Depressive Symptoms as a Modifiable Risk Factor for Cognitive Impairment: Severity-Graded Associations and Clinical Implications from a 9-Year Longitudinal Study in Chinese Older Adults

Depressive Symptoms as a Modifiable Risk Factor for Cognitive Impairment: Severity-Graded Associations and Clinical Implications from a 9-Year Longitudinal Study in Chinese Older Adults | 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 Depressive Symptoms as a Modifiable Risk Factor for Cognitive Impairment: Severity-Graded Associations and Clinical Implications from a 9-Year Longitudinal Study in Chinese Older Adults Qiang Zhang, Yifan Zhao, Peilu Zhang, Xinsong Ren, Dongwei Yan This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9460240/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 Depression and cognitive impairment often co-occur in older adults, but their clinical implications for the management of depression have not been fully explored. Determining whether the severity of depression can predict cognitive outcomes will help inform the development of tiered treatment strategies for depression in later life. Methods This study analyzed 31,570 individual follow-up records from 12,494 independent participants aged ≥ 45 years in the China Health and Retirement Longitudinal Study (CHARLS, 2011–2020). Depressive symptoms were assessed using the CESD-10 scale. We developed a machine learning-based risk stratification system to identify depressed individuals at highest cognitive risk, who are likely to benefit most from intensified interventions. Results Depressive symptoms (CESD-10 score ≥ 10) were significantly associated with cognitive impairment (OR = 2.07, 95% CI: 1.97–2.18), with incidence rates of cognitive impairment of 39.6% in the symptomatic group versus 24.1% in the asymptomatic group. A significant dose-response relationship was observed: statistically significant associations were found across all categories, normal mood (20.8%), mild symptoms (29.5%), moderate symptoms (36.6%), and severe symptoms (46.0%), with a trend P-value < 0.001. Even subclinical symptoms (CESD-10 score 6–10) significantly increased the risk. The quartile risk stratification system demonstrated excellent discriminatory power, with observed rates of cognitive impairment ranging from 4.6% (low risk) to 53.1% (high risk). Conclusion The severity of depressive symptoms is significantly and gradually associated with the risk of cognitive impairment, which has important implications for the clinical management of depression in older adults. The findings support the integration of cognitive function monitoring into the depression care system and the dynamic adjustment of intervention intensity based on depression severity and cognitive risk characteristics. Depressive Symptoms Cognitive Impairment Dose-Response Relationship Risk Stratification Modifiable Risk Factors Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 1. Introduction Late-life depression poses a significant clinical challenge, affecting approximately 10%-15% of older adults worldwide and associated with high morbidity, mortality, and healthcare resource utilization [ 1 , 2 ]. In addition to its direct impact on quality of life and functional status, emerging evidence suggests that depression may increase the risk of subsequent cognitive decline and dementia [ 3 , 4 ]. This potential association has profound implications for how clinicians manage older adults with depression. The relationship between depression and cognitive function in older adults is complex and may be bidirectional [ 5 ]. To explain how chronic depression accelerates cognitive aging, several pathophysiological mechanisms have been proposed, including hippocampal damage resulting from hypothalamic-pituitary-adrenal axis dysfunction, chronic systemic inflammation, cerebrovascular dysfunction, and reduced neurotrophic support [ 6 , 7 ]. Meta-analysis evidence suggests that late-life depression may increase the risk of dementia by approximately 1.5 to 2 times [ 3 , 4 ]. Crucially, unlike most risk factors for cognitive impairment (such as age, genetic factors, and educational level), depression can potentially be modulated through pharmacological treatment and psychological interventions [ 8 ]. However, several key issues remain inadequately addressed. First, existing evidence primarily originates from Western populations, and data from large-scale Chinese cohort studies remain insufficient, given the significant differences in cultural context, healthcare systems, and epidemiological characteristics compared to Western settings [ 9 ]. Second, although the association between depression and cognitive function has been documented in the literature, the dose-response relationship between depression severity and cognitive outcomes has not yet been systematically elucidated. Clarifying this gradient relationship is crucial for developing clinical recommendations for tiered diagnosis and treatment. Third, there remains a scarcity of guidelines for clinicians on how to incorporate cognitive risk factors into treatment plans when managing older adults with depression. This study aims to address these research gaps using data from the China Health and Retirement Longitudinal Study (CHARLS). Specific objectives include: (1) quantifying the association between depressive symptoms and cognitive impairment in a large, nationally representative Chinese cohort; (2) describing the dose-response relationship between depression severity and cognitive outcomes to inform tiered clinical management; (3) establishing a clinically applicable risk stratification system to identify which older adults with depression face the highest cognitive risk and are most in need of enhanced monitoring and intervention; and (4) exploring the implications of these findings for the clinical management of late-life depression. 2. Methods 2.1 Study Population Data were drawn from the China Health and Retirement Longitudinal Study (CHARLS), a nationally representative longitudinal survey of Chinese adults aged 45 and older [ 10 ]. CHARLS employed a multistage stratified probability proportional to size sampling strategy, covering 28 provinces, 150 counties and districts, and 450 villages and towns across China. The baseline survey was conducted in 2011, with follow-up surveys conducted in 2013, 2015, 2018, and 2020. The initial sample comprised 49,015 observations across all waves. After excluding participants with missing cognitive assessment data (n = 8,456), missing depression scores (n = 5,892), incomplete covariate data (n = 2,660), and outliers in age (n = 337), the final analysis sample comprised 31,570 individual-wave observations from 12,494 unique participants ( Supplementary Figure S1 ). The study protocol was approved by the Biomedical Ethics Review Committee of Peking University, and all participants provided written informed consent. 2.2 Measurement Instruments 2.2.1 Assessment of Depressive Symptoms Depressive symptoms were assessed using the Center for Epidemiological Studies Depression Scale (CES-D-10), a well-established tool for screening depression in older adults [ 11 ]. Participants reported the frequency of depressive symptoms over the past week using a 4-point scale (0 = rarely, 3 = most of the time). The total score ranged from 0 to 30 points, with higher scores indicating more severe depressive symptoms. In the preliminary analysis, clinically significant depressive symptoms were defined as a CESD-10 score ≥ 10, a threshold widely adopted in epidemiological studies [ 12 ]. The CESD-10 is used solely as a screening tool to assess the burden of depressive symptoms and does not constitute a basis for the clinical diagnosis of major depressive disorder; therefore, the term “depressive symptoms” is used throughout this paper. In the dose-response analysis, the severity of depressive symptoms was categorized as mild (0–5 points), moderate (6–10 points), severe (11–15 points), and very severe (> 15 points). To partially address the issue of persistent depression, we conducted a supplementary analysis distinguishing persistent depressive symptoms (CESD-10 scores ≥ 10 in two or more consecutive assessment waves) from single-episode symptoms ( Supplementary Table S9 ). 2.2.2 Cognitive Function Assessment Cognitive function was assessed using a comprehensive assessment scale adapted from the Health and Retirement Study (HRS). The assessment included (1) episodic memory (immediate and delayed word recall from a 10-word list, scored 0–20); (2) mental status (serial subtraction by 7, date orientation, and a drawing task, scored 0–11); and (3) executive function (naming task). The total cognitive score was calculated as the sum of the scores for each subscale (score range: 0–31). Cognitive impairment was defined as a score below the 31st percentile after adjustment for age and education level, consistent with previous findings from the CHARLS study [ 13 ]. This definition is based on epidemiological criteria derived from statistical distributions and does not correspond to clinical diagnostic criteria for mild cognitive impairment (MCI; e.g., the Petersen criteria) or dementia (e.g., the DSM-5 criteria). Sensitivity analyses using the 25th and 33rd percentiles yielded consistent results ( Supplementary Table S7 ). 2.2.3 Covariates Covariates included sociodemographic factors (age, sex, education, rural/urban residence, marital status), health behaviors (smoking, alcohol consumption, physical exercise, sleep duration), chronic conditions (hypertension, diabetes, heart disease, stroke, and chronic disease count), functional status (activities of daily living, instrumental activities of daily living, history of falls), physical measurements (body mass index, systolic and diastolic blood pressure, grip strength, walking speed), sensory function (vision, hearing), and self-rated health. Time-varying covariates were updated at each wave, and longitudinal features including years since baseline and total number of visits were incorporated. A complete list of 45 features is provided in Supplementary Table S2 . 2.3 Construction of the Risk Stratification Model To translate the association between depressive symptoms and cognitive function into clinically actionable guidance, the study employed an ensemble machine learning approach to construct a risk stratification model. This model integrated the predictive results of multiple algorithms (histogram-based gradient boosting, additional tree classifier, and random forest classifier) using bootstrap aggregation and performance-weighted averaging. Model hyperparameters are detailed in Supplementary Table S4 . A four-tier risk stratification system was established based on predicted probability thresholds: (1) low risk (predicted probability < 0.25); (2) moderate risk (0.25 ≤ predicted probability < 0.55); (3) high risk (predicted probability ≥ 0.55); (4) Uncertain (cognitive uncertainty ≥ 90th percentile). The “Uncertain” category refers to cases where there is significant disagreement among models; such cases require individualized clinical assessment rather than algorithm-driven recommendations. 2.4 Statistical Analysis The association between depressive symptoms and cognitive impairment was quantified using odds ratios (OR) and their 95% confidence intervals. The Cochran-Armitage test was used to evaluate the dose-response relationship across different severity levels of depression. Model performance was evaluated using 5-fold stratified cross-validation, with the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and Brier score as primary evaluation metrics. Model comparisons were performed using the DeLong test [ 14 ]. Receiver operating characteristic curve analysis was used to assess clinical utility at threshold probabilities [ 15 ]. Subgroup analyses were stratified by sex, age group (≤ 70 years vs. >70 years), and urban/rural residence, and interaction tests were performed. All analyses were conducted using Python 3.10 in conjunction with scikit-learn, NumPy, and SciPy. Statistical significance was defined as a two-sided P < 0.05. 3. Results 3.1 Sample Characteristics and Associations with Depression Table 1 presents the characteristics of the study population stratified by depression status. The analysis sample comprised 31,570 individual follow-up observations from 12,494 unique participants, with a mean age of 67.8 years (standard deviation = 6.1). Approximately 44.0% were women, 56.2% resided in rural areas, and the average years of education was 4.7. The prevalence of clinically significant depressive symptoms (CESD-10 score ≥ 10) was 34.2%. Participants with depressive symptoms differed significantly from those without depressive symptoms across multiple dimensions. Symptomatic individuals were more likely to be female (53.6% vs. 39.0%, P < 0.001), have lower educational attainment (3.8 years vs. 5.2 years, P < 0.001), were more likely to reside in rural areas (65.3% vs. 51.5%, P < 0.001), and had a higher prevalence of chronic diseases (0.9 vs. 0.8, P < 0.001). A key finding was that participants with depressive symptoms had a significantly higher prevalence of cognitive impairment (39.6% vs. 24.1%, P < 0.001), with an unadjusted odds ratio of 2.07 (95% CI: 1.97–2.18). These results indicate that depressive symptoms are associated with more than a twofold increased risk of cognitive impairment in this population. Table 1 Baseline characteristics of study participants stratified by depressive symptom status (n = 31,570 observations from 12,494 unique participants) Characteristic Total (n = 31,570) No Depression (n = 20,757) Depression (n = 10,813) P-value Demographics Age, years, mean ± SD 67.8 ± 6.1 67.8 ± 6.1 67.8 ± 6.0 < 0.001 Female, n (%) 13,890 (44.0) 8,092 (39.0) 5,798 (53.6) < 0.001 Education, years, mean ± SD 4.7 ± 4.4 5.2 ± 4.4 3.8 ± 4.2 < 0.001 Rural residence, n (%) 17,747 (56.2) 10,686 (51.5) 7,061 (65.3) < 0.001 Health Status Chronic diseases, n, mean ± SD 0.8 ± 0.9 0.8 ± 0.9 0.9 ± 1.0 < 0.001 CESD-10 score, mean ± SD 8.1 ± 6.2 4.4 ± 2.8 15.2 ± 4.5 < 0.001 Cognitive Outcomes Cognition score, mean ± SD 11.5 ± 3.6 12.0 ± 3.5 10.4 ± 3.6 < 0.001 Cognitive impairment, n (%) 9,286 (29.4) 4,999 (24.1) 4,287 (39.6) < 0.001 Note: Depression defined as CESD-10 ≥ 10. Data are from 12,494 unique participants across 5 survey waves (2011–2020). 