Bridge symptoms connecting depressive, sleep, and cognitive domains in rural older adults with type 2 diabetes: a symptom network approach Authors: Xueyan Liu, Zhaotai Wang, Yingjuan Cao Published: 2026 ABSTRACT AbstractBackgroundRural older adults with type 2 diabetes mellitus (T2DM) commonly experience depressive symptoms, sleep disturbances, and cognitive impairment, which frequently co-occur and interact. However, the key symptoms linking these domains remain unclear, and network analysis may help identify central and bridge symptoms within this cross-domain system.MethodsRural older adults with T2DM were assessed for depressive symptoms using the Geriatric Depression Scale-15, cognitive performance using the Montreal Cognitive Assessment—Basic, and sleep quality using the Pittsburgh Sleep Quality Index. We constructed a network structure, and calculated index of strength and bridge expected influence for each symptom. Furthermore, a comparative analysis of the network structure across gender was conducted.ResultsA total of 885 participants were included. The estimated network revealed a stable and interconnected symptom system spanning depressive symptoms, sleep disturbances, and cognitive deficits. Symptoms related to abstraction and calculation and helplessness emerged as the most central nodes. Abstraction ability, perceived memory problems, and concentration difficulties functioned as key bridge symptoms linking cognitive, depressive, and sleep-related domains. Sensitivity analyses supported the robustness of the network, and no significant differences in edge weights were observed between male and female participants.ConclusionsThis study highlights a connected cross-domain symptom system underlying depression, sleep disturbance, and cognitive impairment in older adults with T2DM. Central and bridge symptoms may represent pivotal mechanisms through which affective and sleep-related problems interact with cognitive decline. Targeting these high-impact symptoms may offer a theoretically informed strategy for disrupting maladaptive symptom interactions and improving integrated mental and cognitive health outcomes.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12877-026-07559-8.Keywords: Cognitive performance, Sleep quality, Depressive symptoms, Network analysis, Type 2 diabetes mellitus, Older adults Introduction Cognitive impairment represents a growing public health challenge in ageing societies, particularly among older adults living with chronic metabolic conditions [34]. Rather than occurring in isolation, cognitive decline in later life is frequently embedded within a constellation of psychological and behavioral disturbances that jointly shape cognitive trajectories. Older adults with type 2 diabetes mellitus (T2DM) constitute a particularly vulnerable group. In this population, metabolic dysregulation, affective symptoms, and lifestyle-related factors often converge, accelerating cognitive deterioration and increasing the risk of dementia [15]. T2DM has been consistently identified as a major risk factor for cognitive impairment and dementia, with evidence linking diabetes to faster cognitive decline, increased progression from mild cognitive impairment to Alzheimer’s disease, and adverse structural brain changes [14, 43]. However, cognitive outcomes in T2DM cannot be fully explained by metabolic mechanisms alone. Psychological and behavioral factors, particularly depressive symptoms and sleep disturbances, are highly prevalent in this population and have been independently associated with poorer cognitive performance across multiple domains. Depressive symptoms are common among older adults with T2DM and are robustly associated with both the onset and progression of cognitive impairment [44, 50]. Longitudinal evidence suggests a bidirectional relationship, whereby depressive symptoms increase vulnerability to subsequent cognitive decline, while declining cognitive function may further exacerbate depressive experiences [4, 22]. Sleep disturbance is likewise highly prevalent in older adults with T2DM and has been linked to impaired glycemic control, heightened depressive symptoms, and deficits in memory, attention, and executive function [35, 41]. Importantly, depressive symptoms and sleep problems frequently co-occur and may interact synergistically, amplifying their detrimental effects on cognitive functioning (R. Ding et al., [13]. Despite substantial evidence linking depression, sleep, and cognition, existing research has predominantly examined these associations at the construct or syndrome level, typically relying on total scale scores and variable-centered analytic approaches (J. Chen et al., [7]. Such methods implicitly assume that symptoms within a construct are homogeneous and exert uniform effects on cognitive outcomes. This assumption may obscure clinically meaningful heterogeneity, as specific depressive symptoms and distinct components of sleep disturbance may play disproportionately influential roles in cognitive functioning and may serve as critical links between affective, sleep-related, and cognitive domains. Depressive symptoms, sleep