Early Risk Assessment Research Articles

Inter-rater reliability of the Early Assessment Risk List Version 3 (EARL-V3) for children displaying antisocial behavior

BackgroundIdentification of risks is foundational to the development of strategies to prevent conduct problems in children. The Early Assessment Risk List Version 3 (EARL-V3) is a Structured Professional Judgment (SPJ) instrument designed for application to children exhibiting antisocial behavior. The revised EARL-V3 (2021) encourages users to be responsive to gender and cultural issues and is designed to assist in treatment planning. While previous versions of the EARL have undergone extensive testing, reliability has not been established for the new version. MethodsThe present study yields Intraclass Correlation Coefficients (ICCs) for the three EARL-V3 subscales and the total score. Additional analyses are based on Gwet’s AC1 method for examining individual items. Four rater pairs assessed a selection of 124 children (50 girls, 74 boys; six to 12 years). Case files included child and caregiver interviews and clinical assessments. ResultsWe found moderate to excellent agreement for two of the three subscales and the total score, and poor to moderate agreement for one subscale. The individual items ranged from fair to almost perfect agreement. The Overall Clinical Risk Judgement revealed fair to moderate inter-rater agreement. The results largely remained when analyzed based on gender. Some differences were seen for individual risk factors. ConclusionThe EARL-V3 has a good level of inter-rater reliability, comparable or superior to previous versions of the EARL and to other SPJ instruments.

Towards equitable AI in oncology.

Artificial intelligence (AI) stands at the threshold of revolutionizing clinical oncology, with considerable potential to improve early cancer detection and risk assessment, and to enable more accurate personalized treatment recommendations. However, a notable imbalance exists in the distribution of the benefits of AI, which disproportionately favour those living in specific geographical locations and in specific populations. In this Perspective, we discuss the need to foster the development of equitable AI tools that are both accurate in and accessible to a diverse range of patient populations, including those in low-income to middle-income countries. We also discuss some of the challenges and potential solutions in attaining equitable AI, including addressing the historically limited representation of diverse populations in existing clinical datasets and the use of inadequate clinical validation methods. Additionally, we focus on extant sources of inequity including the type of model approach (such as deep learning, and feature engineering-based methods), the implications of dataset curation strategies, the need for rigorous validation across a variety of populations and settings, and the risk of introducing contextual bias that comes with developing tools predominantly in high-income countries.

Construction of an adverse outcome pathway framework for arsenic-induced lung cancer using a network-based approach

Environmental pollutants are considered as a cause of tumorigenesis, but approaches to assess their risk of causing tumors remain insufficient. As an alternative approach, the adverse outcome pathway (AOP) framework is used to assess the risk of tumors caused by environmental pollutants. Arsenic is a pollutant associated with lung cancer, but early assessment of lung cancer risk is lacking. Therefore, we applied the AOP framework to arsenic-induced lung cancer. A systematic review revealed increased risks of lung cancer following exposure to a range of arsenic concentrations in drinking water (OR = 1.83, 95 % CI = 1.46–2.30). We obtained, from public databases, genes related to risk of arsenic-induced lung cancer. Then, Cox and LASSO regressions were used to screen target genes from the risk genes. Subsequently, target genes, phenotypes, and pathways were used to construct the computational AOP network, which was determined by Cytoscape to have 156 edges and 45 nodes. Further, target genes, phenotypes, and pathways were used as molecular initiating events and key events to construct the AOP framework depending on upstream and downstream relationships. In the AOP framework, by Weight of Evidence, arsenic exposure increased levels of EGFR, activated the PI3K/AKT pathway, regulated cell proliferation by promoting the G1/S phase transition, and caused generation of lung cancers. External validation was achieved through arsenite-induced, malignant transformed human bronchial epithelial (HBE) cells. Overall, these results, by integration into existing data to construct an AOP framework, provide insights into the assessment of lung cancer risk for arsenic exposure. Special attention needs to be focused on populations with low-dose arsenic exposure.