3.2 Dose-Response Relationship Between Depression Severity and Cognitive Impairment Figure 1 illustrates the longitudinal cognitive trajectories and dose-response relationship between depression severity and cognitive impairment. Figure 1 A shows that, over the 9-year follow-up period, cognitive decline trajectories differed significantly according to baseline depression severity. The group with severe depressive symptoms (CESD-10 > 15) exhibited the most rapid cognitive decline, whereas the group with a normal mood (CESD-10 0–5) maintained relatively stable cognitive function. These trajectory differences suggest that depression may accelerate the process of cognitive aging, rather than merely reflecting coexisting cognitive impairment. Figure 1 B confirms a significant dose-response relationship between depression severity categories and the incidence of cognitive impairment. The incidence of cognitive impairment increased progressively with increasing severity: normal mood group (20.8%), mild symptom group (29.5%), moderate symptom group (36.6%), and severe symptom group (46.0%). The Cochran-Armitage test revealed a significant linear trend ( P < 0.001). For each increase in depression severity grade, the prevalence of cognitive impairment correspondingly increased by approximately 8–10 percentage points. Notably, even subclinical depressive symptoms (CESD-10 scores of 6–10, below the conventional threshold for clinical significance) were associated with a significantly increased risk of cognitive impairment (29.5% vs. 20.8%), suggesting that early intervention at the subclinical stage may be beneficial. Detailed dose-response statistics, including odds ratios, are presented in Supplementary Table S3. 3.3 Performance of Risk Stratification Models Table 2 presents a comparison of the performance of various machine learning models. The ensemble model achieved an AUC of 0.781 (95% confidence interval: 0.776–0.788), with a sensitivity of 73.5% and a specificity of 66.9%; a Brier score of 0.198 indicates good calibration. Its performance was comparable to or slightly better than that of the base models (logistic regression AUC = 0.781, random forest AUC = 0.780, gradient boosting AUC = 0.781). Cross-validation results ( Supplementary Figure S4 ) demonstrated good model stability, with a 5-fold cross-validation AUC of 0.779 ± 0.008. The calibration curve (Fig. 5 A) confirms good consistency between predicted probabilities and observed frequencies. Receiver operating characteristic (ROC) curve analysis (Fig. 5 B) shows that within the clinically relevant threshold probability range of 20%-50%, this strategy yields superior net benefit compared to the “treat all” and “do not treat” strategies. Table 2 Performance comparison of machine learning models for cognitive impairment prediction. Model AUC (95% CI) Sens Spec PPV NPV F1 Brier P* Ensemble 0.781 (0.776–0.788) 0.735 0.669 0.481 0.858 0.581 0.198 Ref Logistic Reg. 0.781 (0.770–0.782) 0.710 0.695 0.492 0.852 0.581 0.193 0.013 Random Forest 0.780 (0.766–0.778) 0.690 0.712 0.500 0.846 0.580 0.191 < 0.001 Gradient Boost 0.781 (0.771–0.783) 0.720 0.686 0.489 0.855 0.583 0.195 0.078 Note: Sens = Sensitivity; Spec = Specificity; PPV = Positive Predictive Value; NPV = Negative Predictive Value; Brier = Brier Score. *P-value from DeLong test comparing each model to the Ensemble model. 3.4 Clinical Risk Stratification of Elderly Patients with Depression Table 3 presents the validation results of the four-tier risk stratification. The distribution of risk categories is as follows: low risk (14.0%), moderate risk (42.3%), high risk (33.7%), and uncertain (10.0%). The observed incidence of cognitive impairment demonstrated excellent discriminatory power across risk strata: low-risk group (4.6%), moderate-risk group (19.3%), high-risk group (53.1%), and uncertain group (27.3%). The monotonically increasing trend from low to high risk (4.6% → 19.3% → 53.1%) confirms the clinical validity of the stratification system, with the difference in the incidence of cognitive impairment between the low-risk and high-risk groups exceeding a 10-fold increase (OR = 23.62). Notably, the severity of depression (mean CESD − 10) increased progressively with rising risk levels: 3.9 in the low-risk group, 6.2 in the moderate-risk group, 12.2 in the high-risk group, and 8.3 in the uncertain group, indicating that the burden of depressive symptoms is a key driver of cognitive risk classification. The “uncertain” category (accounting for 10.0% of cases) exhibited unique phenotypic characteristics (Fig. 3 ): while depression scores were elevated to a degree comparable to high-risk individuals, functional status paradoxically remained normal, suggesting complex clinical presentations that are difficult to classify using simple algorithms. The incidence of cognitive impairment in this group was moderate (27.3%), necessitating management through individualized expert assessment rather than standardized treatment protocols. Table 3 Risk stratification validation: observed cognitive impairment rates and clinical characteristics by risk stratum. Risk Stratum N (%) CI Cases (%) OR (vs Low) Mean Age Mean CESD-10 Mean Edu Low 4,418 (14.0%) 202 (4.6%) 1.00 (Ref) 64.9 3.9 2.8 Medium 13,367 (42.3%) 2,579 (19.3%) 4.99 66.7 6.2 2.2 High 10,631 (33.7%) 5,644 (53.1%) 23.62 70.3 12.2 1.5 Uncertain 3,158 (10.0%) 862 (27.3%) 7.84 67.8 8.3 1.9 Note: CI = Cognitive Impairment; OR = Odds Ratio; Edu = Education (years). The "Uncertain" category identifies cases with high model disagreement (≥ 90th percentile epistemic uncertainty) warranting individualized clinical evaluation. 3.5 Clinical Decision Pathway for Depression Management Figure 4 illustrates a comprehensive clinical decision pathway that translates risk stratification into actionable recommendations for the management of older adults with depression. This pathway converts a four-tiered system into a tiered care plan based on the World Health Organization's "Integrated Care for Older People (ICOPE) Guidelines" and the guiding principles of The Lancet Commission on Dementia Prevention [ 2 ]. (1) Low-risk patients with depression: Implement standard depression treatment protocols and conduct annual cognitive function monitoring, with a focus on alleviating depressive symptoms through routine lifestyle counseling. (2) Moderate-risk patients with depression: Intensify follow-up to every 6 months, including re-screening for depression and optimized management of vascular risk factors. Evidence-based lifestyle interventions (exercise, cognitive training, sleep hygiene) are recommended. (3) High-risk patients with depression: Prioritize aggressive treatment for depression, refer for specialized cognitive assessment and neuropsychological testing, and consider neuroprotective interventions. Regular monitoring is required (every 3 months). (4) Uncertain cases: Require multidisciplinary team consultation, comprehensive cognitive function testing, and a 3-month follow-up before definitive risk stratification. These patients require individualized clinical judgment rather than adherence to algorithm-driven standardized treatment protocols. These recommendations represent expert consensus based on existing guidelines and require prospective validation through implementation studies before widespread clinical application. 3.6 Subgroup Analysis Supplementary Table S5 presents the results of stratified analyses by sex, age group, and urban/rural residence. The association between depressive symptoms and cognitive function remained consistent across all subgroups, with odds ratios ranging from 1.89 to 2.45. No significant effect modification was observed ( P -values for interactions with all stratification variables were > 0.05), indicating that the increased cognitive risk associated with depressive symptoms is robust across different demographic subgroups. The risk stratification model maintained stable discriminatory performance across all subgroups (area under the curve ranging from 0.76 to 0.80). 3.7 Calibration and Clinical Utility Figure 5 presents model calibration and decision curve analysis. Panel A shows the calibration curve, demonstrating good agreement between predicted probabilities and observed frequencies across deciles. The smooth calibration line closely tracks the perfect calibration diagonal, with narrow confidence bands indicating stable calibration. Panel B presents decision curve analysis results. The BARNN + model provided superior net benefit compared to treat-all and treat-none strategies across the clinically relevant threshold probability range of 20–50%. The net benefit advantage was most pronounced in the 30–45% threshold range, which corresponds to typical clinical decision thresholds for cognitive screening referrals. Figure 6 evaluates the predictive performance of different models for cognitive impairment using ROC curves. The AUC values for the ensemble model, logistic regression, and random forest were 0.781, 0.781, and 0.780, respectively, indicating that all three models possess above-average discriminatory ability and perform similarly. The risk stratification system based on the models showed that the actual incidence rates of cognitive impairment in the low-, medium-, high-risk, and uncertain groups were 4.6%, 19.3%, 53.1%, and 27.3%, respectively. Compared with the low-risk group, the high-risk group had a significantly increased risk of developing cognitive impairment (OR = 23.62), suggesting that this stratification system can effectively identify high-risk individuals. 4. Discussion This study leverages data from the China Health and Retirement Longitudinal Study (CHARLS), a large-scale, nationally representative longitudinal survey of Chinese adults aged 45 and above, to systematically explore the association between depressive symptoms and cognitive impairment. Over a 9-year follow-up period, we analyzed 31,570 individual-wave observations from 12,494 unique participants, providing robust evidence of a significant severity-graded association between depressive symptoms and cognitive impairment in Chinese older adults. To our knowledge, this is the largest longitudinal study of its kind in China to date, and it is the first to translate epidemiological findings into a clinically actionable risk stratification system tailored to elderly patients with depression, effectively bridging the gap between research and clinical practice. 4.1 The Severity-Graded Association Between Depressive Symptoms and Cognitive Impairment: Confirmation and Novelty Our core finding—that depressive symptoms double the risk of cognitive impairment (OR = 2.07, 95% CI: 1.97–2.18)—is highly consistent with meta-analytic evidence from Western populations [ 3 , 4 ]. Meanwhile, this study validates this association in a large, heterogeneous Chinese cohort. Despite significant differences in cultural backgrounds, healthcare systems, and epidemiological characteristics between China and Western countries [ 9 ], this consistency suggests that the association between depressive symptoms and cognitive impairment may have cross-population universal significance. The dose-response relationship clearly illustrated in Fig. 1 B holds significant clinical implications. The prevalence of cognitive impairment gradually increased from 20.8% (normal mood state) to 46.0% (major depression), suggesting that the intensity of intervention should be adjusted according to the severity of depression. Notably, even subclinical depressive symptoms (CESD-10 score of 6–10) are associated with an increased risk of cognitive impairment (29.5% vs. 20.8%), suggesting that symptoms below the diagnostic threshold warrant clinical attention [ 16 ]. The clear dose-response relationship between the severity of depression and cognitive impairment reinforces causal inferences and holds significant clinical implications. If this association is indeed causal, depression screening and treatment could serve as a primary prevention strategy for cognitive decline. The study found that even mild depressive symptoms are associated with an increased risk of cognitive impairment (36.1% vs. 48.3% in asymptomatic individuals), suggesting that early intervention at the subclinical stage may be beneficial. The longitudinal trajectories in Fig. 1 A provide mechanistic insights into the management of mood disorders. Analysis of cognitive decline trajectories based on baseline depression severity suggests that chronic depressive symptoms may accelerate the process of cognitive aging through cumulative neurotoxic effects, such as inflammatory responses, abnormal cortisol regulation, and reduced neurotrophic support [ 6 , 7 ]. From the perspective of depression diagnosis and treatment, these findings underscore the importance of incorporating cognitive monitoring into longitudinal management strategies for older adults with depression. 4.2 Clinical Implications: From Risk Stratification to Individualized Management One of the core contributions of this study is the construction of a clinically operable four-tier risk stratification system, which accurately fills the critical gap in translating epidemiological evidence into clinical guidance. Based on an ensemble machine learning model (AUC = 0.781), this system classifies elderly patients with depression into four groups: low-risk, moderate-risk, high-risk, and uncertain-risk, with the actual incidence of cognitive impairment being 4.6%, 19.3%, 53.1%, and 27.3%, respectively (Table 3 ). The more than 10-fold difference in the risk of cognitive impairment between the low-risk and high-risk groups (OR = 23.62) provides a clear basis for optimizing the allocation of clinical resources and directing intensive interventions to the population that would benefit the most. Based on this stratification system, this study proposes a corresponding clinical decision pathway (Fig. 4 ), which is fully consistent with the core principles of the World Health Organization's “Integrated Care for Older People (ICOPE) Guidelines” and The Lancet Commission on Dementia Prevention [ 2 ]. For low-risk patients, standard depression treatment and annual cognitive monitoring are sufficient; moderate-risk patients require enhanced follow-up every 6 months, optimized management of vascular risk factors, and adjunctive lifestyle interventions; high-risk patients need priority for active depression treatment, referral for specialized cognitive assessment, and close monitoring every 3 months; while the 10.0% “uncertain-risk group,” which exhibits a unique phenotype of severe depressive symptoms but intact functional status (Fig. 3 ), requires multidisciplinary consultation and individualized assessment. This stratified treatment plan achieves precise matching between intervention intensity, depression severity, and cognitive risk, surpassing the traditional “one-size-fits-all”management model. The definition of the “ncertain-risk group” represents an important methodological innovation, which directly addresses the “black box” problem in the clinical application of artificial intelligence (AI). As clearly shown in Fig. 3 , although patients in this group have depression scores comparable to those in the high-risk group, their functional status remains intact, making them difficult to accurately classify by algorithms. By explicitly identifying such cases and recommending individualized judgment, this framework fully reflects the regulatory principle that AI should “enhance rather than replace” clinical decision-making [ 17 ], establishing a human-machine collaborative clinical decision-making model and improving the credibility and application value of the risk stratification system. In addition, the study results strongly call for the integration of cognitive function monitoring into routine care for elderly depression. Current clinical assessments of elderly depression mostly focus on symptom remission and functional recovery, with relative neglect of cognitive outcomes. Given the strong association between the two, we recommend the use of a comprehensive cognitive scale based on the Health and Retirement Study (HRS) as a routine follow-up indicator for all elderly patients with depression. Particularly for high-risk individuals, preliminary evidence suggests that active depression treatment (especially selective serotonin reuptake inhibitors) may delay cognitive decline [ 8 ], highlighting the importance of prioritizing intensive treatment in this subgroup. 