disturbances, and cognitive impairment are theorized to interact through interconnected biological, psychological, and behavioral pathways. Negative affect may disrupt sleep patterns, poor sleep may impair attention and memory, and cognitive decline may exacerbate depressive experiences, creating feedback loops that reinforce symptom co-occurrence [1, 37]. This conceptual framework suggests that these domains cannot be fully understood in isolation and that symptom-level interactions may drive cognitive vulnerability. Network analysis offers a methodological approach that aligns with this conceptual framework. By representing depressive, sleep, and cognitive components as nodes and their conditional associations as edges, network models can identify central symptoms that exert strong influence within each domain, as well as bridge symptoms that connect domains. This symptom-level perspective allows for a detailed examination of cross-domain interactions and provides insights into potential targets for intervention that traditional variable-centered methods may overlook. However, existing network-based studies have primarily focused on single domains such as depression or sleep, or have examined cognitive function separately, and relatively few studies have simultaneously examined depressive symptoms, sleep quality, and cognitive performance within a unified symptom network. Furthermore, network studies in populations with T2DM remain limited, and research specifically focusing on rural older adults with T2DM is particularly scarce. In addition, the symptom-level pathways linking cognitive and depressive domains, as well as potential bridge symptoms connecting these domains, remain insufficiently understood in this high-risk population. Understanding these symptom-level interconnections is especially important in rural older adults with T2DM, who often face compounded risks due to limited healthcare access, lower educational attainment, and reduced availability of specialized mental health and cognitive services. Identifying central and bridge symptoms within a cross-domain symptom system may help inform low-cost, scalable, and symptom-focused assessment and intervention strategies that are feasible within rural primary care and community health settings. Therefore, the present study aimed to examine a cross-domain node-level network encompassing depressive symptoms, sleep quality components, and cognitive performance in rural older adults with T2DM using a cross-sectional network analysis approach. By identifying central and bridge nodes within and across affective, sleep-related, and cognitive domains, this study aimed to characterize the interrelationships among these variables and to provide a potential theoretical basis for future symptom-oriented assessment and intervention in this high-risk population. This study extends existing network-based research by integrating cognitive performance into depression–sleep symptom networks, focusing on a rural older population with T2DM, and identifying central and bridge symptoms across cognitive, affective, and sleep domains, thereby providing a more comprehensive symptom-level understanding of their interrelationships in this vulnerable population. Methods Design This was a cross-sectional observational study using symptom network analysis conducted among rural older adults with T2DM. Data were collected through face-to-face assessments as part of a community-based health investigation focusing on psychological well-being, sleep quality, and cognitive performance. The study formed part of a dementia health action programme implemented by the Chinese government, aiming to promote early identification and management of cognitive impairment among older adults at the community level. The reporting of this study adheres to the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines. Setting and participants Participants were recruited from rural communities using a convenience sampling strategy, selected for its feasibility in accessing older adults with T2DM in geographically dispersed rural areas. Recruitment was facilitated by local healthcare providers and community health workers to ensure adequate participation and data quality. Eligible participants were rural residents with a confirmed diagnosis of T2DM who had established health records in local healthcare facilities. Inclusion criteria were as follows: (1) age ≥ 60 years; (2) diagnosis of T2DM; (3) residence in a rural area for at least 6 months; (4) ability to complete face-to-face interviews and assessments; (5) provision of written informed consent. Participants were excluded if they had: (1) severe visual, hearing, or communication impairments that hindered assessment; (2) a diagnosis of dementia or other severe neurological disorders; (3) a history of severe mental disorders (e.g., schizophrenia or bipolar disorder); (4) severe somatic diseases or organ dysfunctions that significantly impaired cognitive assessment or participation in the study procedures. Ethical considerations This study was approved by the Ethics Committee of the School of Nursing and Rehabilitation, Shandong