD cohorts, respectively (see Table S2 for additional statistics). Authors were blind to participant Aβ status until data lock and did not participate in establishing SUVR cutoffs, nor did they participate in the analysis of PET scans.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eAnalytical Approach\u003c/h2\u003e \u003cp\u003eSeparate ML models were trained to predict cognitive impairment (CI; combining MCI and pAD) or PET Aβ status by each cognitive test and BBM (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). All models were trained and tested using leave-one-out cross-validation. During the inference phase, the training loop model was applied to the left-out subject\u0026rsquo;s data to produce a probability indicating their CI or Aβ status. This process was repeated for all subjects in the sample. The binary classification cutoff value for Aβ status was determined for each model using the concordance probability method (\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e). For each model, we report the mean performance metrics across all iterations. We also report model performance for a 3-class classification paradigm that specifies positive and negative CI or Aβ-status classes, but also an \u0026lsquo;indeterminate\u0026rsquo; class. The two thresholds were determined by simultaneously optimizing sensitivity and specificity while minimizing the percentage of subjects in the indeterminate range (see Supplementary Materials). The 3-class classification paradigm was developed to capture the subjects for whom confidence of CI or Aβ status was relatively lower and provide a potential clinical workflow directing these subjects for confirmatory testing.\u003c/p\u003e \u003cp\u003eGiven the numerous variables derived from the DCR, a random forest classifier was employed, incorporating features from both the Command and Copy Clock tasks, immediate and delayed recall tests, and the participant\u0026rsquo;s age. Clock features included those extracted by the DCTclock algorithm (\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e), newly developed clock-drawing process metric features, and temporal and spatial features extracted from the Apple Pencil (e.g., altitude, azimuth, pressure). Immediate and delayed recall features included the number of words repeated/remembered correctly, and acoustic and linguistic metrics calculated from the speech recordings (\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e). Recursive feature elimination (\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e) was applied independently to each model training process to identify the subset of DCR features that yielded optimal performance for the CI and Aβ-status classification models, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eFor other cognitive tests and BBMs, logistic regressions were trained using each test\u0026rsquo;s most predictive variable combined with the participant\u0026rsquo;s age and/or APOE status. For other cognitive tests, predictors included the MMSE total score, the RAVLT long-delay score, and the Cognivue Clarity average test score (\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e). BBM models used APS, Aβ42/40, pTau-181, and pTau-217 levels as input features. An age-only logistic regression was also trained as a control.\u003c/p\u003e \u003cp\u003eCritically, both MMSE and RAVLT were part of the BH \u003cem\u003ea priori\u003c/em\u003e criteria for cognitive cohort definition and, therefore, were positively biased in their CI classification performance. These analyses, while circular, were included to provide a hypothetical \u0026lsquo;ceiling\u0026rsquo; to assess CI classification performance. Biased results are highlighted in red in the corresponding figures (see figure legends).\u003c/p\u003e \u003cp\u003eWe performed bootstrap non-inferiority procedures (\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e) to compare DCR against the other tests for CI and Aβ-status classification, and to compare unimodal (i.e., BBM or DCA separately) models to ensemble models (i.e., combinations of BBMs and DCAs). First, separately for each cognitive assessment or BBM, we took the mean prediction probability for each participant across 10 cross-validated model training iterations. Then, on each of the 5,000 bootstrapping iterations, we sampled the target-positive and target-negative groups (e.g., with and without CI) with replacement and calculated the difference in the area under the receiver operating characteristic curve (AUC) between the two models. Inferiority, non-inferiority, and superiority were established if the 95% confidence interval of the bootstrapped differences was lower than, within, or higher than the AUC difference plus the margin M, respectively. The choice of M depends on several factors, including a clinician\u0026rsquo;s willingness to accept a test despite lower efficacy, potentially due to lower risks and costs, easier administration, etc. Currently, BBMs are not widely adopted in clinical practice, and guidelines for their clinical acceptance have not been firmly established. Thus, we used a relatively liberal threshold of M\u0026thinsp;=\u0026thinsp;0.1 to assess non-inferiority.\u003c/p\u003e \u003cp\u003eFinally, we evaluated whether the combination of DCAs and BBMs produced additive improvements for Aβ-status predictions. Leave-one-out cross-validation procedures were used to generate subject-wise test predictions for each test, after which logistic regressions were trained on a given ensemble of DCA and BBM to predict Aβ status. Further analyses examined the effects of adding the APOE genotype to the ensembles. These experiments were conducted to mirror recently suggested clinical workflows for AD diagnosis that recommend CI assessment followed by BBMs and eventual confirmatory PET scans or cerebrospinal fluid evaluation (\u003cspan additionalcitationids=\"CR44 CR45\" citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cem\u003eCognitive Impairment (CI) Classification Performance\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eFigure 2A displays the mean AUC for each cognitive assessment and BBM for cohort classification (see Supplementary Materials for ROC curves). Cohort classification by MMSE and RAVLT is inherently circular and thus biased, as performance on these assessments was part of the cohort definition criteria (red borders; see Supplementary Materials). Nonetheless, the DCR 3-class model performance [AUC=0.89; confidence interval (CI)=0.88,0.9] was comparable to those of the MMSE (AUC=0.82; CI=0.808,0.824) and RAVLT (AUC=0.89; CI=0.885,0.896) and superior to Cognivue Clarity (AUC=0.74; CI=0.730,0.746). All BBMs performed worse at classifying cognitive cohorts when compared to the DCAs (Figure 2A). Model performance was worse for A\u0026beta;42/40 (AUC=0.63; CI=0.620,0.639) and pTau-181 (AUC=0.66; CI=0.651,0.673), while pTau-217 (AUC=0.72; CI=0.706,0.726) approached the classification performance of Cognivue, albeit significantly lower than DCR, MMSE, or RAVLT (see Table S3 for additional statistics). The APS performance (AUC=0.60, CI=0.56,0.61) was slightly better than A\u0026beta;42/40 but did not surpass either pTau-217 or pTau-181. An age-only model demonstrated the worst overall performance (AUC=0.63, CI=0.62,0.64). Non-inferiority analyses indicated that for cognitive cohort classification, the DCR was superior to all BBMs, marginally superior to Cognivue, and equivalent to MMSE and RAVLT despite their biased performance (Figure 2B).