Predicting the onset of diabetes-related complications after a diabetes diagnosis with machine learning algorithms.

AIMS: Using machine learning algorithms and administrative data, we aimed to predict the risk of being diagnosed with several diabetes-related complications after one-, two- and three-year post-diabetes diagnosis. METHODS: We used longitudinal data from administrative registers of 610,019 individuals in Catalonia with a diagnosis of diabetes and checked the presence of several complications after diabetes onset from 2013 to 2017: hypertension, renal failure, myocardial infarction, cardiovascular disease, retinopathy, congestive heart failure, cerebrovascular disease, peripheral vascular disease and stroke. Four different machine learning (ML) algorithms (logistic regression (LR), Decision tree (DT), Random Forest (RF), and Extreme Gradient Boosting (XGB)) will be used to assess their prediction performance and to evaluate the prediction accuracy of complications changes over the period considered. RESULTS: 610,019 people with diabetes were included. After three years since diabetes diagnosis, the area under the curve values ranged from 60% (retinopathy) to 69% (congestive heart failure), whereas accuracy rates varied between 60% (retinopathy) to 75% (hypertension). RF was the most relevant technique for hypertension, myocardial and retinopathy, and LR for the rest of the comorbidities. The Shapley additive explanations values showed that age was associated with an elevated risk for all diabetes-related complications except retinopathy. Gender, other comorbidities, co-payment levels and age were the most relevant factors for comorbidity diagnosis prediction. CONCLUSIONS: Our ML models allow for the identification of individuals newly diagnosed with diabetes who are at increased risk of developing diabetes-related complications. The prediction performance varied across complications but within acceptable ranges as prediction tools.

[Diagnosis and treatment strategy of periodontitis with diabetes].

The number of diabetic patients visiting stomatology for periodontal disease is increasing, and the symptoms are relatively severe, and often complications increase the complexity of periodontal treatment. This article briefly describes the research progress and clinical manifestations of the epidemiology and related pathological mechanisms of periodontitis with diabetes, focusing on the treatment and providing reference for stomatologists in the clinical diagnosis and treatment of patients with diabetic periodontitis.

Isolated renal mucormycosis masquerading as emphysematous pyelonephritis.

Mucormycosis is an acute, life-threatening infection and isolated renal involvement is rare. Due to the angioinvasive nature of the disease, it is rapidly progressive and can be lethal if not managed expeditiously. In patients with underlying conditions of immunosuppression, diabetes mellitus, transplantation, COVID-19, intravenous drug and substance use and pyelonephritis, which is unable to be controlled via regular antibiotics, mucormycosis must be considered on the differential and antifungals must be empirically started. Most cases are often diagnosed on histopathology, which causes delayed treatment and resolution. We present a case of emphysematous pyelonephritis diagnosed on imaging and was later found to have mucormycosis on histopathological examination.

Improved prognosis with integrated care management including early rhythm control and healthy lifestyle modification in patients with concurrent atrial fibrillation and diabetes mellitus: a nationwide cohort study.

BACKGROUND: Patients with concurrent atrial fibrillation (AF) and diabetes mellitus (DM) [AF-DM] have a high risk of cardiovascular and diabetes-related complications, but are less engaged in a comprehensive treatment approach. We evaluated the association of early rhythm control (ERC), lifestyle modification (LSM), and a combination of ERC and LSM with cardiovascular or diabetes-related complication risk in patients with AF-DM (type 2). METHODS: From the National Health Information Database, 47,940 patients diagnosed with AF-DM in 2009-2016 were included. We defined ERC as rhythm control therapy within two years of AF diagnosis and LSM as adherence to ≥ 2 of the healthy behaviors among non-current smoking, non-drinking, and regular exercise. We compared the primary (ischemic stroke) and secondary (macro- and microvascular complications, glycemic emergency, and all-cause death) outcomes in four groups: non-ERC and non-LSM (group 1), LSM only (group 2), ERC only (group 3), and both ERC and LSM (group 4). RESULTS: Of total, 10,617 (22%), 26,730 (55.8%), 2,903 (6.1%), and 7,690 (16.0%) were classified into groups 1 to 4, in sequence. The mean duration from AF diagnosis to ERC was 25.6 ± 75.5 days. During 4.0 (interquartile range: 2.5-6.2) years' follow-up, groups 2 and 3 were associated with 23% and 33% lower risks of stroke than group 1, respectively. Group 4 was associated with the lowest risk of stroke: hazard ratio (HR) 0.58, 95% confidence interval (CI) 0.51-0.67, p < 0.001. Regarding secondary outcomes, the lowest risks were also observed in group 4; macro- and microvascular complications, glycemic emergency, and all-cause death had HRs (95% CIs) of 0.63 (0.56-0.70), 0.88 (0.82-0.94), 0.72 (0.62-0.84), and 0.80 (0.73-0.87), respectively, all p < 0.001. CONCLUSIONS: For AF-DM patients, ERC and LSM exert a synergistic effect in preventing cardiovascular and diabetes-related complications with the greatest lowered risk of stroke. A comprehensive treatment approach should be pursued in AF-DM patients.