University (Approval No.: 2023-R-113). All participants were fully informed about the study before participation and provided verbal consent. Measures Depressive symptoms Depressive symptoms were assessed using the 15-item Geriatric Depression Scale (GDS-15), a widely used screening instrument specifically developed for older adults [29]. The Chinese version of the GDS-15 has been validated and demonstrated good reliability and validity in community-dwelling older populations [27]. The GDS-15 consists of 15 dichotomous items (yes/no), with total scores ranging from 0 to 15, where higher scores indicate more severe depressive symptoms. For reference, a GDS-15 score ≥ 5 indicates possible depression symptoms [19]. In the present study, the GDS-15 demonstrated good internal consistency, with a Cronbach’s α of 0.82. Basic information sheet Participants’ basic socio-demographic and clinical information was collected using a structured questionnaire, including age, gender, education level, marital status, living arrangement, duration of diabetes, and comorbid chronic conditions (Supplementary Table S1). Sleep quality Sleep quality was measured using the Pittsburgh Sleep Quality Index (PSQI), a widely used instrument for assessing subjective sleep quality over the past month [2]. The Chinese version of the PSQI has been validated and shown to have good reliability and validity [28]. The PSQI comprises 19 self-rated items that generate seven component scores, including subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of sleep medication, and daytime dysfunction. Each component is scored from 0 to 3, yielding a global score ranging from 0 to 21, with higher scores indicating poorer sleep quality. For reference, a PSQI score ≥ 5 suggests poor sleep quality [2]. In this study, the PSQI demonstrated good internal consistency, with a Cronbach’s α of 0.85. Cognitive performance Cognitive function was assessed using the Montreal Cognitive Assessment—Basic (MoCA-B), which was specifically developed for older adults with low levels of education and literacy. The Chinese version of the MoCA-B has been validated and shown to be suitable for use in community-dwelling older adults [33]. The MoCA-B assesses multiple cognitive domains, including memory, executive function, attention, language, visuospatial ability, and orientation, with a total score ranging from 0 to 30. Higher scores indicate better cognitive function. The MoCA-B was administered through face-to-face interviews by trained assessors. For cognitive performance interpretation, education‑stratified cut‑offs for the MoCA‑Basic were used based on a previous validation study in Chinese older adults, with optimal thresholds of 19 for those with ≤ 6 years of education, 22 for those with 7–12 years, and 24 for those with > 12 years of education [8]. In the present study, the scale demonstrated acceptable internal consistency, with a Cronbach’s α of 0.80. Statistical analysis Continuous variables are presented as means with standard deviations (SDs), and categorical variables are presented as frequencies with corresponding percentages. All statistical analyses were conducted using R software (Version 4.4.1). A two-tailed significance level of α = 0.05 was applied for all statistical tests. Network estimation Network analyses were conducted using R software (Version 4.4.1). In the network model, each node represented a depressive symptom, a sleep quality component, or a cognitive domain score. Full descriptions of the nodes are provided in Supplementary Table S2. Because the network included dichotomous, ordinal, and continuous variables, correlation matrices were estimated using the cor_auto function in the qgraph package, which automatically selects appropriate correlations (e.g., polychoric, tetrachoric, or Pearson correlations) based on variable type. The resulting correlation matrix was then used as input for the Gaussian Graphical Model (GGM). Edges represent partial correlations between nodes after controlling for all other variables in the network. To obtain a sparse and interpretable network, the graphical least absolute shrinkage and selection operator (GLASSO) was applied with model selection based on the Extended Bayesian Information Criterion (EBIC) [17]. The EBIC tuning parameter was set to 0.5, balancing sensitivity and specificity [16]. Network visualization was performed using the “qgraph” package. Positive associations are depicted by green edges, whereas negative associations are depicted by red edges. Thicker edges indicate stronger associations. Centrality and bridge centrality estimation Bridge centrality indicators were used to identify nodes that link depressive symptoms, sleep quality, and cognitive performance within the network. Given that traditional centrality metrics such as closeness and betweenness may yield unstable estimates in psychological networks [16], this study focused on strength and bridge expected influence (BEI), which are considered more robust and interpretable in symptom networks. Network stability and accuracy The accuracy and stability of the estimated network were evaluated using