Predicting Cognitive Decline in Early Alzheimer’s: Insights from East Asian Cohorts

Predicting Cognitive Decline in Early Alzheimer’s: Insights from East Asian Cohorts | 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 Predicting Cognitive Decline in Early Alzheimer’s: Insights from East Asian Cohorts Yao-Hwei Fang, Yung-Shuan Lin, Kazuaki Uchida, Wei-Ju Lee, Chih-Cheng Hsu, and 11 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8629600/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background While cognitive decline is inevitable in early Alzheimer’s disease (AD) and amnestic mild cognitive impairment (aMCI), the rate of decline varies widely. Aims This study aimed to identify groups of individuals following similar trajectories of cognitive decline, and develop a model predicting those at risk of rapid decline. Methods A longitudinal study of 251 clinic-based early AD/aMCI converters from Taiwan, with external validation in an independent Japanese cohort. Group-based trajectory modeling identified cognitive decline patterns using Mini-Mental State Examination (MMSE) scores. Baseline assessments included demographics, lifestyle factors, comorbidities, functional capacities, neuropsychiatric symptoms, neuropsychological tests, and clinical characteristics. Backward logistic regression was used to identify predictors of rapid cognitive decline. Results Two cognitive trajectories were identified: a rapid decline group (two-year MMSE decline of 5.8 ± 5.1 points) and a slower decline group (1.4 ± 2.6 points). Baseline MMSE score, instrumental activities of daily living (IADL) score, and apolipoprotein E ( APOE ) ε4 allele count were significant predictors of rapid cognitive decline. The prediction model demonstrated good discrimination in the Taiwanese cohort (The area under the curve (AUC) = 0.841, 95% confidence intervals (CI): 0.771–0.892) with sensitivity of 0.769 and specificity of 0.728, and showed acceptable discrimination (AUC = 0.715, 95% CI: 0.649–0.780) with good recalibration in the Japanese validation cohort. Conclusions This is the first clinic-based prediction model for cognitive decline in East Asian early AD/aMCI with external validation. Baseline MMSE, IADL, and APOE genotype may assist clinicians in risk stratification, disease monitoring, and individualized care planning in aging populations. Amnestic mild cognitive impairment Cognitive function Early stage of Alzheimer's disease Risk prediction model Trajectory analysis Figures Figure 1 Figure 2 Background Cognitive decline in Alzheimer’s disease (AD) and amnestic mild cognitive impairment (aMCI) is highly variable across individuals [ 1 ]. Early AD and aMCI converters (those converting from aMCI to AD) are interconnected within the AD continuum. A key concern for families and physicians is predicting the rate of cognitive decline. Meta-analytic data from clinic-based cohorts suggested an average Mini-Mental State Examination (MMSE) decline of ≈ 3.3 points per year [ 2 ], whereas community studies report almost half that rate (≈ 1.5 points) [ 3 ]. This discrepancy likely reflects differences in sampling frame–clinic samples over-represent patients with more advanced disease or comorbidities–while trajectory-modeling studies that follow newly diagnosed AD cases show that most individuals decline at an even slower rate the averages reported in either clinic or community samples [ 3 , 4 ]. Such heterogeneity motivates the search for robust predictors of rapid deterioration. Existing evidence can be grouped into three inter-related domains. Demographic factors—earlier age at symptom onset, male sex, and lower educational attainment—have been linked to faster cognitive loss, although these associations often weaken after adjustment for baseline cognition, implying confounding by disease stage [ 5 – 7 ]. Clinical and genetic markers, including extrapyramidal signs, behavioral disturbances, and carriage of the apolipoprotein E ( APOE ) ε4 allele, predict steeper decline in some studies but not others, suggesting that sample age structure, follow-up duration, and diagnostic criteria condition their impact [ 8 – 11 ]. Evidence on vascular and lifestyle influences is mixed. Several studies report that hypertension, diabetes, smoking, or sedentary behaviors accelerate cognitive loss, whereas others find weak or null associations. Such heterogeneity likely reflects differences in how vascular risk is measured and treated, baseline disease severity, follow-up length, and analytic methods, rather than any single setting-related factor [ 12 , 13 ]. Identifying individuals at risk for rapid cognitive decline aids proactive interventions. We hypothesized that individuals with early AD or aMCI could be categorized into distinct groups by cognitive trajectories, with shared baseline characteristics influencing the rate of decline. This study aimed to identify latent cognitive decline groups based on MMSE scores using group-based trajectory modeling (GBTM) in a longitudinal clinical cohort of Han Chinese. We also developed a prediction model for rapid cognitive decline incorporating demographics, comorbidities, functional capacities, neuropsychological tests (NPT), neuropsychiatric symptoms (NPS), and APOE ε4 genotype. Extended follow-up data were used to test the hypothesis that distinct trajectories of cognitive decline exist. The developed prediction model was validated in an independent Japanese cohort from the National Center for Geriatrics and Gerontology (NCGG), making it the first clinic-based prediction model for cognitive decline in East Asian populations with external validation. Methods Study design and participants We recruited individuals with aMCI (n = 344) and AD (n = 975) from the outpatient clinics of Taipei Veterans General Hospital (TVGH) and Taichung Veterans General Hospital (TCVGH) as the development cohort. Supplementary Fig. 1 presents the patient enrollment flowchart. AD diagnoses followed the 2011 National Institute on Aging—Alzheimer’s Association clinical criteria for probable AD [ 14 ] while aMCI was diagnosed per the revised 2004 consensus criteria [ 15 ], defined as a cut-off of 1.5 standard deviations below the age-adjusted norm on the Wechsler Memory Scale III logical memory test [ 16 ]. Participants in the developmental cohort were aged 50–90 years and required a reliable caregiver with at least 10 hours of weekly contact. To focus on early AD and aMCI converters, we included individuals with a baseline MMSE score ≥ 20, corresponding approximately to a Clinical Dementia Rating (CDR) of ≤ 1 [ 17 ]. Exclusion criteria included fewer than two annual follow-ups, aMCI cases that did not convert to AD by the second-year follow-up, and the presence of other significant neurological condition affecting cognition. Exclusion conditions included acute confusion due to systemic disease, major depressive disorder (per Diagnostic and Statistical Manual of Mental Disorders 5th edition), probable vascular dementia, normal pressure hydrocephalus, progressive supranuclear palsy, and significant head trauma with persistent neurologic deficits, or known structural brain abnormalities. Ultimately, 251 out of the 1319 early AD and aMCI converters met the inclusion criteria. Baseline and annual data collection Cognitive state was assessed using the MMSE, and dementia severity via CDR [ 18 ]. Thirty candidate variables were selected based on prior research related to cognitive decline, including demographics, lifestyle factors, comorbidities, functional capacities, NPT, NPS, and clinical characteristics. Demographics included age, sex, education (years). Lifestyle factors included BMI, regular exercise frequency (days/week), and smoking status (≥ 5 cigarettes/day). Overall health status (comorbidity) was assessed by the total number of self-reported physician-diagnosed chronic diseases, including hypertension, diabetes, cardiovascular diseases, etc. (Supplementary Table 1). Functional capacities were assessed using the activity of daily living (ADL) [ 19 ] and instrumental ADL (IADL) scales [ 20 ], in which lower scores indicated better function. NPT included the 12-item memory recall test [ 21 ], 15-item modified Boston Naming Test (mBNT) [ 22 ], the verbal fluency test [ 23 ], and the forward and backward digit span task [ 24 ]. The instruments assessed short-term memory, language, executive function, attention and working memory, respectively. Depressive symptoms were assessed using the 15-item Geriatric Depression Scale (GDS-15) [ 25 ]. Higher NPT scores indicated better performance. NPS were evaluated with the Neuropsychiatric Inventory-Questionnaire (NPI-Q), which measures 12 behavioral and psychological symptoms of dementia, including depression, anxiety, apathy, sleep, appetite, agitation, irritability, disinhibition, elation, motor disturbance, delusions, and hallucinations [ 26 ]. Higher NPI-Q scores indicated worse NPS. Clinical characteristics included baseline diagnosis, CDR, disease duration, and APOE ε4 allele count. Disease duration referred to the time from the initial cognitive dysfunction to study enrollment. APOE ε4 allele count (0, 1, or 2 alleles) was included as a continuous covariate. At each annual follow-up, MMSE and CDR scores were reassessed and profiling of health status and GDS-15 were also updated, because the trajectories of multimorbidity and depression affect cognitive decline patterns [ 27 ]. Statistical analysis The statistical workflow (Supplementary Fig. 2) comprised three steps: (1) Trajectory identification, (2) Feature selection, and (3) Model building and evaluation. For trajectory identification, cognitive trajectories were determined using MMSE scores as a continuous measure of cognitive function [ 28 ]. GBTM was applied to classify early AD/aMCI subjects into distinct patterns of cognitive decline over two years. Potential confounders, identified from the existing literature [ 29 – 31 ], were adjusted in the model. These included time-fixed covariates (baseline age, sex, years of education, and APOE ε4 allele count) and time-varying covariates (health status and depressive symptoms at each follow-up). See Supplementary Methods “Trajectory Identification” for more details. Feature selection followed a two-step process [ 32 ]. First, univariable scanning using Chi-square tests for categorical variables and t-tests for continuous variables that retained with P < 0.002 (Bonferroni corrected). This was followed by backward logistic regression (LR) with Bayesian Information Criterion (BIC) to eliminate redundant variables affected by multicollinearity [ 33 ]. The refined model was to predict cognitive decline trajectories in early AD/aMCI subjects. Model performance was assessed using 10-fold cross-validation, with the area under the curve (AUC) reported alongside 95% confidence intervals (CI) and standard error [ 34 ]. The optimal classification threshold was determined using Bayes Minimum Risk theory, and predictive performance was evaluated based on accuracy, sensitivity, and specificity [ 35 ]. To enhance clinical applicability, we developed a nomogram, assigning predictor-based scores to estimate an individual’s risk of rapid cognitive decline. All analyses were conducted using SAS (Proc Traj) [ 36 ], and R packages, including cvAUC, pROC, and ggplot2 for trajectory modeling, predictive performance assessment, and data visualization [ 37 ]. The validation study The validation cohort comprised 322 early AD subjects from the Memory Clinic at the NCGG in Japan. Detailed information on the validation cohort is provided in Supplementary Methods. Participants aged ≥ 65 with an AD diagnosis at baseline and at least three clinic visits were included. To validate the prediction model for cognitive progression, we determined the optimal threshold for identifying rapid cognitive decline, a critical factor for both model performance and clinical applicability (see Supplementary Methods “threshold determination” for more details). LR coefficients derived from the development cohort were applied to predict the risk of rapid cognitive decline in the validation cohort. Model performance was evaluated using AUC and a calibration plot. Results Demographics, lifestyle, and clinical characteristics In development cohort, 251 participants with a baseline MMSE score ≥ 20 were included in the analysis, consisting of 209 early AD subjects (116 males, 93 females) and 42 aMCI converters (19 males, 23 females). Among AD subjects, 22.5% (47/209) had a baseline CDR of 0.5, while all aMCI subjects had a CDR of 0.5 (Table 1). AD subjects were older than aMCI subjects (79.1 ± 6.8 vs. 74.2 ± 8.4 years, P < 0.001). Baseline MMSE were lower in AD subjects than in aMCI subjects (22.9 ± 2.0 vs. 25.3 ± 2.4, P < 0.001). No significant differences were found in sex, education, health status, disease duration, or APOE ε4 allele count between AD/aMCI subjects. Trajectory identification Model selection for GBTM (Supplementary Fig. 3) identified a two-trajectory model as the best-fitting solution. Both trajectories exhibited linear cognitive decline over time (Supplementary Table 2). Fig. 1(a) displays individual trajectories classified into two groups. Group 1 (G1), consisting of 94 subjects (37.5%) with a mean annual MMSE decline of 2.6 ± 1.3 points, exhibited faster cognitive decline (“fast-decliners”). In contrast, Group 2 (G2), comprising 157 subjects (62.5%) with a mean annual MMSE decline of 0.7 ± 1.0 points, exhibited slower decline (“slow-decliners”). Supplementary Table 3 shows the estimated coefficients of the trajectory model in the development cohort. APOE ε4 allele count predicted cognitive trajectories in early AD/aMCI subjects, while age, sex, education, and health status were not significant predictors. Feature selection Table 2 compares baseline variables between the cognitive trajectory groups. The fast-decliner group (G1) had a significantly higher proportion of early AD cases (93.6%) than the slow-decline group (G2) (77.1%). Age, sex, education, BMI, exercise habits, smoking, hypertension, diabetes, cardiovascular disease, health status, disease duration, depressive symptoms, and NPS (except hallucinations) did not differ between the two groups. However, APOE ε4 allele count, CDR, NPT (MMSE, recall test, verbal fluency, and 15-item mBNT), functional capacities (ADL and IADL scores), and hallucinations (NPS) were significant (P < 0.002, accounting for multiple comparisons) in the univariable scanning. After applying the backward LR method (Table 3, Model 2), only baseline MMSE score, IADL score, and APOE ε4 allele count remained as significant predictors of cognitive decline in early AD/aMCI convertors. The results showed that more APOE ε4 alleles, higher IADL and lower MMSE scores were significantly associated with the fast-decline trajectory. Correlations among the three predictors were low (0.09–0.27) and collectively explained 43.0% of the variability in cognitive decline trajectories [38]. Additionally, redundant variables, including recall, verbal fluency, 15-item mBNT, ADL score, and hallucinations, were not statistically significant in the backward LR method. Hallucinations showed a strong effect (odds ratio = 2.41) but weak significance (P = 0.096), possibly due to the small sample size (Table 3, Model 1). In Model 3 (Table 3), age, sex, and education were included as covariates for further adjustment. Sensitivity analysis using Model 3 confirmed consistent results. A scoring chart and nomogram were created using Model 3 regression coefficients (Fig. 2). Model building and evaluation The optimal classification threshold was set based on the fast-decliner group’s prior probability (τ = 37.5%) [35]. The prediction model (Model 3) achieved an AUC of 0.841, sensitivity of 0.769, and specificity of 0.728 (95% CI: 0.779–0.950, 0.657–0.886, and 0.621–0.845, respectively). The LR model was further compared with machine learning methods, including support vector machine and random forest [39, 40]. Results showed that the LR model achieved a higher AUC in predicting rapid progression of early AD/aMCI (Supplementary Table 4). Validation study analysis The validation cohort consisted of 322 Japanese early AD subjects. Baseline demographics (age, sex, and education), APOE ε4 allele count, IADL, and MMSE score were included (Table 1). Compared with the Taiwanese development cohort, Japanese participants were younger, had a lower