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eA\u0026beta;\u0026nbsp;\u003c/em\u003e\u003cem\u003eStatus Classification Performance\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eFigures 3 displays the AUC curves and mean AUC values for A\u0026beta; classification by each cognitive assessment and BBM. MMSE (AUC=0.71; CI=0.697,0.717), RAVLT (AUC=0.72; CI=0.715,0.734), and Cognivue (AUC=0.70; CI=0.690,0.711) were overall poor predictors of A\u0026beta; status. DCR 3-class model (AUC=0.87; CI=0.844,0.892) outperformed all other cognitive tests. pTau-217 (AUC=0.89; CI=0.881,0.893) outperformed A\u0026beta;42/40 (AUC=0.81; CI=0.802,0.821) and pTau-181 (AUC=0.78; CI=0.765,0.785), both of which had lower performance than the DCR. APS (AUC=0.85, CI=0.84,0.86) outperformed A\u0026beta;42/40 and pTau-181 but was less performant than pTau-217 or DCR. An age-only model performed the worst (AUC=0.65, CI=0.64,0.66). These findings held when the population was limited to CU or CI (MCI+pAD) cohorts (Figure S2, Table S3). Overall, DCR was non-inferior to all BBMs and substantially outperformed the other cognitive assessments in A\u0026beta; classification (Figure 3).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eCombination of BBMs and Cognitive Assessments\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eFigure 4 shows the benefit of combining each cognitive assessment with each BBM in classifying A\u0026beta; status with a secondary logistic regression. All cognitive assessments added value to A\u0026beta;42/40 by increasing the AUC, with DCR adding more than double the others (AUC=0.87; CI=0.853,0.892). Combining pTau-181 or pTau-217, which on their own outperformed A\u0026beta;42/40 (Figure 2), with MMSE, RAVLT, or Cognivue did not improve their performance. However, combining DCR with any BBM improved their classification performance. Combining the DCR with pTau-217, which by itself had the strongest classification performance (AUC=0.89; CI=0.881,0.893), increased the AUC to 0.91 (CI=0.90,0.911). See Tables S4 and S5 for further statistics.\u003c/p\u003e\n\u003cp\u003eAdding the APOE genotype to the DCR/BBM ensembles produced additional performance gains for classifying A\u0026beta; status (Table S6). Figure 5 shows the results from bootstrap non-inferiority analyses for DCR and a combination of all BBMs (base models) compared to these ensembles. DCR combined with a panel of all BBMs and APOE genotype produced the strongest results (AUC=0.94). Despite the significant gains produced by ensemble models, the DCR was equivalent or non-inferior to all other complex models in classifying A\u0026beta;-PET. Table S7 lists thresholds for DCR-based CI and A\u0026beta;-PET status classification.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe access to pharmacologic and non-pharmacologic DMTs for AD is exciting but highlights the challenge of early diagnosis. A positive amyloid PET scan is a hallmark of AD but can also appear in other conditions (\u003cspan additionalcitationids=\"CR48\" citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e) and normal aging (\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e). Further, some individuals with AD-related pathology on PET never develop MCI or have MCI that never progresses to dementia (\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e). The presence of biomarkers alone is not sufficient for DMT eligibility or to establish a diagnosis of MCI or mild dementia due to AD (\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e, \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e). Thus, recently developed guidelines advocate for establishing CI as the first step (\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e, \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e), which is challenging in the time-constrained setting of primary care.\u003c/p\u003e \u003cp\u003eDCAs have long been hailed as streamlining cognitive evaluation and offering efficient screening and triage tools in primary-care settings. However, to date, they have not fulfilled expectations. Our findings indicate that simply digitizing a paper-and-pencil test and computationally assessing the results is not enough. ML-enabled analysis of the process through which an assessment is completed (\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e) and assessing multiple cognitive domains using graphomotor (\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e) and voice/speech metrics (\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e) is critical for DCAs to add substantial value, particularly in primary care settings (\u003cspan additionalcitationids=\"CR20 CR21\" citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e). One such \u0026lsquo;new generation\u0026rsquo; DCA, the Digital Clock and Recall\u0026trade; (DCR), is superior to the MMSE (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e) and Mini-Cog (\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e) for detecting CI despite being completed in a fraction of the time, has great performance in detecting verbal memory impairment on the RAVLT (\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e), and can detect functional impairment (\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e). Furthermore, by borrowing methodology from biology and \u0026lsquo;multiplexing\u0026rsquo; results through simultaneous processing of data from a single source through multiple ML models, it may be possible to deploy ML models based on different subsets of DCR metrics to concurrently evaluate different outcomes and thus concurrently \u003cem\u003epredict\u003c/em\u003e CI and Aβ-PET status (Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and S1). Confirming our hypothesis, we found that DCR-based models successfully identified both CI \u003cem\u003eand\u003c/em\u003e Aβ status. Furthermore, we demonstrated that the DCR was the only cognitive assessment that increased the power of BBMs to predict Aβ-PET status, thus creating a paradigm of additive model improvements as more is learned about the person\u0026rsquo;s health.