Blood hemoglobin A1c might predict adverse differences in heart rate variability in a diabetic population: Evidence from the Midlife in the United States (MIDUS) study.

Background: Cardiac autonomic neuropathy in population with diabetes mellitus (DM) is frequent and linked with high risk of cardiovascular mortality. However, studies on whether blood hemoglobin A1c (HbA1c) levels are related to adverse differences in heart rate variability (HRV) in individuals with DM are scarce. Aim: We aimed to investigate the association of blood HbA1c levels with adverse differences in HRV, which is an indicator of cardiac autonomic control, in adult individuals with and without DM. Methods: Data were collected from the Midlife in the United States (MIDUS) study, and 928 individuals were analyzed for the relationship between blood HbA1c levels and HRV through a cross-sectional analysis. Results: Participants with DM had significantly higher HRV than those without DM. The smooth curve suggested inverse relationships between blood HbA1c levels and HF- and LF-HRV seen in participants with DM but not in those without DM after controlling for all covariates (age, sex, BMI, smoking, number of drinking years and exercise). Furthermore, linear regression analysis demonstrated that elevated blood HbA1c levels did contribute to adverse differences in HF-HRV (Sß= -0.118; 95% CI -0.208, -0.027; P=0.012) and LF-HRV (Sß= -0.097; 95% CI -0.177, -0.017; P=0.019) after controlling for these covariates in participants with DM, while in participants without DM, blood HbA1c was not significantly related to adverse differences in HF-HRV (Sß=0.095; 95% CI -0.059, 0.248; P=0.228) or LF-HRV (Sß=0.043; 95% CI -0.103, 0.189; P=0.565). DM has a significant modifying effect on associations between blood HbA1c and adverse differences in HF-HRV (P for interaction=0.019) and LF-HRV (P for interaction=0.029). Conclusions: We reported strong evidence that elevated blood levels of HbA1c were associated with adverse differences in HRV in the diabetic population but not in the nondiabetic population. This finding supported that long-term hyperglycemia is related to autonomic nerve injury in the diabetic population. Blood HbA1c might be a good indicator of cardiac autonomic neuropathy.

Newly diagnosed diabetes based on an oral glucose tolerance test predicts cardiovascular outcomes in patients with coronary artery disease: An observational study.

Diabetes is prevalent in patients with coronary artery disease (CAD). Using the oral glucose tolerance test (OGTT), abnormal glucose regulation can be detected early in CAD patients without known diabetes. In the present study, we assessed the impact of abnormal glucose regulation on the long-term cardiovascular outcomes of patients with established CAD. Patients hospitalized for a scheduled angiography due to angina were enrolled in Taichung Veterans General Hospital. Fasting plasma glucose (FPG) and 2-hour postload glucose (2hPG) were assessed using the OGTT. Hemoglobin A1c (HbA1c) and other biochemical analyses were assessed using fasting blood samples. During a median follow-up period of 4.6 years, a composite of all-cause mortality, nonfatal myocardial infarction, and nonfatal stroke was recorded as the primary endpoint. In 682 enrolled patients who completed the follow-up, there were 16 myocardial infarction events, 12 stroke events, and 58 deaths as composite endpoints. According to FPG and 2hPG, patients with newly diagnosed diabetes had a 2-fold higher risk for the composite endpoint than those in the normal glucose group (hazard ratio [HR], 2.011; 95% confidence interval (CI), 1.101-3.673; P = .023); however, prediabetes was not significantly associated with the composite endpoint (HR, 1.452; 95% CI, 0.788-2.675; P = .232). On the other hand, patients with diabetes diagnosed by FPG and HbA1c did not have a significantly higher risk for the composite endpoint than those in the normal glucose group (HR, 1.321; 95% CI, 0.686-2.545; P = .405). A 2hPG ≥7.8 mmol/L was a significant predictor for the composite endpoint (odds ratio, 1.743; 95% CI, 1.060-2.863; P = .028) after adjusting for age, sex, and estimated glomerular filtration rate. Diabetes, but not prediabetes, detected via OGTT is associated with a significantly increased risk for the composite endpoint in patients with established CAD. The 2hPG provided a greater predictive power for the composite endpoint than fasting glucose and HbA1c.

