Diabetic retinopathy remains a leading cause of preventable blindness among individuals with type 2 diabetes mellitus and is driven by the increasing global burden of diabetes. Early identification of individuals at high risk is essential for timely intervention. In recent years, artificial intelligence, including machine learning and deep learning, has emerged as a promising approach for prognostic modelling. However, evidence on artificial intelligence-driven models for predicting diabetic retinopathy risk remains fragmented. This systematic review aims to synthesise current evidence on artificial intelligence-driven prognostic models for predicting the risk of diabetic retinopathy among individuals with type 2 diabetes mellitus, focusing on model characteristics, predictor variables, performance, and validation strategies.

A comprehensive search of PubMed, Scopus, Web of Science, and ScienceDirect was performed for studies published between January 2016 and December 2025. Eligible studies included those applying artificial intelligence-based models for diabetic retinopathy risk prediction in adult populations with type 2 diabetes mellitus and reporting model performance metrics. Data were extracted and synthesised narratively, and methodological quality was assessed using the Newcastle-Ottawa Scale.

A total of 1040 records were identified, with eight studies included after screening. A wide range of artificial intelligence algorithms was applied, with ensemble models such as extreme gradient boosting and random forest demonstrating superior performance. Key predictors consistently included glycated haemoglobin levels, duration of diabetes, blood pressure, lipid profile, and renal function markers, alongside emerging metabolomic biomarkers. Model performance ranged from moderate to excellent, with area under the receiver operating characteristic curve values between 0.68 and 0.97. Most studies employed internal validation techniques such as cross-validation or training and testing data splits, while external validation was largely absent. Overall methodological quality was high, although variability in study design, predictor selection, and reporting was observed.

Artificial intelligence-driven prognostic models show substantial potential in predicting diabetic retinopathy risk among individuals with type 2 diabetes mellitus, particularly through the integration of clinical and highdimensional data. Despite promising predictive performance, limitations in external validation, heterogeneity, and reporting of clinical utility hinder translation into practice. Future research should prioritise external validation, standardised reporting frameworks, and clinically interpretable models to enhance applicability in real-world healthcare settings.

Tuberculosis (TB) remains a major occupational risk for healthcare workers (HCWs), who face higher exposure and potential challenges in achieving optimal treatment outcomes. Despite extensive national TB control strategies in Malaysia, evidence on treatment success and its determinants among HCWs is limited. This study aimed to determine the proportion of TB treatment success and identify associated factors among HCWs with active TB in Selangor from 2014 to 2023 using the National Tuberculosis Registry (NTBR) system.

A retrospective cross-sectional study was conducted using registry data from NTBR involving all HCWs diagnosed with active TB and registered in Selangor between 2014 and 2023. Universal sampling was applied to include all eligible cases with complete data. Variables were extracted across sociodemographic, occupational, sociocultural, and clinical domains. TB treatment success was defined as "cured" or "treatment completed", while unsuccessful outcomes included death, treatment failure, and not evaluated. Data were cleaned, deidentified, and analysed using SPSS version 29. Simple logistic regression identified variables with p<0.25 for inclusion in multiple logistic regression Adjusted odds ratios (AOR) and 95% confidence intervals (CI) were reported. Model diagnostics included the Hosmer- Lemeshow test, ROC (Receiver Operating Characteristic) analysis, multicollinearity checks, interaction terms, and Cook's distance.

A total of 313 HCWs with active TB were included. The TB treatment success rate was 86.9%, with 51.5% cured and 48.5% completing treatment. In multivariable analysis, three factors remained significantly associated with treatment success. HCWs aged <34 years had higher odds of treatment success (AOR=5.071; 95% CI: 2.009-12.801). Conversely, doctors (AOR=0.024; 95% CI: 0.003-0.218) and supporting staff (AOR=0.366; 95% CI: 0.143-0.936) showed significantly lower success compared to paramedics. Advanced chest X-ray (CXR) findings demonstrated strong positive associations: moderate (AOR=14.686; 95% CI: 5.873 - 36.724) and far advanced disease (AOR=10.783; 95% CI: 1.909-60.904). Commonly reported risk factors such as HIV, diabetes, smoking, BCG (Bacillus Calmette-Guérin) status, and comorbidities were not significantly associated. Model performance was robust (ROC: 88.3%; Hosmer-Lemeshow p=0.858).