\u003c/p\u003e \u003cp\u003eThe DCR outperformed the MMSE and Cognivue and was non-inferior to the RAVLT for cognitive impairment classification despite its shorter time to complete(~\u0026thinsp;3 min) and despite the biased results for MMSE and RAVLT due to their inclusion in cohort classification by study organizers. Moreover, the DCR outperformed Aβ42/40 and pTau-181 and was non-inferior to pTau-217 for Aβ classification. Rentz et al. (\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e) previously showed that DCTclock, which is a part of the DCR, successfully discriminates CU individuals from those with MCI or mild dementia (AUC\u0026thinsp;=\u0026thinsp;0.86) and predicts Aβ and tau PET burden in preclinical adults. Our results confirm and expand those findings with a much larger sample size that also includes CI individuals.\u003c/p\u003e \u003cp\u003eOur findings demonstrate the utility of a new generation of DCAs, like the DCR, for addressing the challenges exposed by DMTs and the growing demand for early diagnosis of MCI due to AD. First, they offer PCPs a sensitive tool that can seamlessly integrate into their workflow and enable time-efficient and cost-effective means to triage patients for treatments or confirmatory testing. Recent studies at the University of Massachusetts and Indiana University have demonstrated the feasibility and acceptability of the implementation of the DCR in primary-care settings (\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e, \u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e). For the many patients with MCI due to AD who may be ineligible for pharmacologic DMTs (\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e, \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e), or those who choose not to take medications, early diagnosis is still essential to enable timely lifestyle interventions, which can decrease dementia cases by ~\u0026thinsp;45% (\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eSecond, the finding that the DCR performed strongly at both CI and PET Aβ-status classification has important implications for patients and healthcare systems. For a patient, it can take several days to receive the result of a blood test, but an appropriate DCA can assess CI in a few minutes, deliver results immediately at the point of care, and provide an indication for acquiring PET or CSF biomarkers. For healthcare systems, such DCAs enable early patient identification at scale, with low-cost personnel and resources, and without overwhelming laboratory or imaging facilities. It is important to note that many of the recent publications showing the sensitivity and specificity of BBMs for detecting Aβ status were conducted on populations that were already established as cognitively impaired (\u003cspan additionalcitationids=\"CR61 CR62\" citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e63\u003c/span\u003e) (see (\u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e64\u003c/span\u003e) for a review).\u003c/p\u003e \u003cp\u003eThird, DCAs offer important advantages in lower-income countries, which have the highest projected dementia prevalence. The 3-minute DCR is delivered via a commercially available tablet, is non-invasive, does not require specialized personnel or medical facilities, and a single device can screen numerous patients. Such a solution could increase access to new therapies and help mitigate healthcare disparities due to lower socioeconomic status or living in remote or rural areas.\u003c/p\u003e \u003cp\u003eFinally, DCAs like the DCR can streamline recruitment for AD clinical trials, which cost more than other therapeutic areas, compounded by recruitment costs and the time required for participant identification. During 1995\u0026ndash;2021, the costs of developing AD drugs were estimated at \u003cspan\u003e$\u003c/span\u003e42.5\u0026nbsp;billion (\u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e65\u003c/span\u003e), with 50\u0026ndash;70% of the costs devoted to participant identification and screening (\u003cspan citationid=\"CR66\" class=\"CitationRef\"\u003e66\u003c/span\u003e). Novel therapeutic pipelines go beyond amyloid and tau as primary targets (20\u0026ndash;25% during 2019\u0026ndash;2022), enabling evaluation of comparative effectiveness among treatments. DCAs like the DCR can offer time- and cost-efficient solutions to streamline screening but can also enable scalable, objective longitudinal monitoring of patients\u0026rsquo; treatment response, potentially capturing not only cognitive but also pathological outcomes.\u003c/p\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eCombining Digital Cognitive Assessments (DCAs) and Blood-Based Biomarkers (BBMs)\u003c/h2\u003e \u003cp\u003eBBMs are useful for screening, diagnosis, and treatment-response monitoring (\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e, \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e, \u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e67\u003c/span\u003e) and represent a major advance in the evaluation of patients with AD (\u003cspan additionalcitationids=\"CR69\" citationid=\"CR68\" class=\"CitationRef\"\u003e68\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e70\u003c/span\u003e). Several studies have shown associations between biomarker positivity and cognitive decline (\u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e67\u003c/span\u003e, \u003cspan citationid=\"CR71\" class=\"CitationRef\"\u003e71\u003c/span\u003e, \u003cspan citationid=\"CR72\" class=\"CitationRef\"\u003e72\u003c/span\u003e). Ultimately, BBMs can facilitate the identification and monitoring of DMT-eligible patients. However, as noted by Alcolea et al., \u0026ldquo;...blood biomarkers for neurodegeneration are not anticipated to substitute clinical judgment\u0026rdquo; (\u003cspan citationid=\"CR73\" class=\"CitationRef\"\u003e73\u003c/span\u003e) (see also (\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e)). Accordingly, our results show that relative to cognitive assessments, BBMs are poor predictors of CI and that combining appropriate DCAs with BBMs can help detect CI and improve the prediction of Aβ-PET status over and above BBMs or DCAs alone. These synergistic results are particularly compelling considering both the recently revised AD diagnostic criteria (\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e) and the \u0026ldquo;clinical-biological construct\u0026rdquo; proposed for AD (\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e). Research into racial and ethnic differences in BBMs is only just now being conducted. Next-generation DCAs like the DCR have been shown to be less biased by demographic groups (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e) and may prove to be a strong first-line screening tool for AD pathology. Further research, including the development of normative datasets and cutoffs for various demographic groups, is required for BBMs to reach their full potential and enable understanding of the complex relationship between BBMs, cognition, and pathology.