Factors Influencing the Utilization of Diabetes Complication Tests Under the COVID-19 Pandemic: Machine Learning Approach.

Objective: There are still not enough studies on the prediction of non-utilization of a complication test or a glycated hemoglobin test for preventing diabetes complications by using large-scale community-based big data. This study identified the ratio of not taking a diabetes complication test (fundus examination and microprotein urination test) among adult diabetic patients over 19 years using a national survey conducted in South Korea and developed a model for predicting the probability of not taking a diabetes complication test based on it. Methods: This study analyzed 25,811 subjects who responded that they had been diagnosed with diabetes by a doctor in the 2020 Community Health Survey. Outcome variables were defined as the utilization of the microprotein urination test and the fundus examination during the past year. This study developed a model for predicting the utilization of a diabetes complication test using logistic regression analysis and nomogram to understand the relationship of predictive factors on the utilization of a diabetes complication test. Results: The results of this study confirmed that age, education level, the recognition of own blood glucose level, current diabetes treatment, diabetes management education, not conducting the glycated hemoglobin test in the past year, smoking, single-person household, subjectively good health, and living in the rural area were independently related to the non-utilization of diabetes complication test after the COVID-19 pandemic. Conclusion: Additional longitudinal studies are required to confirm the causality of the non-utilization of diabetes complication screening tests.

Coverage of diabetes complications screening in rural Eastern Cape, South Africa: A cross-sectional survey.

BACKGROUND: There is a paucity of data on the coverage of diabetes mellitus (DM) complications screening in primary healthcare facilities in South Africa (SA). This study assesses the extent of screening for DM complications among individuals with type 2 DM attending primary health facilities in rural Eastern Cape (EC), SA. METHODS: The study adopted a descriptive, cross-sectional design and obtained data from 372 individuals with type 2 diabetes attending six selected primary healthcare centres (PHCs) in two EC districts. Demographic and clinical data were obtained through questionnaire-based interviews and reviews of medical records. We assessed the extent of screening for estimated glomerular filtration rate (eGFR), fasting lipogram, eye examination, foot examination and glycated hemoglobin (HbA1c) in the past year. RESULTS: Participants mean age was 62 (standard deviation [s.d.] ± 11) years, and their mean duration of diagnosis was 9 (s.d. ± 8) years. In the past year, HbA1c result was available for 71 (19.1%) of the participants; 60 (16.1%) had eGFR results, while only 33 (8.9%) had documented lipid results. In total, 52 (14.0%) had carried out eye examinations, while only 9 (2.3%) had undergone foot examinations in the past year. About two-thirds of the participants (59.9%) had not undergone any form of complication screening in the past year, and none had undergone the complete screening panel. CONCLUSION: The coverage of screening for DM complications was low across all indicators. Studies to understand barriers to and facilitators of DM complications screening at PHCs are required. Also, interventions to improve diabetes complication screening in the region are needed and should target the primary healthcare providers.

Deep Personal Multitask Prediction of Diabetes Complication with Attentive Interactions Predicting Diabetes Complications by Multitask-Learning.

Objective: Diabetic complications have brought a tremendous burden for diabetic patients, but the problem of predicting diabetic complications is still unresolved. Our aim is to explore the relationship between hemoglobin A1C (HbA1c), insulin (INS), and glucose (GLU) and diabetic complications in combination with individual factors and to effectively predict multiple complications of diabetes. Methods: This was a real-world study. Data were collected from 40,913 participants with an average age of 48 years from the Department of Endocrinology of Ruijin Hospital in Shanghai. We proposed deep personal multitask prediction of diabetes complication with attentive interactions (DPMP-DC) to predict the five complication models of diabetes, including diabetic retinopathy, diabetic nephropathy, diabetic peripheral neuropathy, diabetic foot disease, and diabetic cardiovascular disease. Results: Our model has an accuracy rate of 88.01% for diabetic retinopathy, 89.58% for diabetic nephropathy, 85.77% for diabetic neuropathy, 80.56% for diabetic foot disease, and 82.48% for diabetic cardiovascular disease. The multitasking accuracy of multiple complications is 84.67%, and the missed diagnosis rate is 9.07%. Conclusion: We put forward the method of interactive integration with individual factors of patients for the first time in diabetic complications, which reflect the differences between individuals. Our multitask model using the hard sharing mechanism provides better prediction than prior single prediction models.

