Prediction of type 2 diabetes mellitus onset using logistic regression-based scorecards

Predicting Type 2 diabetes onset age using machine learning: A case study in KSA

The rising prevalence of Type 2 Diabetes (T2D) in Saudi Arabia presents significant healthcare challenges. Estimating the age at onset of T2D can aid early interventions, potentially reducing complications due to late diagnoses. This study, conducted at King Abdulaziz Medical University Hospital, aims to predict the age at onset of T2D using Multiple Linear Regression (MLR), Artificial Neural Networks (ANN), Random Forest (RF), Support Vector Regression (SVR), and Decision Tree Regression (DTR). It also seeks to identify key predictors influencing the age at onset of T2D in Saudi Arabia, which ranks 7th globally in prevalence. Medical records from 1,000 diabetic patients from 2018 to 2022 that contain demographic, lifestyle, and lipid profile data are used to develop the models. The average onset age was 65 years, with the most common onset range between 40 and 90 years. The MLR and RF models provided the best fit, achieving R2 values of 0.90 and 0.89, root mean square errors (RMSE) of 0.07 and 0.01, and mean absolute errors (MAE) of 0.05 and 0.13, respectively, using the logarithmic transformation of the onset age. Key factors influencing the age at onset included triglycerides (TG), total cholesterol (TC), high-density lipoprotein (HDL), ferritin, body mass index (BMI), systolic blood pressure (SBP), white blood cell count (WBC), diet, and vitamin D levels. This study is the first in Saudi Arabia to employ MLR, ANN, RF, SVR, and DTR models to predict T2D onset age, providing valuable tools for healthcare practitioners to monitor and design intervention strategies aimed at reducing the impact of T2D in the region.

Robust predictive framework for diabetes classification using optimized machine learning on imbalanced datasets

Introduction

Diabetes prediction using clinical datasets is crucial for medical data analysis. However, class imbalances, where non-diabetic cases dominate, can significantly affect machine learning model performance, leading to biased predictions and reduced generalization.

Methods

A novel predictive framework employing cutting-edge machine learning algorithms and advanced imbalance handling techniques was developed. The framework integrates feature engineering and resampling strategies to enhance predictive accuracy.

Results

Rigorous testing was conducted on three datasets-PIMA, Diabetes Dataset 2019, and BIT_2019-demonstrating the robustness and adaptability of the methodology across varying data environments.

Discussion

The experimental results highlight the critical role of model selection and imbalance mitigation in achieving reliable and generalizable diabetes predictions. This study offers significant contributions to medical informatics by proposing a robust data-driven framework that addresses class imbalance challenges, thereby advancing diabetes prediction accuracy.

Genetics-driven risk predictions leveraging the Mendelian randomization framework

Accurate predictive models of future disease onset are crucial for effective preventive healthcare, yet longitudinal data sets linking early risk factors to subsequent health outcomes are limited. To overcome this challenge, we introduce a novel framework, Predictive Risk modeling using Mendelian Randomization (PRiMeR), which utilizes genetic effects as supervisory signals to learn disease risk predictors without relying on longitudinal data. To do so, PRiMeR leverages risk factors and genetic data from a healthy cohort, along with results from genome-wide association studies of diseases of interest. After training, the learned predictor can be used to assess risk for new patients solely based on risk factors. We validate PRiMeR through comprehensive simulations and in future type 2 diabetes predictions in UK Biobank participants without diabetes, using follow-up onset labels for validation. Moreover, we apply PRiMeR to predict future Alzheimer's disease onset from brain imaging biomarkers and future Parkinson's disease onset from accelerometer-derived traits. Overall, with PRiMeR we offer a new perspective in predictive modeling, showing it is possible to learn risk predictors leveraging genetics rather than longitudinal data.