proportion of men and had lower MMSE scores and BMI. Diabetes and cardiovascular disease were more prevalent in the Taiwanese group. No significant differences were observed between the cohorts in GDS-15 scores, years of education, hypertension, or APOE ε4 allele count. Data harmonization ensured consistency between cohorts, except for IADL (details in Supplementary Methods). Supplementary Fig. 4 shows an optimal threshold probability of 0.3 for defining rapid cognitive decline (χ 2 = 62.7, P = 1.66×10 -14 ). This threshold was comparable to the development cohort’s classification using Bayes Minimum Risk theory (37.5% fast-decliners, 62.5% slow-decliners). Figure 1(b) illustrates cognitive trajectories of the validation cohort. Baseline mean MMSE scores were lower among slow-decliners in the validation cohort (22.6 ± 2.1) compared to those in the development cohort (24.2 ± 2.0), likely due to differences in inclusion criteria. The validation cohort included only AD participants aged 65 and older whereas the development cohort also included aMCI cases. In both cohorts, fast-decliners demonstrated a greater mean annual MMSE decline compared to slow-decliners. Model 3’s calibration was assessed (Supplementary Fig. 5(a)), showing a calibration slope of 0.52 (95% CI: 0.35–0.70) and an intercept of -0.21 (95% CI: -0.58–0.17), indicating poor alignment. Mis-calibration likely resulted from cohort differences in subject characteristics, disease prevalence, and treatment policies. Therefore, model recalibration was necessary. After logistic recalibration [41], model performance improved (Supplementary Fig. 5 (b)). The recalibrated model yielded an AUC of 0.715 (95% CI: 0.649–0.780) for distinguishing fast-decliners from slow-decliners. Subsequent follow-up for cognitive decline We analyzed follow-up data to determine whether slow-decliners maintained slower MMSE decline rates beyond 2 years. Of the 251 early AD/aMCI participants, 173 (68.9%) completed three-year follow-up, 124 (49.4%) four-year, 81 (32.3%) five-year follow-up, and 28 (11.2%) six-year, respectively. Supplementary Table 5 shows that slow-decliners exhibited a slower MMSE decline rate (1.6 points/year between the second and fifth follow-up) compared to fast-decliners (1.9 points/year). Disease duration did not differ significantly between groups (3.7 years for fast-decliners vs. 3.3 years for slow-decliners) (Table 2). Discussion This study assessed cognitive trajectories in 251 early AD/aMCI convertors using GBTM based on MMSE scores. Two distinct cognitive trajectories were identified: 37.5% were fast-decliners, while 62.5% were slow-decliners. This finding is consistent with previous research showing more gradual cognitive decline in clinical settings[ 3 , 4 ]. Our study advances prior work by integrating clinical measures and genetic predictors into a prediction model for rapid cognitive decline in early AD/aMCI. This model was externally validated in a Japanese cohort, addressing limitations of earlier models lacking external validation or genetic data. Fast decliners had more APOE ε4 alleles, higher baseline CDR, worse cognitive performance, and more severe hallucinations than slow-decliners. These findings align with studies linking worse initial NPT scores, APOE ε4 alleles [ 11 ], impaired IADL [ 42 ], and NPS like hallucinations to faster decline [ 43 ]. Despite several significant predictors in univariable screening, only baseline MMSE, IADL, and APOE ε4 allele count remained significant in the final model. Other variables (Table 2 ) were excluded as they did not improve model fit based on BIC. MMSE, the primary predictor, already assesses key cognitive domains (memory, executive function, language, attention, and visuospatial skills). As a result, recall, verbal fluency, and 15-item mBNT were considered redundant. This overlap was evident in Pearson correlations with MMSE (recall: 0.43, verbal fluency: 0.30, 15-item mBNT: 0.24, all P < 0.001). Our study found that neither comorbidities nor years of education influenced the rate of cognitive decline (Table 2 ). This contrasts with studies linking worse medical conditions and vascular risk factors to faster decline [ 12 ]. The impact of education on cognitive decline rate remains debated likely due to heterogeneous design and variations in patient populations [ 3 , 4 , 7 ]. Additionally, we found age showed no association with cognitive decline, aligning with recent evidence that late-life cognitive loss reflects pathological rather than normative aging [ 44 ]. MMSE is central to cognitive assessment, yet in aMCI stage, mild functional impairments emerge [ 45 , 46 ]. IADL decline often precedes cognitive deterioration, particularly in those receiving cholinesterase inhibitor with MMSE scores of 20–26 [ 42 ]. Worse initial IADL may indicate reduced awareness of cognitive decline, delaying interventions or lifestyle modifications. This may explain why worse initial IADL were associated with faster decline in our study. Our findings highlight the need for routine functional assessments in early AD/aMCI. APOE ε4 genotyping provides clinical insights, having been recognized as a major AD risk gene for over two decades. While APOE ε4 increases AD risk and results in earlier disease onset [ 47 ], its impact on cognitive decline remains inconsistent in the literature. Some studies link APOE ε4 to faster decline [ 48 , 49 ], whereas others report no effect [ 50 ]. These discrepancies may result from variations in baseline conditions and follow-up periods. Predicting cognitive trajectories, particularly diagnostic conversion from healthy to MCI or AD, remains challenging due to variations in study designs and data types [ 29 ]. Wu et al. used latent class growth analysis to identify cognitive decline trajectories in a Han Chinese population but achieved only modest performance (AUC = 0.66, sensitivity = 0.73, specificity = 0.58), with limited generalizability due to the lack of independent validation. In contrast, Bhagwat et al. used ML with imaging data in Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset, achieving higher performance (AUC = 0.97). While our model performed lower than Bhagwat et al., incorporating imaging could improve accuracy but is constrained by cost and feasibility in clinical application. Our extended follow-up data supports the hypothesis that fast-decliners and slow-decliners represent distinct trajectory groups, rather than different disease stages (Supplementary Fig. 6). Viewing slow-decliners as an earlier stage of the fast-decliner trajectory does not fully explain the persistent difference in MMSE decline rates. Additionally, Supplementary Fig. 6 clarifies why lower baseline MMSE scores were associated with more rapid cognitive decline. This study has three key strengths. First, it is the first clinic-based prediction model for cognitive decline in East Asian individuals with early AD/aMCI, featuring external validation and addressing the limitations of previous models that lacked validation. Second, to reduce heterogeneity in the aMCI group, we included only individuals who converted to AD within two years. Additionally, given that cognitive decline in AD is slower in the early stages [ 5 ], we restricted the development cohort to early AD/aMCI cases with MMSE ≥ 20 to minimize variability. Third, extended follow-up data enhance the reliability of findings. Our results support the hypothesis that fast- and slow-decliners represent distinct trajectory groups rather than different disease stages. However, two limitations existed. First, generalizability may be limited due to the clinic-based cohort and relatively small number of aMCI participants. Second, while external validation was performed, the lack of chronic disease health assessments in the Japanese cohort prevented full validation of the trajectory model, potentially affecting broader applicability. Conclusion This study presents a clinically applicable tool with good accuracy for predicting rapid cognitive decline in individuals with early AD/aMCI. MMSE, IADL, and APOE ε4 genotype emerged as key predictors, all readily available in dementia clinics. Developed and validated in two independent East Asian cohorts, this model provides valuable insights into future cognitive progression, supporting proactive healthcare planning for clinicians and patients. Declarations Ethics approval This study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Boards of TVGH and TCVGH. Validation data from the NCGG Memory Clinic were used with approval from the NCGG Ethics Committee (Approval No. 1611). Informed consent Written informed consent was obtained from all participants, and their anonymity was maintained according to the approved protocols. Competing interests The authors declare no conflict of interest. Consent for publication All authors read and approved the manuscript for submission and gave consent for publication. Availability of data and materials The datasets analyzed and R code are available from the corresponding author upon reasonable request. Funding sources The work was supported by grants from Academia Sinica of Taiwan (AS-BD-108-2; PH-111-GP-08, AS-KPQ-111-KNT); the Ministry of Science and Technology of Taiwan (112-2314-B-075-036-MY2, 112-2634-F-A49-003-,112-2321-B-A49-021-, and 112-2321-B-001-008-); Taipei Veterans General Hospital (V113C-047); Taichung Veterans General Hospital (TCVGH-1043402C, TCVGH-1053403C, and TCVGH-1063403C); The Brain Research Center, National Yang Ming University, from the Featured Areas Research Center Program within the framework of the Higher Education Sprout Project by the Ministry of Education in Taiwan; the National Health Research Institutes of Taiwan (PH-108-SP-01, PH-109-SP-01, PH-110-SP-01, PH-111-SP-01, and PH-112-SP-01); the Ministry of Science and Technology of Taiwan (108-2314-B-400-038, 109- 2740-B-400-002, 109-2314-B-400-014, 110-2740-B-400-002, 110-2314-B- 400-009, 111-2740-B-400-002, NSTC 112-2740-B-400-005, and NSTC 113-2740-B-400-005); the Research Funding for Longevity Sciences from the National Center for Geriatrics and Gerontology (22-2 and 22-23). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Authors' contributions SJ Wang, JL Fuh, CA Hsiung, IS Chang, and CC Hsu designed the study. SJ Wang, JL Fuh, WJ Lee, and YC Liao collected data in Taiwan. K Uchida, T Sugimoto, Y Kuroda, T Sakurai, and H Arai collected validation data in Japan. JL Fuh, YS Lin, and YH Fang searched the literature. YH Fang, TJ Hsieh, and TY Chen analyzed the data. JL Fuh, CA Hsiung, YS Lin, and YH Fang interpreted the data. YH Fang, YS Lin, JL Fuh, and CA Hsiung wrote, reviewed, and/or revised the manuscript. Acknowledgments The authors thank the Taiwan Biobank and Academia Sinica of Taiwan for their support for this work. References Livingston G, Sommerlad A, Orgeta V, Costafreda SG, Huntley J, Ames D, Ballard C, Banerjee S, Burns A, Cohen-Mansfield J, Cooper C, Fox N, Gitlin LN, Howard R, Kales HC, Larson EB, Ritchie K, Rockwood K, Sampson EL, Samus Q, Schneider LS, Selbaek G, Teri L, Mukadam N (2017) Dementia prevention, intervention, and care. Lancet 390:2673-2734. https://doi.org/10.1016/S0140-6736(17)31363-6 Han L, Cole M, Bellavance F, McCusker J, Primeau F (2000) Tracking cognitive decline in Alzheimer's disease using the mini-mental state examination: a meta-analysis. Int Psychogeriatr 12:231-247. https://doi.org/10.1017/s1041610200006359 Leoutsakos JM, Forrester SN, Corcoran CD, Norton MC, Rabins PV, Steinberg MI, Tschanz JT, Lyketsos CG (2015) Latent classes of course in Alzheimer's disease and predictors: the Cache County Dementia Progression Study. Int J Geriatr Psychiatry 30:824-832. https://doi.org/10.1002/gps.4221 Haaksma ML, Calderón-Larrañaga A, Olde Rikkert MGM, Melis RJF, Leoutsakos J-MS (2018) Cognitive and functional progression in Alzheimer disease: A prediction model of latent classes. International Journal of Geriatric Psychiatry 33:1057-1064. https://doi.org/10.1002/gps.4893 Xie SX, Ewbank DC, Chittams J, Karlawish JH, Arnold SE, Clark CM (2009) Rate of decline in Alzheimer disease measured by a Dementia Severity Rating Scale. Alzheimer Dis Assoc Disord 23:268-274. https://doi.org/10.1097/WAD.0b013e318194a324 Tschanz JT, Corcoran CD, Schwartz S, Treiber K, Green RC, Norton MC, Mielke MM, Piercy K, Steinberg M, Rabins PV, Leoutsakos J-M, Welsh-Bohmer KA, Breitner JCS, Lyketsos CG (2011) Progression of cognitive, functional, and neuropsychiatric symptom domains in a population cohort with Alzheimer dementia: the Cache County Dementia Progression study. Am J Geriatr Psychiatry 19:532-542. https://doi.org/10.1097/JGP.0b013e3181faec23 Andel R, Vigen C, Mack WJ, Clark LJ, Gatz M (2006) The effect of education and occupational complexity on rate of cognitive decline in Alzheimer's patients. Journal of the International Neuropsychological Society 12:147-152. Scarmeas N, Hadjigeorgiou GM, Papadimitriou A, Dubois B, Sarazin M, Brandt J, Albert M, Marder K, Bell K, Honig LS, Wegesin D, Stern Y (2004) Motor signs during the course of Alzheimer disease. Neurology 63:975. https://doi.org/10.1212/01.WNL.0000138440.39918.0C Wilkosz PA, Seltman HJ, Devlin B, Weamer EA, Lopez OL, DeKosky ST, Sweet RA (2010) Trajectories of cognitive decline in Alzheimer's disease. Int Psychogeriatr 22:281-290. https://doi.org/10.1017/s1041610209991001 Rouse HJ, Small BJ, Schinka JA, Loewenstein DA, Duara R, Potter H (2021) Mild behavioral impairment as a predictor of cognitive functioning in older adults. International Psychogeriatrics 33:285-293. https://doi.org/Pii S1041610220000678 10.1017/S1041610220000678 Cosentino S, Scarmeas N, Helzner E, Glymour MM, Brandt J, Albert M, Blacker D, Stern Y (2008) APOE ε4 allele predicts faster cognitive decline in mild Alzheimer disease. Neurology 70:1842-1849. https://doi.org/10.1212/01.wnl.0000304038.37421.cc Lee WJ, Liao YC, Wang YF, Lin YS, Wang SJ, Fuh JL (2020) Summative Effects of Vascular Risk Factors on the Progression of Alzheimer Disease. J Am Geriatr Soc 68:129-136. https://doi.org/10.1111/jgs.16181 Solfrizzi V, Capurso C, D'Introno A, Colacicco AM, Santamato A, Ranieri M, Fiore P, Capurso A, Panza F (2008) Lifestyle-related factors in predementia and dementia syndromes. Expert Rev Neurother 8:133-158. https://doi.org/10.1586/14737175.8.1.133 McKhann GM, Knopman DS, Chertkow H, Hyman BT, Jack CR, Jr., Kawas CH, Klunk WE, Koroshetz WJ, Manly JJ, Mayeux R, Mohs RC, Morris JC, Rossor MN, Scheltens P, Carrillo MC, Thies B, Weintraub S, Phelps CH (2011) The diagnosis of dementia due to Alzheimer's disease: recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimers Dement 7:263-269. https://doi.org/10.1016/j.jalz.2011.03.005 Winblad B, Palmer K, Kivipelto M, Jelic V, Fratiglioni L, Wahlund LO, Nordberg A, Backman L, Albert M, Almkvist O, Arai H, Basun H, Blennow K, de Leon M, DeCarli C, Erkinjuntti T, Giacobini E, Graff C, Hardy J, Jack C, Jorm A, Ritchie K, van Duijn C, Visser P, Petersen RC (2004) Mild cognitive impairment--beyond controversies, towards a consensus: report of the International Working Group on Mild Cognitive Impairment. J Intern Med 256:240-246. https://doi.org/10.1111/j.1365-2796.2004.01380.x Wechsler D (1997) WMS-III: Wechsler Memory Scale Administration and Scoring Manual. Psychological Corporation. Perneczky R, Wagenpfeil S, Komossa K, Grimmer T, Diehl J, Kurz A (2006) Mapping scores onto stages: mini-mental state examination and clinical dementia rating. Am J Geriatr Psychiatry 14:139-144. https://doi.org/10.1097/01.JGP.0000192478.82189.a8 Morris JC (1993) The Clinical Dementia Rating (CDR): current version and scoring rules. Neurology 43:2412-2414. https://doi.org/10.1212/wnl.43.11.2412-a Katz S, Ford AB, Moskowitz RW, Jackson BA, Jaffe MW (1963) Studies of illness in the aged: the index of ADL: a standardized measure of biological and psychosocial function. Jama 185:914-919. https://doi.org/10.1001/jama.1963.03060120024016 Lawton MP, Brody EM (1969) Assessment of older people: self-maintaining and instrumental activities of daily living. The gerontologist 9:179-186. Vanderploeg RD, Schinka JA, Jones T, Small BJ, Graves AB, Mortimer JA (2000) Elderly norms for the Hopkins Verbal Learning Test-Revised. Clin Neuropsychol 14:318-324. https://doi.org/10.1076/1385-4046(200008)14:3;1-P;FT318 Mack WJ, Freed DM, Williams BW, Henderson VW (1992) Boston Naming Test: shortened versions for use in Alzheimer's disease. J Gerontol 