the “bootnet” package. A non-parametric bootstrap procedure with 1,000 resamples was conducted to estimate 95% confidence intervals (CIs) for edge weights. Narrower CIs indicate higher precision. The stability of centrality indices was examined using a case-dropping bootstrap procedure. The Correlation Stability (CS) coefficient was calculated, with values ≥ 0.25 indicating acceptable stability and values ≥ 0.50 indicating good stability. Network comparison To explore potential gender differences in the network structure of depressive symptoms, sleep quality, and cognitive performance, networks were compared using the R package “NetworkComparisonTest” with 1,000 permutations. Results Study sample A total of 885 rural older adults with T2DM were included in the analysis. Eligible participants were aged 60 years or older, with a mean age of 69.74 (SD = 5.91) years, and 64.2% were female. Most participants were married or living with a surviving spouse (77.9%), had low educational attainment (42.7% were illiterate), and were primarily engaged in farming (93.2%). The mean duration of diabetes was 8.04 years (SD = 5.85). Mean scores were 5.00 (SD = 2.63) for the GDS-15, 9.35 (SD = 3.17) for the PSQI, and 14.51 (SD = 5.35) for the MoCA-B (Table 1). Table 1Socio-demographic characteristics of participantsCharacteristicN = 885Age (years) (n, %) 60–69427 (48.2) ≥ 70458 (51.8)Gender (n, %) Male316 (35.7) Female569 (64.2)Marital status (n, %) Spouse survived689 (77.9) Widowed/Single196 (22.1)Education level (years) (n, %) Illiterate387 (42.7) 1–6271 (30.6) 7–9153 (17.3) ≥ 1082 (9.3)Residence (n, %) Living with spouse545 (61.6) Live with spouse and children126 (14.2) Living with or adjacent to children86 (9.7) Living alone128 (14.5)Occupation (n, %) Famers825 (93.2) Non-farmers60 (6.8)Smoking status (n, %) Never smoking591 (66.8) Former smoking164 (18.5) Smoking130 (14.7)Alcohol consumption (n, %) Never drinking613 (69.3) Former drinking139 (15.7) Drinking133 (15.0) Diabetes duration (years) (mean, sd)8.04 (5.853) GDS-15 (mean, sd)5 (2.633) PSQI (mean, sd)9.35 (3.167) MoCA-B (mean, sd)14.51 (5.352) Socio-demographic characteristics of participants Network structure The estimated symptom network is presented in Fig. 1. The network consisted of 31 nodes representing depressive symptoms, sleep quality components, and cognitive domains, with 195 non-zero edges among the 465 possible connections, indicating a moderately dense and well-connected structure. Overall, associations were more strongly clustered within domains, while meaningful connections were also observed across depressive, sleep-related, and cognitive domains, suggesting an integrated cross-domain symptom system. Within the cognitive performance subnetwork, strong associations were observed between MoCA-B4 (Calculation) and MoCA-B5 (Abstraction), as well as between MoCA-B1 (Executive function) and MoCA-B5 (Abstraction), indicating close interrelations among higher-order cognitive processes. In the sleep quality subnetwork, PSQI5 (Sleep disturbances) showed strong associations with PSQI7 (Daytime dysfunction), and PSQI1 (Subjective sleep quality) was also strongly linked to PSQI7, reflecting the central role of daytime dysfunction in the experience of poor sleep. Within the depressive symptoms subnetwork, the strongest association was observed between GDS3 (Emptiness) and GDS4 (Boredom), followed by the association between GDS6 (Anticipatory anxiety) and GDS8 (Helplessness). Bootstrap difference tests of edge weights are shown in Figure S1. Network structure of cognitive function, sleep quality, and depressive symptoms among rural older adults with type 2 diabetes mellitus and mild cognitive impairment. Note: MoCA-B1 = Executive, MoCA-B2 = Verbal fluency, MoCA-B3 = Orientation, MoCA-B4 = Calculation, MoCA-B5 = Abstraction, MoCA-B6 = Delayed recall, MoCA-B7 = Visuospatial, MoCA-B8 = Naming, MoCA-B9 = Attention, PSQI1 = Subjective Sleep Quality, PSQI2 = Sleep Latency, PSQI3 = Sleep Duration, PSQI4 = Habitual Sleep Efficiency, PSQI5 = Sleep Disturbances, PSQI6 = Use of Sleeping Medication, PSQI7 = Daytime Dysfunction, GDS1 = Life dissatisfaction, GDS2 = Activity withdrawal, GDS3 = Emptiness, GDS4 = Boredom, GDS5 = Low positive affect, GDS6 = Anticipatory anxiety, GDS7 = Low happiness, GDS8 = Helplessness, GDS9 = Social withdrawal, GDS10 = Perceived memory problem, GDS11 = Reduced life appreciation, GDS12 = Hopelessness, GDS13 = Perceived irreversibility, GDS14 = Perceived inferiority, GDS15 = Concentration difficulty Network centrality Centrality analyses based on node strength are presented in Fig. 2. Across the entire network, MoCA-B5 (Abstraction), MoCA-B4 (Calculation), and GDS8 (Helplessness) emerged as the most highly connected nodes, indicating that these symptoms occupied central positions within the cross-domain symptom system. Among these, MoCA-B5 (Abstraction) demonstrated the highest strength, suggesting a particularly influential role in linking cognitive, affective, and sleep-related components. In contrast, PSQI6 (Use of sleeping medication) and GDS5 (Low positive affect) exhibited the lowest strength values, indicating