TB treatment is successful among HCWs in Selangor (86.9%), which is near to the WHO (World Health Organization) End TB Strategy. The odds of treatment success were much higher among younger HCWs, whereas doctors and other supporting staff reported had lower adjusted odds in comparison with paramedics. Moderate and far advanced CXR findings at diagnosis were independently associated with higher treatment success; however, this finding should be interpreted cautiously because of the small number of far advanced cases and the possibility of estimate instability and residual confounding. Conventional risk factors like human immunodeficiency virus infection, diabetes mellitus, smoking, and BCG status were found not associated significantly and may be due to the small subgroups and the nature of this registry data limitation. These conclusions indicate the fact that occupational role and clinical presentation are significant factors defining TB treatment outcomes among HCWs. Rolespecific and targeted adherence support enhanced psychosocial interventions, and better NTBR data coverage can further improve the outcome of treatment and support the TB control in the Malaysian healthcare labour force.

Effective management of hypertension often requires pharmacological intervention, which can involve either single pill combination therapy (SPC) or multiple pill combination therapy (MPC). However, the choice between SPC and MPC can be influenced by patient demographics, clinical characteristics, comorbid conditions and physician preferences. This study was aimed at exploring the prevalence of SPC versus MPC usage among people with hypertension in a tertiary hospital and its associated factors.

A cross-sectional study using a validated questionnaire was conducted among the hypertensive patients aged 18 years and on antihypertensive medication(s) for at least six months attending various speciality clinics at Hospital Sultan Abdul Aziz Shah, Selangor.

The prevalence of patients using SPC was 27.9% (78/280). In univariate analysis, factors that were found to be associated with the use of SPC were presence of diabetes mellitus (p < 0.001), presence of chronic kidney disease (p = 0.033), physician's own preference for SPC (p = 0.027) and duration of hypertension (p = 0.003). The independent factors associated with SPC use was the presence of chronic kidney disease, physician's own preference for SPC and absence of diabetes mellitus (p < 0.05).

The use of SPC is only 27.9%. The independent factors associated with SPC use was the presence of chronic kidney disease, physician's own preference for SPC and absence of diabetes mellitus. Hence, this study's findings can be applied by healthcare providers who should consider these factors when prescribing hypertension treatments. Tailoring the treatment plan to the individual patient's needs and circumstances can enhance adherence and efficacy.

Diabetic foot ulcers (DFUs) are a serious complication of diabetes mellitus. DFUs have a profound impact on patients' health-related quality of life (HRQoL). This study aimed to evaluate age-related variations in HRQoL and to identify factors associated with physical and mental health among patients with recurrent DFUs.

A cross-sectional study was conducted among 245 patients with type 2 diabetes mellitus. HRQoL was measured using the Indonesian version of the SF-36, whereas depression, diabetes distress, and self-care adherence were evaluated using the Beck Depression Inventory II (BDI-II), the Diabetes Distress Scale-17 (DDS-17), and the Summary of Diabetes Self-Care Activities (SDSCA), respectively. Data were analysed using bivariate tests and multivariate linear regression.

Older adults (≥60 years) had significantly lower physical functioning and general health scores (p < 0.05). Depression (β = -0.48, p < 0.001) and diabetes distress (β = -0.25, p = 0.003) were independent predictors of poorer physical HRQoL, whereas higher self-care adherence improved HRQoL (β = 0.18, p = 0.014). Age was not a significant factor after adjustment. Physical HRQoL is poorer among older adults with recurrent DFUs, primarily mediated by psychosocial distress.

Our findings demonstrated that psychosocial conditions, especially depression and diabetes-related distress, significantly contribute to poor HRQoL. There is a relationship between depression, distress, and poor quality of life. Moreover, depression and distress independently predicted lower physical QoL, and better self-care is a predictor of QoL. Integrated management addressing psychological well-being and self-care behaviours may enhance the overall patient outcomes.

Chronic kidney disease (CKD) is a common and important complication among patients with, contributing to substantial morbidity and adverse clinical outcomes. This study aimed to determine the factors associated with CKD among patients with T2DM in Johor, Malaysia.