47:P154-158. https://doi.org/10.1093/geronj/47.3.p154 Harrison JE, Buxton P, Husain M, Wise R (2000) Short test of semantic and phonological fluency: normal performance, validity and test-retest reliability. Br J Clin Psychol 39:181-191. https://doi.org/10.1348/014466500163202 Humpstone HJ (1919) Memory Span Tests. Psychol Clin 12:196-200. Yesavage JA, Sheikh JI (1986) 9/Geriatric Depression Scale (GDS). Clinical Gerontologist 5:165-173. https://doi.org/10.1300/J018v05n01_09 Kaufer DI, Cummings JL, Ketchel P, Smith V, MacMillan A, Shelley T, Lopez OL, DeKosky ST (2000) Validation of the NPI-Q, a brief clinical form of the Neuropsychiatric Inventory. J Neuropsychiatry Clin Neurosci 12:233-239. https://doi.org/10.1176/jnp.12.2.233 Cui K, Zhao RH, Ding N, Xu XN, Gu WQ, He J, Ge XY (2025) The trajectories of depression and multimorbidity affect the patterns of cognitive decline: a prospective cohort study. Age Ageing 54. https://doi.org/ARTN afaf036 10.1093/ageing/afaf036 Taniguchi Y, Kitamura A, Murayama H, Amano H, Shinozaki T, Yokota I, Seino S, Nofuji Y, Nishi M, Yokoyama Y, Matsuyama Y, Fujiwara Y, Shinkai S (2017) Mini-Mental State Examination score trajectories and incident disabling dementia among community-dwelling older Japanese adults. Geriatr Gerontol Int 17:1928-1935. https://doi.org/10.1111/ggi.12996 Wu YF, Jia MN, Xiang CY, Lin SW, Jiang ZQ, Fang Y (2022) Predicting the long-term cognitive trajectories using machine learning approaches: A Chinese nationwide longitudinal database. Psychiat Res 310. https://doi.org/ARTN 114434 10.1016/j.psychres.2022.114434 Uchida K, Sugimoto T, Tange C, Nishita Y, Shimokata H, Saji N, Kuroda Y, Matsumoto N, Kishino Y, Ono R, Akisue T, Otsuka R, Sakurai T (2024) Association between abdominal adiposity and cognitive decline in older adults: a 10-year community-based study. J Nutr Health Aging 28. https://doi.org/ARTN 100175 10.1016/j.jnha.2024.100175 Forbes M, Lotfaliany M, Mohebbi M, Reynolds CF, Woods RL, Orchard S, Chong T, Agustini B, O'Neil A, Ryan J, Berk M (2024) Depressive symptoms and cognitive decline in older adults. International Psychogeriatrics. https://doi.org/10.1017/S1041610224000541 Heinze G, Dunkler D (2017) Five myths about variable selection. Transpl Int 30:6-10. https://doi.org/10.1111/tri.12895 Hoffmann VS, Schönecker S, Amin M, Reidler P, Brauer A, Kopczak A, Wunderlich S, Poli S, Althaus K, Müller S, Mansmann U, Kellert L (2024) A novel prediction score determining individual clinical outcome 3 months after juvenile stroke (PREDICT-score). J Neurol. https://doi.org/10.1007/s00415-024-12552-5 LeDell E, Petersen M, van der Laan M (2015) Computationally efficient confidence intervals for cross-validated area under the ROC curve estimates. Electron J Stat 9:1583-1607. https://doi.org/10.1214/15-Ejs1035 Dal Pozzolo A, Caelen O, Johnson RA, Bontempi G (2015) Calibrating Probability with Undersampling for Unbalanced Classification. 2015 Ieee Symposium Series on Computational Intelligence (Ieee Ssci):159-166. https://doi.org/10.1109/Ssci.2015.33 Jones BL, Nagin DS, Roeder K (2001) A SAS Procedure Based on Mixture Models for Estimating Developmental Trajectories. Sociological Methods & Research 29:374-393. https://doi.org/10.1177/0049124101029003005 Team RC (2020) R: A language and environment for statistical computing. In. R Foundation for Statistical Computing, Vienna, Austria. Hu B, Shao J, Palta M (2006) Pseudo-R 2 in logistic regression model. Statistica Sinica:847-860. Menze BH, Kelm BM, Masuch R, Himmelreich U, Bachert P, Petrich W, Hamprecht FA (2009) A comparison of random forest and its Gini importance with standard chemometric methods for the feature selection and classification of spectral data. Bmc Bioinformatics 10. https://doi.org/Artn 213 10.1186/1471-2105-10-213 Wu YF, Wang X, Gu CM, Zhu JM, Fang Y (2023) Investigating predictors of progression from mild cognitive impairment to Alzheimer's disease based on different time intervals. Age Ageing 52. https://doi.org/ARTN afad182 10.1093/ageing/afad182 Winter A, Aberle DR, Hsu W (2019) External validation and recalibration of the Brock model to predict probability of cancer in pulmonary nodules using NLST data. Thorax 74:551-563. https://doi.org/10.1136/thoraxjnl-2018-212413 Wattmo C, Wallin ÅK, Minthon L (2013) Progression of mild Alzheimer’s disease: knowledge and prediction models required for future treatment strategies. Alzheimer's Research & Therapy 5:44. https://doi.org/10.1186/alzrt210 Wilson RS, Gilley DW, Bennett DA, Beckett LA, Evans DA (2000) Hallucinations, delusions, and cognitive decline in Alzheimer's disease. J Neurol Neurosurg Psychiatry 69:172-177. https://doi.org/10.1136/jnnp.69.2.172 Wilson RS, Wang TH, Yu L, Bennett DA, Boyle PA (2020) Normative Cognitive Decline in Old Age. Ann Neurol 87:816-829. https://doi.org/10.1002/ana.25711 Brown PJ, Devanand DP, Liu X, Caccappolo E (2011) Functional impairment in elderly patients with mild cognitive impairment and mild Alzheimer disease. Arch Gen Psychiatry 68:617-626. https://doi.org/10.1001/archgenpsychiatry.2011.57 Montero-Odasso M, Speechley M, Muir-Hunter SW, Pieruccini-Faria F, Sarquis-Adamson Y, Hachinski V, Bherer L, Borrie M, Wells J, Garg AX, Tian Q, Ferrucci L, Bray NW, Cullen S, Mahon J, Titus J, Camicioli R (2020) Dual decline in gait speed and cognition is associated with future dementia: evidence for a phenotype. Age Ageing 49:995-1002. https://doi.org/10.1093/ageing/afaa106 Farrer LA, Cupples LA, Haines JL, Hyman B, Kukull WA, Mayeux R, Myers RH, Pericak-Vance MA, Risch N, van Duijn CM (1997) Effects of age, sex, and ethnicity on the association between apolipoprotein E genotype and Alzheimer disease. A meta-analysis. APOE and Alzheimer Disease Meta Analysis Consortium. Jama 278:1349-1356. O'Hara R, Sommer B, Way N, Kraemer HC, Taylor J, Murphy G (2008) Slower speed-of-processing of cognitive tasks is associated with presence of the apolipoprotein epsilon4 allele. J Psychiatr Res 42:199-204. https://doi.org/10.1016/j.jpsychires.2006.12.001 Hayden KM, Reed BR, Manly JJ, Tommet D, Pietrzak RH, Chelune GJ, Yang FM, Revell AJ, Bennett DA, Jones RN (2011) Cognitive decline in the elderly: an analysis of population heterogeneity. Age Ageing 40:684-689. https://doi.org/10.1093/ageing/afr101 Kleiman T, Zdanys K, Black B, Rightmer T, Grey M, Garman K, Macavoy M, Gelernter J, van Dyck C (2006) Apolipoprotein E epsilon4 allele is unrelated to cognitive or functional decline in Alzheimer's disease: retrospective and prospective analysis. Dement Geriatr Cogn Disord 22:73-82. https://doi.org/10.1159/000093316 Tables Table 1 Differences in baseline demographics between early AD and aMCI in development cohort (N = 251) and the validation cohort (N = 322) Development cohort Validation cohort Total (n = 251) Early AD (n = 209) aMCI (n = 42) Early AD (n = 322) Continuous variable Mean SD Mean SD Mean SD P-value a Mean SD Age, years 78.3 7.3 79.1 6.8 74.2 8.4 <.001* 77.1 5.6 Education, years 10.9 4.5 10.8 4.5 11.2 4.7 0.601 10.8 2.4 Disease duration, years 3.5 3.3 3.6 3.5 2.7 2.5 0.055 - b - MMSE score 23.3 2.3 22.9 2.0 25.3 2.4 <.001* 22.7 2.1 BMI 23.9 3.5 24.0 3.5 23.4 3.9 0.412 22.2 3.5 Exercise habits, days/week 4.1 3.0 4.1 3.0 4.0 3.0 0.855 - - GDS-15 3.3 3.0 3.3 3.0 3.8 3.0 0.251 3.7 2.7 Health status 1.8 1.6 1.8 1.6 1.8 1.4 0.845 - - Categorical variable Number % Number % Number % Number % Sex, male 136 53.8 116 55.5 19 45.2 0.223 97 30.1 Smoking status, ≥5 cigarettes/day 8 3.2 7 3.4 1 2.4 0.999 - - Hypertension 147 58.6 122 59.2 25 59.5 0.999 165 51.2 Diabetes 63 25.1 53 25.9 10 23.8 0.934 47 14.6 Cardiovascular disease 66 26.3 55 26.7 11 26.2 0.999 56 17.4 APOE ε4 allele 0 163 64.9 140 67.0 23 54.8 0.193 167 51.9 1 79 31.5 63 30.1 16 38.1 106 32.9 2 9 3.6 6 2.9 3 7.1 26 8.1 Global CDR 0.5 89 35.5 47 22.5 42 100.0 <.001* 14 4.3 1 162 64.5 162 77.5 0 0 2 0.6 a Chi-square test or t-test was used to assess differences in demographic characteristics between early AD and aMCI groups. b “-” indicates that variable information is not available. *P-value < 0.05. Abbreviations: AD, Alzheimer’s disease; aMCI, amnestic mild cognitive impairment; CDR, Clinical Dementia Rating; GDR-15, 15-item Geriatric Depression Scale; MMSE, Mini-Mental State Examination. Table 2 Comparison of two cognitive trajectory groups on baseline variables in the development cohort (N = 251) G1 (n = 94) G2 (n = 157) Demographics Mean (SD) or N (%) Mean (SD) or N (%) P-value a Age, years 79.0 (7.3) 77.8 (7.3) 0.231 Sex, male 54 (57.4) 81 (51.6) 0.442 Education, years 10.7 (4.3) 11.0 (4.6) 0.674 Lifestyle factors BMI 23.6 (3.2) 24.1 (3.7) 0.299 Exercise habits, days/week 4.0(3.2) 4.1(2.9) 0.819 Smoking status, ≥5 cigarettes/day 3 (3.2) 5 (3.2) 0.999 Clinical characteristics Baseline diagnosis AD 88 (93.6) 121 (77.1) 0.001** aMCI 6 (6.4) 36 (22.9) Global CDR 0.5 18 (19.1) 71 (45.2) <.001** 1 76 (80.9) 86 (54.7) APOE ε4 allele 0 49 (52.1) 114 (72.6) 0.001** 1 40 (42.6) 39 (24.9) 2 5 (5.3) 4 (2.5) Disease duration, years 3.7 (3.3) 3.3 (3.4) 0.364 Comorbidities Health status 2.0 (1.9) 1.7 (1.3) 0.118 Hypertension 56 (59.6) 91 (59.1) 0.999 Diabetes 26 (28.0) 37 (24.0) 0.592 Cardiovascular disease 27 (28.7) 39 (25.3) 0.660 Functional capacities ADL score 2.5 (3.9) 1.0 (2.5) <.001** IADL score 7.5 (6.4) 4.0 (4.6) <.001** Neuropsychological tests GDS-15 3.1 (3.2) 3.5 (3.0) 0.260 MMSE score 21.8 (1.8) 24.2 (2.0) <.001** Recall (after 15 min) 1.2 (1.8) 2.6 (2.5) <.001** Forward digit span 9.2 (2.7) 9.7 (2.5) 0.102 Backward digit span 4.7 (1.8) 5.3 (2.0) 0.025* Verbal fluency 7.2 (2.9) 8.4 (2.5) <.001** mBNT 12.3 (2.3) 13.2 (1.6) 0.001** Severity of NPS (Frequency × Intensity) Delusions 0.5 (0.9) 0.4 (0.8) 0.201 Hallucinations 0.3 (0.7) 0.1 (0.4) 0.002** Agitation 0.5 (0.8) 0.4 (0.7) 0.100 Depression 0.6 (0.8) 0.6 (0.8) 0.982 Anxiety 0.6 (0.9) 0.5 (0.9) 0.250 Euphoria 0.2 (0.6) 0.1 (0.5) 0.380 Apathy 0.6 (0.9) 0.4 (0.8) 0.062 Disinhibition 0.4 (0.7) 0.4 (0.8) 0.656 Irritability 0.7 (0.8) 0.6 (0.9) 0.502 Motor disturbance 0.4 (0.9) 0.2 (0.5) 0.005* Night-time behaviors 0.4 (0.9) 0.4 (0.7) 0.706 Appetite 0.3 (0.7) 0.4 (0.8) 0.419 a P-value for Chi-square test or t-test was used to assess differences in variables between fast-decliners (G1) and slow-decliners (G2). *P-value < 0.05. ** P-value < 0.002 (Bonferroni correction for 30 comparisons). Abbreviations: SD, standard deviation; CDR, Clinical Dementia Rating; GDR-15, 15-item Geriatric Depression Scale; IADL, Instrumental Activities of Daily Living; mBNT, 15-item modified Boston Naming Test; MMSE, Mini-Mental State Examination; NPS, neuropsychiatric symptoms. Table 3 Odds ratios of predictors in the logistic regression model for rapid cognitive decline in the development cohort (N = 251) OR 95% CI P-value OR 95% CI P-value OR 95% CI P-value Predictor Model 1 a Model 2 b Model 3 c APOE ε4 allele count 2.70 1.45-5.18 0.002* 3.14 1.72–5.91 <.001* 2.85 1.57–5.34 <.001* MMSE score 0.58 0.47-0.70 <.001* 0.56 0.46–0.66 <.001* 0.54 0.45–0.65 <.001* IADL score 1.08 1.00-1.17 0.004* 1.13 1.07–1.21 <.001* 1.13 1.06–1.21 <.001* ADL score 1.04 0.91-1.21 0.586 - - - - - - Recall (after 15 min) 0.92 0.77-1.08 0.302 - - - - - - Verbal fluency 0.97 0.85-1.10 0.593 - - - - - - 15-item mBNT 0.88 0.74-1.04 0.140 - - - - - - Hallucinations 2.41 0.88-7.12 0.096 - - - - - - Note: The analysis included the full sample of 94 fast-decliners (G1) and 157 slow-decliners (G2) from the development cohort. a Model 1: A logistic regression model for rapid cognitive decline constructed using functional capacity measures (ADL and IADL scores), neuropsychological tests (MMSE score, recall test, verbal fluency test, and mBNT), neuropsychiatric symptoms (hallucinations), and APOE ε4 allele count. b Model 2: An unadjusted regression model that included only baseline MMSE, IADL scores, and APOE ε4 allele count as predictors. c Model 3: Derived from Model 2 with additional adjustments for age, sex, and education. *P-value < 0.05. Abbreviations: CI, Confidence Interval; IADL, Instrumental Activities of Daily Living; mBNT, modified Boston naming test; MMSE, Mini-Mental State Examination; NPS, neuropsychiatric symptoms; NPT, neuropsychological tests; OR, Odds Ratio for fast-decliners. Additional Declarations No competing interests reported. Supplementary Files Supplementarymaterial20251226.docx Cite Share Download PDF Status: Posted Version 1 posted 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-8629600","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":584624456,"identity":"b55566e5-ee97-4ef7-a722-526562a68398","order_by":0,"name":"Yao-Hwei Fang","email":"","orcid":"","institution":"National Health Research Institutes","correspondingAuthor":false,"prefix":"","firstName":"Yao-Hwei","middleName":"","lastName":"Fang","suffix":""},{"id":584624457,"identity":"e789936b-f0e5-4a40-b1ab-0142218f94cd","order_by":1,"name":"Yung-Shuan Lin","email":"","orcid":"","institution":"Taipei Veterans General Hospital","correspondingAuthor":false,"prefix":"","firstName":"Yung-Shuan","middleName":"","lastName":"Lin","suffix":""},{"id":584624458,"identity":"34c9d338-3222-4e6a-81e9-ba62c4f7c265","order_by":2,"name":"Kazuaki Uchida","email":"","orcid":"","institution":"National Center for Geriatrics and Gerontology","correspondingAuthor":false,"prefix":"","firstName":"Kazuaki","middleName":"","lastName":"Uchida","suffix":""},{"id":584624459,"identity":"e9240b58-05f6-4bc4-ab17-123a15f404bb","order_by":3,"name":"Wei-Ju Lee","email":"","orcid":"","institution":"Taichung Veterans General Hospital","correspondingAuthor":false,"prefix":"","firstName":"Wei-Ju","middleName":"","lastName":"Lee","suffix":""},{"id":584624460,"identity":"42ed153e-ff2f-4564-94bf-0cb69b475ce3","order_by":4,"name":"Chih-Cheng Hsu","email":"","orcid":"","institution":"National Health Research Institutes","correspondingAuthor":false,"prefix":"","firstName":"Chih-Cheng","middleName":"","lastName":"Hsu","suffix":""},{"id":584624461,"identity":"b357ff5f-4744-4a19-a304-7b54e57ab70a","order_by":5,"name":"Tsung-Jen Hsieh","email":"","orcid":"","institution":"National Health Research Institutes","correspondingAuthor":false,"prefix":"","firstName":"Tsung-Jen","middleName":"","lastName":"Hsieh","suffix":""},{"id":584624462,"identity":"358d17ab-bdff-4ac3-9f76-ec3637168243","order_by":6,"name":"Tzu-Yu Chen","email":"","orcid":"","institution":"National Health Research Institutes","correspondingAuthor":false,"prefix":"","firstName":"Tzu-Yu","middleName":"","lastName":"Chen","suffix":""},{"id":584624464,"identity":"3e244277-651a-4059-9653-ffbb0095f287","order_by":7,"name":"Yi-Chu Liao","email":"","orcid":"","institution":"Taipei Veterans General Hospital","correspondingAuthor":false,"prefix":"","firstName":"Yi-Chu","middleName":"","lastName":"Liao","suffix":""},{"id":584624469,"identity":"fd494eed-3a69-4fd1-a3f6-6006df306783","order_by":8,"name":"Yujiro Kuroda","email":"","orcid":"","institution":"National Center for Geriatrics and Gerontology","correspondingAuthor":false,"prefix":"","firstName":"Yujiro","middleName":"","lastName":"Kuroda","suffix":""},{"id":584624470,"identity":"d1a0d81e-6be6-4739-9049-f203799d6101","order_by":9,"name":"Taiki Sugimoto","email":"","orcid":"","institution":"National Center for Geriatrics and Gerontology","correspondingAuthor":false,"prefix":"","firstName":"Taiki","middleName":"","lastName":"Sugimoto","suffix":""},{"id":584624473,"identity":"1036fa9b-fc0f-4c4a-93d8-eadfbe9986ad","order_by":10,"name":"I-Shou Chang","email":"","orcid":"","institution":"National Health Research Institutes","correspondingAuthor":false,"prefix":"","firstName":"I-Shou","middleName":"","lastName":"Chang","suffix":""},{"id":584624475,"identity":"6371f9a2-4eb1-49b6-9f7e-9d270a40c899","order_by":11,"name":"Takashi Sakurai","email":"","orcid":"","institution":"National Center for Geriatrics and Gerontology","correspondingAuthor":false,"prefix":"","firstName":"Takashi","middleName":"","lastName":"Sakurai","suffix":""},{"id":584624478,"identity":"891029cf-6d68-4b6c-820c-43035032a78a","order_by":12,"name":"Hidenori Arai","email":"","orcid":"","institution":"National Center for Geriatrics and Gerontology","correspondingAuthor":false,"prefix":"","firstName":"Hidenori","middleName":"","lastName":"Arai","suffix":""},{"id":584624481,"identity":"bb20b242-d63b-446a-aa96-da5f05f398b7","order_by":13,"name":"Shuu-Jiun Wang","email":"","orcid":"","institution":"Taipei Veterans General Hospital","correspondingAuthor":false,"prefix":"","firstName":"Shuu-Jiun","middleName":"","lastName":"Wang","suffix":""},{"id":584624483,"identity":"5e923eac-c4ba-4caf-83c5-94e83574966c","order_by":14,"name":"Chao A. Hsiung","email":"","orcid":"","institution":"National Health Research Institutes","correspondingAuthor":false,"prefix":"","firstName":"Chao","middleName":"A.","lastName":"Hsiung","suffix":""},{"id":584624486,"identity":"d04b9f36-cd8c-4477-9bf3-c93176b9a2dd","order_by":15,"name":"Jong-Ling Fuh","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA3ElEQVRIiWNgGAWjYBACxnbGB8wMFQwMBsRraWY2YGY4A9TCBhNiw6ceBJiBWhjbSNHC3MzMJl0477Dddvnegw8YftkkNsj34HckI0jLzG2Hk3e28SUbMPalJTaw8RDSwn9MmndbWrLBMR4zCcaew8YMbLwbCNvCOwesxfwHCVoabOxAtjAw/DgsR4wWZmueYzYJBsfykiUSG9Lk2NjyP+DVYtjezHibp0bC3uDw2YMfPvyx4eFnPpaAX0sDhE5sYOABkm1ExKQ8lLZnAGlh+ENI/SgYBaNgFIxEAABbNzvoa8PgnAAAAABJRU5ErkJggg==","orcid":"","institution":"Taipei Veterans General Hospital","correspondingAuthor":true,"prefix":"","firstName":"Jong-Ling","middleName":"","lastName":"Fuh","suffix":""}],"badges":[],"createdAt":"2026-01-18 06:53:21","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8629600/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8629600/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":101787285,"identity":"d62f4d0f-5082-4f38-85f0-140a268e216c","added_by":"auto","created_at":"2026-02-03 15:47:35","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":660849,"visible":true,"origin":"","legend":"\u003cp\u003eObserved individual MMSE trajectories for fast- and slow-decliners in (a) the developemtn cohort and (b) the validation cohort.