4.3 Public Health Implications: Toward Geriatric Mental Health and Cognitive Prevention The results of this study have far-reaching guiding significance for public health policies in the context of China's rapid aging. The study shows that the prevalence of clinically significant depressive symptoms in this cohort is as high as 34.2%, indicating that depression has become a major public health burden among Chinese older adults. If the association between the two is indeed causal, then depression screening and treatment may become a core strategy for preventing cognitive decline at the population level, which is of extremely high practical urgency in the context of the increasing burden of dementia in China [ 9 ]. A key public health recommendation of this study is to promote the in-depth integration of mental health and brain health management in geriatric care. The traditional model often treats depression and cognitive impairment in isolation, but this study confirms a strong synergistic effect between the two. Therefore, we advocate the establishment of a collaborative care model, integrating depression screening into cognitive risk assessment and cognitive assessment into depression follow-up, to achieve “simultaneous physical and mental treatment” and shift from passive treatment to active prevention, thereby improving the overall clinical outcomes of older adults. Meanwhile, the method for quantifying prediction uncertainty in this study has universal implications for the application of AI in the medical field. By identifying 10% of cases with unreliable model predictions, we demonstrate how to apply AI-assisted decision-making in a responsible and transparent manner. This approach effectively alleviates clinicians' trust concerns about the AI “black box” and promotes the rational and safe application of AI tools in geriatric care. As AI penetrates deeper into the field of geriatrics, such transparency will be crucial for improving tool effectiveness and ensuring medical quality. 4.4 Methodological Innovations The model demonstrates statistically significant improvements over traditional machine learning methods(Figure 6 ). Although the absolute AUC differences appear small (ΔAUC = 0.006–0.009) [ 20 ], these improvements are clinically significant at the population level. For a screening program covering 100,000 people, such improvements could accurately reclassify thousands of patients. Furthermore, the interpretability approach based on SHAP analysis (Fig. 2 ) addresses a key barrier in clinical AI applications—the “black box” problem [ 21 ]. By decomposing individual predictions into feature-level contributions, clinicians can understand why patients are classified into specific risk categories. The study found that depression severity, age, and educational level consistently emerged as the primary predictors ( Supplementary Figure S3 ), a finding consistent with clinical intuition and epidemiological evidence [ 2 , 4 ] that is expected to enhance clinicians’ confidence. A key innovation of this study lies in the integration of techniques for quantifying cognitive uncertainty. Traditional machine learning models provide only point estimates without conveying prediction confidence, which may lead to inappropriate clinical decisions. Our method identifies 10% of “uncertain” cases—cases where the model’s prediction is unreliable and requires expert review. This feature aligns with regulatory guidelines emphasizing transparency in AI medical devices [ 17 ] and supports human-machine collaborative decision-making mechanisms. The four-tier risk stratification system provides actionable guidance for clinical practice: low-risk individuals can undergo routine follow-up; moderate-risk individuals require enhanced monitoring; high-risk individuals should be referred for comprehensive evaluation and intervention; and uncertain cases require expert consultation before clinical decisions are made. This refined stratification mechanism enables personalized treatment pathways and optimizes the allocation of healthcare resources. 4.5 Study Limitations and Future Directions This study has several limitations that require careful interpretation. First, the definition of cognitive impairment was based on epidemiological percentile thresholds (below the 31st percentile after adjusting for age and education), rather than clinical diagnostic criteria (such as the Petersen criteria or DSM-5 criteria). Although sensitivity analyses using the 25th and 33rd percentiles yielded consistent results, future studies still need to use clinically confirmed data for validation to further promote clinical translation. Second, the CESD-10 scale is only a symptom screening tool and cannot be used as a basis for the clinical diagnosis of major depressive disorder, nor can it distinguish between depressive subtypes. The CHARLS data also lack information on antidepressant use, and these factors may affect the observed association and limit the generalizability of the conclusions. Future studies should include clinical diagnostic data and treatment information to further clarify the complex relationships. Third, although the sample is nationally representative, the extrapolation of the study conclusions to other populations (such as elderly groups in low- and middle-income countries and populations with different cultural backgrounds) still requires more validation. At the same time, the observational study design makes it difficult to establish a clear causal relationship; future randomized controlled trials are needed to directly verify whether depression treatment can reduce the risk of cognitive impairment. Future research can focus on three key directions: first, validate the risk stratification system in independent cohorts and real clinical settings to evaluate its practical application value; second, design randomized controlled trials to explore the preventive effects of various depression intervention strategies, including pharmacotherapy, psychotherapy, and collaborative care, on cognitive impairment; third, further explore the potential pathophysiological mechanisms by which depression severity affects cognitive decline, providing a scientific basis for the development of targeted interventions. 5. Conclusions In summary, based on a large-scale national cohort, this study provides strong evidence of a significant, severity-graded association between depressive symptoms and cognitive impairment in Chinese older adults. The clear dose-response relationship provides a solid basis for implementing stratified clinical management and adjusting intervention intensity according to depression severity. The risk stratification system and clinical decision pathway constructed in this study provide clinicians with directly operable guidance, helping to achieve individualized care and optimal allocation of medical resources. The results support the integration of cognitive function monitoring into the routine diagnosis and treatment process of elderly depression, and emphasize the importance of depression treatment as a potential strategy for preventing cognitive decline. Faced with the severe challenge of population aging in China, the findings of this study provide key scientific support and practical pathways for improving the quality of care for elderly depression and reducing the burden of cognitive impairment. Declarations Ethics Approval: The study protocol was approved by the Biomedical Ethics Review Committee of Peking University. All participants provided written informed consent. Funding Declaration This study relies on the Shandong Province Education Teaching Planning Project (Innovation and Practice of“Combination of Medicine, Health and Nutrition, Integration of Teaching, Production and Research”Parenting Mode of Higher Vocational Intelligent Recreation Professional Group, NO. B20G590805) Author Contribution Q. Z. and P.-L. Z. conducted the whole study conception and design. Y.-F. Z. and X.-S. R. prepared the draft of the article and revised it critically for important intellectual content. D. -W. Y. approved the submitted version. All authors have read and agreed to the published version of the manuscript. Acknowledgments: We thank the China Health and Retirement Longitudinal Study (CHARLS) team for providing access to the data. Data Availability The data used in this study are publicly available from the China Health and Retirement Longitudinal Study (CHARLS) repository (https://charls.pku.edu.cn/). The detailed data access procedures and usage policies are available on the official website of CHARLS. References Alexopoulos GS. Mechanisms and treatment of late-life depression. Transl Psychiatry. 2019;9(1):188. 10.1038/s41398-019-0514-6 . Livingston G, Huntley J, Sommerlad A, et al. Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet. 2020;396(10248):413–46. 10.1016/S0140-6736(20)30367-6 . Diniz BS, Butters MA, Albert SM, Dew MA, Reynolds CF. Late-life depression and risk of vascular dementia and Alzheimer's disease: systematic review and meta-analysis. Br J Psychiatry. 2013;202(5):329–35. 10.1192/bjp.bp.112.118307 . Ownby RL, Crocco E, Acevedo A, John V, Loewenstein D. Depression and risk for Alzheimer disease: systematic review, meta-analysis, and metaregression analysis. Arch Gen Psychiatry. 2006;63(5):530–8. 10.1001/archpsyc.63.5.530 . Bennett S, Thomas AJ. Depression and dementia: cause, consequence or coincidence? Maturitas. 2014;79(2):184–90. 10.1016/j.maturitas.2014.05.009 . Byers AL, Yaffe K. Depression and risk of developing dementia. Nat Rev Neurol. 2011;7(6):323–31. 10.1038/nrneurol.2011.60 . Sapolsky RM. Glucocorticoids and hippocampal atrophy in neuropsychiatric disorders. Arch Gen Psychiatry. 2000;57(10):925–35. 10.1001/archpsyc.57.10.925 . Bartels C, Wagner M, Wolfsgruber S, Ehrenreich H, Schneider A. Impact of SSRI therapy on risk of conversion from mild cognitive impairment to Alzheimer's dementia in individuals with previous depression. Am J Psychiatry. 2018;175(3):232–41. 10.1176/appi.ajp.2017.17040404 . Jia L, Du Y, Chu L, et al. Prevalence, risk factors, and management of dementia and mild cognitive impairment in adults aged 60 years or older in China: a cross-sectional study. Lancet Public Health. 2020;5(12):e661–71. 10.1016/S2468-2667(20)30185-7 . Zhao Y, Hu Y, Smith JP, Strauss J, Yang G. Cohort profile: the China Health and Retirement Longitudinal Study (CHARLS). Int J Epidemiol. 2014;43(1):61–8. 10.1093/ije/dys203 . Andresen EM, Malmgren JA, Carter WB, Patrick DL. Screening for depression in well older adults: evaluation of a short form of the CES-D. Am J Prev Med. 1994;10(2):77–84. 10.1016/s0749-3797(18)30622-6 . Cheng ST, Chan AC. The Center for Epidemiologic Studies Depression Scale in older Chinese: thresholds for long and short forms. Int J Geriatr Psychiatry. 2005;20(5):465–70. 10.1002/gps.1314 . Lei X, Smith JP, Sun X, Zhao Y. Gender differences in cognition in China and reasons for change over time: evidence from CHARLS. J Econ Ageing. 2014;4:46–55. 10.1016/j.jeoa.2013.11.001 . DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44(3):837–45. 10.2307/2531595 . Vickers AJ, Elkin EB. Decision curve analysis: a novel method for evaluating prediction models. Med Decis Mak. 2006;26(6):565–74. 10.1177/0272989X06295361 . Köhler S, van Boxtel M, Jolles J, Verhey F. Depressive symptoms and risk for dementia: a 9-year follow-up of the Maastricht Aging Study. Am J Geriatr Psychiatry. 2011;19(10):902–5. 10.1097/JGP.0b013e31821f1b6a . US Food and Drug Administration. Artificial Intelligence/Machine Learning (AI/ML)-Based Software as a Medical Device (SaMD) Action Plan. FDA; 2021. Katon W, Pedersen HS, Ribe AR, et al. Effect of depression and diabetes mellitus on the risk for dementia: a national population-based cohort study. JAMA Psychiatry. 2015;72(6):612–9. 10.1001/jamapsychiatry.2015.0082 . Cherbuin N, Kim S, Anstey KJ. Dementia risk estimates associated with measures of depression: a systematic review and meta-analysis. BMJ Open. 2015;5(12):e008853. 10.1136/bmjopen-2015-008853 . Steyerberg EW, Vickers AJ, Cook NR, et al. Assessing the performance of prediction models: a framework for traditional and novel measures. Epidemiology. 2010;21(1):128–38. 10.1097/EDE.0b013e3181c30fb2 . Lundberg SM, Lee SI. A unified approach to interpreting model predictions. Adv Neural Inf Process Syst. 2017;30:4765–74. Miller AH, Maletic V, Raison CL. Inflammation and its discontents: the role of cytokines in the pathophysiology of major depression. Biol Psychiatry. 2009;65(9):732–41. 10.1016/j.biopsych.2008.11.029 . Stern Y, Cognitive reserve. Neuropsychologia. 2009;47(10):2015–28. 10.1016/j.neuropsychologia.2009.03.004 . Taylor WD, Aizenstein HJ, Alexopoulos GS. The vascular depression hypothesis: mechanisms linking vascular disease with depression. Mol Psychiatry. 2013;18(9):963–74. 10.1038/mp.2013.20 . Gallagher D, Kiss A, Lanctot K, Herrmann N. Depression and risk of Alzheimer dementia: a longitudinal analysis to determine predictors of increased risk among older adults with depression. Am J Geriatr Psychiatry. 2018;26(8):819–27. 10.1016/j.jagp.2018.05.002 . Saczynski JS, Beiser A, Seshadri S, Auerbach S, Wolf PA, Au R. Depressive symptoms and risk of dementia: the Framingham Heart Study. Neurology. 2010;75(1):35–41. 