\u003c/p\u003e \u003cp\u003eRecent studies have explored the value of combining multiple biomarkers as well as genetic factors to increase the accuracy of AD diagnosis. For example, the addition of pTau-181 and APOE improved the Aβ classification ability of plasma Aβ42/40 (\u003cspan citationid=\"CR72\" class=\"CitationRef\"\u003e72\u003c/span\u003e). In another study, combining measures of memory and executive function with pTau-217 and APOE improved the ability of pTau-217 to predict dementia due to AD within 4 years (from AUC of 0.83 to 0.91) (\u003cspan citationid=\"CR74\" class=\"CitationRef\"\u003e74\u003c/span\u003e). Our analyses yielded similar results with AUC\u0026thinsp;=\u0026thinsp;0.91 for the combination of DCR and pTau-217 without APOE genotyping. This greatly improves upon the reported accuracy of specialists\u0026rsquo; clinical diagnosis (AUC\u0026thinsp;=\u0026thinsp;0.71) (\u003cspan citationid=\"CR74\" class=\"CitationRef\"\u003e74\u003c/span\u003e). Our results highlight the utility of combining DCAs with a panel of BBMs and genetic profiling to produce a holistic assessment of a person\u0026rsquo;s likelihood for CI and AD-related pathology.\u003c/p\u003e \u003c/div\u003e"},{"header":"Conclusions","content":"\u003cp\u003eAt present, as summarized in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e, PCPs wait for a patient-initiated complaint to complete a \u0026lsquo;for-cause\u0026rsquo; evaluation, which then leads to a specialist referral. At that point, patients tend to be cognitively impaired, resulting in most patients with MCI remaining undiagnosed (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e). Specialist evaluation leads to the diagnosis, but treatment often occurs late and cannot meaningfully prevent or reduce disability. Only about 8% of patients in such a workflow may be eligible for DMTs (\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eNew-generation, AI-enabled DCAs, such as the DCR, offer efficient and effective tools for concurrently identifying cognitive impairment and AD-related pathology. They integrate easily into clinical workflows in primary care and will facilitate the early identification of individuals with MCI or early dementia due to AD, empowering PCPs to help their patients and their families make more effective and timely decisions.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eA\u0026beta;\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Amyloid-beta\u003c/p\u003e\n\u003cp\u003eAD\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Alzheimer\u0026rsquo;s Disease\u003c/p\u003e\n\u003cp\u003eAPOE\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Apolipoprotein E\u003c/p\u003e\n\u003cp\u003eAPS\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Amyloid Probability Score\u003c/p\u003e\n\u003cp\u003eAUC\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Area Under the Receiver Operating Characteristic Curve\u003c/p\u003e\n\u003cp\u003eBBM\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Blood-Based Biomarker\u003c/p\u003e\n\u003cp\u003eBH\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Bio-Hermes\u003c/p\u003e\n\u003cp\u003eCSF\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Cerebrospinal Fluid\u003c/p\u003e\n\u003cp\u003eCU\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Cognitively Unimpaired\u003c/p\u003e\n\u003cp\u003eDCA\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Digital Cognitive Assessment\u003c/p\u003e\n\u003cp\u003eDCR\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Digital Clock and Recall\u003c/p\u003e\n\u003cp\u003eDMT\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Disease-Modifying Treatment\u003c/p\u003e\n\u003cp\u003eFAQ\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Functional Activities Questionnaire\u003c/p\u003e\n\u003cp\u003eFDA\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Food and Drug Administration\u003c/p\u003e\n\u003cp\u003eGAP\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Global Alzheimer\u0026rsquo;s Platform\u003c/p\u003e\n\u003cp\u003eMCI\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Mild Cognitive Impairment\u003c/p\u003e\n\u003cp\u003eML\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Machine Learning\u003c/p\u003e\n\u003cp\u003eMMSE\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Mini-Mental State Examination\u003c/p\u003e\n\u003cp\u003eMRI\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Magnetic Resonance Imaging\u003c/p\u003e\n\u003cp\u003epAD\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Probable Alzheimer\u0026rsquo;s Dementia\u003c/p\u003e\n\u003cp\u003ePCP\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Primary Care Provider\u003c/p\u003e\n\u003cp\u003ePET\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Positron Emission Tomography\u003c/p\u003e\n\u003cp\u003epTau-181\u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Phosphorylated Tau-181\u003c/p\u003e\n\u003cp\u003epTau-217\u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Phosphorylated Tau-217\u003c/p\u003e\n\u003cp\u003eRAVLT\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Rey Auditory Verbal Learning Test\u003c/p\u003e\n\u003cp\u003eROC\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Receiver Operating Characteristic\u003c/p\u003e\n\u003cp\u003eSUVR \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Standardized Uptake Value Ratio\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study was performed in accordance with the Declaration of Helsinki and its later amendments. The study procedures were explained to participants verbally and through written informed consent that was approved by the local IRB of each site participating in the GAP consortium (see the Bio-Hermes study website(75) for a list of study sites). If, in the opinion of the site principal investigator, the participant did not have the capacity to sign the informed consent form, a legally authorized representative was used to grant consent on behalf of the participant.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data underlying the findings of this study were collected as part of the Bio-Hermes-001 study (ClinicalTrials.gov Identifier: NCT04733989) and are governed by the Global Alzheimer\u0026apos;s Platform (GAP) consortium agreement. Data will be made available via the Alzheimer\u0026rsquo;s Disease Data Initiative (ADDI) Workbench in the future and at the discretion of GAP. All requests for data access should be made directly to GAP. The code used to calculate the reported results is available from Linus Health, Inc. upon reasonable request and with the permission of Linus Health, Inc. Usage restrictions apply to the availability of this code, which is not immediately publicly available.