\u003c/p\u003e\n\u003cp\u003eFast-decliners (G1) included 94 individuals in the development cohort and 98 individuals in the validation cohort. Slow-decliners (G2) included 157 individuals in the development cohort and 224 individuals in the validation cohort. The plots illustrate individual cognitive trajectories within each trajectory group (each line represents a single subject), with the overall mean trajectory for each group displayed as a bold black line.\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-8629600/v1/b68752dfa93a63cc8017dc34.png"},{"id":101787286,"identity":"e8957faf-423f-4be5-9ae2-98cf90612f9f","added_by":"auto","created_at":"2026-02-03 15:47:35","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":367623,"visible":true,"origin":"","legend":"\u003cp\u003eNomogram of the prediction model estimates the risk of rapid cognitive decline.\u003c/p\u003e\n\u003cp\u003eUsing the scoring chart derived from the logistic regression model (top 7 rows), each predictor is assigned a value along the 'Points' axis, calculated as the product of the predictor’s value and its regression coefficient. Summing these points yields a 'Total Points' score, which is then used to estimate the risk of rapid cognitive decline by drawing a vertical line from the 'Total Points' axis to the 'Risk Score' axis (range 0–1).\u003c/p\u003e\n\u003cp\u003eAbbreviations: APOE, Apolipoprotein E; IADL, instrumental activities of daily living; MMSE, Mini-Mental State Examination.\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-8629600/v1/d36662ce9aa192d0eb22bf8a.png"},{"id":103664180,"identity":"2ce3b437-29a0-4bef-849f-4dfc1f5473f4","added_by":"auto","created_at":"2026-02-28 18:54:27","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1764281,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8629600/v1/d21a1db2-307e-436f-96fd-3216b215034d.pdf"},{"id":101787288,"identity":"bc06d56a-e2cd-4026-9937-f9b2f30bacb0","added_by":"auto","created_at":"2026-02-03 15:47:35","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":999752,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementarymaterial20251226.docx","url":"https://assets-eu.researchsquare.com/files/rs-8629600/v1/74605517c5ab124126805ca4.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Predicting Cognitive Decline in Early Alzheimer’s: Insights from East Asian Cohorts","fulltext":[{"header":"Background","content":"\u003cp\u003eCognitive decline in Alzheimer\u0026rsquo;s disease (AD) and amnestic mild cognitive impairment (aMCI) is highly variable across individuals [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Early AD and aMCI converters (those converting from aMCI to AD) are interconnected within the AD continuum. A key concern for families and physicians is predicting the rate of cognitive decline. Meta-analytic data from clinic-based cohorts suggested an average Mini-Mental State Examination (MMSE) decline of \u0026asymp;\u0026thinsp;3.3 points per year [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e], whereas community studies report almost half that rate (\u0026asymp;\u0026thinsp;1.5 points) [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. This discrepancy likely reflects differences in sampling frame\u0026ndash;clinic samples over-represent patients with more advanced disease or comorbidities\u0026ndash;while trajectory-modeling studies that follow newly diagnosed AD cases show that most individuals decline at an even slower rate the averages reported in either clinic or community samples [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Such heterogeneity motivates the search for robust predictors of rapid deterioration. Existing evidence can be grouped into three inter-related domains. Demographic factors\u0026mdash;earlier age at symptom onset, male sex, and lower educational attainment\u0026mdash;have been linked to faster cognitive loss, although these associations often weaken after adjustment for baseline cognition, implying confounding by disease stage [\u003cspan additionalcitationids=\"CR6\" citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Clinical and genetic markers, including extrapyramidal signs, behavioral disturbances, and carriage of the apolipoprotein E (\u003cem\u003eAPOE\u003c/em\u003e) ε4 allele, predict steeper decline in some studies but not others, suggesting that sample age structure, follow-up duration, and diagnostic criteria condition their impact [\u003cspan additionalcitationids=\"CR9 CR10\" citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Evidence on vascular and lifestyle influences is mixed. Several studies report that hypertension, diabetes, smoking, or sedentary behaviors accelerate cognitive loss, whereas others find weak or null associations. Such heterogeneity likely reflects differences in how vascular risk is measured and treated, baseline disease severity, follow-up length, and analytic methods, rather than any single setting-related factor [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e13\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eIdentifying individuals at risk for rapid cognitive decline aids proactive interventions. We hypothesized that individuals with early AD or aMCI could be categorized into distinct groups by cognitive trajectories, with shared baseline characteristics influencing the rate of decline. This study aimed to identify latent cognitive decline groups based on MMSE scores using group-based trajectory modeling (GBTM) in a longitudinal clinical cohort of Han Chinese. We also developed a prediction model for rapid cognitive decline incorporating demographics, comorbidities, functional capacities, neuropsychological tests (NPT), neuropsychiatric symptoms (NPS), and \u003cem\u003eAPOE\u003c/em\u003e ε4 genotype. Extended follow-up data were used to test the hypothesis that distinct trajectories of cognitive decline exist. The developed prediction model was validated in an independent Japanese cohort from the National Center for Geriatrics and Gerontology (NCGG), making it the first clinic-based prediction model for cognitive decline in East Asian populations with external validation.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy design and participants\u003c/h2\u003e \u003cp\u003eWe recruited individuals with aMCI (n\u0026thinsp;=\u0026thinsp;344) and AD (n\u0026thinsp;=\u0026thinsp;975) from the outpatient clinics of Taipei Veterans General Hospital (TVGH) and Taichung Veterans General Hospital (TCVGH) as the development cohort. Supplementary Fig.\u0026nbsp;1 presents the patient enrollment flowchart. AD diagnoses followed the 2011 National Institute on Aging\u0026mdash;Alzheimer\u0026rsquo;s Association clinical criteria for probable AD [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e14\u003c/span\u003e] while aMCI was diagnosed per the revised 2004 consensus criteria [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e15\u003c/span\u003e], defined as a cut-off of 1.5 standard deviations below the age-adjusted norm on the Wechsler Memory Scale III logical memory test [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. Participants in the developmental cohort were aged 50\u0026ndash;90 years and required a reliable caregiver with at least 10 hours of weekly contact.\u003c/p\u003e \u003cp\u003eTo focus on early AD and aMCI converters, we included individuals with a baseline MMSE score\u0026thinsp;\u0026ge;\u0026thinsp;20, corresponding approximately to a Clinical Dementia Rating (CDR) of \u0026le;\u0026thinsp;1 [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. Exclusion criteria included fewer than two annual follow-ups, aMCI cases that did not convert to AD by the second-year follow-up, and the presence of other significant neurological condition affecting cognition. Exclusion conditions included acute confusion due to systemic disease, major depressive disorder (per Diagnostic and Statistical Manual of Mental Disorders 5th edition), probable vascular dementia, normal pressure hydrocephalus, progressive supranuclear palsy, and significant head trauma with persistent neurologic deficits, or known structural brain abnormalities. Ultimately, 251 out of the 1319 early AD and aMCI converters met the inclusion criteria.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eBaseline and annual data collection\u003c/h3\u003e\n\u003cp\u003eCognitive state was assessed using the MMSE, and dementia severity via CDR [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Thirty candidate variables were selected based on prior research related to cognitive decline, including demographics, lifestyle factors, comorbidities, functional capacities, NPT, NPS, and clinical characteristics.\u003c/p\u003e \u003cp\u003eDemographics included age, sex, education (years). Lifestyle factors included BMI, regular exercise frequency (days/week), and smoking status (\u0026ge;\u0026thinsp;5 cigarettes/day). Overall health status (comorbidity) was assessed by the total number of self-reported physician-diagnosed chronic diseases, including hypertension, diabetes, cardiovascular diseases, etc. (Supplementary Table\u0026nbsp;1). Functional capacities were assessed using the activity of daily living (ADL) [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e19\u003c/span\u003e] and instrumental ADL (IADL) scales [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e20\u003c/span\u003e], in which lower scores indicated better function. NPT included the 12-item memory recall test [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e21\u003c/span\u003e], 15-item modified Boston Naming Test (mBNT) [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e22\u003c/span\u003e], the verbal fluency test [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e23\u003c/span\u003e], and the forward and backward digit span task [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. The instruments assessed short-term memory, language, executive function, attention and working memory, respectively. Depressive symptoms were assessed using the 15-item Geriatric Depression Scale (GDS-15) [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. Higher NPT scores indicated better performance. NPS were evaluated with the Neuropsychiatric Inventory-Questionnaire (NPI-Q), which measures 12 behavioral and psychological symptoms of dementia, including depression, anxiety, apathy, sleep, appetite, agitation, irritability, disinhibition, elation, motor disturbance, delusions, and hallucinations [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. Higher NPI-Q scores indicated worse NPS. Clinical characteristics included baseline diagnosis, CDR, disease duration, and \u003cem\u003eAPOE\u003c/em\u003e ε4 allele count. Disease duration referred to the time from the initial cognitive dysfunction to study enrollment. \u003cem\u003eAPOE\u003c/em\u003e ε4 allele count (0, 1, or 2 alleles) was included as a continuous covariate.\u003c/p\u003e \u003cp\u003eAt each annual follow-up, MMSE and CDR scores were reassessed and profiling of health status and GDS-15 were also updated, because the trajectories of multimorbidity and depression affect cognitive decline patterns [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e27\u003c/span\u003e].\u003c/p\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eStatistical analysis\u003c/h2\u003e \u003cp\u003eThe statistical workflow (Supplementary Fig.\u0026nbsp;2) comprised three steps: (1) Trajectory identification, (2) Feature selection, and (3) Model building and evaluation.\u003c/p\u003e \u003cp\u003eFor trajectory identification, cognitive trajectories were determined using MMSE scores as a continuous measure of cognitive function [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. GBTM was applied to classify early AD/aMCI subjects into distinct patterns of cognitive decline over two years. Potential confounders, identified from the existing literature [\u003cspan additionalcitationids=\"CR30\" citationid=\"CR31\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e31\u003c/span\u003e], were adjusted in the model. These included time-fixed covariates (baseline age, sex, years of education, and \u003cem\u003eAPOE\u003c/em\u003e ε4 allele count) and time-varying covariates (health status and depressive symptoms at each follow-up). See Supplementary Methods \u0026ldquo;Trajectory Identification\u0026rdquo; for more details.\u003c/p\u003e \u003cp\u003eFeature selection followed a two-step process [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. First, univariable scanning using Chi-square tests for categorical variables and t-tests for continuous variables that retained with P\u0026thinsp;\u0026lt;\u0026thinsp;0.002 (Bonferroni corrected). This was followed by backward logistic regression (LR) with Bayesian Information Criterion (BIC) to eliminate redundant variables affected by multicollinearity [\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. The refined model was to predict cognitive decline trajectories in early AD/aMCI subjects. Model performance was assessed using 10-fold cross-validation, with the area under the curve (AUC) reported alongside 95% confidence intervals (CI) and standard error [\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. The optimal classification threshold was determined using Bayes Minimum Risk theory, and predictive performance was evaluated based on accuracy, sensitivity, and specificity [\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e35\u003c/span\u003e]. To enhance clinical applicability, we developed a nomogram, assigning predictor-based scores to estimate an individual\u0026rsquo;s risk of rapid cognitive decline.\u003c/p\u003e \u003cp\u003eAll analyses were conducted using SAS (Proc Traj) [\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e36\u003c/span\u003e], and R packages, including cvAUC, pROC, and ggplot2 for trajectory modeling, predictive performance assessment, and data visualization [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e37\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eThe validation study\u003c/h3\u003e\n\u003cp\u003eThe validation cohort comprised 322 early AD subjects from the Memory Clinic at the NCGG in Japan. Detailed information on the validation cohort is provided in Supplementary Methods. Participants aged\u0026thinsp;\u0026ge;\u0026thinsp;65 with an AD diagnosis at baseline and at least three clinic visits were included.\u003c/p\u003e \u003cp\u003eTo validate the prediction model for cognitive progression, we determined the optimal threshold for identifying rapid cognitive decline, a critical factor for both model performance and clinical applicability (see Supplementary Methods \u0026ldquo;threshold determination\u0026rdquo; for more details). LR coefficients derived from the development cohort were applied to predict the risk of rapid cognitive decline in the validation cohort. Model performance was evaluated using AUC and a calibration plot.