relatively peripheral roles within the network. Strength values. Note: MoCA-B1 = Executive, MoCA-B2 = Verbal fluency, MoCA-B3 = Orientation, MoCA-B4 = Calculation, MoCA-B5 = Abstraction, MoCA-B6 = Delayed recall, MoCA-B7 = Visuospatial, MoCA-B8 = Naming, MoCA-B9 = Attention, PSQI1 = Subjective Sleep Quality, PSQI2 = Sleep Latency, PSQI3 = Sleep Duration, PSQI4 = Habitual Sleep Efficiency, PSQI5 = Sleep Disturbances, PSQI6 = Use of Sleeping Medication, PSQI7 = Daytime Dysfunction, GDS1 = Life dissatisfaction, GDS2 = Activity withdrawal, GDS3 = Emptiness, GDS4 = Boredom, GDS5 = Low positive affect, GDS6 = Anticipatory anxiety, GDS7 = Low happiness, GDS8 = Helplessness, GDS9 = Social withdrawal, GDS10 = Perceived memory problem, GDS11 = Reduced life appreciation, GDS12 = Hopelessness, GDS13 = Perceived irreversibility, GDS14 = Perceived inferiority, GDS15 = Concentration difficulty Bridge expected influence analyses further identified key symptoms that connected depressive, sleep-related, and cognitive domains (Fig. 3). MoCA-B5 (Abstraction), GDS10 (Perceived memory problem), and GDS15 (Concentration difficulty) showed the highest bridge expected influence values, highlighting their roles as cross-domain connectors within the symptom system. Bootstrap difference tests for bridge expected influence are presented in Figure S2. Bridge expected influence values. Note: MoCA-B1 = Executive, MoCA-B2 = Verbal fluency, MoCA-B3 = Orientation, MoCA-B4 = Calculation, MoCA-B5 = Abstraction, MoCA-B6 = Delayed recall, MoCA-B7 = Visuospatial, MoCA-B8 = Naming, MoCA-B9 = Attention, PSQI1 = Subjective Sleep Quality, PSQI2 = Sleep Latency, PSQI3 = Sleep Duration, PSQI4 = Habitual Sleep Efficiency, PSQI5 = Sleep Disturbances, PSQI6 = Use of Sleeping Medication, PSQI7 = Daytime Dysfunction, GDS1 = Life dissatisfaction, GDS2 = Activity withdrawal, GDS3 = Emptiness, GDS4 = Boredom, GDS5 = Low positive affect, GDS6 = Anticipatory anxiety, GDS7 = Low happiness, GDS8 = Helplessness, GDS9 = Social withdrawal, GDS10 = Perceived memory problem, GDS11 = Reduced life appreciation, GDS12 = Hopelessness, GDS13 = Perceived irreversibility, GDS14 = Perceived inferiority, GDS15 = Concentration difficulty Network stability and accuracy Network stability and accuracy analyses demonstrated that the estimated network was robust. The correlation stability coefficients for both node strength and bridge expected influence exceeded 0.50, indicating good stability of the centrality metrics (Figure S3). In addition, non-parametric bootstrap analyses revealed relatively narrow 95% confidence intervals for edge weights, supporting the precision and reliability of the estimated network structure (Figure S4). Network comparison Gender-specific networks are shown in Figure S5. Network Comparison Tests indicated no statistically significant differences in overall network structure between male and female participants (network structure invariance index = 0.166, P = 0.218). Similarly, global network strength did not differ significantly between females (4.97) and males (3.02), with a non-significant global strength difference of 1.95 (P = 0.243) (Figure S6). At the edge level, no individual connections showed statistically significant differences between gender groups after permutation testing, indicating a largely invariant symptom network structure across gender. Discussion This study applied network analysis to elucidate the complex interrelationships among depressive symptoms, sleep quality, and cognitive performance in rural older adults with T2DM. A structurally stable symptom network was identified, revealing tightly connected cognitive and depressive symptom clusters. Within this network, abstraction, calculation, and helplessness emerged as central symptoms, while abstraction, perceived memory problems, and concentration difficulty functioned as key bridge symptoms linking cognitive and depressive domains. No significant gender differences in network structure or edge weights were observed. To our knowledge, no prior study has simultaneously examined depressive, sleep, and cognitive symptoms in this population, highlighting the novelty of these findings. Together, these findings provide a nuanced, symptom-level understanding of how specific cognitive and affective symptoms—rather than global depression or sleep scores—co-occur and are interconnected in rural older adults with T2DM. These findings also suggest practical targets for assessment and intervention in clinical and community settings. Central symptoms Consistent with prior symptom network studies in older populations [38, 49], stronger connections were predominantly observed within symptom domains, whereas cross-domain associations were comparatively weaker. Within the cognitive domain, abstraction and calculation occupied highly central positions, indicating their broad associations with other cognitive symptoms. Impairments in higher-order conceptual reasoning appear to play an important role in the cognitive network of rural older adults with T2DM. Abstraction reflects the ability to identify conceptual commonalities, apply abstract rules, and reason beyond