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAJ, KT, CTS, CH, JS, and ST are employees of Linus Health and declare ownership of shares or share options in the company. DB is a co-founder of Linus Health and declares ownership of shares or share options in the company. APL is a co-founder of Linus Health and declares ownership of shares or share options in the company. APL serves as a paid member of the scientific advisory boards for Neuroelectrics, Magstim Inc., TetraNeuron, AscenZion, Bitbrain, Skin2Neuron, and MedRhythms.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe Bio-Hermes-001 study was organized by the Global Alzheimer\u0026apos;s Platform (GAP) and funded by the Alzheimer\u0026rsquo;s Drug Discovery Foundation (ADDF). Neither GAP nor ADDF had any influence on the analysis, decision to publish, or manuscript preparation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026apos; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAJ defined the aims of the analysis, interpreted the data, and drafted and revised the manuscript for content, including medical writing. KT played a major role in the design of the analysis, as well as the analysis and interpretation of data, and revised the manuscript for content. CTS contributed to the analysis and interpretation of data and revised the manuscript for content. JGO revised the manuscript for content. RB revised the manuscript for content. CH analyzed the data and revised the manuscript for content. JS interpreted the data and revised the manuscript for content. DB contributed to the study concept and design, definition of aims of the analysis, interpretation of data, and revision of the manuscript. ST played a major role in the design and interpretation of the analysis, and in drafting and revising the manuscript for content, including medical writing. APL contributed to the study concept and design, played a major role in the conception and definition of aims and interpretation of data, and drafted and revised the manuscript for content, including medical writing. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe thank the participants, organizers, and staff of the Bio-Hermes-001 study.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eNichols E, Steinmetz JD, Vollset SE, Fukutaki K, Chalek J, Abd-Allah F, et al. Estimation of the global prevalence of dementia in 2019 and forecasted prevalence in 2050: an analysis for the Global Burden of Disease Study 2019. The Lancet Public Health. 2022 Feb;7(2):e105\u0026ndash;25.\u003c/li\u003e\n\u003cli\u003eNational Institute on Aging [Internet]. [cited 2023 Feb 23]. Alzheimer\u0026rsquo;s Disease Fact Sheet. Available from: https://www.nia.nih.gov/health/alzheimers-disease-fact-sheet\u003c/li\u003e\n\u003cli\u003eSalari N, Lotfi F, Abdolmaleki A, Heidarian P, Rasoulpoor S, Fazeli J, et al. The global prevalence of mild cognitive impairment in geriatric population with emphasis on influential factors: a systematic review and meta-analysis. BMC Geriatr. 2025 May 6;25(1):313.\u003c/li\u003e\n\u003cli\u003eMattke S, Jun H, Chen E, Liu Y, Becker A, Wallick C. Expected and diagnosed rates of mild cognitive impairment and dementia in the U.S. Medicare population: observational analysis. Alz Res Therapy. 2023 Jul 22;15(1):128.\u003c/li\u003e\n\u003cli\u003eAlzheimer\u0026rsquo;s Association. 2022 Alzheimer\u0026rsquo;s disease facts and figures. Alzheimers Dement. 2022 Apr;18(4):700\u0026ndash;89.\u003c/li\u003e\n\u003cli\u003eFood and Drug Administration (FDA). FDA. FDA; 2023 [cited 2023 Jul 22]. FDA Converts Novel Alzheimer\u0026rsquo;s Disease Treatment to Traditional Approval. Available from: https://www.fda.gov/news-events/press-announcements/fda-converts-novel-alzheimers-disease-treatment-traditional-approval\u003c/li\u003e\n\u003cli\u003eCenter for Drug Evaluation and Research. FDA approves treatment for adults with Alzheimer\u0026rsquo;s disease. FDA [Internet]. 2024 Jul 2 [cited 2025 Feb 11]; Available from: https://www.fda.gov/drugs/news-events-human-drugs/fda-approves-treatment-adults-alzheimers-disease\u003c/li\u003e\n\u003cli\u003eKulmala J, Ngandu T, Havulinna S, Lev\u0026auml;lahti E, Lehtisalo J, Solomon A, et al. The Effect of Multidomain Lifestyle Intervention on Daily Functioning in Older People. J Am Geriatr Soc. 2019 Jun;67(6):1138\u0026ndash;44.\u003c/li\u003e\n\u003cli\u003eChowdhary N, Barbui C, Anstey KJ, Kivipelto M, Barbera M, Peters R, et al. Reducing the Risk of Cognitive Decline and Dementia: WHO Recommendations. Front Neurol. 2021;12:765584.\u003c/li\u003e\n\u003cli\u003eLehtisalo J, Palmer K, Mangialasche F, Solomon A, Kivipelto M, Ngandu T. Changes in Lifestyle, Behaviors, and Risk Factors for Cognitive Impairment in Older Persons During the First Wave of the Coronavirus Disease 2019 Pandemic in Finland: Results From the FINGER Study. Frontiers in Psychiatry [Internet]. 2021 [cited 2022 Aug 18];12. Available from: https://www.frontiersin.org/articles/10.3389/fpsyt.2021.624125\u003c/li\u003e\n\u003cli\u003eSolomon A, Handels R, Wimo A, Antikainen R, Laatikainen T, Lev\u0026auml;lahti E, et al. Effect of a Multidomain Lifestyle Intervention on Estimated Dementia Risk. J Alzheimers Dis. 2021;82(4):1461\u0026ndash;6.\u003c/li\u003e\n\u003cli\u003eOrnish D, Madison C, Kivipelto M, Kemp C, McCulloch CE, Galasko D, et al. Effects of intensive lifestyle changes on the progression of mild cognitive impairment or early dementia due to Alzheimer\u0026rsquo;s disease: a randomized, controlled clinical trial. Alz Res Therapy. 2024 Jun 7;16(1):122.\u003c/li\u003e\n\u003cli\u003eLivingston G, Huntley J, Liu KY, Costafreda SG, Selb\u0026aelig;k G, Alladi S, et al. Dementia prevention, intervention, and care: 2024 report of the Lancet standing Commission. The Lancet [Internet]. 2024 Jul 31 [cited 2024 Aug 8];0(0). Available from: https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(24)01296-0/fulltext\u003c/li\u003e\n\u003cli\u003eCommissioner O of the. FDA. FDA; 2025 [cited 2025 May 19]. FDA Clears First Blood Test Used in Diagnosing Alzheimer\u0026rsquo;s Disease. Available from: https://www.fda.gov/news-events/press-announcements/fda-clears-first-blood-test-used-diagnosing-alzheimers-disease\u003c/li\u003e\n\u003cli\u003eBeckman Coulter Receives FDA Breakthrough Device Designation for Alzheimer\u0026rsquo;s Disease Blood Test [Internet]. [cited 2025 May 19]. Available from: httpss://www.beckmancoulter.com/about-beckman-coulter/newsroom/press-releases/2025/q1/2025-jan28-bec-receives-fda-breakthrough-device-designation\u003c/li\u003e\n\u003cli\u003eQuanterix. Breaking Ground in Alzheimer\u0026rsquo;s Diagnosis: Simoa\u0026reg; p-Tau 217 Blood Test Receives FDA Breakthrough Device Designation [Internet]. Quanterix. 