Plasma infrared fingerprinting with machine learning enables single-measurement multi-phenotype health screening

Infrared spectroscopy is a powerful technique for probing the molecular profiles of complex biofluids, offering a promising avenue for high-throughput in vitro diagnostics. While several studies showcased its potential in detecting health conditions, a large-scale analysis of a naturally heterogeneous potential patient population has not been attempted. Using a population-based cohort, here we analyze 5,184 blood plasma samples from 3,169 individuals using Fourier transform infrared (FTIR) spectroscopy. Applying a multi-task classification to distinguish between dyslipidemia, hypertension, prediabetes, type 2 diabetes, and healthy states, we find that the approach can accurately single out healthy individuals and characterize chronic multimorbid states. We further identify the capacity to forecast the development of metabolic syndrome years in advance of onset. Dataset-independent testing confirms the robustness of infrared signatures against variations in sample handling, storage time, and measurement regimes. This study provides the framework that establishes infrared molecular fingerprinting as an efficient modality for populational health diagnostics.

Geno-GCN: A Genome-specific Graph Convolutional Network for Diabetes Prediction

Drawing inspiration from convolutional neural networks, graph convolutional networks (GCNs) have been implemented in various applications. Yet, the integration of GCNs into clinical settings, particularly in the context of complex health conditions like diabetes, remains distant. In this paper, we introduce a genome-specific graph convolutional network (Geno-GCN) with a multi-graph aggregator to predict the risk of developing Type 2 diabetes based on whole genome sequencing data. Geno-GCN consolidates both positive and negative influences from graphs formulated from diabetes risk factors. This is achieved through a negative sample strategy combined with multi-view aggregators. We assessed Geno-GCN using Australia's largest genome bank and benchmarked it against rule-based methods, bioinformatics tools, and other state-of-the-art machine-learning techniques. The results demonstrated the superior efficacy and robustness of our method, which consistently outperformed competitors across all evaluation metrics. Geno-GCN also exhibited the closest alignment with actual labels, showcasing its potential in large population studies.

Analysis of risk factors and clinical implications for diabetes in first-degree relatives in the northeastern region of China

Background

The prevalence of diabetes has risen fast with a considerable weighted prevalence of undiagnosed diabetes or uncontrolled diabetes. Then it becomes more necessary to timely screen out and monitor high-risk populations who are likely to be ignored during the COVID-19 pandemic. To classify and find the common risks of undiagnosed diabetes and uncontrolled diabetes, it's beneficial to put specific risk control measures into effect for comprehensive primary care. Especially, there is a need for accurate yet accessible prediction models.

Objective

Based on a cross-sectional study and secondary analysis on the health examination held in Changchun City (2016), we aimed to evaluate the factors associated with hyperglycemia, analyze the management status of T2DM, and determine the best cutoff value of incidence of diabetes in the first-degree relatives to suggest the necessity of early diagnosis of diabetes after first screening.

Results

A total of 5658 volunteers were analyzed. Prevalence of T2DM and impaired fasting glucose were 8.4% (n=477) and 11.5% (n=648), respectively. There were 925 participants (16.3%) with a family history of T2DM in their first-degree relatives. Multivariable analysis demonstrated that family history was associated with hyperglycemia. Among the 477 patients with T2DM, 40.9% had not been previously diagnosed. The predictive equation was calculated with the following logistic regression parameters with 0.71 (95% CI: 0.67-0.76) of the area under the ROC curve, 64.0% of sensitivity and 29% of specificity (P < 0.001): P = \frac{1}{1 + e^{-z}}, where z = -3.08 + [0.89 (Family history-group) + 0.69 (age-group)+ 0.25 (BMI-group)]. Positive family history was associated with the diagnosis of T2DM, but not glucose level in the diagnosed patients. The best cutoff value of incidence of diabetes in the first-degree relatives was 9.55% (P < 0.001).

Conclusions

Family history of diabetes was independently associated with glucose dysfunction. Classification by the first-degree relatives with diabetes is prominent for targeting high-risk population. Meanwhile, positive family history of diabetes was associated with diabetes being diagnosed rather than the glycemic control in patients who had been diagnosed. It's necessary to emphasize the linkage between early diagnosis and positive family history for high proportions of undiagnosed T2DM.

Carpal Tunnel Syndrome and Trigger Finger May Be an Early Symptom of Preclinic Type 2 Diabetes

Background

Type 2 diabetes (T2D) is a major risk factor for carpal tunnel syndrome (CTS) and trigger finger (TF), but less is known regarding the risk of developing T2D after being diagnosed with CTS or TF. CTS and TF could be early signs of preclinical T2D, and early detection of T2D is crucial to prevent complications and morbidity. Therefore, we investigate the association between CTS/TF and T2D in an adult population without previous T2D using big data registers in Sweden.