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003eDemographics, lifestyle, and clinical characteristics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn\u0026nbsp;development\u0026nbsp;cohort, 251 participants with a baseline MMSE score \u0026ge; 20 were included in the analysis, consisting of 209 early AD subjects (116 males, 93 females) and 42 aMCI converters (19 males, 23 females). Among AD subjects,\u0026nbsp;22.5% (47/209) had a baseline CDR of 0.5,\u0026nbsp;while\u0026nbsp;all aMCI subjects had a CDR of 0.5 (Table 1). AD subjects were older than aMCI subjects (79.1 \u0026plusmn; 6.8 vs. 74.2 \u0026plusmn; 8.4 years, P \u0026lt;\u0026nbsp;0.001). Baseline MMSE were lower in AD subjects than in aMCI subjects (22.9 \u0026plusmn; 2.0 vs. 25.3 \u0026plusmn; 2.4, P\u0026nbsp;\u0026lt;\u0026nbsp;0.001).\u0026nbsp;No significant differences were found in sex, education, health status, disease duration, or \u003cem\u003eAPOE\u003c/em\u003e \u0026epsilon;4 allele count between AD/aMCI subjects.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTrajectory identification\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eModel selection for GBTM (Supplementary Fig. 3) identified a two-trajectory model as the best-fitting solution. Both trajectories exhibited linear cognitive decline over time (Supplementary Table 2). Fig. 1(a) displays individual trajectories classified into two groups. Group 1 (G1), consisting of 94 subjects (37.5%) with a mean annual MMSE decline of 2.6 \u0026plusmn; 1.3 points, exhibited faster cognitive decline (\u0026ldquo;fast-decliners\u0026rdquo;). In contrast, Group 2 (G2), comprising 157 subjects (62.5%) with a mean annual MMSE decline of 0.7 \u0026plusmn; 1.0 points, exhibited slower decline (\u0026ldquo;slow-decliners\u0026rdquo;).\u0026nbsp;Supplementary\u0026nbsp;Table 3 shows the estimated coefficients of the trajectory model\u0026nbsp;in the\u0026nbsp;development\u0026nbsp;cohort. \u003cem\u003eAPOE\u003c/em\u003e \u0026epsilon;4 allele count predicted cognitive trajectories in early AD/aMCI subjects,\u0026nbsp;while age, sex, education, and health status were\u0026nbsp;not significant predictors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFeature selection\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTable 2 compares baseline variables between the cognitive trajectory groups. The fast-decliner group (G1) had a significantly higher proportion of early AD cases (93.6%) than the slow-decline group (G2) (77.1%). Age, sex, education, BMI, exercise habits,\u0026nbsp;smoking, hypertension, diabetes, cardiovascular disease, health status, disease duration, depressive symptoms, and NPS (except hallucinations) did not differ between the two groups. However,\u0026nbsp;\u003cem\u003eAPOE\u003c/em\u003e \u0026epsilon;4 allele count, CDR, NPT\u0026nbsp;(MMSE,\u0026nbsp;recall test, verbal fluency, and\u0026nbsp;15-item\u0026nbsp;mBNT), functional capacities\u0026nbsp;(ADL and\u0026nbsp;IADL scores), and hallucinations (NPS)\u0026nbsp;were significant (P\u0026nbsp;\u0026lt;\u0026nbsp;0.002, accounting for multiple comparisons) in the univariable scanning.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAfter applying the backward LR method (Table 3, Model 2), only baseline MMSE score, IADL score, and\u0026nbsp;\u003cem\u003eAPOE\u003c/em\u003e \u0026epsilon;4 allele count remained as significant predictors of cognitive decline in early AD/aMCI convertors. The results showed that more\u0026nbsp;\u003cem\u003eAPOE\u003c/em\u003e \u0026epsilon;4 alleles,\u0026nbsp;higher IADL and lower MMSE scores were significantly associated with the fast-decline trajectory.\u0026nbsp;Correlations among the three predictors were low (0.09\u0026ndash;0.27) and collectively\u0026nbsp;explained 43.0% of the variability in cognitive decline trajectories\u0026nbsp;[38].\u0026nbsp;Additionally,\u0026nbsp;redundant variables, including recall, verbal fluency,\u0026nbsp;15-item\u0026nbsp;mBNT, ADL score, and hallucinations, were not statistically significant in the backward LR method. Hallucinations showed a strong effect (odds ratio = 2.41) but weak significance (P = 0.096), possibly due to the small sample size (Table 3, Model 1).\u003c/p\u003e\n\u003cp\u003eIn Model 3 (Table 3), age, sex, and education were included as covariates for further adjustment. Sensitivity analysis using Model 3 confirmed consistent results.\u0026nbsp;A scoring chart and nomogram were created using Model 3 regression coefficients (Fig. 2).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eModel building and evaluation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe optimal classification threshold was set based on the fast-decliner group\u0026rsquo;s prior probability (\u0026tau; = 37.5%) [35]. The prediction\u0026nbsp;model (Model 3) achieved an AUC of 0.841, sensitivity of 0.769, and specificity of 0.728\u0026nbsp;(95% CI: 0.779\u0026ndash;0.950, 0.657\u0026ndash;0.886, and 0.621\u0026ndash;0.845, respectively).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe LR model was further compared with machine learning methods, including support vector machine and random forest [39, 40]. Results showed that the LR model achieved a higher AUC in predicting rapid progression of early AD/aMCI (Supplementary\u0026nbsp;Table 4).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eValidation study analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe validation cohort consisted of 322 Japanese early AD subjects. Baseline demographics (age, sex, and education), \u003cem\u003eAPOE\u003c/em\u003e \u0026epsilon;4\u0026nbsp;allele count, IADL, and MMSE score were included (Table 1). Compared with the Taiwanese development cohort, Japanese participants were\u0026nbsp;younger, had a lower proportion of men and had lower MMSE scores and BMI. Diabetes and cardiovascular disease were more prevalent in the Taiwanese group. No significant differences were observed between the cohorts in GDS-15 scores, years of education, hypertension, or \u003cem\u003eAPOE\u003c/em\u003e \u0026epsilon;4 allele count. Data harmonization ensured consistency between cohorts, except for IADL (details in Supplementary Methods).\u003c/p\u003e\n\u003cp\u003eSupplementary\u0026nbsp;Fig. 4 shows an optimal threshold probability of 0.3 for defining rapid cognitive decline (\u0026chi;\u003csup\u003e2\u003c/sup\u003e = 62.7, P\u0026nbsp;= 1.66\u0026times;10\u003csup\u003e-14\u003c/sup\u003e).\u0026nbsp;This\u0026nbsp;threshold\u0026nbsp;was comparable to the development cohort\u0026rsquo;s classification using Bayes Minimum Risk theory (37.5% fast-decliners, 62.5% slow-decliners).\u0026nbsp;Figure 1(b) illustrates cognitive trajectories of the validation cohort. Baseline mean MMSE scores were lower among slow-decliners in the validation cohort (22.6 \u0026plusmn; 2.1) compared to those in the development cohort (24.2 \u0026plusmn; 2.0), likely due to differences in inclusion criteria. The validation cohort included only AD participants aged 65 and older whereas the development cohort also included aMCI cases. In both cohorts, fast-decliners demonstrated a greater mean annual MMSE decline compared to slow-decliners.\u003c/p\u003e\n\u003cp\u003eModel 3\u0026rsquo;s calibration was assessed (Supplementary\u0026nbsp;Fig. 5(a)), showing a calibration slope of 0.52 (95% CI: 0.35\u0026ndash;0.70) and an intercept of -0.21 (95% CI: -0.58\u0026ndash;0.17), indicating poor alignment. Mis-calibration likely resulted from cohort differences in subject characteristics, disease prevalence, and treatment policies. Therefore, model recalibration was necessary. After logistic recalibration [41], model performance improved (Supplementary\u0026nbsp;Fig. 5 (b)). The recalibrated model yielded an AUC of 0.715 (95% CI: 0.649\u0026ndash;0.780) for distinguishing fast-decliners from slow-decliners.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSubsequent follow-up for cognitive decline\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe analyzed follow-up data to determine whether slow-decliners maintained slower MMSE decline rates beyond 2 years. Of the 251 early AD/aMCI participants, 173 (68.9%) completed three-year follow-up, 124 (49.4%) four-year, 81 (32.3%) five-year follow-up, and 28 (11.2%) six-year, respectively. Supplementary Table 5 shows that slow-decliners exhibited a slower MMSE decline rate (1.6 points/year between the second and fifth follow-up) compared to fast-decliners (1.9 points/year). Disease duration did not differ significantly between groups (3.7 years for fast-decliners vs. 3.3 years for slow-decliners) (Table 2).\u0026nbsp;\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study assessed cognitive trajectories in 251 early AD/aMCI convertors using GBTM based on MMSE scores. Two distinct cognitive trajectories were identified: 37.5% were fast-decliners, while 62.5% were slow-decliners. This finding is consistent with previous research showing more gradual cognitive decline in clinical settings[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Our study advances prior work by integrating clinical measures and genetic predictors into a prediction model for rapid cognitive decline in early AD/aMCI. This model was externally validated in a Japanese cohort, addressing limitations of earlier models lacking external validation or genetic data.\u003c/p\u003e \u003cp\u003eFast decliners had more \u003cem\u003eAPOE\u003c/em\u003e ε4 alleles, higher baseline CDR, worse cognitive performance, and more severe hallucinations than slow-decliners. These findings align with studies linking worse initial NPT scores, \u003cem\u003eAPOE\u003c/em\u003e ε4 alleles [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e11\u003c/span\u003e], impaired IADL [\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e42\u003c/span\u003e], and NPS like hallucinations to faster decline [\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e43\u003c/span\u003e]. Despite several significant predictors in univariable screening, only baseline MMSE, IADL, and \u003cem\u003eAPOE\u003c/em\u003e ε4 allele count remained significant in the final model. Other variables (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e) were excluded as they did not improve model fit based on BIC. MMSE, the primary predictor, already assesses key cognitive domains (memory, executive function, language, attention, and visuospatial skills). As a result, recall, verbal fluency, and 15-item mBNT were considered redundant. This overlap was evident in Pearson correlations with MMSE (recall: 0.43, verbal fluency: 0.30, 15-item mBNT: 0.24, all P\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e \u003cp\u003eOur study found that neither comorbidities nor years of education influenced the rate of cognitive decline (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). This contrasts with studies linking worse medical conditions and vascular risk factors to faster decline [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. The impact of education on cognitive decline rate remains debated likely due to heterogeneous design and variations in patient populations [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Additionally, we found age showed no association with cognitive decline, aligning with recent evidence that late-life cognitive loss reflects pathological rather than normative aging [\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e44\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eMMSE is central to cognitive assessment, yet in aMCI stage, mild functional impairments emerge [\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e45\u003c/span\u003e, \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e46\u003c/span\u003e]. IADL decline often precedes cognitive deterioration, particularly in those receiving cholinesterase inhibitor with MMSE scores of 20\u0026ndash;26 [\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e42\u003c/span\u003e]. Worse initial IADL may indicate reduced awareness of cognitive decline, delaying interventions or lifestyle modifications. This may explain why worse initial IADL were associated with faster decline in our study. Our findings highlight the need for routine functional assessments in early AD/aMCI. \u003cem\u003eAPOE\u003c/em\u003e ε4 genotyping provides clinical insights, having been recognized as a major AD risk gene for over two decades. While \u003cem\u003eAPOE\u003c/em\u003e ε4 increases AD risk and results in earlier disease onset [\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e47\u003c/span\u003e], its impact on cognitive decline remains inconsistent in the literature. Some studies link \u003cem\u003eAPOE\u003c/em\u003e ε4 to faster decline [\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e48\u003c/span\u003e, \u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e49\u003c/span\u003e], whereas others report no effect [\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e50\u003c/span\u003e]. These discrepancies may result from variations in baseline conditions and follow-up periods.\u003c/p\u003e \u003cp\u003ePredicting cognitive trajectories, particularly diagnostic conversion from healthy to MCI or AD, remains challenging due to variations in study designs and data types [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. Wu et al. used latent class growth analysis to identify cognitive decline trajectories in a Han Chinese population but achieved only modest performance (AUC\u0026thinsp;=\u0026thinsp;0.66, sensitivity\u0026thinsp;=\u0026thinsp;0.73, specificity\u0026thinsp;=\u0026thinsp;0.58), with limited generalizability due to the lack of independent validation. In contrast, Bhagwat et al. used ML with imaging data in Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset, achieving higher performance (AUC\u0026thinsp;=\u0026thinsp;0.97). While our model performed lower than Bhagwat et al., incorporating imaging could improve accuracy but is constrained by cost and feasibility in clinical application.\u003c/p\u003e \u003cp\u003eOur extended follow-up data supports the hypothesis that fast-decliners and slow-decliners represent distinct trajectory groups, rather than different disease stages (Supplementary Fig.\u0026nbsp;6). Viewing slow-decliners as an earlier stage of the fast-decliner trajectory does not fully explain the persistent difference in MMSE decline rates. Additionally, Supplementary Fig.\u0026nbsp;6 clarifies why lower baseline MMSE scores were associated with more rapid cognitive decline.\u003c/p\u003e \u003cp\u003eThis study has three key strengths. First, it is the first clinic-based prediction model for cognitive decline in East Asian individuals with early AD/aMCI, featuring external validation and addressing the limitations of previous models that lacked validation. Second, to reduce heterogeneity in the aMCI group, we included only individuals who converted to AD within two years. Additionally, given that cognitive decline in AD is slower in the early stages [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e], we restricted the development cohort to early AD/aMCI cases with MMSE\u0026thinsp;\u0026ge;\u0026thinsp;20 to minimize variability. Third, extended follow-up data enhance the reliability of findings. Our results support the hypothesis that fast- and slow-decliners represent distinct trajectory groups rather than different disease stages. However, two limitations existed. First, generalizability may be limited due to the clinic-based cohort and relatively small number of aMCI participants. Second, while external validation was performed, the lack of chronic disease health assessments in the Japanese cohort prevented full validation of the trajectory model, potentially affecting broader applicability.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis study presents a clinically applicable tool with good accuracy for predicting rapid cognitive decline in individuals with early AD/aMCI. MMSE, IADL, and \u003cem\u003eAPOE\u003c/em\u003e \u0026epsilon;4 genotype emerged as key predictors, all readily available in dementia clinics. Developed and validated in two independent East Asian cohorts, this model provides valuable insights into future cognitive progression, supporting proactive healthcare planning for clinicians and patients.