This was a registry-based crosssectional study using secondary data from the Malaysian National Diabetes Registry (NDR). The study included patients with T2DM captured in the Johor NDR during the 2020-2021 audit cycle. CKD status was determined by calculating the estimated glomerular filtration rate (eGFR) using the CKD Epidemiology Collaboration (CKD-EPI) formula based on serum creatinine and relevant patient variables extracted from the registry. Patients were operationally classified as having CKD if the available registry-derived eGFR was <60 mL/min/1.73 m². Sociodemographic, clinical, biochemical, diabetes-related complication, and medication-related variables were analysed. Univariate analysis was performed using Chisquare test or Fisher's exact test, and variables with p<0.05 were entered into multivariable logistic regression to identify factors independently associated with CKD among patients with T2DM.

A total of 10,117 patients with T2DM were included, of whom 2,387 (23.6%) had CKD. Factors independently associated with CKD were age ≥60 years (aOR: 2.97, 95% CI: 2.60, 3.38), hypertension (aOR: 1.37, 95% CI: 1.10, 1.70), diabetes duration >10 years (aOR: 1.84, 95% CI: 1.58, 2.14), abnormal triglyceride (aOR: 1.59, 95% CI: 1.42, 1.79), abnormal high-density lipoprotein (HDL) (aOR: 1.26, 95% CI: 1.13, 1.40), diabetic retinopathy (aOR: 1.22, 95% CI: 1.06, 1.40), cerebrovascular disease (aOR: 1.40, 95% CI: 1.06, 1.85), and diabetic foot ulcer (aOR: 2.40, 95% CI: 1.68, 3.41). Medication-related associations were observed but should be interpreted cautiously because of potential confounding by indication and reverse causation.

CKD affected nearly one-quarter of patients with T2DM in this Johor registry-based study. Older age, hypertension, longer diabetes duration, lipid abnormalities, diabetic retinopathy, cerebrovascular disease, and diabetic foot ulcer were independently associated with CKD. These findings provide useful real-world evidence to support earlier risk identification and future development of a locally derived CKD risk prediction model among Malaysian patients with T2DM.

Chronic kidney disease (CKD) and cardiovascular disease (CVD) are major causes of morbidity and mortality in people with type 2 diabetes (T2D). Contemporary CKD guidelines increasingly support incorporation of glucagon-like peptide-1 receptor agonists (GLP-1RAs) as an additional pillar of cardiorenal therapy alongside renin-angiotensin-aldosterone system (RAAS) blockade, sodium-glucose co-transporter 2 inhibitors (SGLT2i) and finerenone. Semaglutide is the first GLP-1RA with kidney-outcome data from a dedicated trial in T2D-CKD: the Evaluate Renal Function with Semaglutide Once Weekly (FLOW) study. In FLOW, once-weekly subcutaneous semaglutide 1 mg reduced major kidney disease events by 24%, major adverse cardiovascular events by 18% and all-cause mortality by 20% versus placebo on a background of guideline-directed therapy. Here a multidisciplinary expert panel summarises Australian Therapeutic Goods Administration indications and Pharmaceutical Benefits Scheme restrictions, highlights common access barriers and outlines approaches to shared care between primary and tertiary services for use of GLP-1RAs. Guidance is also provided on treatment sequencing of GLP-1RA alongside RAAS blockade, SGLT2i and finerenone. Evidence from key GLP-1RA trials, in addition to FLOW, is reviewed and translated into practical, implementable prescribing strategies for Australian clinicians. The panel also identified priority patient groups for GLP-1RA therapy, including those with residual albuminuria, heart failure with preserved ejection fraction, established CVD, obesity and Aboriginal and Torres Strait Islander peoples at high cardiorenal risk. Remaining evidence gaps, particularly for non-diabetic CKD, advanced CKD, dialysis and transplant populations and equity of access, are identified as priorities for future research and policy.