2024 [cited 2025 May 19]. Available from: https://www.quanterix.com/breaking-ground-in-alzheimers-diagnosis-simoa-p-tau-217-blood-test-receives-fda-breakthrough-device-designation/\u003c/li\u003e\n\u003cli\u003eRoche granted FDA Breakthrough Device Designation for blood test to support earlier Alzheimer\u0026rsquo;s disease diagnosis [Internet]. [cited 2025 May 19]. Available from: https://www.roche.com/media/releases/med-cor-2024-04-11\u003c/li\u003e\n\u003cli\u003eSpear Bio. FDA breakthrough device designation of novel pTau 217 blood test [Internet]. Spear Bio. 2025 [cited 2025 May 19]. Available from: https://spear.bio/blog/2025/01/13/spear-bio-secures-fda-breakthrough-device-designation-for-its-novel-ptau-217-blood-test-advancing-scalable-solutions-for-early-alzheimers-disease-diagnosis/\u003c/li\u003e\n\u003cli\u003eLibon DJ, Swenson R, Lamar M, Price CC, Baliga G, Pascual-Leone A, et al. The Boston Process Approach and Digital Neuropsychological Assessment: Past Research and Future Directions. Loewenstein D, editor. JAD. 2022 Jun 14;87(4):1419\u0026ndash;32.\u003c/li\u003e\n\u003cli\u003eLibon DJ, Matusz EF, Cosentino S, Price CC, Swenson R, Vermeulen M, et al. Using digital assessment technology to detect neuropsychological problems in primary care settings. Front Psychol. 2023 Nov 17;14:1280593.\u003c/li\u003e\n\u003cli\u003eLibon DJ, Swenson R, Price CC, Lamar M, Cosentino S, Bezdicek O, et al. Digital assessment of cognition in neurodegenerative disease: a data driven approach leveraging artificial intelligence. Front Psychol. 2024 Jul 5;15:1415629.\u003c/li\u003e\n\u003cli\u003eDoerr AJ, Orwig TA, McNulty M, Sison SDM, Paquette DR, Leung R, et al. Digital Assessment of Cognitive Health in Outpatient Primary Care: Usability Study. JMIR Form Res. 2025 Mar 12;9:e66695.\u003c/li\u003e\n\u003cli\u003eJannati A, Toro-Serey C, Gomes-Osman J, Banks R, Ciesla M, Showalter J, et al. Digital Clock and Recall is superior to the Mini-Mental State Examination for the detection of mild cognitive impairment and mild dementia. Alz Res Therapy. 2024 Jan 2;16(1):2.\u003c/li\u003e\n\u003cli\u003eBanks R, Higgins C, Greene BR, Jannati A, Gomes-Osman J, Tobyne S, et al. Clinical classification of memory and cognitive impairment with multimodal digital biomarkers. Alzheimer\u0026rsquo;s \u0026amp; Dementia: Diagnosis, Assessment \u0026amp; Disease Monitoring. 2024;16(1):e12557.\u003c/li\u003e\n\u003cli\u003eGomes-Osman J, Borson S, Toro-Serey C, Banks R, Ciesla M, Jannati A, et al. Digital Clock and Recall: a digital, process-driven evolution of the Mini-Cog. Front Hum Neurosci. 2024 Aug 26;18:1337851.\u003c/li\u003e\n\u003cli\u003eBeauregard DW, Mohs R, Dwyer J, Hollingshead S, Smith K, Bork J, et al. Bio-Hermes: A Validation Study to Assess a Meaningful Relationship Between Blood and Digital Biomarkers with A\u0026beta; PET Scans for Alzheimer\u0026rsquo;s Disease. Alzheimer\u0026rsquo;s \u0026amp; Dementia. 2022;18(S5):e063676.\u003c/li\u003e\n\u003cli\u003eMohs RC, Beauregard D, Dwyer J, Gaudioso J, Bork J, MaGee‐Rodgers T, et al. The Bio‐Hermes Study: Biomarker database developed to investigate blood‐based and digital biomarkers in community‐based, diverse populations clinically screened for Alzheimer\u0026rsquo;s disease. Alzheimer\u0026rsquo;s \u0026amp; Dementia. 2024 Feb 28;alz.13722.\u003c/li\u003e\n\u003cli\u003eBorson S, Scanlan J, Brush M, Vitaliano P, Dokmak A. The mini-cog: a cognitive \u0026ldquo;vital signs\u0026rdquo; measure for dementia screening in multi-lingual elderly. Int J Geriatr Psychiatry. 2000 Nov;15(11):1021\u0026ndash;7.\u003c/li\u003e\n\u003cli\u003eSchmidt M. Rey auditory verbal learning test: A handbook. Vol. 17. Western Psychological Services Los Angeles, CA; 1996.\u003c/li\u003e\n\u003cli\u003eCahn-Hidalgo D, Estes PW, Benabou R. Validity, reliability, and psychometric properties of a computerized, cognitive assessment test (Cognivue\u0026reg;). WJP. 2020 Jan 19;10(1):1\u0026ndash;11.\u003c/li\u003e\n\u003cli\u003eSouillard-Mandar W, Penney D, Schaible B, Pascual-Leone A, Au R, Davis R. DCTclock: Clinically-Interpretable and Automated Artificial Intelligence Analysis of Drawing Behavior for Capturing Cognition. Frontiers in Digital Health [Internet]. 2021 [cited 2022 Feb 1];3. Available from: https://www.ncbi.nlm.nih.gov/labs/pmc/articles/PMC8553980/\u003c/li\u003e\n\u003cli\u003eRentz DM, Papp KV, Mayblyum DV, Sanchez JS, Klein H, Souillard-Mandar W, et al. Association of Digital Clock Drawing With PET Amyloid and Tau Pathology in Normal Older Adults. Neurology. 2021 Apr 6;96(14):e1844\u0026ndash;54.\u003c/li\u003e\n\u003cli\u003eBelloy ME, Andrews SJ, Le Guen Y, Cuccaro M, Farrer LA, Napolioni V, et al. APOE Genotype and Alzheimer Disease Risk Across Age, Sex, and Population Ancestry. JAMA Neurology [Internet]. 2023 Nov 6 [cited 2023 Dec 7]; Available from: https://doi.org/10.1001/jamaneurol.2023.3599\u003c/li\u003e\n\u003cli\u003eWolk DA, Dickerson BC, the Alzheimer\u0026rsquo;s Disease Neuroimaging Initiative, Weiner M, Aiello M, Aisen P, et al. Apolipoprotein E (APOE) genotype has dissociable effects on memory and attentional-executive network function in Alzheimer\u0026rsquo;s disease. Proceedings of the National Academy of Sciences. 2010 Jun 1;107(22):10256\u0026ndash;61.\u003c/li\u003e\n\u003cli\u003eMonane M, Johnson KG, Snider BJ, Turner RS, Drake JD, Maraganore DM, et al. A blood biomarker test for brain amyloid impacts the clinical evaluation of cognitive impairment. Ann Clin Transl Neurol. 2023 Oct;10(10):1738\u0026ndash;48.\u003c/li\u003e\n\u003cli\u003ePalmqvist S, Janelidze S, Quiroz YT, Zetterberg H, Lopera F, Stomrud E, et al. Discriminative Accuracy of Plasma Phospho-tau217 for Alzheimer Disease vs Other Neurodegenerative Disorders. JAMA. 2020 Aug 25;324(8):772\u0026ndash;81.\u003c/li\u003e\n\u003cli\u003eKarikari TK, Pascoal TA, Ashton NJ, Janelidze S, Benedet AL, Rodriguez JL, et al. Blood phosphorylated tau 181 as a biomarker for Alzheimer\u0026rsquo;s disease: a diagnostic performance and prediction modelling study using data from four prospective cohorts. The Lancet Neurology. 2020;19(5):422\u0026ndash;33.\u003c/li\u003e\n\u003cli\u003eClark CM, Schneider JA, Bedell BJ, Beach TG, Bilker WB, Mintun MA, et al. Use of florbetapir-PET for imaging \u0026beta;-amyloid pathology. Jama. 2011;305(3):275\u0026ndash;83.\u003c/li\u003e\n\u003cli\u003eWolz R, Aljabar P, Hajnal JV, Hammers A, Rueckert D, Alzheimer\u0026rsquo;s Disease Neuroimaging Initiative. LEAP: learning embeddings for atlas propagation. Neuroimage. 2010 Jan 15;49(2):1316\u0026ndash;25.\u003c/li\u003e\n\u003cli\u003eGrecchi E, Foley C, Gispert JD, Wolz R. P3‐434: CENTILOID PET SUVR ANALYSIS USING THE SUPRATENTORIAL WHITE MATTER AS REFERENCE REGION. Alzheimer\u0026rsquo;s \u0026amp;amp; Dementia [Internet]. 2018 Jul [cited 2023 Aug 26];14(7S_Part_24). Available from: https://alz-journals.onlinelibrary.wiley.com/doi/10.1016/j.jalz.2018.06.1797\u003c/li\u003e\n\u003cli\u003eLiu X. Classification accuracy and cut point\u0026thinsp;selection. Stat Med. 2012 Oct 15;31(23):2676\u0026ndash;86.\u003c/li\u003e\n\u003cli\u003ePedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, et al. Scikit-learn: Machine Learning in Python. Journal of Machine Learning Research. 2011;12:2825\u0026ndash;30.