Methods

Data were collected by crosslinking five nationwide Swedish registers. Individuals aged 40-85 years on December 31, 2010, without prior overt diabetes, were included (n = 3,948,517) and followed up from baseline (ie, a diagnosis of CTS or TF) or January 1, 2011, for controls, until a diagnosis of T2D, prescription of oral antidiabetics or insulin, or end of follow-up four years after baseline. Multivariate Cox regression models were created to calculate hazard ratios for T2D.

Results

In total, 37,346 (0.95%) patients were diagnosed with CTS, whereof 1329 (3.46%) developed T2D. There were 17,432 (0.44%) patients who developed TF, whereof 639 (3.67%) developed T2D. Among the controls, 2.73% developed T2D. Compared with controls, there was an increased risk of developing T2D after being diagnosed with either CTS (HR 1.35; 95% confidence interval 1.28-1.43) or TF (HR 1.21; 95% confidence interval 1.12-1.31).

Conclusion

Compared with controls, a diagnosis of CTS or TF was associated with 35% and 21% higher risk for later T2D, respectively, which might indicate the existence of undetected T2D in this population.

Risk of diabetes and expected years in life without diabetes among adults from an urban community in India: findings from a retrospective cohort

BACKGROUND

Diabetes prevalence has increased over the past few decades, and the shift of the burden of diabetes from the older population to the younger population has increased the exposure of longer durations in a morbid state. The study aimed at ascertaining the likelihood of progression to diabetes and to estimate the onset of diabetes within the urban community of Mumbai.

METHODS

This study utilized an observational retrospective non-diabetic cohort comprising 1629 individuals enrolled in a health security scheme. Ten years of data were extracted from electronic medical records, and the life table approach was employed to assess the probability of advancing to diabetes and estimate the expected number of years lived without a diabetes diagnosis.

RESULTS

The study revealed a 42% overall probability of diabetes progression, with age and gender variations. Males (44%) show higher probabilities than females (40%) of developing diabetes. Diabetes likelihood rises with age, peaking in males aged 55-59 and females aged 65-69. Males aged 30-34 exhibit a faster progression (10.6 years to diagnosis) compared to females (12.3 years).

CONCLUSION

The study's outcomes have significant implications for the importance of early diabetes detection. Progression patterns suggest that younger cohorts exhibit a comparatively slower rate of progression compared to older cohorts.

A novel interpretable deep transfer learning combining diverse learnable parameters for improved T2D prediction based on single-cell gene regulatory networks

Accurate deep learning (DL) models to predict type 2 diabetes (T2D) are concerned not only with targeting the discrimination task but also with learning useful feature representation. However, existing DL tools are far from perfect and do not provide appropriate interpretation as a guideline to explain and promote superior performance in the target task. Therefore, we provide an interpretable approach for our presented deep transfer learning (DTL) models to overcome such drawbacks, working as follows. We utilize several pre-trained models including SEResNet152, and SEResNeXT101. Then, we transfer knowledge from pre-trained models via keeping the weights in the convolutional base (i.e., feature extraction part) while modifying the classification part with the use of Adam optimizer to deal with classifying healthy controls and T2D based on single-cell gene regulatory network (SCGRN) images. Another DTL models work in a similar manner but just with keeping weights of the bottom layers in the feature extraction unaltered while updating weights of consecutive layers through training from scratch. Experimental results on the whole 224 SCGRN images using five-fold cross-validation show that our model (TFeSEResNeXT101) achieving the highest average balanced accuracy (BAC) of 0.97 and thereby significantly outperforming the baseline that resulted in an average BAC of 0.86. Moreover, the simulation study demonstrated that the superiority is attributed to the distributional conformance of model weight parameters obtained with Adam optimizer when coupled with weights from a pre-trained model.