Diabetic kidney disease (DKD) is among the most prevalent complications of diabetes and is commonly accompanied by hypertension as a comorbid condition. The coexistence of these two conditions markedly augments the risk of cardiovascular disease. Although traditional risk factors such as dyslipidemia, smoking, and decreased renal function have been shown to be associated with increased cardiovascular events, evidence for the higher-risk subgroup of "DKD with hypertension" remains relatively limited. This study aimed to retrospectively analyze the influencing factors of adverse cardiovascular events in this population and assess the independent predictive value of different clinical indicators.

In this single-center retrospective cohort study, 204 patients diagnosed with DKD and hypertension were enrolled between January 2021 and January 2024. Baseline data, including patient demographics and clinical indicators, were collected. Patients were divided into an event group and a non-event group based on whether they experienced cardiovascular events during the 12-month follow-up period. Independent risk factors were screened using univariate and multivariate logistic regression, and the effects of prior hyperlipidemia and smoking on event-free survival were assessed using Kaplan-Meier curves.

Compared with the non-event group, patients in the event group had significantly higher levels of low-density lipoprotein (LDL), urinary protein, and kidney injury molecule-1 (KIM-1), and significantly lower levels of high-density lipoprotein (HDL), along with a greater prevalence of prior hyperlipidemia and smoking (all p < 0.05). Univariate analysis showed that prior hyperlipidemia, smoking, elevated LDL, reduced HDL, and increased KIM-1 were all associated with cardiovascular events. Further multivariate analysis confirmed that prior hyperlipidemia, smoking, HDL, LDL, and KIM-1 were all independent predictors of cardiovascular events (all p < 0.05). According to Kaplan-Meier curve analysis, patients with a history of hyperlipidemia and a smoking habit had significantly lower event-free survival than those without these characteristics (p = 0.034 and p = 0.024).

In DKD patients with hypertension, prior hyperlipidemia, smoking, elevated LDL, reduced HDL, and increased KIM-1 were key independent risk factors for cardiovascular events. The adverse association of dyslipidemia and smoking with event-free survival was evident early in the follow-up period. These results suggest that this high-risk population should receive enhanced lipid management and smoking cessation interventions, and the potential value of renal tubular injury markers (such as KIM-1) in cardiovascular risk assessment should be investigated to facilitate early identification of high-risk individuals and the development of individualized prevention strategies.

Patients who develop mild hemorrhagic transformation (HT) after intravenous thrombolysis (IVT) for acute ischemic stroke (AIS) exhibit substantial heterogeneity in functional recovery. However, a systematic evaluation of predictors of functional recovery in this specific population remains lacking. This study aimed to identify the clinical and nutritional factors associated with functional recovery in these patients.

This dual-center retrospective cohort study consecutively enrolled 204 patients with AIS who developed mild HT after IVT between March 2020 and March 2025. Based on the modified Rankin Scale (mRS) score at 3 months, patients were classified into a good functional recovery group (mRS ≤2, n = 153) and a poor functional recovery group (mRS ≥3, n = 51). Demographic, clinical, and laboratory data were collected. Nutritional status was evaluated using the Controlling Nutritional Status (CONUT) score and the Prognostic Nutritional Index (PNI). Univariate and multivariate logistic regression analyses were performed to identify independent predictors. The predictive performance of individual factors and a combined model was assessed using receiver operating characteristic (ROC) curve analysis.

Univariate analysis demonstrated that age, baseline National Institutes of Health Stroke Scale (NIHSS) score, PNI, CONUT score, arrival-to-thrombolysis time, and history of diabetes mellitus (HD) were significantly associated with poor functional recovery (all p < 0.05). Multivariate analysis identified HD, older age, a higher baseline NIHSS score, and a higher CONUT score as independent risk factors for poor functional recovery (all p < 0.05). A combined model incorporating these four variables demonstrated the highest discriminative performance, with an area under the curve (AUC) of 0.840 (95% CI: 0.783-0.897). At the optimal probability cut-off of 0.156, the model achieved a specificity of 92.2% and a sensitivity of 64.5%.

In patients with AIS who develop mild HT after IVT, baseline neurological deficit severity, nutritional status, HD, and age are independent determinants of functional recovery. A model integrating these factors demonstrates superior discriminative potential and may facilitate early risk stratification and personalized management in this patient population.