Comparative proteomic analysis of saliva from chewing tobacco users and oral cancer patients reveals shared biomarkers: A case control observational study

ObjectiveThe aim of this study was to compare the salivary proteomic profile of smokeless tobacco users with that of non-users and oral cancer patients using Liquid Chromatography-Mass Spectrometry/ Mass Spectrometry (LC-MS/MS). MethodsSaliva samples from 65 participants were collected in three groups: control (25 participants), smokeless tobacco users (25 participants), and oral cancer (15 participants). ResultsThe analysis revealed 343 protein groups with significantly altered abundance in the saliva samples (P < 0.05). Among these, 43 out of 51 dysregulated proteins in the smokeless tobacco group were also dysregulated in the oral cancer group. Notably, Apolipoprotein A1 (ApoA1) and Pon1 were found to be significantly increased in both smokeless tobacco users and oral cancer patients (p < 0.05). Furthermore, six out of the 20 most significantly altered proteins were mitochondrial proteins, and all of these were decreased relative to controls in both smokeless tobacco users and cancer samples. ConclusionThe proteomic profile of users of chewing (smokeless) tobacco (SLT) shows substantial overlap in the altered pathways and dysregulated proteins with those altered in oral cancer samples, suggesting that SLT use induces a shift toward an oncogenic state. Specifically indicated pathways included blood microparticles, platelet α-granules and protease inhibitors as well as indicators of oxidative stress and exogenous compound processing. What differentiates oral cancer samples from SLT users is enrichment of alterations related to cytoskeletal organisation and tissue remodelling. Clinical SignificanceThe findings emphasize the importance of salivary proteomic profiles because changes in certain proteins may be indicators for early oral cancer identification and risk assessment in smokeless tobacco users.

Forecasting Meteorological Drought Conditions in South Korea Using a Data-Driven Model with Lagged Global Climate Variability

Drought prediction is crucial for early risk assessment, preventing negative impacts and the timely implementation of mitigation measures for sustainable water management. This study investigated the relationship between climate variations in three seas and the prediction of December meteorological droughts in South Korea, using the Standardized Precipitation Evapotranspiration Index (SPEI). Climate indices with multiple time lags were integrated into multiple linear regression (MLR) and Random Forest (RF) models and evaluated using Pearson’s correlation coefficients (PCCs) and the Root Mean Square Error (RMSE). The results indicated that the MLR model outperformed RF model in the western inland region with a PCC of 0.52 for predicting SPEI-2. On the other hand, the RF model effectively predicted drought states of ‘moderate drought’ or worse (SPEI &lt; −1) nationwide, achieving an average hit rate of 47.17% and Heidke skill score (HSS) of 0.56, particularly excelling in coastal areas. Nino 3.4 turned out to be the most influential factor for short-period extreme droughts (SPEI-2) with a three-month lag, contributed by the Pacific, Atlantic, and Indian Oceans. For periods of four months or longer, climate variations had a lower predictive value. However, integrating autocorrelation functions to account for the previous month’s drought status improved the accuracy. A HYBRID model, which blends linear and nonlinear approaches, further enhanced reliability, making the proposed model more applicable for drought forecasting in neighboring countries and valuable for South Korea’s drought monitoring system to support sustainable water management.

Heart Disease Prediction Using Machine Learning

Heart disease remains a leading cause of mortality worldwide, necessitating improved methods for early detection and risk assessment. This paper reviews and analyzes the application of machine learning techniques in heart disease prediction, focusing on five primary algorithms: Naïve Bayes, k-Nearest Neighbor (KNN), Decision Tree, Artificial Neural Network (ANN), and Random Forest. By examining existing studies and datasets, we evaluate the effectiveness of these algorithms in predicting heart disease risk. Our analysis demonstrates that machine learning models can significantly enhance the accuracy and efficiency of heart disease prediction compared to traditional diagnostic methods. The Random Forest algorithm exhibited the highest overall performance, with studies reporting accuracy rates up to 95% in identifying potential heart disease cases. This review highlights the potential of machine learning in revolutionizing cardiovascular healthcare by enabling more personalized risk assessments and facilitating early intervention strategies. The integration of these advanced predictive models into clinical practice could substantially improve patient outcomes and reduce the global burden of heart disease. Keywords: Cardiovascular Risk Prediction, Machine Learning Algorithms, Electronic Health Records, Random Forest, Artificial Neural Networks, Feature Importance, Clinical Decision Support, Personalized Medicine, Predictive Analytics in Healthcare, Early Disease Detection.