concrete stimuli—functions that are widely regarded as core components of executive cognition and are particularly sensitive to early frontal system dysfunction [30]. In rural Chinese settings, this vulnerability may be further amplified by lower educational attainment, reduced lifelong cognitive stimulation, and limited access to cognitively enriching environments, all of which constrain cognitive reserve and increase reliance on already compromised frontal systems [5, 46]. Although previous neuroimaging studies suggest that dorsolateral and medial prefrontal regions and frontoparietal networks support abstraction and calculation, these neurobiological interpretations are speculative, given that our study relied solely on questionnaire and cognitive task data [9, 32]. Consequently, deficits in abstraction may represent a potential early indicator of diabetes-related cognitive dysfunction in this population. Calculation represents a complementary but distinct executive process, engaging distributed neural systems supporting numerical manipulation, sustained attention, and working memory [36]. Functional neuroimaging studies indicate that mental calculation consistently recruits bilateral frontal and parietal cortices, reflecting its reliance on executive monitoring and attentional control [24]. Diabetes-related alterations in prefrontal cortical thickness and frontoparietal connectivity may undermine the efficiency of these processes, rendering calculation particularly vulnerable under metabolic stress [31]. Among rural older adults, limited educational opportunities and reduced engagement in cognitively demanding activities may further diminish compensatory capacity, amplifying calculation difficulties that require sustained executive resources [11, 12]. Within the depressive symptom cluster, helplessness emerged as a central node. Helplessness reflects a persistent perception of diminished control over health, daily functioning, and future outcomes—a psychological state highly salient in individuals managing chronic metabolic disease [26]. In rural contexts, this perception may be reinforced by limited access to specialized healthcare, insufficient diabetes education, and constrained psychosocial support, all of which undermine confidence in disease self-management [39, 48]. Network studies of late-life depression further indicate that symptoms reflecting diminished agency, such as helplessness, often occupy central positions and show strong associations with other depressive symptoms [45]. Taken together, the centrality of abstraction, calculation, and helplessness suggests that executive cognitive vulnerability and perceived loss of control may represent important components within the interconnected symptom network linking cognitive and affective domains. From a clinical perspective, identifying these central symptoms highlights potential priority targets for assessment and intervention. For example, screening for abstraction and calculation deficits can help detect early executive dysfunction, while addressing helplessness through supportive counseling or structured disease management programs may improve adherence and emotional well-being. Bridge symptoms From a network perspective, bridge symptoms may represent potential pathways connecting symptom domains, and identifying these specific symptoms provides important insights for understanding symptom co-occurrence across domains [40]. Perceived memory problems, while related to everyday cognitive experiences, are commonly embedded within depressive symptomatology and often reflect negatively biased self-appraisal rather than objective cognitive impairment. In later life, depressive symptoms have been shown to amplify subjective cognitive complaints independently of objectively measured cognitive performance, likely through mechanisms involving affective bias, reduced self-efficacy, and rumination [21]. Importantly, depressive cognitive complaints are closely intertwined with sleep disturbances, particularly insomnia and non-restorative sleep, which exacerbate fatigue, attentional lapses, and negative self-evaluation [23, 25]. Within this framework, perceived memory problems may serve as a hypothesized conduit linking depressive affect, sleep-related problems, and perceived cognitive decline. Similarly, concentration difficulty represents a core depressive symptom characterized by sustained attentional inefficiency and mental fatigue rather than isolated executive dysfunction. Difficulty concentrating has been consistently associated with sleep fragmentation and reduced cognitive efficiency in attention and processing speed domains [20]. Consequently, concentration difficulty may represent a potential interface through which sleep disturbances and depressive symptoms are associated with broader cognitive performance. In rural older adults with T2DM, the bridging role of these depressive cognitive symptoms may be particularly pronounced. Chronic metabolic dysregulation, vascular burden, low-grade inflammation, limited healthcare access, and lower health literacy may exacerbate these symptom patterns, highlighting the importance