\u003c/li\u003e\n\u003cli\u003eHansson O, Edelmayer RM, Boxer AL, Carrillo MC, Mielke MM, Rabinovici GD, et al. The Alzheimer\u0026rsquo;s Association appropriate use recommendations for blood biomarkers in Alzheimer\u0026rsquo;s disease. Alzheimer\u0026rsquo;s \u0026amp; Dementia. 2022;18(12):2669\u0026ndash;86.\u003c/li\u003e\n\u003cli\u003eTherriault J, Janelidze S, Benedet AL, Ashton NJ, Arranz Mart\u0026iacute;nez J, Gonzalez-Escalante A, et al. Diagnosis of Alzheimer\u0026rsquo;s disease using plasma biomarkers adjusted to clinical probability. Nat Aging. 2024 Nov;4(11):1529\u0026ndash;37.\u003c/li\u003e\n\u003cli\u003eUdeh‐Momoh CT, Mielke MM, Schindler SE, Hansson O, Khachaturian AS, Weiss J. Bridging the Gap: The Global CEOi Collaborative Workgroup for Adoption of Alzheimer\u0026rsquo;s disease Blood‐Based Biomarkers in Clinical Practice. Alzheimer\u0026rsquo;s \u0026amp; Dementia. 2023 Dec;19(S24):e082822.\u003c/li\u003e\n\u003cli\u003eSchindler SE, Galasko D, Pereira AC, Rabinovici GD, Salloway S, Su\u0026aacute;rez-Calvet M, et al. Acceptable performance of blood biomarker tests of amyloid pathology \u0026mdash; recommendations from the Global CEO Initiative on Alzheimer\u0026rsquo;s Disease. Nat Rev Neurol. 2024 Jun 12;1\u0026ndash;14.\u003c/li\u003e\n\u003cli\u003eDiaz-Galvan P, Przybelski SA, Lesnick TG, Schwarz CG, Senjem ML, Gunter JL, et al. \u0026beta;-Amyloid Load on PET Along the Continuum of Dementia With Lewy Bodies. Neurology. 2023 Jul 11;101(2):e178\u0026ndash;88.\u003c/li\u003e\n\u003cli\u003eTan RH, Kril JJ, Yang Y, Tom N, Hodges JR, Villemagne VL, et al. Assessment of amyloid \u0026beta; in pathologically confirmed frontotemporal dementia syndromes. Alzheimers Dement (Amst). 2017 May 29;9:10\u0026ndash;20.\u003c/li\u003e\n\u003cli\u003eGurol ME, Becker JA, Fotiadis P, Riley G, Schwab K, Johnson KA, et al. Florbetapir-PET to diagnose cerebral amyloid angiopathy: A prospective study. Neurology. 2016 Nov 8;87(19):2043\u0026ndash;9.\u003c/li\u003e\n\u003cli\u003eHaller S, Montandon ML, Lilja J, Rodriguez C, Garibotto V, Herrmann FR, et al. PET amyloid in normal aging: direct comparison of visual and automatic processing methods. Sci Rep. 2020 Oct 7;10(1):16665.\u003c/li\u003e\n\u003cli\u003eSantaCruz KS, Sonnen JA, Pezhouh MK, Desrosiers MF, Nelson PT, Tyas SL. Alzheimer disease pathology in subjects without dementia in 2 studies of aging: the Nun Study and the Adult Changes in Thought Study. J Neuropathol Exp Neurol. 2011 Oct;70(10):832\u0026ndash;40.\u003c/li\u003e\n\u003cli\u003eJack CR, Andrews JS, Beach TG, Buracchio T, Dunn B, Graf A, et al. Revised criteria for diagnosis and staging of Alzheimer\u0026rsquo;s disease: Alzheimer\u0026rsquo;s Association Workgroup. Alzheimers Dement. 2024 Aug;20(8):5143\u0026ndash;69.\u003c/li\u003e\n\u003cli\u003eDubois B, Villain N, Schneider L, Fox N, Campbell N, Galasko D, et al. Alzheimer Disease as a Clinical-Biological Construct\u0026mdash;An International Working Group Recommendation. JAMA Neurol. 2024 Dec 1;81(12):1304.\u003c/li\u003e\n\u003cli\u003eFrisoni GB, Festari C, Massa F, Ramusino MC, Orini S, Aarsland D, et al. European intersocietal recommendations for the biomarker-based diagnosis of neurocognitive disorders. The Lancet Neurology. 2024 Mar 1;23(3):302\u0026ndash;12.\u003c/li\u003e\n\u003cli\u003eCiesla M, Toro-Serey C, Jannati A, Banks RE, Gomes-Osman J, Showalter J, et al. Detecting functional impairment with the Digital Clock and Recall. Journal of Alzheimer\u0026rsquo;s Disease. 2024;102(2):329\u0026ndash;37.\u003c/li\u003e\n\u003cli\u003eFowler NR, Hammers DB, Perkins AJ, Summanwar D, Higbie A, Swartzell K, et al. Feasibility and Acceptability of Implementing a Digital Cognitive Assessment for Alzheimer Disease and Related Dementias in Primary Care. Ann Fam Med. 2025 Apr 29;240293.\u003c/li\u003e\n\u003cli\u003eSummanwar D, Fowler NR, Hammers DB, Perkins AJ, Brosch JR, Willis DR. Agile Implementation of a Digital Cognitive Assessment for Dementia in Primary Care. Ann Fam Med. 2025 Apr 29;240294.\u003c/li\u003e\n\u003cli\u003eCummings J, Apostolova L, Rabinovici GD, Atri A, Aisen P, Greenberg S, et al. Lecanemab: Appropriate Use Recommendations. J Prev Alz Dis [Internet]. 2023 [cited 2023 Jul 25]; Available from: https://link.springer.com/article/10.14283/jpad.2023.30\u003c/li\u003e\n\u003cli\u003ePittock RR, Aakre J, Castillo AM, Ramanan VK, Kremers WK, Jack CR, et al. Eligibility for Anti-Amyloid Treatment in a Population-Based Study of Cognitive Aging. Neurology [Internet]. 2023 Aug 16 [cited 2023 Aug 23]; Available from: https://www.neurology.org/lookup/doi/10.1212/WNL.0000000000207770\u003c/li\u003e\n\u003cli\u003eBrum WS, Cullen NC, Janelidze S, Ashton NJ, Zimmer ER, Therriault J, et al. A two-step workflow based on plasma p-tau217 to screen for amyloid \u0026beta; positivity with further confirmatory testing only in uncertain cases. Nat Aging. 2023 Aug 31;3(9):1079\u0026ndash;90.\u003c/li\u003e\n\u003cli\u003eManjavong M, Kang JM, Diaz A, Ashford MT, Eichenbaum J, Aaronson A, et al. Performance of Plasma Biomarkers Combined with Structural MRI to Identify Candidate Participants for Alzheimer\u0026rsquo;s Disease-Modifying Therapy. J Prev Alzheimers Dis. 2024;11(5):1198\u0026ndash;205.\u003c/li\u003e\n\u003cli\u003eCano A, Capdevila M, Puerta R, Arranz J, Montrreal L, de Rojas I, et al. Clinical value of plasma pTau181 to predict Alzheimer\u0026rsquo;s disease pathology in a large real-world cohort of a memory clinic. EBioMedicine. 2024 Oct;108:105345.\u003c/li\u003e\n\u003cli\u003eArranz J, Zhu N, Rubio-Guerra S, Rodr\u0026iacute;guez-Baz \u0026Iacute;, Ferrer R, Carmona-Iragui M, et al. Diagnostic performance of plasma pTau217, pTau181, A\u0026beta;1-42 and A\u0026beta;1-40 in the LUMIPULSE automated platform for the detection of Alzheimer disease. Alzheimers Res Ther. 2024 Jun 26;16(1):139.\u003c/li\u003e\n\u003cli\u003eGarcia-Escobar G, Manero RM, Fern\u0026aacute;ndez-Lebrero A, Ois A, Navalpotro-G\u0026oacute;mez I, Puente-Periz V, et al. Blood Biomarkers of Alzheimer\u0026rsquo;s Disease and Cognition: A Literature Review. Biomolecules. 2024 Jan 11;14(1):93.\u003c/li\u003e\n\u003cli\u003eCummings JL, Goldman DP, Simmons‐Stern NR, Ponton E. The costs of developing treatments for Alzheimer\u0026rsquo;s disease: A retrospective exploration. Alzheimer\u0026rsquo;s \u0026amp; Dementia. 2022 Mar;18(3):469\u0026ndash;77.\u003c/li\u003e\n\u003cli\u003eCummings J, Lee G, Nahed P, Kambar MEZN, Zhong K, Fonseca J, et al. Alzheimer\u0026rsquo;s disease drug development pipeline: 2022. Alzheimer\u0026rsquo;s \u0026amp; Dementia: Translational Research \u0026amp; Clinical Interventions. 2022;8(1):e12295.\u003c/li\u003e\n\u003cli\u003eBucci M, Chiotis K, Nordberg A, Alzheimer\u0026rsquo;s Disease Neuroimaging Initiative. Alzheimer\u0026rsquo;s disease profiled by fluid and imaging markers: tau PET best predicts cognitive decline. Molecular Psychiatry. 