The role of heat shock proteins (HSPs) in type 2 diabetes mellitus pathophysiology

Type 2 diabetes mellitus (T2DM) is a metabolic disorder characterized by sustained hyperglycemia caused by impaired insulin signaling and secretion. Metabolic stress, caused by an inappropriate diet, is one of the major hallmarks provoking inflammation, endoplasmic reticulum (ER) stress, and mitochondrial dysfunction. Heat shock proteins (HSPs) are a group of highly conserved proteins that have a crucial role in chaperoning damaged and misfolded proteins to avoid disruption of cellular homeostasis under stress conditions. To do this, HSPs interact with diverse intra-and extracellular pathways among which are the insulin signaling, insulin secretion, and apoptosis pathways. Therefore, HSP dysfunction, e.g. HSP70, may lead to disruption of the pathways responsible for insulin secretion and uptake. Consistently, the altered expression of other HSPs and genetic polymorphisms in HSP-producing genes in diabetic subjects has made HSPs hot research in T2DM. This paper provides a comprehensive overview of the role of different HSPs in T2DM pathogenesis, affected cellular pathways, and the potential therapeutic strategies targeting HSPs in T2DM.

Multi-omics Analysis Reveals the Crucial Mediators of DJB in the Treatment of Type 2 Diabetes

PURPOSE

Duodenal-jejunal bypass (DJB) has a definite hypoglycemic effect; however, the intrinsic mechanisms remain unclear. The purpose of this study was to determine whether DJB may cause changes in the gut microbiota and metabolite of portal venous blood and to explore the effects of DJB on blood glucose metabolism.

METHODS

T2DM was induced in rats with a high-fat diet and a low dose of streptozotocin, which were randomly divided into two groups: Sham operation and DJB.

RESULTS

DJB significantly improved several diabetic parameters. 16S rRNA analyses showed that the compositions of the gut microbiota were significantly different between the two groups. The results of metabolomics showed that DJB could significantly regulate the metabolites, among which diaminopimelic acid and isovaleric acid had a significant down-regulation in the DJB group. Transcriptomic analysis showed that DJB can regulate the expression of hepatic genes related to abnormal glucose metabolism, such as Ltc4s, Alox15, Ggt1, Gpat3, and Cyp2c24. Correlation analyses showed that diaminopimelic acid was positively associated with Allobaculum, Serratia, and Turicibacter. There was a significant correlation between diaminopimelic acid and Gpat3, and its Spearman correlation coefficient was the highest among metabolite-DEG pairs (ρ=0.97).

DISCUSSIONS

These results suggest an important cue of the relation between the diaminopimelic acid, Gpat3, and gut microbiome in the mechanism by which DJB can improve glucose metabolism.

Application of Artificial Intelligence in Assessing the Self-Management Practices of Patients with Type 2 Diabetes

The use of Artificial intelligence in healthcare has evolved substantially in recent years. In medical diagnosis, Artificial intelligence algorithms are used to forecast or diagnose a variety of life-threatening illnesses, including breast cancer, diabetes, heart disease, etc. The main objective of this study is to assess self-management practices among patients with type 2 diabetes in rural areas of Pakistan using Artificial intelligence and machine learning algorithms. Of particular note is the assessment of the factors associated with poor self-management activities, such as non-adhering to medications, poor eating habits, lack of physical activities, and poor glycemic control (HbA1c %). The sample of 200 participants was purposefully recruited from the medical clinics in rural areas of Pakistan. The artificial neural network algorithm and logistic regression classification algorithms were used to assess diabetes self-management activities. The diabetes dataset was split 80:20 between training and testing; 80% (160) instances were used for training purposes and 20% (40) instances were used for testing purposes, while the algorithms' overall performance was measured using a confusion matrix. The current study found that self-management efforts and glycemic control were poor among diabetes patients in rural areas of Pakistan. The logistic regression model performance was evaluated based on the confusion matrix. The accuracy of the training set was 98%, while the test set's accuracy was 97.5%; each set had a recall rate of 79% and 75%, respectively. The output of the confusion matrix showed that only 11 out of 200 patients were correctly assessed/classified as meeting diabetes self-management targets based on the values of HbA1c < 7%. We added a wide range of neurons (32 to 128) in the hidden layers to train the artificial neural network models. The results showed that the model with three hidden layers and Adam's optimisation function achieved 98% accuracy on the validation set. This study has assessed the factors associated with poor self-management activities among patients with type 2 diabetes in rural areas of Pakistan. The use of a wide range of neurons in the hidden layers to train the artificial neural network models improved outcomes, confirming the model's effectiveness and efficiency in assessing diabetes self-management activities from the required data attributes.