Myeloperoxidase and its derivative hypochlorous acid combined clinical indicators predict new-onset atrial fibrillation in sepsis: a case-control study

BackgroudNew-onset atrial fibrillation (NOAF) is a common complication of sepsis and linked to higher death rates in affected patients. The lack of effective predictive tools hampers early risk assessment for the development of NOAF. This study aims to develop practical and effective predictive tools for identifying the risk of NOAF.MethodsThis case-control study retrospectively analyzed patients with sepsis admitted to the emergency department of Xinhua Hospital, Shanghai Jiao Tong University School of Medicine from September 2017 to January 2023. Based on electrocardiographic reports and electrocardiogram monitoring records, patients were categorized into NOAF and non-NOAF groups. Laboratory tests, including myeloperoxidase (MPO) and hypochlorous acid (HOCl), were collected, along with demographic data and comorbidities. Least absolute shrinkage and selection operator regression and multivariate logistic regression analyses were employed to identify predictors. The area under the curve (AUC) was used to evaluate the predictive model’s performance in identifying NOAF.ResultsA total of 389 patients with sepsis were included in the study, of which 63 developed NOAF. MPO and HOCl levels were significantly higher in the NOAF group compared to the non-NOAF group. Multivariate logistic regression analysis identified MPO, HOCl, tumor necrosis factor-α (TNF-α), white blood cells (WBC), and the Acute Physiology and Chronic Health Evaluation II (APACHE II) score as independent risk factors for NOAF in sepsis. Additionally, a nomogram model developed using these independent risk factors achieved an AUC of 0.897.ConclusionThe combination of MPO and its derivative HOCl with clinical indicators improves the prediction of NOAF in sepsis. The nomogram model can serve as a practical predictive tool for the early identification of NOAF in patients with sepsis.

Role of the Stress Index in Predicting Mortality among Patients with Traumatic Femoral Fractures.

Traumatic femoral fractures, often resulting from high-energy impacts such as traffic accidents, necessitate immediate management to avoid severe complications. The Stress Index (SI), defined as the glucose-to-potassium ratio, serves as a predictor of mortality and adverse outcomes in various trauma contexts. This study aims to evaluate the prognostic value of the SI in patients with traumatic femoral fractures. This retrospective cohort study included adult trauma patients aged 20 or above with traumatic femoral fractures from the Trauma Registry System at a level 1 trauma center in southern Taiwan between 1 January 2009 and 31 December 2022. At the emergency room, serum electrolyte levels were assessed using baseline laboratory testing. By dividing blood glucose (mg/dL) by potassium (mEq/L), the SI was calculated. The best cut-off value of the SI for predicting mortality was determined using the Area Under the Curve (AUC) of Receiver Operating Characteristic (ROC). A total of 3717 patients made up the final group, of which 3653 survived and 64 died. In comparison to survivors, deceased patients had substantially higher blood glucose levels (199.3 vs. 159.0 mg/dL, p < 0.001) and SIs (53.1 vs. 41.6, p < 0.001). The optimal SI cut-off value for predicting mortality was 49.7, with a sensitivity of 53.1% and a specificity of 78.7% (AUC = 0.609). High SI was associated with increased mortality (4.2% vs. 1.0%, p < 0.001) and longer hospital stays (12.8 vs. 9.5 days, p < 0.001). The adjusted odds ratios of mortality, controlled by comorbidities, the Glasgow Coma Scale, and the Injury Severity Score, were significantly higher in patients with a higher SI (AOR 2.05, p = 0.016) than those with a lower SI. Elevated SI upon admission correlates with higher mortality and extended hospital stay in patients with traumatic femoral fractures. Although the SI has a moderate predictive value, it remains a useful early risk assessment tool, necessitating further prospective, multi-center studies for validation and standardization.