Suboptimal glycaemic control may delay surgery and increase perioperative risk, yet national guidance provides limited detail on how preoperative diabetes optimisation should be organised. This study evaluated a multidisciplinary perioperative diabetes optimisation pathway in a tertiary hospital.

This prospective single-centre cohort included 174 consecutive adults referred to The Queen Elizabeth Hospital Birmingham, United Kingdom, between 1 January and 31 December 2024, with glycated haemoglobin (HbA1c) >69 mmol/mol (>8.5%) before elective surgery. The pathway comprised diabetologist assessment, individualised treatment adjustment, diabetes specialist nurse follow-up, education, repeat HbA1c measurement and communication with surgical teams. Primary outcomes were change in HbA1c and achievement of HbA1c ≤69 mmol/mol (≤8.5%). Secondary outcomes included progression to surgery and factors associated with target achievement and follow-up HbA1c. Multivariable logistic and linear regression examined target achievement and follow-up HbA1c, respectively.

Paired data were available for 149/174 patients (85.6%). Mean HbA1c decreased from 96.2 ± 14.0 to 73.1 ± 14.1 mmol/mol, a reduction of 23.1 mmol/mol (2.1 percentage points; p < 0.001); 77/149 patients (51.7%) achieved the target. Higher baseline HbA1c was associated with lower odds of target achievement (odds ratio 0.96 per mmol/mol, 95% confidence interval 0.94-0.99; p = 0.006), although patients with the highest baseline values achieved the largest absolute reductions. After adjustment for baseline HbA1c and other covariates, each additional diabetes-team appointment was associated with a 0.58 mmol/mol lower follow-up HbA1c (95% confidence interval -0.98 to -0.18; p = 0.005). Overall, 124/174 patients (71.3%) proceeded to surgery.

Referral to the pathway was associated with substantial improvement in preoperative HbA1c; just over half of patients achieved the preoperative target and most proceeded to surgery. Baseline HbA1c identified patients less likely to reach the fixed target despite larger absolute reductions, while more frequent specialist contact was modestly associated with lower follow-up HbA1c. Coordinated specialist follow-up, including pragmatic basal insulin initiation and titration where clinically indicated, provides a reproducible hospital-based model for preoperative diabetes optimisation.

To investigate the association between oral health abnormality phenotypes based on previous oral health records and heart failure(HF)-related events during follow-up in patients with type 2 diabetes mellitus (T2DM), and to explore their implications for nursing assessment.

A single-center retrospective cohort study was conducted involving 120 patients with T2DM. According to baseline oral examination records, including periodontal pocket probing depth, gingival bleeding, oral hygiene index (OHI), plaque index (PLI), and tooth loss, the patients were classified into three groups: no obvious periodontal abnormality, mild periodontal pocket deepening, and marked periodontal pocket deepening. General clinical data, oral health indicators, inflammatory markers, lipid profiles, and renal function indicators were collected. Correlation analysis, Kaplan-Meier analysis, and Cox proportional hazards regression models were used to evaluate the relationship between oral health abnormality phenotypes and HF-related events.

During follow-up, a total of 46 patients developed HF-related events, with an incidence rate of 38.33%. Kaplan-Meier analysis showed a statistically significant difference in HF event-free survival among the three groups(P=0.019). Oral health indicators were positively correlated with inflammatory markers and negatively correlated with estimated glomerular filtration rate (eGFR) (P＜0.05). Cox regression analysis showed that oral health abnormality phenotype stratification was associated with HF-related events in the unadjusted model and the age-adjusted model. However, after further adjustment for eGFR, this association was markedly attenuated, and the model estimates became unstable. Sensitivity analysis showed that no stable association was observed between oral health abnormality phenotype stratification and HF-related events in the subgroup with eGFR ≥60 mL/(min·1.73 m²).

Oral health abnormality phenotypes based on previous oral health records showes a potential association with HF-related events in patients with T2DM, but this relationship is substantially influenced by renal function status. Oral health abnormalities may more likely reflect an overall adverse clinical condition characterized by renal impairment, increased inflammatory burden, and metabolic disturbance. Therefore, oral health status may serve as a useful clue for comprehensive nursing assessment in chronic disease management. However, the current evidence is insufficient to support it as a stable independent predictor of HF-related events.