10.1212/WNL.0b013e3181e62138 . GBD 2019 Dementia Forecasting Collaborators. Estimation of the global prevalence of dementia in 2019 and forecasted prevalence in 2050: an analysis for the Global Burden of Disease Study 2019. Lancet Public Health. 2022;7(2):e105–25. 10.1016/S2468-2667(21)00249-8 . Chen X, Giles J, Yao Y, et al. The path to healthy ageing in China: a Peking University-Lancet Commission. Lancet. 2022;400(10367):1967–2006. 10.1016/S0140-6736(22)01546-X . Huang Y, Wang Y, Wang H, et al. Prevalence of mental disorders in China: a cross-sectional epidemiological study. Lancet Psychiatry. 2019;6(3):211–24. 10.1016/S2215-0366(18)30511-X . Barnes DE, Yaffe K, Byers AL, McCormick M, Schaefer C, Whitmer RA. Midlife vs late-life depressive symptoms and risk of dementia: differential effects for Alzheimer disease and vascular dementia. Arch Gen Psychiatry. 2012;69(5):493–8. 10.1001/archgenpsychiatry.2011.1481 . Collins GS, Reitsma JB, Altman DG, Moons KGM. Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD): the TRIPOD Statement. BMJ. 2015;350:g7594. 10.1136/bmj.g7594 . Breiman L. Random forests. Mach Learn. 2001;45(1):5–32. 10.1023/A:1010933404324 . Additional Declarations No competing interests reported. Supplementary Files BARNNSupplementary.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-9460240","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":635044957,"identity":"6f0b4af6-0ecc-413b-82d8-e1a9b7953bee","order_by":0,"name":"Qiang Zhang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAz0lEQVRIiWNgGAWjYBACAwYGxgcJFRI8/OyNjQ8/EKmF2eDDGQsZyZ7DzcYSRGphk5zZVmFjcCO9TYCHGC3m7D0G0jxsEjwMNx+2MUgw2MnpNhDQYtlzxsCYh0eCh3F2YtuDAoZkY7MDhBx2I3dDMo+EBA+zdGK7gQTDgcRtxGg5zGMgwcMmebAN6DzitGxsnJEgAXQbI5FaLHvOf2b4cEAC6JtEYCAbEOEXc/a29B+J/+rs7Y8ff/jwQ4WdHEEt6O4kTfkoGAWjYBSMAhwAAC/fQHdvjNPJAAAAAElFTkSuQmCC","orcid":"","institution":"Jining Polytechnic","correspondingAuthor":true,"prefix":"","firstName":"Qiang","middleName":"","lastName":"Zhang","suffix":""},{"id":635044958,"identity":"4cdbda16-fa9e-4d66-b63a-9facd080ec8d","order_by":1,"name":"Yifan Zhao","email":"","orcid":"","institution":"Jining Polytechnic","correspondingAuthor":false,"prefix":"","firstName":"Yifan","middleName":"","lastName":"Zhao","suffix":""},{"id":635044959,"identity":"883b5990-8877-46ed-8558-112f64d79c81","order_by":2,"name":"Peilu Zhang","email":"","orcid":"","institution":"Jining Polytechnic","correspondingAuthor":false,"prefix":"","firstName":"Peilu","middleName":"","lastName":"Zhang","suffix":""},{"id":635044960,"identity":"51d3c770-f939-4c8c-bb18-87565b2ae45d","order_by":3,"name":"Xinsong Ren","email":"","orcid":"","institution":"nursing of traditional chinese medicine clinic, Binzhou People’s Hospital","correspondingAuthor":false,"prefix":"","firstName":"Xinsong","middleName":"","lastName":"Ren","suffix":""},{"id":635044963,"identity":"1e4fccb2-aadd-455e-ba61-5e652c6c8b57","order_by":4,"name":"Dongwei Yan","email":"","orcid":"","institution":"Shandong University of Traditional Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Dongwei","middleName":"","lastName":"Yan","suffix":""}],"badges":[],"createdAt":"2026-04-19 07:53:26","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9460240/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9460240/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":109067506,"identity":"8c60db8c-ba67-459a-9f82-e5aa37d91bea","added_by":"auto","created_at":"2026-05-12 09:54:55","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":407007,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eLongitudinal cognitive trajectories and dose-response relationship between depressive symptom severity and cognitive impairment. (A) Cognitive score trajectories over 9-year follow-up stratified by baseline depression severity. Shaded areas represent 95% confidence intervals. (B) Cognitive impairment rates across depression severity categories with 95% confidence intervals. P for trend \u0026lt; 0.001 (Cochran-Armitage test).\u003c/em\u003e\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-9460240/v1/9cc8758b15d6d7e62601c73d.png"},{"id":108941421,"identity":"4d87e94a-6af0-4f1e-a000-6dddb73261cb","added_by":"auto","created_at":"2026-05-11 05:35:07","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":133769,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eSHAP-based individual prediction explanations for representative high-risk and low-risk cases, demonstrating how depression severity contributes to cognitive risk classification.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-9460240/v1/aa72b5e10f18160968f0058d.png"},{"id":108977591,"identity":"a92c8132-6b87-434f-8fbf-9bbc6a49ef29","added_by":"auto","created_at":"2026-05-11 11:32:15","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":91845,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003ePhenotypic profile of risk groups using Z-score deviations from low-risk reference. The uncertain group exhibits elevated depression scores but preserved functional status, suggesting complex clinical presentations requiring individualized evaluation.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-9460240/v1/6ff7c583e50f9afe65a3ff18.png"},{"id":108977556,"identity":"53e9964f-e219-4316-84eb-c400c3b695c0","added_by":"auto","created_at":"2026-05-11 11:32:05","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":268921,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eClinical decision pathway for managing depressed older adults based on cognitive risk stratification. The pathway integrates depression screening, risk assessment, and severity-graded management recommendations.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-9460240/v1/e90450b4c18629d49e04059b.png"},{"id":108941423,"identity":"d4e1ea7d-d426-4dcf-8d5a-377e668bbf78","added_by":"auto","created_at":"2026-05-11 05:35:07","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":152205,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eModel calibration and decision curve analysis. (A) Calibration curve comparing predicted probabilities to observed frequencies. (B) Decision curve analysis showing net benefit across threshold probabilities.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-9460240/v1/fab7ed4d920e4e87a3675463.png"},{"id":108977894,"identity":"52600c32-0905-414b-8f43-9019ba5bdeb3","added_by":"auto","created_at":"2026-05-11 11:33:23","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":141608,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eROC curves and risk stratification validation. (A) Receiver operating characteristic curves comparing model performance. (B) Observed cognitive impairment rates across risk strata.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-9460240/v1/2949593719816093653ea8d9.png"},{"id":109069121,"identity":"324f306a-4d65-4df7-8173-4b0d08dff82d","added_by":"auto","created_at":"2026-05-12 10:20:01","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1480070,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9460240/v1/d0394236-00b8-4376-8f43-82c9c7a48065.pdf"},{"id":108941420,"identity":"1fa3ae8c-8f87-4716-8052-0af6f35f2279","added_by":"auto","created_at":"2026-05-11 05:35:07","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":457220,"visible":true,"origin":"","legend":"","description":"","filename":"BARNNSupplementary.docx","url":"https://assets-eu.researchsquare.com/files/rs-9460240/v1/953d97c5f07193637391545e.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Depressive Symptoms as a Modifiable Risk Factor for Cognitive Impairment: Severity-Graded Associations and Clinical Implications from a 9-Year Longitudinal Study in Chinese Older Adults","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eLate-life depression poses a significant clinical challenge, affecting approximately 10%-15% of older adults worldwide and associated with high morbidity, mortality, and healthcare resource utilization [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. In addition to its direct impact on quality of life and functional status, emerging evidence suggests that depression may increase the risk of subsequent cognitive decline and dementia [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. This potential association has profound implications for how clinicians manage older adults with depression.\u003c/p\u003e \u003cp\u003eThe relationship between depression and cognitive function in older adults is complex and may be bidirectional [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. To explain how chronic depression accelerates cognitive aging, several pathophysiological mechanisms have been proposed, including hippocampal damage resulting from hypothalamic-pituitary-adrenal axis dysfunction, chronic systemic inflammation, cerebrovascular dysfunction, and reduced neurotrophic support [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Meta-analysis evidence suggests that late-life depression may increase the risk of dementia by approximately 1.5 to 2 times [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Crucially, unlike most risk factors for cognitive impairment (such as age, genetic factors, and educational level), depression can potentially be modulated through pharmacological treatment and psychological interventions [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eHowever, several key issues remain inadequately addressed. First, existing evidence primarily originates from Western populations, and data from large-scale Chinese cohort studies remain insufficient, given the significant differences in cultural context, healthcare systems, and epidemiological characteristics compared to Western settings [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Second, although the association between depression and cognitive function has been documented in the literature, the dose-response relationship between depression severity and cognitive outcomes has not yet been systematically elucidated. Clarifying this gradient relationship is crucial for developing clinical recommendations for tiered diagnosis and treatment. Third, there remains a scarcity of guidelines for clinicians on how to incorporate cognitive risk factors into treatment plans when managing older adults with depression.\u003c/p\u003e \u003cp\u003eThis study aims to address these research gaps using data from the China Health and Retirement Longitudinal Study (CHARLS). Specific objectives include: (1) quantifying the association between depressive symptoms and cognitive impairment in a large, nationally representative Chinese cohort; (2) describing the dose-response relationship between depression severity and cognitive outcomes to inform tiered clinical management; (3) establishing a clinically applicable risk stratification system to identify which older adults with depression face the highest cognitive risk and are most in need of enhanced monitoring and intervention; and (4) exploring the implications of these findings for the clinical management of late-life depression.\u003c/p\u003e"},{"header":"2. Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Study Population\u003c/h2\u003e \u003cp\u003eData were drawn from the China Health and Retirement Longitudinal Study (CHARLS), a nationally representative longitudinal survey of Chinese adults aged 45 and older [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. CHARLS employed a multistage stratified probability proportional to size sampling strategy, covering 28 provinces, 150 counties and districts, and 450 villages and towns across China. The baseline survey was conducted in 2011, with follow-up surveys conducted in 2013, 2015, 2018, and 2020.\u003c/p\u003e \u003cp\u003eThe initial sample comprised 49,015 observations across all waves. After excluding participants with missing cognitive assessment data (n\u0026thinsp;=\u0026thinsp;8,456), missing depression scores (n\u0026thinsp;=\u0026thinsp;5,892), incomplete covariate data (n\u0026thinsp;=\u0026thinsp;2,660), and outliers in age (n\u0026thinsp;=\u0026thinsp;337), the final analysis sample comprised 31,570 individual-wave observations from 12,494 unique participants (\u003cb\u003eSupplementary Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e\u003c/b\u003e). The study protocol was approved by the Biomedical Ethics Review Committee of Peking University, and all participants provided written informed consent.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Measurement Instruments\u003c/h2\u003e \u003cdiv id=\"Sec5\" class=\"Section3\"\u003e \u003ch2\u003e2.2.1 Assessment of Depressive Symptoms\u003c/h2\u003e \u003cp\u003eDepressive symptoms were assessed using the Center for Epidemiological Studies Depression Scale (CES-D-10), a well-established tool for screening depression in older adults [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Participants reported the frequency of depressive symptoms over the past week using a 4-point scale (0\u0026thinsp;=\u0026thinsp;rarely, 3\u0026thinsp;=\u0026thinsp;most of the time). The total score ranged from 0 to 30 points, with higher scores indicating more severe depressive symptoms.\u003c/p\u003e \u003cp\u003eIn the preliminary analysis, clinically significant depressive symptoms were defined as a CESD-10 score\u0026thinsp;\u0026ge;\u0026thinsp;10, a threshold widely adopted in epidemiological studies [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. The CESD-10 is used solely as a screening tool to assess the burden of depressive symptoms and does not constitute a basis for the clinical diagnosis of major depressive disorder; therefore, the term \u0026ldquo;depressive symptoms\u0026rdquo; is used throughout this paper. In the dose-response analysis, the severity of depressive symptoms was categorized as mild (0\u0026ndash;5 points), moderate (6\u0026ndash;10 points), severe (11\u0026ndash;15 points), and very severe (\u0026gt;\u0026thinsp;15 points). To partially address the issue of persistent depression, we conducted a supplementary analysis distinguishing persistent depressive symptoms (CESD-10 scores\u0026thinsp;\u0026ge;\u0026thinsp;10 in two or more consecutive assessment waves) from single-episode symptoms (\u003cb\u003eSupplementary Table S9\u003c/b\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section3\"\u003e \u003ch2\u003e2.2.2 Cognitive Function Assessment\u003c/h2\u003e \u003cp\u003eCognitive function was assessed using a comprehensive assessment scale adapted from the Health and Retirement Study (HRS). The assessment included (1) episodic memory (immediate and delayed word recall from a 10-word list, scored 0\u0026ndash;20); (2) mental status (serial subtraction by 7, date orientation, and a drawing task, scored 0\u0026ndash;11); and (3) executive function (naming task). The total cognitive score was calculated as the sum of the scores for each subscale (score range: 0\u0026ndash;31).