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch3\u003eEthics approval\u003c/h3\u003e\n\u003cp\u003eThis study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Boards of TVGH and TCVGH. Validation data from the NCGG Memory Clinic were used with approval from the NCGG Ethics Committee (Approval No. 1611).\u0026nbsp;\u003c/p\u003e\n\u003ch3\u003eInformed consent\u003c/h3\u003e\n\u003cp\u003eWritten informed consent was obtained from all participants, and their anonymity was maintained according to the approved protocols.\u003c/p\u003e\n\u003ch3\u003eCompeting interests\u003c/h3\u003e\n\u003cp\u003eThe authors declare no conflict of interest.\u003c/p\u003e\n\u003ch3\u003eConsent for publication\u003c/h3\u003e\n\u003cp\u003eAll authors read and approved the manuscript for submission and gave consent for publication.\u003c/p\u003e\n\u003ch3\u003eAvailability of data and materials\u003c/h3\u003e\n\u003cp\u003eThe datasets analyzed and R\u0026nbsp;code are available from the corresponding author upon reasonable request.\u003c/p\u003e\n\u003ch3\u003eFunding sources\u003c/h3\u003e\n\u003cp\u003eThe work was supported by grants from Academia Sinica of Taiwan (AS-BD-108-2; PH-111-GP-08,\u0026nbsp;AS-KPQ-111-KNT); the Ministry of Science and Technology of Taiwan (112-2314-B-075-036-MY2, 112-2634-F-A49-003-,112-2321-B-A49-021-, and 112-2321-B-001-008-); Taipei Veterans General Hospital (V113C-047); Taichung Veterans General Hospital (TCVGH-1043402C, TCVGH-1053403C, and TCVGH-1063403C); The Brain Research Center, National Yang Ming University, from the Featured Areas Research Center Program within the framework of the Higher Education Sprout Project by the Ministry of Education in Taiwan; the National Health Research Institutes of Taiwan (PH-108-SP-01, PH-109-SP-01, PH-110-SP-01, PH-111-SP-01, and PH-112-SP-01); the Ministry of Science and Technology of Taiwan (108-2314-B-400-038, 109- 2740-B-400-002, 109-2314-B-400-014, 110-2740-B-400-002, 110-2314-B- 400-009, 111-2740-B-400-002, NSTC 112-2740-B-400-005, and NSTC 113-2740-B-400-005); the Research Funding for Longevity Sciences from the\u0026nbsp;National Center for Geriatrics and Gerontology (22-2 and 22-23). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.\u003c/p\u003e\n\u003ch3\u003eAuthors\u0026apos; contributions\u003c/h3\u003e\n\u003cp\u003eSJ Wang, JL Fuh, CA Hsiung, IS Chang, and CC Hsu designed the study. SJ Wang, JL Fuh, WJ Lee, and YC Liao collected data in Taiwan. K Uchida, T Sugimoto, Y Kuroda, T Sakurai, and H Arai collected validation data in Japan. JL Fuh, YS Lin, and YH Fang searched the literature. YH Fang, TJ Hsieh, and TY Chen analyzed the data. JL Fuh, CA Hsiung, YS Lin, and YH Fang interpreted the data. YH Fang, YS Lin, JL Fuh, and CA Hsiung wrote, reviewed, and/or revised the manuscript.\u003c/p\u003e\n\u003ch3\u003eAcknowledgments\u003c/h3\u003e\n\u003cp\u003eThe authors thank the Taiwan Biobank and Academia Sinica of Taiwan for their support for this work.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eLivingston G, Sommerlad A, Orgeta V, Costafreda SG, Huntley J, Ames D, Ballard C, Banerjee S, Burns A, Cohen-Mansfield J, Cooper C, Fox N, Gitlin LN, Howard R, Kales HC, Larson EB, Ritchie K, Rockwood K, Sampson EL, Samus Q, Schneider LS, Selbaek G, Teri L, Mukadam N (2017) Dementia prevention, intervention, and care. Lancet 390:2673-2734. https://doi.org/10.1016/S0140-6736(17)31363-6\u003c/li\u003e\n\u003cli\u003eHan L, Cole M, Bellavance F, McCusker J, Primeau F (2000) Tracking cognitive decline in Alzheimer\u0026apos;s disease using the mini-mental state examination: a meta-analysis. Int Psychogeriatr 12:231-247. https://doi.org/10.1017/s1041610200006359\u003c/li\u003e\n\u003cli\u003eLeoutsakos JM, Forrester SN, Corcoran CD, Norton MC, Rabins PV, Steinberg MI, Tschanz JT, Lyketsos CG (2015) Latent classes of course in Alzheimer\u0026apos;s disease and predictors: the Cache County Dementia Progression Study. Int J Geriatr Psychiatry 30:824-832. https://doi.org/10.1002/gps.4221\u003c/li\u003e\n\u003cli\u003eHaaksma ML, Calder\u0026oacute;n-Larra\u0026ntilde;aga A, Olde Rikkert MGM, Melis RJF, Leoutsakos J-MS (2018) Cognitive and functional progression in Alzheimer disease: A prediction model of latent classes. International Journal of Geriatric Psychiatry 33:1057-1064. https://doi.org/10.1002/gps.4893\u003c/li\u003e\n\u003cli\u003eXie SX, Ewbank DC, Chittams J, Karlawish JH, Arnold SE, Clark CM (2009) Rate of decline in Alzheimer disease measured by a Dementia Severity Rating Scale. Alzheimer Dis Assoc Disord 23:268-274. https://doi.org/10.1097/WAD.0b013e318194a324\u003c/li\u003e\n\u003cli\u003eTschanz JT, Corcoran CD, Schwartz S, Treiber K, Green RC, Norton MC, Mielke MM, Piercy K, Steinberg M, Rabins PV, Leoutsakos J-M, Welsh-Bohmer KA, Breitner JCS, Lyketsos CG (2011) Progression of cognitive, functional, and neuropsychiatric symptom domains in a population cohort with Alzheimer dementia: the Cache County Dementia Progression study. Am J Geriatr Psychiatry 19:532-542. https://doi.org/10.1097/JGP.0b013e3181faec23\u003c/li\u003e\n\u003cli\u003eAndel R, Vigen C, Mack WJ, Clark LJ, Gatz M (2006) The effect of education and occupational complexity on rate of cognitive decline in Alzheimer\u0026apos;s patients. Journal of the International Neuropsychological Society 12:147-152.\u003c/li\u003e\n\u003cli\u003eScarmeas N, Hadjigeorgiou GM, Papadimitriou A, Dubois B, Sarazin M, Brandt J, Albert M, Marder K, Bell K, Honig LS, Wegesin D, Stern Y (2004) Motor signs during the course of Alzheimer disease. Neurology 63:975. https://doi.org/10.1212/01.WNL.0000138440.39918.0C\u003c/li\u003e\n\u003cli\u003eWilkosz PA, Seltman HJ, Devlin B, Weamer EA, Lopez OL, DeKosky ST, Sweet RA (2010) Trajectories of cognitive decline in Alzheimer\u0026apos;s disease. Int Psychogeriatr 22:281-290. https://doi.org/10.1017/s1041610209991001\u003c/li\u003e\n\u003cli\u003eRouse HJ, Small BJ, Schinka JA, Loewenstein DA, Duara R, Potter H (2021) Mild behavioral impairment as a predictor of cognitive functioning in older adults. International Psychogeriatrics 33:285-293. https://doi.org/Pii S1041610220000678 10.1017/S1041610220000678\u003c/li\u003e\n\u003cli\u003eCosentino S, Scarmeas N, Helzner E, Glymour MM, Brandt J, Albert M, Blacker D, Stern Y (2008) APOE \u0026epsilon;4 allele predicts faster cognitive decline in mild Alzheimer disease. Neurology 70:1842-1849. https://doi.org/10.1212/01.wnl.0000304038.37421.cc\u003c/li\u003e\n\u003cli\u003eLee WJ, Liao YC, Wang YF, Lin YS, Wang SJ, Fuh JL (2020) Summative Effects of Vascular Risk Factors on the Progression of Alzheimer Disease. J Am Geriatr Soc 68:129-136. https://doi.org/10.1111/jgs.16181\u003c/li\u003e\n\u003cli\u003eSolfrizzi V, Capurso C, D\u0026apos;Introno A, Colacicco AM, Santamato A, Ranieri M, Fiore P, Capurso A, Panza F (2008) Lifestyle-related factors in predementia and dementia syndromes. Expert Rev Neurother 8:133-158. https://doi.org/10.1586/14737175.8.1.133\u003c/li\u003e\n\u003cli\u003eMcKhann GM, Knopman DS, Chertkow H, Hyman BT, Jack CR, Jr., Kawas CH, Klunk WE, Koroshetz WJ, Manly JJ, Mayeux R, Mohs RC, Morris JC, Rossor MN, Scheltens P, Carrillo MC, Thies B, Weintraub S, Phelps CH (2011) The diagnosis of dementia due to Alzheimer\u0026apos;s disease: recommendations from the National Institute on Aging-Alzheimer\u0026apos;s Association workgroups on diagnostic guidelines for Alzheimer\u0026apos;s disease. Alzheimers Dement 7:263-269. https://doi.org/10.1016/j.jalz.2011.03.005\u003c/li\u003e\n\u003cli\u003eWinblad B, Palmer K, Kivipelto M, Jelic V, Fratiglioni L, Wahlund LO, Nordberg A, Backman L, Albert M, Almkvist O, Arai H, Basun H, Blennow K, de Leon M, DeCarli C, Erkinjuntti T, Giacobini E, Graff C, Hardy J, Jack C, Jorm A, Ritchie K, van Duijn C, Visser P, Petersen RC (2004) Mild cognitive impairment--beyond controversies, towards a consensus: report of the International Working Group on Mild Cognitive Impairment. J Intern Med 256:240-246. https://doi.org/10.1111/j.1365-2796.2004.01380.x\u003c/li\u003e\n\u003cli\u003eWechsler D (1997) WMS-III: Wechsler Memory Scale Administration and Scoring Manual. Psychological Corporation.\u003c/li\u003e\n\u003cli\u003ePerneczky R, Wagenpfeil S, Komossa K, Grimmer T, Diehl J, Kurz A (2006) Mapping scores onto stages: mini-mental state examination and clinical dementia rating. Am J Geriatr Psychiatry 14:139-144. https://doi.org/10.1097/01.JGP.0000192478.82189.a8\u003c/li\u003e\n\u003cli\u003eMorris JC (1993) The Clinical Dementia Rating (CDR): current version and scoring rules. Neurology 43:2412-2414. https://doi.org/10.1212/wnl.43.11.2412-a\u003c/li\u003e\n\u003cli\u003eKatz S, Ford AB, Moskowitz RW, Jackson BA, Jaffe MW (1963) Studies of illness in the aged: the index of ADL: a standardized measure of biological and psychosocial function. Jama 185:914-919. https://doi.org/10.1001/jama.1963.03060120024016\u003c/li\u003e\n\u003cli\u003eLawton MP, Brody EM (1969) Assessment of older people: self-maintaining and instrumental activities of daily living. The gerontologist 9:179-186.\u003c/li\u003e\n\u003cli\u003eVanderploeg RD, Schinka JA, Jones T, Small BJ, Graves AB, Mortimer JA (2000) Elderly norms for the Hopkins Verbal Learning Test-Revised. Clin Neuropsychol 14:318-324. https://doi.org/10.1076/1385-4046(200008)14:3;1-P;FT318\u003c/li\u003e\n\u003cli\u003eMack WJ, Freed DM, Williams BW, Henderson VW (1992) Boston Naming Test: shortened versions for use in Alzheimer\u0026apos;s disease. J Gerontol 47:P154-158. https://doi.org/10.1093/geronj/47.3.p154\u003c/li\u003e\n\u003cli\u003eHarrison JE, Buxton P, Husain M, Wise R (2000) Short test of semantic and phonological fluency: normal performance, validity and test-retest reliability. Br J Clin Psychol 39:181-191. https://doi.org/10.1348/014466500163202\u003c/li\u003e\n\u003cli\u003eHumpstone HJ (1919) Memory Span Tests. Psychol Clin 12:196-200.\u003c/li\u003e\n\u003cli\u003eYesavage JA, Sheikh JI (1986) 9/Geriatric Depression Scale (GDS). Clinical Gerontologist 5:165-173. https://doi.org/10.1300/J018v05n01_09\u003c/li\u003e\n\u003cli\u003eKaufer DI, Cummings JL, Ketchel P, Smith V, MacMillan A, Shelley T, Lopez OL, DeKosky ST (2000) Validation of the NPI-Q, a brief clinical form of the Neuropsychiatric Inventory. J Neuropsychiatry Clin Neurosci 12:233-239. https://doi.org/10.1176/jnp.12.2.233\u003c/li\u003e\n\u003cli\u003eCui K, Zhao RH, Ding N, Xu XN, Gu WQ, He J, Ge XY (2025) The trajectories of depression and multimorbidity affect the patterns of cognitive decline: a prospective cohort study. Age Ageing 54. https://doi.org/ARTN afaf036 10.1093/ageing/afaf036\u003c/li\u003e\n\u003cli\u003eTaniguchi Y, Kitamura A, Murayama H, Amano H, Shinozaki T, Yokota I, Seino S, Nofuji Y, Nishi M, Yokoyama Y, Matsuyama Y, Fujiwara Y, Shinkai S (2017) Mini-Mental State Examination score trajectories and incident disabling dementia among community-dwelling older Japanese adults. Geriatr Gerontol Int 17:1928-1935. https://doi.org/10.1111/ggi.12996\u003c/li\u003e\n\u003cli\u003eWu YF, Jia MN, Xiang CY, Lin SW, Jiang ZQ, Fang Y (2022) Predicting the long-term cognitive trajectories using machine learning approaches: A Chinese nationwide longitudinal database. Psychiat Res 310. https://doi.org/ARTN 114434 10.1016/j.psychres.2022.114434\u003c/li\u003e\n\u003cli\u003eUchida K, Sugimoto T, Tange C, Nishita Y, Shimokata H, Saji N, Kuroda Y, Matsumoto N, Kishino Y, Ono R, Akisue T, Otsuka R, Sakurai T (2024) Association between abdominal adiposity and cognitive decline in older adults: a 10-year community-based study. J Nutr Health Aging 28. https://doi.org/ARTN 100175 10.1016/j.jnha.2024.100175\u003c/li\u003e\n\u003cli\u003eForbes M, Lotfaliany M, Mohebbi M, Reynolds CF, Woods RL, Orchard S, Chong T, Agustini B, O\u0026apos;Neil A, Ryan J, Berk M (2024) Depressive symptoms and cognitive decline in older adults. International Psychogeriatrics. https://doi.org/10.1017/S1041610224000541\u003c/li\u003e\n\u003cli\u003eHeinze G, Dunkler D (2017) Five myths about variable selection. Transpl Int 30:6-10. https://doi.org/10.1111/tri.12895\u003c/li\u003e\n\u003cli\u003eHoffmann VS, Sch\u0026ouml;necker S, Amin M, Reidler P, Brauer A, Kopczak A, Wunderlich S, Poli S, Althaus K, M\u0026uuml;ller S, Mansmann U, Kellert L (2024) A novel prediction score determining individual clinical outcome 3 months after juvenile stroke (PREDICT-score). J Neurol. https://doi.org/10.1007/s00415-024-12552-5\u003c/li\u003e\n\u003cli\u003eLeDell E, Petersen M, van der Laan M (2015) Computationally efficient confidence intervals for cross-validated area under the ROC curve estimates. Electron J Stat 9:1583-1607. https://doi.org/10.1214/15-Ejs1035\u003c/li\u003e\n\u003cli\u003eDal Pozzolo A, Caelen O, Johnson RA, Bontempi G (2015) Calibrating Probability with Undersampling for Unbalanced Classification. 2015 Ieee Symposium Series on Computational Intelligence (Ieee Ssci):159-166. https://doi.org/10.1109/Ssci.2015.33\u003c/li\u003e\n\u003cli\u003eJones BL, Nagin DS, Roeder K (2001) A SAS Procedure Based on Mixture Models for Estimating Developmental Trajectories. Sociological Methods \u0026amp; Research 29:374-393. https://doi.org/10.1177/0049124101029003005\u003c/li\u003e\n\u003cli\u003eTeam RC (2020) R: A language and environment for statistical computing. In. R Foundation for Statistical Computing, Vienna, Austria.\u003c/li\u003e\n\u003cli\u003eHu B, Shao J, Palta M (2006) Pseudo-R 2 in logistic regression model. Statistica Sinica:847-860.\u003c/li\u003e\n\u003cli\u003eMenze BH, Kelm BM, Masuch R, Himmelreich U, Bachert P, Petrich W, Hamprecht FA (2009) A comparison of random forest and its Gini importance with standard chemometric methods for the feature selection and classification of spectral data. Bmc Bioinformatics 10. https://doi.org/Artn 213 10.1186/1471-2105-10-213\u003c/li\u003e\n\u003cli\u003eWu YF, Wang X, Gu CM, Zhu JM, Fang Y (2023) Investigating predictors of progression from mild cognitive impairment to Alzheimer\u0026apos;s disease based on different time intervals. Age Ageing 52. https://doi.org/ARTN afad182 10.1093/ageing/afad182\u003c/li\u003e\n\u003cli\u003eWinter A, Aberle DR, Hsu W (2019) External validation and recalibration of the Brock model to predict probability of cancer in pulmonary nodules using NLST data. Thorax 74:551-563. https://doi.org/10.1136/thoraxjnl-2018-212413\u003c/li\u003e\n\u003cli\u003eWattmo C, Wallin \u0026Aring;K, Minthon L (2013) Progression of mild Alzheimer\u0026rsquo;s disease: knowledge and prediction models required for future treatment strategies. Alzheimer\u0026apos;s Research \u0026amp; Therapy 5:44. https://doi.org/10.1186/alzrt210\u003c/li\u003e\n\u003cli\u003eWilson RS, Gilley DW, Bennett DA, Beckett LA, Evans DA (2000) Hallucinations, delusions, and cognitive decline in Alzheimer\u0026apos;s disease. J Neurol Neurosurg Psychiatry 69:172-177. https://doi.org/10.1136/jnnp.69.2.172\u003c/li\u003e\n\u003cli\u003eWilson RS, Wang TH, Yu L, Bennett DA, Boyle PA (2020) Normative Cognitive Decline in Old Age. Ann Neurol 87:816-829. https://doi.org/10.1002/ana.25711\u003c/li\u003e\n\u003cli\u003eBrown PJ, Devanand DP, Liu X, Caccappolo E (2011) Functional impairment in elderly patients with mild cognitive impairment and mild Alzheimer disease. Arch Gen Psychiatry 68:617-626. https://doi.org/10.1001/archgenpsychiatry.2011.57\u003c/li\u003e\n\u003cli\u003eMontero-Odasso M, Speechley M, Muir-Hunter SW, Pieruccini-Faria F, Sarquis-Adamson Y, Hachinski V, Bherer L, Borrie M, Wells J, Garg AX, Tian Q, Ferrucci L, Bray NW, Cullen S, Mahon J, Titus J, Camicioli R (2020) Dual decline in gait speed and cognition is associated with future dementia: evidence for a phenotype. Age Ageing 49:995-1002. https://doi.org/10.1093/ageing/afaa106\u003c/li\u003e\n\u003cli\u003eFarrer LA, Cupples LA, Haines JL, Hyman B, Kukull WA, Mayeux R, Myers RH, Pericak-Vance MA, Risch N, van Duijn CM (1997) Effects of age, sex, and ethnicity on the association between apolipoprotein E genotype and Alzheimer disease. A meta-analysis. APOE and Alzheimer Disease Meta Analysis Consortium. Jama 278:1349-1356.