Towards automatic forecasting of lung nodule diameter with tabular data and CT imaging

Objective:This study aims to forecast the progression of lung cancer by estimating the future diameter of lung nodules. Methods:This approach uses as input the tabular data, axial images from tomography scans, and both data types, employing a ResNet50 model for image feature extraction and direct analysis of patient information for tabular data. The data are processed through a neural network before prediction. In the training phase, class weights are assigned based on the rarity of different types of nodules within the dataset, in alignment with nodule management guidelines. Results:Tabular data alone yielded the most accurate results, with a mean absolute deviation of 0.99 mm. For malignant nodules, the best performance, marked by a deviation of 2.82 mm, was achieved using tabular data applying Lung-RADS class weights during training. The tabular data results highlight the influence of using the initial nodule size as an input feature. These results surpass the literature reference of 348-day volume doubling time for malignant nodules. Conclusion:The developed predictive model is optimized for integration into a clinical workflow after detecting, segmenting, and classifying nodules. It provides accurate growth forecasts, establishing a more objective basis for determining follow-up intervals. Significance:With lung cancer’s low survival rates, the capacity for precise nodule growth prediction represents a significant breakthrough. This methodology promises to revolutionize patient care and management, enhancing the chances for early risk assessment and effective intervention.

Research perspective in the clinical management of Kawasaki disease.

This study examines research perspective in the clinical diagnosis, treatment, and prevention of cardiovascular complications in Kawasaki Disease (KD). Starting with an overview of the disease, it introduces KD's clinical manifestations, etiology, epidemiological features, and its impact on the cardiovascular system. Subsequently, the study discusses in detail the diagnostic methods, pathological mechanisms, and treatment strategies for KD, including foundational and emerging approaches such as high-dose intravenous immunoglobulin and aspirin therapy, biologic therapy, and corticosteroid pulse therapy. Additionally, it outlines strategies for preventing cardiovascular complications, including early risk assessment and long-term management. The study also explores the intersection of the COVID-19 pandemic with an increase in KD-like symptoms, emphasizing the need for further studies on the association between SARS-CoV-2 and KD. Lastly, it explores future research directions to enhance understanding of KD and improve patient outcomes and quality of life. This study provides valuable insights into the comprehensive treatment and management of KD and highlights avenues for future research.

Long-term association of remnant cholesterol with all-cause and cardiovascular disease mortality: a nationally representative cohort study.

Despite reducing low-density lipoprotein cholesterol (LDL-C) to the normal range, residual cardiovascular risk remain. Remnant cholesterol (RC) exerts a potential residual risk for cardiovascular disease (CVD) prevention, and the long-term longitudinal association between RC and mortality has yet to be well elucidated. This study examined a nationally representative sample of 13,383 adults aged 20 years or older (mean age 45.7 and 52% women) who participated in the NHANES III (from1988 to1994). Causes of death were ascertained by linkage to death records through December 31, 2019. The relations of RC with all-cause and CVD mortality were tested using weighted Cox proportional hazard models. Through a median follow-up of 26.6 years, 5,044 deaths were reported, comprising 1,741 deaths of CVD [1,409 deaths of ischemic heart disease (IHD) and 332 deaths of stroke] and 1,126 of cancer. Compared to those with RC <14.26 mg/dl (lowest quartile), participants with RC ≥29.80 mg/dl (highest quartile) had multivariable-adjusted HRs of 1.23 (95% CI: 1.07-1.42) for all-cause mortality, 1.22 (95% CI: 0.97-1.53) for CVD mortality, and 1.32 (95% CI: 1.03-1.69) for IHD mortality, and 0.89 (95% CI: 0.55-1.43) for stroke mortality, and 1.17 (95% CI 0.90-1.52) for cancer mortality. We observed that elevated RC levels increased CVD risk and IHD mortality despite LDL-C being in the normal range. Elevated blood RC was associated with an increased long-term risk of all-cause, CVD, and IHD mortality. These associations were independent of socioeconomic factors, lifestyles, and history of diseases, and remained robust across the LDL-C stratum. Measuring RC levels might favor clinical assessment of early CVD risk. Further investigation is needed to elucidate the optimal range of RC levels for cardiovascular disease health in the general population.