\u003c/p\u003e \u003cp\u003eCognitive impairment was defined as a score below the 31st percentile after adjustment for age and education level, consistent with previous findings from the CHARLS study [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. This definition is based on epidemiological criteria derived from statistical distributions and does not correspond to clinical diagnostic criteria for mild cognitive impairment (MCI; e.g., the Petersen criteria) or dementia (e.g., the DSM-5 criteria). Sensitivity analyses using the 25th and 33rd percentiles yielded consistent results (\u003cb\u003eSupplementary Table S7\u003c/b\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section3\"\u003e \u003ch2\u003e2.2.3 Covariates\u003c/h2\u003e \u003cp\u003eCovariates included sociodemographic factors (age, sex, education, rural/urban residence, marital status), health behaviors (smoking, alcohol consumption, physical exercise, sleep duration), chronic conditions (hypertension, diabetes, heart disease, stroke, and chronic disease count), functional status (activities of daily living, instrumental activities of daily living, history of falls), physical measurements (body mass index, systolic and diastolic blood pressure, grip strength, walking speed), sensory function (vision, hearing), and self-rated health. Time-varying covariates were updated at each wave, and longitudinal features including years since baseline and total number of visits were incorporated. A complete list of 45 features is provided in \u003cb\u003eSupplementary Table S2\u003c/b\u003e.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Construction of the Risk Stratification Model\u003c/h2\u003e \u003cp\u003eTo translate the association between depressive symptoms and cognitive function into clinically actionable guidance, the study employed an ensemble machine learning approach to construct a risk stratification model. This model integrated the predictive results of multiple algorithms (histogram-based gradient boosting, additional tree classifier, and random forest classifier) using bootstrap aggregation and performance-weighted averaging. Model hyperparameters are detailed in \u003cb\u003eSupplementary Table S4\u003c/b\u003e.\u003c/p\u003e \u003cp\u003eA four-tier risk stratification system was established based on predicted probability thresholds: (1) low risk (predicted probability\u0026thinsp;\u0026lt;\u0026thinsp;0.25); (2) moderate risk (0.25\u0026thinsp;\u0026le;\u0026thinsp;predicted probability\u0026thinsp;\u0026lt;\u0026thinsp;0.55); (3) high risk (predicted probability\u0026thinsp;\u0026ge;\u0026thinsp;0.55); (4) Uncertain (cognitive uncertainty\u0026thinsp;\u0026ge;\u0026thinsp;90th percentile). The \u0026ldquo;Uncertain\u0026rdquo; category refers to cases where there is significant disagreement among models; such cases require individualized clinical assessment rather than algorithm-driven recommendations.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e2.4 Statistical Analysis\u003c/h2\u003e \u003cp\u003eThe association between depressive symptoms and cognitive impairment was quantified using odds ratios (OR) and their 95% confidence intervals. The Cochran-Armitage test was used to evaluate the dose-response relationship across different severity levels of depression. Model performance was evaluated using 5-fold stratified cross-validation, with the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and Brier score as primary evaluation metrics. Model comparisons were performed using the DeLong test [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Receiver operating characteristic curve analysis was used to assess clinical utility at threshold probabilities [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. Subgroup analyses were stratified by sex, age group (\u0026le;\u0026thinsp;70 years vs. \u0026gt;70 years), and urban/rural residence, and interaction tests were performed. All analyses were conducted using Python 3.10 in conjunction with scikit-learn, NumPy, and SciPy. Statistical significance was defined as a two-sided \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Sample Characteristics and Associations with Depression\u003c/h2\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e presents the characteristics of the study population stratified by depression status. The analysis sample comprised 31,570 individual follow-up observations from 12,494 unique participants, with a mean age of 67.8 years (standard deviation\u0026thinsp;=\u0026thinsp;6.1). Approximately 44.0% were women, 56.2% resided in rural areas, and the average years of education was 4.7. The prevalence of clinically significant depressive symptoms (CESD-10 score\u0026thinsp;\u0026ge;\u0026thinsp;10) was 34.2%.\u003c/p\u003e \u003cp\u003eParticipants with depressive symptoms differed significantly from those without depressive symptoms across multiple dimensions. Symptomatic individuals were more likely to be female (53.6% vs. 39.0%, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001), have lower educational attainment (3.8 years vs. 5.2 years, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001), were more likely to reside in rural areas (65.3% vs. 51.5%, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001), and had a higher prevalence of chronic diseases (0.9 vs. 0.8, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e \u003cp\u003eA key finding was that participants with depressive symptoms had a significantly higher prevalence of cognitive impairment (39.6% vs. 24.1%, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001), with an unadjusted odds ratio of 2.07 (95% CI: 1.97\u0026ndash;2.18). These results indicate that depressive symptoms are associated with more than a twofold increased risk of cognitive impairment in this population.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eBaseline characteristics of study participants stratified by depressive symptom status (n\u0026thinsp;=\u0026thinsp;31,570 observations from 12,494 unique participants)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCharacteristic\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTotal\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;31,570)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNo Depression\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;20,757)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eDepression\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;10,813)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eP-value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDemographics\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge, years, mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e67.8\u0026thinsp;\u0026plusmn;\u0026thinsp;6.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e67.8\u0026thinsp;\u0026plusmn;\u0026thinsp;6.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e67.8\u0026thinsp;\u0026plusmn;\u0026thinsp;6.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemale, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e13,890 (44.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e8,092 (39.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5,798 (53.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEducation, years, mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.7\u0026thinsp;\u0026plusmn;\u0026thinsp;4.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e5.2\u0026thinsp;\u0026plusmn;\u0026thinsp;4.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.8\u0026thinsp;\u0026plusmn;\u0026thinsp;4.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRural residence, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e17,747 (56.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e10,686 (51.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e7,061 (65.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eHealth Status\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eChronic diseases, n, mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.8\u0026thinsp;\u0026plusmn;\u0026thinsp;0.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.8\u0026thinsp;\u0026plusmn;\u0026thinsp;0.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.9\u0026thinsp;\u0026plusmn;\u0026thinsp;1.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCESD-10 score, mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e8.1\u0026thinsp;\u0026plusmn;\u0026thinsp;6.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4.4\u0026thinsp;\u0026plusmn;\u0026thinsp;2.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e15.2\u0026thinsp;\u0026plusmn;\u0026thinsp;4.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCognitive Outcomes\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCognition score, mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e11.5\u0026thinsp;\u0026plusmn;\u0026thinsp;3.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e12.0\u0026thinsp;\u0026plusmn;\u0026thinsp;3.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e10.4\u0026thinsp;\u0026plusmn;\u0026thinsp;3.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCognitive impairment, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e9,286 (29.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4,999 (24.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4,287 (39.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003e\u003cem\u003eNote: Depression defined as CESD-10\u0026thinsp;\u0026ge;\u0026thinsp;10. Data are from 12,494 unique participants across 5 survey waves (2011\u0026ndash;2020).\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e3.2 Dose-Response Relationship Between Depression Severity and Cognitive Impairment\u003c/h2\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e illustrates the longitudinal cognitive trajectories and dose-response relationship between depression severity and cognitive impairment. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA shows that, over the 9-year follow-up period, cognitive decline trajectories differed significantly according to baseline depression severity. The group with severe depressive symptoms (CESD-10\u0026thinsp;\u0026gt;\u0026thinsp;15) exhibited the most rapid cognitive decline, whereas the group with a normal mood (CESD-10 0\u0026ndash;5) maintained relatively stable cognitive function. These trajectory differences suggest that depression may accelerate the process of cognitive aging, rather than merely reflecting coexisting cognitive impairment.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eB confirms a significant dose-response relationship between depression severity categories and the incidence of cognitive impairment. The incidence of cognitive impairment increased progressively with increasing severity: normal mood group (20.8%), mild symptom group (29.5%), moderate symptom group (36.6%), and severe symptom group (46.0%). The Cochran-Armitage test revealed a significant linear trend (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001). For each increase in depression severity grade, the prevalence of cognitive impairment correspondingly increased by approximately 8\u0026ndash;10 percentage points.\u003c/p\u003e \u003cp\u003eNotably, even subclinical depressive symptoms (CESD-10 scores of 6\u0026ndash;10, below the conventional threshold for clinical significance) were associated with a significantly increased risk of cognitive impairment (29.5% vs. 20.8%), suggesting that early intervention at the subclinical stage may be beneficial. Detailed dose-response statistics, including odds ratios, are presented in Supplementary Table S3.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e3.3 Performance of Risk Stratification Models\u003c/h2\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e presents a comparison of the performance of various machine learning models. The ensemble model achieved an AUC of 0.781 (95% confidence interval: 0.776\u0026ndash;0.788), with a sensitivity of 73.5% and a specificity of 66.9%; a Brier score of 0.198 indicates good calibration. Its performance was comparable to or slightly better than that of the base models (logistic regression AUC\u0026thinsp;=\u0026thinsp;0.781, random forest AUC\u0026thinsp;=\u0026thinsp;0.780, gradient boosting AUC\u0026thinsp;=\u0026thinsp;0.781). Cross-validation results (\u003cb\u003eSupplementary Figure S4\u003c/b\u003e) demonstrated good model stability, with a 5-fold cross-validation AUC of 0.779\u0026thinsp;\u0026plusmn;\u0026thinsp;0.008. The calibration curve (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA) confirms good consistency between predicted probabilities and observed frequencies. Receiver operating characteristic (ROC) curve analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB) shows that within the clinically relevant threshold probability range of 20%-50%, this strategy yields superior net benefit compared to the \u0026ldquo;treat all\u0026rdquo; and \u0026ldquo;do not treat\u0026rdquo; strategies.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePerformance comparison of machine learning models for cognitive impairment prediction.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"9\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAUC (95% CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSens\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSpec\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003ePPV\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eNPV\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eF1\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eBrier\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eP*\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eEnsemble\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.781 (0.776\u0026ndash;0.788)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.735\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.669\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.481\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.858\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.581\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.198\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003eRef\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLogistic Reg.