\u003c/li\u003e\n\u003cli\u003eO\u0026apos;Hara R, Sommer B, Way N, Kraemer HC, Taylor J, Murphy G (2008) Slower speed-of-processing of cognitive tasks is associated with presence of the apolipoprotein epsilon4 allele. J Psychiatr Res 42:199-204. https://doi.org/10.1016/j.jpsychires.2006.12.001\u003c/li\u003e\n\u003cli\u003eHayden KM, Reed BR, Manly JJ, Tommet D, Pietrzak RH, Chelune GJ, Yang FM, Revell AJ, Bennett DA, Jones RN (2011) Cognitive decline in the elderly: an analysis of population heterogeneity. Age Ageing 40:684-689. https://doi.org/10.1093/ageing/afr101\u003c/li\u003e\n\u003cli\u003eKleiman T, Zdanys K, Black B, Rightmer T, Grey M, Garman K, Macavoy M, Gelernter J, van Dyck C (2006) Apolipoprotein E epsilon4 allele is unrelated to cognitive or functional decline in Alzheimer\u0026apos;s disease: retrospective and prospective analysis. Dement Geriatr Cogn Disord 22:73-82. https://doi.org/10.1159/000093316 \u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTable 1 Differences in baseline demographics between early AD and aMCI in development cohort (N = 251) and the validation cohort (N = 322)\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"935\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 255px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"7\" valign=\"top\" style=\"width: 539px;\"\u003e\n \u003cp\u003eDevelopment cohort\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 142px;\"\u003e\n \u003cp\u003eValidation\u0026nbsp;cohort\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 255px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 151px;\"\u003e\n \u003cp\u003eTotal (n = 251)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 151px;\"\u003e\n \u003cp\u003eEarly AD (n = 209)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 151px;\"\u003e\n \u003cp\u003eaMCI (n = 42)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 142px;\"\u003e\n \u003cp\u003eEarly AD (n = 322)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 255px;\"\u003e\n \u003cp\u003eContinuous variable\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003eMean\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eSD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eMean\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eSD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eMean\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eSD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003eP-value\u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003eMean\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003eSD\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eAge, years\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e78.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e7.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e79.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e6.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e74.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e8.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026lt;.001*\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e77.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e5.6\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eEducation, years\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e10.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e4.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e10.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e4.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e11.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e4.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.601 \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e10.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e2.4\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eDisease duration, years\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e2.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e2.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.055 \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e-\u003csup\u003eb\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eMMSE score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e23.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e2.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e22.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e2.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e25.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e2.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026lt;.001*\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e22.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e2.1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eBMI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e23.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e24.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e23.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.412 \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e22.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e3.5\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eExercise habits, days/week\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e4.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e4.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e4.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.855 \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eGDS-15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.251 \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e2.7\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eHealth status\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e1.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e1.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e1.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e1.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e1.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e1.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.845 \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eCategorical variable\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003eNumber\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003eNumber\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003eNumber\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003eNumber\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eSex, male\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e136\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e53.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e116\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e55.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e45.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.223 \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e97\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e30.1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eSmoking status, \u0026ge;5 cigarettes/day\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e2.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.999 \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eHypertension\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e147\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e58.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e122\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e59.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e25\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e59.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.999 \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e165\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e51.2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eDiabetes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e63\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e25.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e53\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e25.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e23.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.934 \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e47\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e14.6\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003eCardiovascular disease\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e66\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e26.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e55\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e26.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e26.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.999 \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e56\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e17.4\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 255px;\"\u003e\n \u003cp\u003e\u003cem\u003eAPOE\u003c/em\u003e \u0026epsilon;4 allele 0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e163\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e64.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e140\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e67.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e23\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e54.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.193 \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e167\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e51.9\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 255px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e79\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e31.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e63\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e30.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e38.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e106\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e32.9\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 255px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e2.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e7.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e26\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e8.1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 255px;\"\u003e\n \u003cp\u003eGlobal CDR \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; 0.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e89\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e35.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e47\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e22.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e42\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e100.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026lt;.001*\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e4.3\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 255px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e162\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e64.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e162\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e77.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 66px;\"\u003e\n \u003cp\u003e0.6\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003csup\u003ea\u003c/sup\u003eChi-square test or t-test was used to assess differences in demographic characteristics between early AD and aMCI groups.\u003c/p\u003e\n\u003cp\u003e\u003csup\u003eb\u003c/sup\u003e\u0026ldquo;-\u0026rdquo; indicates that variable information is\u0026nbsp;not\u0026nbsp;available.\u003c/p\u003e\n\u003cp\u003e*P-value \u0026lt; 0.05.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAbbreviations: AD, Alzheimer\u0026rsquo;s disease; aMCI, amnestic mild cognitive impairment; CDR, Clinical Dementia Rating; GDR-15, 15-item Geriatric Depression Scale; MMSE, Mini-Mental State Examination.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTable 2 Comparison of two cognitive trajectory groups on baseline variables in the development cohort (N = 251)\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eG1 (n = 94)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eG2 (n = 157)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 71px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eDemographics\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eMean (SD) or N (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eMean (SD) or N (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 71px;\"\u003e\n \u003cp\u003eP-value\u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eAge, years\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e79.0 (7.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e77.8 (7.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.231\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eSex, male\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e54 (57.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e81 (51.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.442\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 264px;\"\u003e\n \u003cp\u003eEducation, years\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e10.7 (4.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e11.0 (4.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.674\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 264px;\"\u003e\n \u003cp\u003eLifestyle factors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eBMI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e23.6 (3.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e24.1 (3.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.299\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 264px;\"\u003e\n \u003cp\u003eExercise habits, days/week\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e4.0(3.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e4.1(2.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.819\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eSmoking status, \u0026ge;5 cigarettes/day\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e3 (3.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003e5 (3.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 71px;\"\u003e\n \u003cp\u003e0.999\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 264px;\"\u003e\n \u003cp\u003eClinical characteristics\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eBaseline diagnosis \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; AD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e88 (93.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e121 (77.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.001**\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eaMCI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e6 (6.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e36 (22.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eGlobal CDR \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; 0.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e18 (19.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003e71 (45.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e\u0026lt;.001**\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e76 (80.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003e86 (54.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003e\u003cem\u003eAPOE\u003c/em\u003e \u0026epsilon;4 allele \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; 0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e49 (52.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003e114 (72.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.001**\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; 1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e40 (42.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003e39 (24.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 71px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e5 (5.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003e4 (2.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 71px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eDisease duration, years\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e3.7 (3.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e3.3 (3.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.364 \u0026nbsp; \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 553px;\"\u003e\n \u003cp\u003eComorbidities\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eHealth status\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e2.0 (1.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003e1.7 (1.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 71px;\"\u003e\n \u003cp\u003e0.118\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eHypertension\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e56 (59.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003e91 (59.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 71px;\"\u003e\n \u003cp\u003e0.999\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eDiabetes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e26 (28.