Assessing the interplay between off-target promiscuity, cytotoxicity, and tolerability in rodents to improve the safety profile of novel anti-malarial plasmepsin X inhibitors.

Within drug development, high off-target promiscuity as well as potent cytotoxicity, are associated with a high attrition rate. We investigated the safety profile of novel plasmepsin X (PMX) inhibitors for the treatment of malaria. In our screening cascade, a total of 249 PMX compounds were profiled in a panel of in vitro secondary pharmacology assays containing 44 targets (SafetyScreen44 panel) and in a cytotoxicity assay in HepG2 cells using ATP as an endpoint. Six of the lead compounds were subsequently tested in a 7-d rat toxicology study, and/or in a cardiovascular study in guinea pigs. Overall, compounds with high cytotoxicity in HepG2 cells correlated with high promiscuity (off-target hit rate >20%) in the SafetyScreen44 panel and were associated with poor tolerability in vivo (decedents, morbidity, adverse clinical signs, or severe cardiovascular effects). Some side effects observed in rats or guinea pigs could putatively be linked with hits in the secondary pharmacological profiling, such as the M1 or M2 muscarinic acetylcholine receptor, opioid µ and/or κ receptors or hERG/CaV1.2/Na+ channels, which were common to >50% the compounds tested in vivo. In summary, compounds showing high cytotoxicity and high promiscuity are likely to be poorly tolerated in vivo. However, such associations do not necessarily imply a causal relationship. Identifying the targets that cause these undesirable effects is key for early safety risk assessment. A tiered approach, based on a set of in vitro assays, helps selecting the compounds with highest likelihood of success to proceed to in vivo toxicology studies.

Potential of High Sensitivity C-Reactive Protein in Early Assessment of ASCVD Risk among Prediabetic Patients

Potential of High Sensitivity C-Reactive Protein in Early Assessment of ASCVD Risk among Prediabetic Patients

Maternal ethnicity and socioeconomic deprivation: influence on adverse pregnancy outcomes.

To evaluate the relative importance of ethnicity and socioeconomic deprivation in determining the likelihood and prevalence of placentally derived composite of adverse pregnancy outcomes (CAPO) and composite of severe adverse pregnancy outcomes (CAPO-S). This was a single-center retrospective cohort study of data obtained in a tertiary maternity unit. Data regarding ethnicity and socioeconomic deprivation (as measured with indices of multiple deprivation) were collected for 13 165 singleton pregnancies screened routinely in the first trimester for pre-eclampsia using the Fetal Medicine Foundation combined risk-assessment algorithm. CAPO was defined as the presence of one or more interrelated outcomes associated with placental dysfunction, namely, hypertensive disorders of pregnancy, preterm birth, birth weight ≤ 10th centile and stillbirth. CAPO-S was defined as the presence of one or more of the following: hypertensive disorders of pregnancy at < 37 + 0 weeks, preterm birth at < 34 + 0 weeks, birth weight ≤ 5th centile and stillbirth at < 37 + 0 weeks. The prevalence of CAPO was 16.3% in white women, 29.3% in black women and 29.3% in South Asian women. However, half (51.7%) of all CAPO cases occurred in white women. There was a strong interaction between ethnicity and socioeconomic deprivation, with a correlation coefficient of -0.223. Both ethnicity and socioeconomic deprivation influenced the prevalence of CAPO and CAPO-S, with the contribution of ethnicity being the strongest. Black and Asian ethnicity, as well as socioeconomic deprivation, influence the prevalence of placenta-mediated adverse pregnancy outcomes. Despite this, most adverse pregnancy outcomes occur in white women, who represent the majority of the population and are also affected by socioeconomic deprivation. For these reasons, inclusion of socioeconomic deprivation should be considered in early pregnancy risk assessment for placenta-mediated CAPO. © 2024 The Authors. Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.