Metabolic Status Modulates Choroidal Thickness - A Possible Early Indicator for Diabetic Eye Complications?

OBJECTIVE: To investigate the impact of metabolic status on choroidal thickness (ChT) in healthy subjects, patients with obesity, and type 2 diabetes. DESIGN AND METHODS: Fasting blood glucose, insulin, insulin-like growth factor-1 (IGF-1), and ChT measured by optical coherence tomography were assessed in healthy normal-weight (n=17), obese participants (n=20), and obese participants with T2D (n=16). RESULTS: ChT increased in obese participants and obese participants with T2D as compared to healthy normal-weight participants (P<0.0001). A negative correlation was observed between IGF1 and ChT (r=-0.268, P=0.050) for all cohorts. Furthermore, body mass index (BMI; R2=0.209; P=0.002; beta=0.388) and model assessment-estimated insulin resistance (HOMA-IR; R2=0.074; P=0.015; beta=0.305) were independent variables of ChT, explaining 20.9 and 7.4% of its variance (both p<0.016), whereas age, sex, and IGF-1 were not significant confounders of ChT (p>0.975). CONCLUSION: ChT is associated with metabolic characteristics, i. e., BMI and HOMA-IR. Due to the key role of choroidal function in retinal physiology, future studies are needed to evaluate whether metabolic traits, ChT, and potential metabolic eye complications are mechanistically linked.

Disparities in diabetes-related avoidable hospitalization among diabetes patients with disability using a nationwide cohort study.

Diabetes is an ambulatory care sensitive condition that quality of care can prevent complications development and hospitalization needs. However, diabetes patients with disability face greater challenges with receiving quality diabetes care than those without disabilities. This study examined diabetes-related avoidable hospitalizations (DRAH) focusing on the association with disability. We used nationally representative health insurance cohort data from 2002 to 2013. The study population is people who were newly diagnosed with type 2 diabetes. We measured the cumulated number of DRAH using the Prevention Quality Indicators (PQIs). The variables of interest were disability severity and type. We performed a recurrent events analysis using Cox proportional hazard regression model. Among 49,410 type 2 diabetes patients, 12,231 (24.8%) experienced DRAHs at least once during the follow-up period. Among the total population, 5924 (12.0%) diabetes patients were registered as disabled. The findings report that disability severity was significantly associated with higher risks for DRAH, where severely disabled diabetes patients showed the highest hazard ratio of 2.24 (95% CI 1.80-2.79). Among three DRAH indicators, severely disabled diabetes patients showed increased risks for long-term (AHR 2.21, 95% CI 1.89-2.60) and uncontrolled (AHR 2.28, 95% CI 1.80-2.88) DRAH. In addition, intellectual (AHR 5.52, 95% CI 3.78-8.05) and mental (AHR 3.97, 95% CI 2.29-6.89) disability showed higher risks than other types of disability. In conclusion, diabetes patients with disability are at higher risk for DRAH compared to those without disabilities, and those with intellectual and mental disabilities were more likely to experience DRAH compared to those with physical or other types of disability. These findings call for action to find the more appropriate interventions to improve targeted diabetes primary care for patients with disability. Further research is needed to better understand determinants of increasing risks of DRAH.

Development of the simulation-based German albuminuria screening model (S-GASM) for estimating the cost-effectiveness of albuminuria screening in Germany.

BACKGROUND: Chronic kidney disease is often asymptomatic in its early stages but constitutes a severe burden for patients and causes major healthcare systems costs worldwide. While models for assessing the cost-effectiveness of screening were proposed in the past, they often presented only a limited view. This study aimed to develop a simulation-based German Albuminuria Screening Model (S-GASM) and present some initial applications. METHODS: The model consists of an individual-based simulation of disease progression, considering age, gender, body mass index, systolic blood pressure, diabetes, albuminuria, glomerular filtration rate, and quality of life, furthermore, costs of testing, therapy, and renal replacement therapy with parameters based on published evidence. Selected screening scenarios were compared in a cost-effectiveness analysis. RESULTS: Compared to no testing, a simulation of 10 million individuals with a current age distribution of the adult German population and a follow-up until death or the age of 90 shows that a testing of all individuals with diabetes every two years leads to a reduction of the lifetime prevalence of renal replacement therapy from 2.5% to 2.3%. The undiscounted costs of this intervention would be 1164.10 € / QALY (quality-adjusted life year). Considering saved costs for renal replacement therapy, the overall undiscounted costs would be-12581.95 € / QALY. Testing all individuals with diabetes or hypertension and screening the general population reduced the lifetime prevalence even further (to 2.2% and 1.8%, respectively). Both scenarios were cost-saving (undiscounted, - 7127.10 €/QALY and-5439.23 €/QALY). CONCLUSIONS: The S-GASM can be used for the comparison of various albuminuria testing strategies. The exemplary analysis demonstrates cost savings through albuminuria testing for individuals with diabetes, diabetes or hypertension, and for population-wide screening.