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003e37 (24.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 71px;\"\u003e\n \u003cp\u003e0.592\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eCardiovascular disease\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e27 (28.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003e39 (25.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 71px;\"\u003e\n \u003cp\u003e0.660\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 553px;\"\u003e\n \u003cp\u003eFunctional capacities\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eADL score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e2.5 (3.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e1.0 (2.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e\u0026lt;.001**\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eIADL score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e7.5 (6.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e4.0 (4.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e\u0026lt;.001**\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 553px;\"\u003e\n \u003cp\u003eNeuropsychological tests\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eGDS-15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e3.1 (3.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003e3.5 (3.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 71px;\"\u003e\n \u003cp\u003e0.260\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eMMSE score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e21.8 (1.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e24.2 (2.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e\u0026lt;.001**\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eRecall (after 15 min)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e1.2 (1.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e2.6 (2.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e\u0026lt;.001**\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eForward digit span\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e9.2 (2.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e9.7 (2.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.102\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eBackward digit span\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e4.7 (1.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e5.3 (2.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.025* \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eVerbal fluency\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e7.2 (2.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e8.4 (2.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e\u0026lt;.001**\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003emBNT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e12.3 (2.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e13.2 (1.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.001**\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\" valign=\"top\" style=\"width: 553px;\"\u003e\n \u003cp\u003eSeverity of NPS (Frequency \u0026times; Intensity)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eDelusions\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e0.5 (0.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e0.4 (0.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.201\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eHallucinations\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e0.3 (0.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e0.1 (0.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.002**\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eAgitation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e0.5 (0.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e0.4 (0.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.100\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eDepression\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e0.6 (0.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e0.6 (0.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.982\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eAnxiety\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e0.6 (0.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e0.5 (0.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.250\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eEuphoria\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e0.2 (0.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e0.1 (0.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.380\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eApathy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e0.6 (0.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e0.4 (0.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.062\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eDisinhibition\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e0.4 (0.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e0.4 (0.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.656\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eIrritability\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e0.7 (0.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e0.6 (0.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.502\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eMotor disturbance\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e0.4 (0.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e0.2 (0.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.005* \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eNight-time behaviors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e0.4 (0.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e0.4 (0.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.706\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 264px;\"\u003e\n \u003cp\u003eAppetite\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 113px;\"\u003e\n \u003cp\u003e0.3 (0.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e0.4 (0.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 71px;\"\u003e\n \u003cp\u003e0.419\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003csup\u003ea\u003c/sup\u003eP-value for\u0026nbsp;Chi-square test or t-test was used\u0026nbsp;to assess differences in variables between fast-decliners (G1)\u0026nbsp;and\u0026nbsp;slow-decliners\u0026nbsp;(G2).\u003c/p\u003e\n\u003cp\u003e*P-value\u0026nbsp;\u0026lt; 0.05.\u003c/p\u003e\n\u003cp\u003e**\u0026nbsp;P-value \u0026lt; 0.002 (Bonferroni correction for 30 comparisons).\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eAbbreviations: SD, standard deviation; CDR, Clinical Dementia Rating; GDR-15, 15-item Geriatric Depression Scale; IADL, Instrumental Activities of Daily Living; mBNT, 15-item modified Boston Naming Test; MMSE, Mini-Mental State Examination; NPS, neuropsychiatric symptoms.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 3 Odds ratios of predictors in the logistic regression model for rapid cognitive decline in the development cohort (N = 251)\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"925\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 198px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003eOR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e95% CI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003eP-value\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003eOR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e95% CI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003eP-value\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003eOR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e95% CI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003eP-value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 198px;\"\u003e\n \u003cp\u003ePredictor\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" style=\"width: 236px;\"\u003e\n \u003cp\u003eModel 1\u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" style=\"width: 246px;\"\u003e\n \u003cp\u003eModel 2\u003csup\u003eb\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" style=\"width: 246px;\"\u003e\n \u003cp\u003eModel 3\u003csup\u003ec\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 198px;\"\u003e\n \u003cp\u003e\u003cem\u003eAPOE\u003c/em\u003e \u0026epsilon;4 allele count\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e2.70\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e1.45-5.18\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.002*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e3.14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e1.72\u0026ndash;5.91\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026lt;.001*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e2.85\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e1.57\u0026ndash;5.34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026lt;.001*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 198px;\"\u003e\n \u003cp\u003eMMSE score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e0.58\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e0.47-0.70\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026lt;.001*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e0.56\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e0.46\u0026ndash;0.66\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026lt;.001*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e0.54\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e0.45\u0026ndash;0.65\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026lt;.001*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 198px;\"\u003e\n \u003cp\u003eIADL score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e1.08\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e1.00-1.17\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.004*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e1.13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e1.07\u0026ndash;1.21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026lt;.001*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e1.13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e1.06\u0026ndash;1.21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026lt;.001*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 198px;\"\u003e\n \u003cp\u003eADL score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e1.04\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e0.91-1.21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.586 \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 198px;\"\u003e\n \u003cp\u003eRecall (after 15 min)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e0.92\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e0.77-1.08\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.302 \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 198px;\"\u003e\n \u003cp\u003eVerbal fluency\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e0.97\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e0.85-1.10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.593 \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 198px;\"\u003e\n \u003cp\u003e15-item mBNT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e0.88\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e0.74-1.04\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.140 \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 198px;\"\u003e\n \u003cp\u003eHallucinations\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e2.41\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e0.88-7.12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e0.096 \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 104px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 85px;\"\u003e\n \u003cp\u003e- \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eNote: The analysis included the full sample of 94 fast-decliners (G1) and 157 slow-decliners (G2) from the development cohort.\u003c/p\u003e\n\u003cp\u003e\u003csup\u003ea\u003c/sup\u003eModel 1: A logistic regression model for rapid cognitive decline constructed using functional capacity measures (ADL and IADL scores), neuropsychological tests (MMSE score, recall test, verbal fluency test, and mBNT), neuropsychiatric symptoms (hallucinations), and \u003cem\u003eAPOE\u003c/em\u003e \u0026epsilon;4 allele count.\u003c/p\u003e\n\u003cp\u003e\u003csup\u003eb\u003c/sup\u003eModel 2: An unadjusted regression model that included only baseline MMSE, IADL scores, and \u003cem\u003eAPOE\u003c/em\u003e \u0026epsilon;4 allele count as predictors.\u003c/p\u003e\n\u003cp\u003e\u003csup\u003ec\u003c/sup\u003eModel 3: Derived from Model 2 with additional adjustments for age, sex, and education.\u003c/p\u003e\n\u003cp\u003e*P-value \u0026lt; 0.05.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAbbreviations: CI, Confidence Interval; IADL, Instrumental Activities of Daily Living; mBNT, modified Boston naming test; MMSE, Mini-Mental State Examination; NPS, neuropsychiatric symptoms; NPT, neuropsychological tests; OR, Odds Ratio for fast-decliners.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Amnestic mild cognitive impairment, Cognitive function, Early stage of Alzheimer's disease, Risk prediction model, Trajectory analysis","lastPublishedDoi":"10.21203/rs.3.rs-8629600/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8629600/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eWhile cognitive decline is inevitable in early Alzheimer\u0026rsquo;s disease (AD) and amnestic mild cognitive impairment (aMCI), the rate of decline varies widely.\u003c/p\u003e\u003ch2\u003eAims\u003c/h2\u003e \u003cp\u003eThis study aimed to identify groups of individuals following similar trajectories of cognitive decline, and develop a model predicting those at risk of rapid decline.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eA longitudinal study of 251 clinic-based early AD/aMCI converters from Taiwan, with external validation in an independent Japanese cohort. Group-based trajectory modeling identified cognitive decline patterns using Mini-Mental State Examination (MMSE) scores. Baseline assessments included demographics, lifestyle factors, comorbidities, functional capacities, neuropsychiatric symptoms, neuropsychological tests, and clinical characteristics. Backward logistic regression was used to identify predictors of rapid cognitive decline.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eTwo cognitive trajectories were identified: a rapid decline group (two-year MMSE decline of 5.8\u0026thinsp;\u0026plusmn;\u0026thinsp;5.1 points) and a slower decline group (1.4\u0026thinsp;\u0026plusmn;\u0026thinsp;2.6 points). Baseline MMSE score, instrumental activities of daily living (IADL) score, and apolipoprotein E (\u003cem\u003eAPOE\u003c/em\u003e) ε4 allele count were significant predictors of rapid cognitive decline. The prediction model demonstrated good discrimination in the Taiwanese cohort (The area under the curve (AUC)\u0026thinsp;=\u0026thinsp;0.841, 95% confidence intervals (CI): 0.771\u0026ndash;0.892) with sensitivity of 0.769 and specificity of 0.728, and showed acceptable discrimination (AUC\u0026thinsp;=\u0026thinsp;0.715, 95% CI: 0.649\u0026ndash;0.780) with good recalibration in the Japanese validation cohort.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eThis is the first clinic-based prediction model for cognitive decline in East Asian early AD/aMCI with external validation. Baseline MMSE, IADL, and \u003cem\u003eAPOE\u003c/em\u003e genotype may assist clinicians in risk stratification, disease monitoring, and individualized care planning in aging populations.\u003c/p\u003e","manuscriptTitle":"Predicting Cognitive Decline in Early Alzheimer’s: Insights from East Asian Cohorts","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-03 15:47:30","doi":"10.21203/rs.3.rs-8629600/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"8ae930ac-5e9d-4063-a489-60e0b71112ce","owner":[],"postedDate":"February 3rd, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-02-28T18:54:11+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-03 15:47:30","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8629600","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8629600","identity":"rs-8629600","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}