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.781 (0.770\u0026ndash;0.782)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.710\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.695\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.492\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.852\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.581\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.193\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e0.013\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRandom Forest\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.780 (0.766\u0026ndash;0.778)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.690\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.712\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.500\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.846\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.580\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.191\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGradient Boost\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.781 (0.771\u0026ndash;0.783)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.720\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.686\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.489\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.855\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.583\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.195\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e0.078\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"9\"\u003e\u003cem\u003eNote: Sens\u0026thinsp;=\u0026thinsp;Sensitivity; Spec\u0026thinsp;=\u0026thinsp;Specificity; PPV\u0026thinsp;=\u0026thinsp;Positive Predictive Value; NPV\u0026thinsp;=\u0026thinsp;Negative Predictive Value; Brier\u0026thinsp;=\u0026thinsp;Brier Score. *P-value from DeLong test comparing each model to the Ensemble model.\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003e3.4 Clinical Risk Stratification of Elderly Patients with Depression\u003c/h2\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e presents the validation results of the four-tier risk stratification. The distribution of risk categories is as follows: low risk (14.0%), moderate risk (42.3%), high risk (33.7%), and uncertain (10.0%). The observed incidence of cognitive impairment demonstrated excellent discriminatory power across risk strata: low-risk group (4.6%), moderate-risk group (19.3%), high-risk group (53.1%), and uncertain group (27.3%).\u003c/p\u003e \u003cp\u003eThe monotonically increasing trend from low to high risk (4.6% \u0026rarr; 19.3% \u0026rarr; 53.1%) confirms the clinical validity of the stratification system, with the difference in the incidence of cognitive impairment between the low-risk and high-risk groups exceeding a 10-fold increase (OR\u0026thinsp;=\u0026thinsp;23.62). Notably, the severity of depression (mean CESD\u0026thinsp;\u0026minus;\u0026thinsp;10) increased progressively with rising risk levels: 3.9 in the low-risk group, 6.2 in the moderate-risk group, 12.2 in the high-risk group, and 8.3 in the uncertain group, indicating that the burden of depressive symptoms is a key driver of cognitive risk classification.\u003c/p\u003e \u003cp\u003eThe \u0026ldquo;uncertain\u0026rdquo; category (accounting for 10.0% of cases) exhibited unique phenotypic characteristics (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e): while depression scores were elevated to a degree comparable to high-risk individuals, functional status paradoxically remained normal, suggesting complex clinical presentations that are difficult to classify using simple algorithms. The incidence of cognitive impairment in this group was moderate (27.3%), necessitating management through individualized expert assessment rather than standardized treatment protocols.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eRisk stratification validation: observed cognitive impairment rates and clinical characteristics by risk stratum.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRisk Stratum\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eN (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCI Cases (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eOR (vs Low)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eMean Age\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eMean CESD-10\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eMean Edu\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLow\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e4,418 (14.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e202 (4.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.00 (Ref)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e64.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e3.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e2.8\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedium\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e13,367 (42.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2,579 (19.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.99\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e66.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e6.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e2.2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHigh\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e10,631 (33.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e5,644 (53.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e23.62\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e70.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e12.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e1.5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUncertain\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3,158 (10.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e862 (27.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e7.84\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e67.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e8.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e1.9\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"7\"\u003e\u003cem\u003eNote: CI\u0026thinsp;=\u0026thinsp;Cognitive Impairment; OR\u0026thinsp;=\u0026thinsp;Odds Ratio; Edu\u0026thinsp;=\u0026thinsp;Education (years). The \"Uncertain\" category identifies cases with high model disagreement (\u0026ge;\u0026thinsp;90th percentile epistemic uncertainty) warranting individualized clinical evaluation.\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003e3.5 Clinical Decision Pathway for Depression Management\u003c/h2\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e illustrates a comprehensive clinical decision pathway that translates risk stratification into actionable recommendations for the management of older adults with depression. This pathway converts a four-tiered system into a tiered care plan based on the World Health Organization's \"Integrated Care for Older People (ICOPE) Guidelines\" and the guiding principles of The Lancet Commission on Dementia Prevention [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e].\u003c/p\u003e \u003cp\u003e(1) Low-risk patients with depression: Implement standard depression treatment protocols and conduct annual cognitive function monitoring, with a focus on alleviating depressive symptoms through routine lifestyle counseling.\u003c/p\u003e \u003cp\u003e(2) Moderate-risk patients with depression: Intensify follow-up to every 6 months, including re-screening for depression and optimized management of vascular risk factors. Evidence-based lifestyle interventions (exercise, cognitive training, sleep hygiene) are recommended.\u003c/p\u003e \u003cp\u003e(3) High-risk patients with depression: Prioritize aggressive treatment for depression, refer for specialized cognitive assessment and neuropsychological testing, and consider neuroprotective interventions. Regular monitoring is required (every 3 months).\u003c/p\u003e \u003cp\u003e(4) Uncertain cases: Require multidisciplinary team consultation, comprehensive cognitive function testing, and a 3-month follow-up before definitive risk stratification. These patients require individualized clinical judgment rather than adherence to algorithm-driven standardized treatment protocols.\u003c/p\u003e \u003cp\u003e These recommendations represent expert consensus based on existing guidelines and require prospective validation through implementation studies before widespread clinical application.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003e3.6 Subgroup Analysis\u003c/h2\u003e \u003cp\u003e \u003cb\u003eSupplementary Table S5\u003c/b\u003e presents the results of stratified analyses by sex, age group, and urban/rural residence. The association between depressive symptoms and cognitive function remained consistent across all subgroups, with odds ratios ranging from 1.89 to 2.45. No significant effect modification was observed (\u003cem\u003eP\u003c/em\u003e-values for interactions with all stratification variables were \u0026gt;\u0026thinsp;0.05), indicating that the increased cognitive risk associated with depressive symptoms is robust across different demographic subgroups. The risk stratification model maintained stable discriminatory performance across all subgroups (area under the curve ranging from 0.76 to 0.80).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003e3.7 Calibration and Clinical Utility\u003c/h2\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e presents model calibration and decision curve analysis. \u003cb\u003ePanel A\u003c/b\u003e shows the calibration curve, demonstrating good agreement between predicted probabilities and observed frequencies across deciles. The smooth calibration line closely tracks the perfect calibration diagonal, with narrow confidence bands indicating stable calibration. \u003cb\u003ePanel B\u003c/b\u003e presents decision curve analysis results. The BARNN\u0026thinsp;+\u0026thinsp;model provided superior net benefit compared to treat-all and treat-none strategies across the clinically relevant threshold probability range of 20\u0026ndash;50%. The net benefit advantage was most pronounced in the 30\u0026ndash;45% threshold range, which corresponds to typical clinical decision thresholds for cognitive screening referrals.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e evaluates the predictive performance of different models for cognitive impairment using ROC curves. The AUC values for the ensemble model, logistic regression, and random forest were 0.781, 0.781, and 0.780, respectively, indicating that all three models possess above-average discriminatory ability and perform similarly.\u003c/p\u003e \u003cp\u003eThe risk stratification system based on the models showed that the actual incidence rates of cognitive impairment in the low-, medium-, high-risk, and uncertain groups were 4.6%, 19.3%, 53.1%, and 27.3%, respectively. Compared with the low-risk group, the high-risk group had a significantly increased risk of developing cognitive impairment (OR\u0026thinsp;=\u0026thinsp;23.62), suggesting that this stratification system can effectively identify high-risk individuals.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eThis study leverages data from the China Health and Retirement Longitudinal Study (CHARLS), a large-scale, nationally representative longitudinal survey of Chinese adults aged 45 and above, to systematically explore the association between depressive symptoms and cognitive impairment. Over a 9-year follow-up period, we analyzed 31,570 individual-wave observations from 12,494 unique participants, providing robust evidence of a significant severity-graded association between depressive symptoms and cognitive impairment in Chinese older adults. To our knowledge, this is the largest longitudinal study of its kind in China to date, and it is the first to translate epidemiological findings into a clinically actionable risk stratification system tailored to elderly patients with depression, effectively bridging the gap between research and clinical practice.\u003c/p\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003e4.1 The Severity-Graded Association Between Depressive Symptoms and Cognitive Impairment: Confirmation and Novelty\u003c/h2\u003e \u003cp\u003eOur core finding\u0026mdash;that depressive symptoms double the risk of cognitive impairment (OR\u0026thinsp;=\u0026thinsp;2.07, 95% CI: 1.97\u0026ndash;2.18)\u0026mdash;is highly consistent with meta-analytic evidence from Western populations [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Meanwhile, this study validates this association in a large, heterogeneous Chinese cohort. Despite significant differences in cultural backgrounds, healthcare systems, and epidemiological characteristics between China and Western countries [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e], this consistency suggests that the association between depressive symptoms and cognitive impairment may have cross-population universal significance.\u003c/p\u003e \u003cp\u003eThe dose-response relationship clearly illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eB holds significant clinical implications. The prevalence of cognitive impairment gradually increased from 20.8% (normal mood state) to 46.0% (major depression), suggesting that the intensity of intervention should be adjusted according to the severity of depression. Notably, even subclinical depressive symptoms (CESD-10 score of 6\u0026ndash;10) are associated with an increased risk of cognitive impairment (29.5% vs. 20.8%), suggesting that symptoms below the diagnostic threshold warrant clinical attention [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. The clear dose-response relationship between the severity of depression and cognitive impairment reinforces causal inferences and holds significant clinical implications. If this association is indeed causal, depression screening and treatment could serve as a primary prevention strategy for cognitive decline. The study found that even mild depressive symptoms are associated with an increased risk of cognitive impairment (36.1% vs. 48.3% in asymptomatic individuals), suggesting that early intervention at the subclinical stage may be beneficial.\u003c/p\u003e \u003cp\u003eThe longitudinal trajectories in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA provide mechanistic insights into the management of mood disorders. Analysis of cognitive decline trajectories based on baseline depression severity suggests that chronic depressive symptoms may accelerate the process of cognitive aging through cumulative neurotoxic effects, such as inflammatory responses, abnormal cortisol regulation, and reduced neurotrophic support [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. From the perspective of depression diagnosis and treatment, these findings underscore the importance of incorporating cognitive monitoring into longitudinal management strategies for older adults with depression.