of considering cultural and contextual factors when interpreting symptom interactions in this population [3, 47]. From an intervention and research perspective, network analysis suggests that central and bridge symptoms may represent potential targets for symptom-oriented interventions. Specifically, targeting bridge symptoms such as perceived memory problems and concentration difficulty could help prevent the propagation of emotional distress and sleep disturbances into broader cognitive impairment. This symptom-focused approach allows clinicians to prioritize interventions on those symptoms that exert the most influence across domains, potentially enhancing overall cognitive and mental health outcomes in rural older adults with T2DM. Network comparison of gender Despite prior evidence suggesting gender differences in the relationships among depressive symptoms, sleep quality, and cognitive function in older adults with T2DM (H. F. Chen et al., [6], the present study did not identify significant gender differences in network structure or edge weights. Previous findings regarding gender-specific cognitive and emotional vulnerability in T2DM have been mixed, with observed differences often depending on cognitive domain, metabolic status, and contextual factors [10, 18]. From a network perspective, the lack of observed gender differences in this rural cohort may reflect relative homogeneity in socioeconomic conditions, healthcare access, and lifestyle exposures, which could attenuate gender-specific pathways. Moreover, emerging evidence suggests that sex differences in diabetes-related cognitive decline may manifest primarily under specific biological or environmental stressors rather than representing fixed disparities [10, 42]. In this context, shared metabolic, vascular, sleep-related, and psychosocial stressors may exert comparable influences on symptom interconnections across genders in rural older adults with T2DM. The lack of significant gender differences suggests that interventions targeting central and bridge symptoms may be broadly applicable across male and female rural older adults with T2DM. This highlights the potential utility of symptom-focused strategies that do not need to be strongly sex-specific in this population. Study limitations Several limitations should be acknowledged. First, the cross-sectional design precludes causal inference regarding the relationships among depressive symptoms, sleep quality, and cognitive performance, and longitudinal studies are needed to clarify temporal dynamics. Second, individuals with diagnosed dementia were excluded from this study; therefore, the cognitive distribution of the sample may not fully represent the broader rural older population with T2DM, particularly those with more severe cognitive impairment, which may limit the generalizability of the findings. Third, participants were recruited from rural older adults with T2DM, which may limit the generalizability of the findings to urban or socioeconomically diverse populations. Fourth, depressive symptoms and sleep quality were assessed using self-reported instruments, which may be influenced by subjective perception, mood state, or health literacy. In addition, relevant psychosocial and behavioral factors such as loneliness, life satisfaction, and physical activity were not measured. Some important clinical variables related to T2DM, such as glycemic control, treatment regimen, diabetes duration, and complications, were also not included and may have influenced the network structure. Because network models are inherently dependent on the variables included in the model, the omission of potentially relevant psychological and clinical variables may have led to variable selection bias and may have partially influenced edge weights, centrality indices, and bridge centrality estimates. Therefore, the network structure identified in this study should be interpreted as conditional on the selected variables rather than representing a complete symptom system. Finally, although network analysis provides a useful framework for identifying central and bridge nodes, the results are dependent on the selected measures and sample characteristics. Replication in independent and longitudinal cohorts is warranted to confirm the stability and clinical relevance of the identified network structure. Conclusion In rural older adults with T2DM, this study identified a stable symptom network linking cognitive performance, depressive symptoms, and sleep quality. Abstraction, calculation, and helplessness emerged as central nodes, while abstraction, perceived memory problems, and concentration difficulties functioned as bridge symptoms connecting cognitive and emotional domains. These findings highlight the importance of symptom-level interconnections in understanding cognitive vulnerability among older adults with T2DM. A network-informed perspective may help identify high-priority assessment targets and inform more precise strategies for supporting cognitive and mental health in resource-limited rural settings. Supplementary Information Supplementary Material 1. Supplementary Material 2.