Time series of diabetes attributable mortality from 2008 to 2017.

PURPOSE: Diabetes is a growing health problem. The aim of this study was to capture time trends in mortality associated with diabetes. METHODS: The mortality database of the Veneto region (Italy) includes both the underlying causes of death, and all the diseases mentioned in the death certificate. The annual percent change (APC) in age-standardized rates from 2008 to 2017 was computed by the Joinpoint Regression Program. RESULTS: Overall 453,972 deaths (56,074 with mention of diabetes) were observed among subjects aged ≥ 40 years. Mortality rates declined for diabetes as the underlying cause of death and from diabetes-related circulatory diseases. The latter declined especially in females - 4.4 (CI 95% - 5.3/- 3.4), while in males the APC was - 2.8 (CI 95% - 4.0/- 1.6). CONCLUSION: We observed a significant reduction in mortality during the period 2008-2017 in diabetes either as underlying cause of death or when all mentions of diabetes in the death certificate were considered.

The TUDID Study - Background and Design of a Prospective Cohort.

Prevalence of both type 1 and type 2 diabetes mellitus is growing worldwide and one major cause for morbidity and mortality. However, not every patient develops diabetes-related complications, but causes for the individual susceptibility are still not fully understood. As a platform to address this, we initiated the TUDID (TUebingen DIabetes Database) study, a prospective, monocentric, observational study that includes adults with diabetes mellitus who are treated in the inpatient clinic of a University Hospital in southern Germany. Besides a thorough clinical examination and extensive laboratory tests (with integrated biobanking), major study focuses are the kidneys, the eyes, the vasculature as well as cognition and mood where standardized investigations for early stages for diabetes complications are performed. Analyses of the data generated by this precise characterization of diabetes-related complications will contribute to our understanding of the development and course of such complications, and thus facilitate the implementation of tailored treatment options that can reduce the risk and severity of diabetes-related complications.

Nail changes in diabetes.

Diabetes is a common condition that is increasing in incidence worldwide. Although the skin manifestations of this condition are well described, there is scant literature on the associated nail changes. In this review, we describe the various clinical features of nail changes associated with diabetes, which can be broadly divided into infections, vascular changes, neuropathic manifestations and miscellaneous changes, although there is overlap between them. There is no pathognomonic nail alteration, but it is important for clinicians to be aware of the potential nail manifestations in diabetes as they can facilitate investigations and thereby early diagnosis of diabetes, resulting in holistic management of the patient.

Circular RNA mediated gene regulation in chronic diabetic complications.

Chronic diabetic complications affect multiple organs causing widespread organ damage. Although there are some commonalities, the phenotype of such changes show tissue specific variation. Given this, we examined whether differences in circular RNA (circRNA) mediated gene regulatory mechanisms contribute to changes in gene expression at the basal level and in diabetes. CircRNAs are single-stranded RNA with covalently closed loop structures and act as miRNA sponges, factors of RNA splicing, scaffolding for proteins, regulators of transcription, and modulators of the expression of parental genes, among other roles. We examined heart and retinal tissue from Streptozotocin-induced diabetic mice with established diabetes related tissue damage and tissue from non-diabetic controls. A custom array analysis was performed and the data were analysed. Two major circRNA mediated processes were uniquely upregulated in diabetic heart tissue, namely, positive regulation of endothelial cell migration and regulation of mitochondria: mitochondrial electron transport. In the retina, circRNAs regulating extracellular matrix protein production and endothelial to mesenchymal transition (EndMT) were found to be upregulated. The current study identified regulatory and potential pathogenetic roles of specific circRNA in diabetic retinopathy and cardiomyopathy. Understanding such novel mechanisms, may in the future, be useful to develop RNA based treatment strategies.