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003e4.2 Clinical Implications: From Risk Stratification to Individualized Management\u003c/h2\u003e \u003cp\u003eOne of the core contributions of this study is the construction of a clinically operable four-tier risk stratification system, which accurately fills the critical gap in translating epidemiological evidence into clinical guidance. Based on an ensemble machine learning model (AUC\u0026thinsp;=\u0026thinsp;0.781), this system classifies elderly patients with depression into four groups: low-risk, moderate-risk, high-risk, and uncertain-risk, with the actual incidence of cognitive impairment being 4.6%, 19.3%, 53.1%, and 27.3%, respectively (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). The more than 10-fold difference in the risk of cognitive impairment between the low-risk and high-risk groups (OR\u0026thinsp;=\u0026thinsp;23.62) provides a clear basis for optimizing the allocation of clinical resources and directing intensive interventions to the population that would benefit the most.\u003c/p\u003e \u003cp\u003eBased on this stratification system, this study proposes a corresponding clinical decision pathway (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e), which is fully consistent with the core principles of the World Health Organization's \u0026ldquo;Integrated Care for Older People (ICOPE) Guidelines\u0026rdquo; and The Lancet Commission on Dementia Prevention [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. For low-risk patients, standard depression treatment and annual cognitive monitoring are sufficient; moderate-risk patients require enhanced follow-up every 6 months, optimized management of vascular risk factors, and adjunctive lifestyle interventions; high-risk patients need priority for active depression treatment, referral for specialized cognitive assessment, and close monitoring every 3 months; while the 10.0% \u0026ldquo;uncertain-risk group,\u0026rdquo; which exhibits a unique phenotype of severe depressive symptoms but intact functional status (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e), requires multidisciplinary consultation and individualized assessment. This stratified treatment plan achieves precise matching between intervention intensity, depression severity, and cognitive risk, surpassing the traditional \u0026ldquo;one-size-fits-all\u0026rdquo;management model.\u003c/p\u003e \u003cp\u003eThe definition of the \u0026ldquo;ncertain-risk group\u0026rdquo; represents an important methodological innovation, which directly addresses the \u0026ldquo;black box\u0026rdquo; problem in the clinical application of artificial intelligence (AI). As clearly shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, although patients in this group have depression scores comparable to those in the high-risk group, their functional status remains intact, making them difficult to accurately classify by algorithms. By explicitly identifying such cases and recommending individualized judgment, this framework fully reflects the regulatory principle that AI should \u0026ldquo;enhance rather than replace\u0026rdquo; clinical decision-making [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e], establishing a human-machine collaborative clinical decision-making model and improving the credibility and application value of the risk stratification system.\u003c/p\u003e \u003cp\u003eIn addition, the study results strongly call for the integration of cognitive function monitoring into routine care for elderly depression. Current clinical assessments of elderly depression mostly focus on symptom remission and functional recovery, with relative neglect of cognitive outcomes. Given the strong association between the two, we recommend the use of a comprehensive cognitive scale based on the Health and Retirement Study (HRS) as a routine follow-up indicator for all elderly patients with depression. Particularly for high-risk individuals, preliminary evidence suggests that active depression treatment (especially selective serotonin reuptake inhibitors) may delay cognitive decline [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e], highlighting the importance of prioritizing intensive treatment in this subgroup.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003e4.3 Public Health Implications: Toward Geriatric Mental Health and Cognitive Prevention\u003c/h2\u003e \u003cp\u003eThe results of this study have far-reaching guiding significance for public health policies in the context of China's rapid aging. The study shows that the prevalence of clinically significant depressive symptoms in this cohort is as high as 34.2%, indicating that depression has become a major public health burden among Chinese older adults. If the association between the two is indeed causal, then depression screening and treatment may become a core strategy for preventing cognitive decline at the population level, which is of extremely high practical urgency in the context of the increasing burden of dementia in China [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eA key public health recommendation of this study is to promote the in-depth integration of mental health and brain health management in geriatric care. The traditional model often treats depression and cognitive impairment in isolation, but this study confirms a strong synergistic effect between the two. Therefore, we advocate the establishment of a collaborative care model, integrating depression screening into cognitive risk assessment and cognitive assessment into depression follow-up, to achieve \u0026ldquo;simultaneous physical and mental treatment\u0026rdquo; and shift from passive treatment to active prevention, thereby improving the overall clinical outcomes of older adults.\u003c/p\u003e \u003cp\u003eMeanwhile, the method for quantifying prediction uncertainty in this study has universal implications for the application of AI in the medical field. By identifying 10% of cases with unreliable model predictions, we demonstrate how to apply AI-assisted decision-making in a responsible and transparent manner. This approach effectively alleviates clinicians' trust concerns about the AI \u0026ldquo;black box\u0026rdquo; and promotes the rational and safe application of AI tools in geriatric care. As AI penetrates deeper into the field of geriatrics, such transparency will be crucial for improving tool effectiveness and ensuring medical quality.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003e4.4 Methodological Innovations\u003c/h2\u003e \u003cp\u003eThe model demonstrates statistically significant improvements over traditional machine learning methods(Figure \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). Although the absolute AUC differences appear small (ΔAUC\u0026thinsp;=\u0026thinsp;0.006\u0026ndash;0.009) [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e], these improvements are clinically significant at the population level. For a screening program covering 100,000 people, such improvements could accurately reclassify thousands of patients. Furthermore, the interpretability approach based on SHAP analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e) addresses a key barrier in clinical AI applications\u0026mdash;the \u0026ldquo;black box\u0026rdquo; problem [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. By decomposing individual predictions into feature-level contributions, clinicians can understand why patients are classified into specific risk categories. The study found that depression severity, age, and educational level consistently emerged as the primary predictors (\u003cb\u003eSupplementary Figure S3\u003c/b\u003e), a finding consistent with clinical intuition and epidemiological evidence [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e] that is expected to enhance clinicians\u0026rsquo; confidence.\u003c/p\u003e \u003cp\u003eA key innovation of this study lies in the integration of techniques for quantifying cognitive uncertainty. Traditional machine learning models provide only point estimates without conveying prediction confidence, which may lead to inappropriate clinical decisions. Our method identifies 10% of \u0026ldquo;uncertain\u0026rdquo; cases\u0026mdash;cases where the model\u0026rsquo;s prediction is unreliable and requires expert review. This feature aligns with regulatory guidelines emphasizing transparency in AI medical devices [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e] and supports human-machine collaborative decision-making mechanisms.\u003c/p\u003e \u003cp\u003eThe four-tier risk stratification system provides actionable guidance for clinical practice: low-risk individuals can undergo routine follow-up; moderate-risk individuals require enhanced monitoring; high-risk individuals should be referred for comprehensive evaluation and intervention; and uncertain cases require expert consultation before clinical decisions are made. This refined stratification mechanism enables personalized treatment pathways and optimizes the allocation of healthcare resources.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec23\" class=\"Section2\"\u003e \u003ch2\u003e4.5 Study Limitations and Future Directions\u003c/h2\u003e \u003cp\u003eThis study has several limitations that require careful interpretation. First, the definition of cognitive impairment was based on epidemiological percentile thresholds (below the 31st percentile after adjusting for age and education), rather than clinical diagnostic criteria (such as the Petersen criteria or DSM-5 criteria). Although sensitivity analyses using the 25th and 33rd percentiles yielded consistent results, future studies still need to use clinically confirmed data for validation to further promote clinical translation.\u003c/p\u003e \u003cp\u003eSecond, the CESD-10 scale is only a symptom screening tool and cannot be used as a basis for the clinical diagnosis of major depressive disorder, nor can it distinguish between depressive subtypes. The CHARLS data also lack information on antidepressant use, and these factors may affect the observed association and limit the generalizability of the conclusions. Future studies should include clinical diagnostic data and treatment information to further clarify the complex relationships.\u003c/p\u003e \u003cp\u003eThird, although the sample is nationally representative, the extrapolation of the study conclusions to other populations (such as elderly groups in low- and middle-income countries and populations with different cultural backgrounds) still requires more validation. At the same time, the observational study design makes it difficult to establish a clear causal relationship; future randomized controlled trials are needed to directly verify whether depression treatment can reduce the risk of cognitive impairment.\u003c/p\u003e \u003cp\u003eFuture research can focus on three key directions: first, validate the risk stratification system in independent cohorts and real clinical settings to evaluate its practical application value; second, design randomized controlled trials to explore the preventive effects of various depression intervention strategies, including pharmacotherapy, psychotherapy, and collaborative care, on cognitive impairment; third, further explore the potential pathophysiological mechanisms by which depression severity affects cognitive decline, providing a scientific basis for the development of targeted interventions.\u003c/p\u003e \u003c/div\u003e"},{"header":"5. Conclusions","content":"\u003cp\u003eIn summary, based on a large-scale national cohort, this study provides strong evidence of a significant, severity-graded association between depressive symptoms and cognitive impairment in Chinese older adults. The clear dose-response relationship provides a solid basis for implementing stratified clinical management and adjusting intervention intensity according to depression severity. The risk stratification system and clinical decision pathway constructed in this study provide clinicians with directly operable guidance, helping to achieve individualized care and optimal allocation of medical resources. The results support the integration of cognitive function monitoring into the routine diagnosis and treatment process of elderly depression, and emphasize the importance of depression treatment as a potential strategy for preventing cognitive decline. Faced with the severe challenge of population aging in China, the findings of this study provide key scientific support and practical pathways for improving the quality of care for elderly depression and reducing the burden of cognitive impairment.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e \u003ch2\u003eEthics Approval:\u003c/h2\u003e \u003cp\u003eThe study protocol was approved by the Biomedical Ethics Review Committee of Peking University. All participants provided written informed consent.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eFunding Declaration\u003c/h2\u003e \u003cp\u003eThis study relies on the Shandong Province Education Teaching Planning Project (Innovation and Practice of\u0026ldquo;Combination of Medicine, Health and Nutrition, Integration of Teaching, Production and Research\u0026rdquo;Parenting Mode of Higher Vocational Intelligent Recreation Professional Group, NO. B20G590805)\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eQ. Z. and P.-L. Z. conducted the whole study conception and design. Y.-F. Z. and X.-S. R. prepared the draft of the article and revised it critically for important intellectual content. D. -W. Y. approved the submitted version. All authors have read and agreed to the published version of the manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgments:\u003c/h2\u003e \u003cp\u003eWe thank the China Health and Retirement Longitudinal Study (CHARLS) team for providing access to the data.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe data used in this study are publicly available from the China Health and Retirement Longitudinal Study (CHARLS) repository (https://charls.pku.edu.cn/). The detailed data access procedures and usage policies are available on the official website of CHARLS.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAlexopoulos GS. Mechanisms and treatment of late-life depression. Transl Psychiatry. 2019;9(1):188. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41398-019-0514-6\u003c/span\u003e\u003cspan address=\"10.1038/s41398-019-0514-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLivingston G, Huntley J, Sommerlad A, et al. Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet. 2020;396(10248):413\u0026ndash;46. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/S0140-6736(20)30367-6\u003c/span\u003e\u003cspan address=\"10.1016/S0140-6736(20)30367-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDiniz BS, Butters MA, Albert SM, Dew MA, Reynolds CF. Late-life depression and risk of vascular dementia and Alzheimer's disease: systematic review and meta-analysis. Br J Psychiatry. 2013;202(5):329\u0026ndash;35. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1192/bjp.bp.112.118307\u003c/span\u003e\u003cspan address=\"10.1192/bjp.bp.112.118307\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOwnby RL, Crocco E, Acevedo A, John V, Loewenstein D. Depression and risk for Alzheimer disease: systematic review, meta-analysis, and metaregression analysis. Arch Gen Psychiatry. 2006;63(5):530\u0026ndash;8. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1001/archpsyc.63.5.530\u003c/span\u003e\u003cspan address=\"10.1001/archpsyc.63.5.530\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBennett S, Thomas AJ. Depression and dementia: cause, consequence or coincidence? Maturitas. 2014;79(2):184\u0026ndash;90. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.maturitas.2014.05.009\u003c/span\u003e\u003cspan address=\"10.1016/j.maturitas.2014.05.009\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eByers AL, Yaffe K. Depression and risk of developing dementia. Nat Rev Neurol. 2011;7(6):323\u0026ndash;31. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/nrneurol.2011.60\u003c/span\u003e\u003cspan address=\"10.1038/nrneurol.2011.60\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSapolsky RM. Glucocorticoids and hippocampal atrophy in neuropsychiatric disorders. Arch Gen Psychiatry. 2000;57(10):925\u0026ndash;35. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1001/archpsyc.57.10.925\u003c/span\u003e\u003cspan address=\"10.1001/archpsyc.57.10.925\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBartels C, Wagner M, Wolfsgruber S, Ehrenreich H, Schneider A. Impact of SSRI therapy on risk of conversion from mild cognitive impairment to Alzheimer's dementia in individuals with previous depression. Am J Psychiatry. 2018;175(3):232\u0026ndash;41. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1176/appi.ajp.2017.17040404\u003c/span\u003e\u003cspan address=\"10.1176/appi.ajp.2017.17040404\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJia L, Du Y, Chu L, et al. Prevalence, risk factors, and management of dementia and mild cognitive impairment in adults aged 60 years or older in China: a cross-sectional study. Lancet Public Health. 2020;5(12):e661\u0026ndash;71. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/S2468-2667(20)30185-7\u003c/span\u003e\u003cspan address=\"10.1016/S2468-2667(20)30185-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao Y, Hu Y, Smith JP, Strauss J, Yang G. Cohort profile: the China Health and Retirement Longitudinal Study (CHARLS). Int J Epidemiol. 2014;43(1):61\u0026ndash;8. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/ije/dys203\u003c/span\u003e\u003cspan address=\"10.1093/ije/dys203\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAndresen EM, Malmgren JA, Carter WB, Patrick DL. Screening for depression in well older adults: evaluation of a short form of the CES-D. Am J Prev Med. 1994;10(2):77\u0026ndash;84. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/s0749-3797(18)30622-6\u003c/span\u003e\u003cspan address=\"10.1016/s0749-3797(18)30622-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCheng ST, Chan AC. The Center for Epidemiologic Studies Depression Scale in older Chinese: thresholds for long and short forms. Int J Geriatr Psychiatry. 2005;20(5):465\u0026ndash;70. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/gps.1314\u003c/span\u003e\u003cspan address=\"10.1002/gps.1314\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLei X, Smith JP, Sun X, Zhao Y. Gender differences in cognition in China and reasons for change over time: evidence from CHARLS. J Econ Ageing. 2014;4:46\u0026ndash;55. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.jeoa.2013.11.001\u003c/span\u003e\u003cspan address=\"10.1016/j.jeoa.2013.11.001\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44(3):837\u0026ndash;45. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2307/2531595\u003c/span\u003e\u003cspan address=\"10.2307/2531595\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVickers AJ, Elkin EB. Decision curve analysis: a novel method for evaluating prediction models. Med Decis Mak. 2006;26(6):565\u0026ndash;74. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1177/0272989X06295361\u003c/span\u003e\u003cspan address=\"10.1177/0272989X06295361\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eK\u0026ouml;hler S, van Boxtel M, Jolles J, Verhey F. Depressive symptoms and risk for dementia: a 9-year follow-up of the Maastricht Aging Study. Am J Geriatr Psychiatry. 2011;19(10):902\u0026ndash;5. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1097/JGP.0b013e31821f1b6a\u003c/span\u003e\u003cspan address=\"10.1097/JGP.0b013e31821f1b6a\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUS Food and Drug Administration. Artificial Intelligence/Machine Learning (AI/ML)-Based Software as a Medical Device (SaMD) Action Plan. FDA; 2021.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKaton W, Pedersen HS, Ribe AR, et al. Effect of depression and diabetes mellitus on the risk for dementia: a national population-based cohort study. JAMA Psychiatry. 2015;72(6):612\u0026ndash;9. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1001/jamapsychiatry.2015.0082\u003c/span\u003e\u003cspan address=\"10.1001/jamapsychiatry.2015.0082\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCherbuin N, Kim S, Anstey KJ. Dementia risk estimates associated with measures of depression: a systematic review and meta-analysis. BMJ Open. 2015;5(12):e008853. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1136/bmjopen-2015-008853\u003c/span\u003e\u003cspan address=\"10.1136/bmjopen-2015-008853\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSteyerberg EW, Vickers AJ, Cook NR, et al. Assessing the performance of prediction models: a framework for traditional and novel measures. Epidemiology. 2010;21(1):128\u0026ndash;38. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1097/EDE.0b013e3181c30fb2\u003c/span\u003e\u003cspan address=\"10.1097/EDE.0b013e3181c30fb2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLundberg SM, Lee SI. A unified approach to interpreting model predictions. Adv Neural Inf Process Syst. 2017;30:4765\u0026ndash;74.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMiller AH, Maletic V, Raison CL. Inflammation and its discontents: the role of cytokines in the pathophysiology of major depression. Biol Psychiatry. 2009;65(9):732\u0026ndash;41. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.biopsych.2008.11.029\u003c/span\u003e\u003cspan address=\"10.1016/j.biopsych.2008.11.029\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStern Y, Cognitive reserve. Neuropsychologia. 2009;47(10):2015\u0026ndash;28. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.neuropsychologia.2009.03.004\u003c/span\u003e\u003cspan address=\"10.1016/j.neuropsychologia.2009.03.004\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTaylor WD, Aizenstein HJ, Alexopoulos GS. The vascular depression hypothesis: mechanisms linking vascular disease with depression. Mol Psychiatry. 2013;18(9):963\u0026ndash;74. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/mp.2013.20\u003c/span\u003e\u003cspan address=\"10.1038/mp.2013.20\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGallagher D, Kiss A, Lanctot K, Herrmann N. Depression and risk of Alzheimer dementia: a longitudinal analysis to determine predictors of increased risk among older adults with depression. Am J Geriatr Psychiatry. 2018;26(8):819\u0026ndash;27. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.jagp.2018.05.002\u003c/span\u003e\u003cspan address=\"10.1016/j.jagp.2018.05.002\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSaczynski JS, Beiser A, Seshadri S, Auerbach S, Wolf PA, Au R. Depressive symptoms and risk of dementia: the Framingham Heart Study. Neurology. 2010;75(1):35\u0026ndash;41. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1212/WNL.0b013e3181e62138\u003c/span\u003e\u003cspan address=\"10.1212/WNL.0b013e3181e62138\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGBD 2019 Dementia Forecasting Collaborators. Estimation of the global prevalence of dementia in 2019 and forecasted prevalence in 2050: an analysis for the Global Burden of Disease Study 2019. Lancet Public Health. 2022;7(2):e105\u0026ndash;25. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/S2468-2667(21)00249-8\u003c/span\u003e\u003cspan address=\"10.1016/S2468-2667(21)00249-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen X, Giles J, Yao Y, et al. The path to healthy ageing in China: a Peking University-Lancet Commission. Lancet. 2022;400(10367):1967\u0026ndash;2006. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/S0140-6736(22)01546-X\u003c/span\u003e\u003cspan address=\"10.1016/S0140-6736(22)01546-X\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang Y, Wang Y, Wang H, et al. Prevalence of mental disorders in China: a cross-sectional epidemiological study. Lancet Psychiatry. 2019;6(3):211\u0026ndash;24. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/S2215-0366(18)30511-X\u003c/span\u003e\u003cspan address=\"10.1016/S2215-0366(18)30511-X\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBarnes DE, Yaffe K, Byers AL, McCormick M, Schaefer C, Whitmer RA. Midlife vs late-life depressive symptoms and risk of dementia: differential effects for Alzheimer disease and vascular dementia. Arch Gen Psychiatry. 2012;69(5):493\u0026ndash;8. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1001/archgenpsychiatry.2011.1481\u003c/span\u003e\u003cspan address=\"10.1001/archgenpsychiatry.2011.1481\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCollins GS, Reitsma JB, Altman DG, Moons KGM. Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD): the TRIPOD Statement. BMJ. 2015;350:g7594. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1136/bmj.g7594\u003c/span\u003e\u003cspan address=\"10.1136/bmj.g7594\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBreiman L. Random forests. Mach Learn. 2001;45(1):5\u0026ndash;32. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1023/A:1010933404324\u003c/span\u003e\u003cspan address=\"10.1023/A:1010933404324\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003c/ol\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":"Depressive Symptoms, Cognitive Impairment, Dose-Response Relationship, Risk Stratification, Modifiable Risk Factors","lastPublishedDoi":"10.21203/rs.3.rs-9460240/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9460240/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eDepression and cognitive impairment often co-occur in older adults, but their clinical implications for the management of depression have not been fully explored. Determining whether the severity of depression can predict cognitive outcomes will help inform the development of tiered treatment strategies for depression in later life.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eThis study analyzed 31,570 individual follow-up records from 12,494 independent participants aged\u0026thinsp;\u0026ge;\u0026thinsp;45 years in the China Health and Retirement Longitudinal Study (CHARLS, 2011\u0026ndash;2020). Depressive symptoms were assessed using the CESD-10 scale. We developed a machine learning-based risk stratification system to identify depressed individuals at highest cognitive risk, who are likely to benefit most from intensified interventions.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eDepressive symptoms (CESD-10 score\u0026thinsp;\u0026ge;\u0026thinsp;10) were significantly associated with cognitive impairment (OR\u0026thinsp;=\u0026thinsp;2.07, 95% CI: 1.97\u0026ndash;2.18), with incidence rates of cognitive impairment of 39.6% in the symptomatic group versus 24.1% in the asymptomatic group. A significant dose-response relationship was observed: statistically significant associations were found across all categories, normal mood (20.8%), mild symptoms (29.5%), moderate symptoms (36.6%), and severe symptoms (46.0%), with a trend P-value\u0026thinsp;\u0026lt;\u0026thinsp;0.001. Even subclinical symptoms (CESD-10 score 6\u0026ndash;10) significantly increased the risk. The quartile risk stratification system demonstrated excellent discriminatory power, with observed rates of cognitive impairment ranging from 4.6% (low risk) to 53.1% (high risk).\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eThe severity of depressive symptoms is significantly and gradually associated with the risk of cognitive impairment, which has important implications for the clinical management of depression in older adults. The findings support the integration of cognitive function monitoring into the depression care system and the dynamic adjustment of intervention intensity based on depression severity and cognitive risk characteristics.\u003c/p\u003e","manuscriptTitle":"Depressive Symptoms as a Modifiable Risk Factor for Cognitive Impairment: Severity-Graded Associations and Clinical Implications from a 9-Year Longitudinal Study in Chinese Older Adults","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-11 05:34:57","doi":"10.21203/rs.3.rs-9460240/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":"50f17285-4721-4742-84fe-2dec67a65078","owner":[],"postedDate":"May 11th, 2026","published":true,"recentEditorialEvents":[{"type":"decision","content":"Rejected","date":"2026-05-07T09:34:30+00:00","index":"","fulltext":""},{"type":"reviewerAgreed","content":"293964122966348754134857031903512059417","date":"2026-05-05T21:07:45+00:00","index":40,"fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-05T09:32:17+00:00","index":37,"fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-30T23:55:41+00:00","index":36,"fulltext":""},{"type":"reviewerAgreed","content":"164789094068337497451683624523502369061","date":"2026-04-30T01:07:16+00:00","index":35,"fulltext":""},{"type":"reviewerAgreed","content":"217670557714738774598535589596017606393","date":"2026-04-29T22:50:47+00:00","index":34,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-05-11T05:35:14+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-11 05:34:57","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9460240","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9460240","identity":"rs-9460240","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}