2021;26(10):5888\u0026ndash;98.\u003c/li\u003e\n\u003cli\u003eThijssen EH, La Joie R, Wolf A, Strom A, Wang P, Iaccarino L, et al. Diagnostic value of plasma phosphorylated tau181 in Alzheimer\u0026rsquo;s disease and frontotemporal lobar degeneration. Nat Med. 2020 Mar;26(3):387\u0026ndash;97.\u003c/li\u003e\n\u003cli\u003eLeuzy A, Mattsson‐Carlgren N, Palmqvist S, Janelidze S, Dage JL, Hansson O. Blood‐based biomarkers for Alzheimer\u0026rsquo;s disease. EMBO Mol Med. 2022 Jan 11;14(1):e14408.\u003c/li\u003e\n\u003cli\u003eMattsson-Carlgren N, Palmqvist S. The emerging era of staging Alzheimer\u0026rsquo;s disease pathology using plasma biomarkers. Brain. 2023 May 2;146(5):1740\u0026ndash;2.\u003c/li\u003e\n\u003cli\u003eLi RX, Ma YH, Tan L, Yu JT. Prospective biomarkers of Alzheimer\u0026rsquo;s disease: A systematic review and meta-analysis. Ageing Research Reviews. 2022 Nov 1;81:101699.\u003c/li\u003e\n\u003cli\u003ePalmqvist S, Stomrud E, Cullen N, Janelidze S, Manuilova E, Jethwa A, et al. An accurate fully automated panel of plasma biomarkers for Alzheimer\u0026rsquo;s disease. Alzheimer\u0026rsquo;s \u0026amp; Dementia. 2023 Apr;19(4):1204\u0026ndash;15.\u003c/li\u003e\n\u003cli\u003eAlcolea D, Beeri MS, Rojas JC, Gardner RC, Lle\u0026oacute; A. Blood Biomarkers in Neurodegenerative Diseases: Implications for the Clinical Neurologist. Neurology. 2023 Jul 25;101(4):172\u0026ndash;80.\u003c/li\u003e\n\u003cli\u003ePalmqvist S, Tideman P, Cullen N, Zetterberg H, Blennow K, the Alzheimer\u0026rsquo;s Disease Neuroimaging Initiative, et al. Prediction of future Alzheimer\u0026rsquo;s disease dementia using plasma phospho-tau combined with other accessible measures. Nat Med. 2021 Jun;27(6):1034\u0026ndash;42.\u003c/li\u003e\n\u003cli\u003eGlobal Alzheimer\u0026rsquo;s Platform. Bio-Hermes study [Internet]. [cited 2022 May 11]. Available from: https://globalalzplatform.org/biohermesstudy/\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"alzheimers-research-and-therapy","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"azrt","sideBox":"Learn more about [Alzheimer's Research and Therapy](http://alzres.biomedcentral.com/)","snPcode":"13195","submissionUrl":"https://submission.nature.com/new-submission/13195/3","title":"Alzheimer's Research \u0026 Therapy","twitterHandle":"@AlzheimersRes","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Alzheimer’s disease, mild cognitive impairment, dementia, digital cognitive assessment","lastPublishedDoi":"10.21203/rs.3.rs-6768373/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6768373/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground: \u003c/strong\u003eEarly identification of cognitive impairment \u003cem\u003eand\u003c/em\u003ebrain pathology associated with Alzheimer’s disease (AD) is essential to maximize benefits from lifestyle interventions and emerging pharmacologic disease-modifying treatments (DMT). Digital cognitive assessments (DCAs) can quickly capture an array of metrics that can be used to train machine learning models to concurrently evaluate different outcomes. DCAs have the potential to optimize clinical workflows and enable efficient assessment of cognitive function and the likelihood of a given underlying pathology.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods: \u003c/strong\u003eWe assessed the ability of a next-generation DCA, the Digital Clock and Recall (DCR), to concurrently estimate brain amyloid-beta (Aβ) status and detect cognitive impairment, as compared with traditional cognitive assessments, including the MMSE, RAVLT, a DCA, Cognivue®, and blood-based biomarkers in 930 participants from the Bio-Hermes-001 clinical study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults: \u003c/strong\u003eAβ42/40, pTau-181, and pTau-217 poorly classified cognitive impairment (AUCs: 0.63; 0.66; 0.72, respectively), but accurately classified Aβ status (AUCs: 0.81; 0.78; 0.89, respectively). MMSE, RAVLT, and Cognivue poorly classified Aβ status (AUCs: 0.71, 0.72, 0.70, respectively). However, separate multimodal, DCR-based machine-learning classification models, run in parallel, accurately classified both cognitive impairment (AUC=0.85) \u003cem\u003eand\u003c/em\u003e Aβ status (AUC=0.83).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions: \u003c/strong\u003eDCAs that leverage digital technologies to generate advanced metrics, such as the DCR, enable accurate and efficient detection of cognitive impairment associated with AD pathology. They have the potential to empower health systems and primary care providers to help their patients make timely treatment decisions.\u003c/p\u003e","manuscriptTitle":"Concurrent Detection of Cognitive Impairment and Amyloid Positivity with a Next-Generation Digital Cognitive Assessment","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-05-30 16:24:03","doi":"10.21203/rs.3.rs-6768373/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-07-19T07:42:12+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-07-02T23:39:23+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-06-24T07:24:33+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"303933633910025008953892956300310763554","date":"2025-06-18T20:22:12+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"46747531993330519201218660599435123140","date":"2025-06-12T06:59:32+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-06-11T19:06:57+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-05-29T05:02:37+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-05-29T05:02:13+00:00","index":"","fulltext":""},{"type":"submitted","content":"Alzheimer's Research \u0026 Therapy","date":"2025-05-28T12:58:12+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"alzheimers-research-and-therapy","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"azrt","sideBox":"Learn more about [Alzheimer's Research and Therapy](http://alzres.biomedcentral.com/)","snPcode":"13195","submissionUrl":"https://submission.nature.com/new-submission/13195/3","title":"Alzheimer's Research \u0026 Therapy","twitterHandle":"@AlzheimersRes","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"0562201e-93da-4786-bcb4-eb0c20b4826d","owner":[],"postedDate":"May 30th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-12-15T16:05:50+00:00","versionOfRecord":{"articleIdentity":"rs-6768373","link":"https://doi.org/10.1186/s13195-025-01913-5","journal":{"identity":"alzheimers-research-and-therapy","isVorOnly":false,"title":"Alzheimer's Research \u0026 Therapy"},"publishedOn":"2025-12-11 15:57:42","publishedOnDateReadable":"December 11th, 2025"},"versionCreatedAt":"2025-05-30 16:24:03","video":"","vorDoi":"10.1186/s13195-025-01913-5","vorDoiUrl":"https://doi.org/10.1186/s13195-025-01913-5","workflowStages":[]},"version":"v1","identity":"rs-6768373","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6768373","identity":"rs-6768373","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}