An Advanced Lung Carcinoma Prediction and Risk Screening Model Using Transfer Learning.

Lung cancer, also known as lung carcinoma, has a high death rate, but an early diagnosis can substantially reduce this risk. In the current era, prediction models face challenges such as low accuracy, excessive noise, and low contrast. To resolve these problems, an advanced lung carcinoma prediction and risk screening model using transfer learning is proposed. Our proposed model initially preprocesses lung computed tomography images for noise removal, contrast stretching, convex hull lung region extraction, and edge enhancement. The next phase segments the preprocessed images using the modified Bates distribution coati optimization (B-RGS) algorithm to extract key features. The PResNet classifier then categorizes the cancer as normal or abnormal. For abnormal cases, further risk screening determines whether the risk is low or high. Experimental results depict that our proposed model performs at levels similar to other state-of-the-art models, achieving enhanced accuracy, precision, and recall rates of 98.21%, 98.71%, and 97.46%, respectively. These results validate the efficiency and effectiveness of our suggested methodology in early lung carcinoma prediction and risk assessment.

The Urinary Glycopeptide Profile Differentiates Early Cardiorenal Risk in Subjects Not Meeting Criteria for Chronic Kidney Disease.

Early diagnosis and treatment of chronic kidney disease (CKD) is a worldwide challenge. Subjects with albumin-to-creatinine ratio (ACR) ≥ 30 mg/g and preserved renal function are considered to be at no cardiorenal risk in clinical practice, but prospective clinical studies evidence increased risk, even at the high-normal (HN) ACR range (10-30 mg/g), supporting the need to identify other molecular indicators for early assessment of patients at higher risk. Following our previous studies, here we aim to stratify the normoalbuminuria range according to cardiorenal risk and identify the glycoproteins and N-glycosylation sites associated with kidney damage in subclinical CKD. Glycoproteins were analyzed in urine from hypertensive patients within the HN ACR range compared to control group (C; ACR < 10 mg/g) by mass spectrometry. A different cohort was analyzed for confirmation (ELISA) and sex perspective was evaluated. Patients' follow-up for 8 years since basal urine collection revealed higher renal function decline and ACR progression for HN patients. Differential N-glycopeptides and their N -glycosylation sites were also identified, together with their pathogenicity. N-glycosylation may condition pathological protein deregulation, and a panel of 62 glycoproteins evidenced alteration in normoalbuminuric subjects within the HN range. Haptoglobin-related protein, haptoglobin, afamin, transferrin, and immunoglobulin heavy constant gamma 1 (IGHG1) and 2 (IGHG2) showed increased levels in HN patients, pointing to disturbed iron metabolism and tubular reabsorption and supporting the tubule as a target of interest in the early progression of CKD. When analyzed separately, haptoglobin, afamin, transferrin, and IGHG2 remained significant in HN, in both women and men. At the peptide level, 172 N-glycopeptides showed differential abundance in HN patients, and 26 showed high pathogenicity, 10 of them belonging to glycoproteins that do not show variation between HN and C groups. This study highlights the value of glycosylation in subjects not meeting KDIGO criteria for CKD. The identified N-glycopeptides and glycosylation sites showed novel targets, for both the early assessment of individual cardiorenal risk and for intervention aimed at anticipating CKD progression.

Exploring the role of interleukin-1β and interleukin-6 in the pathophysiology of obsessive-compulsive disorder.

Obsessive-compulsive disorder (OCD) is a highly prevalent neuropsychiatric disorder. Recently, there has been a growing interest in investigating the association between pro-inflammatory cytokines and the pathogenesis of OCD. However, studies targeting interleukin-1β (IL-1β) and interleukin-6 (IL-6) in OCD are limited. Therefore, the present study aimed to explore the potential role of pro-inflammatory cytokines IL-1β and IL-6 in the pathophysiology and development of OCD. This study recruited 58 OCD patients and 30 age-sex-matched healthy controls (HCs). A qualified psychiatrist diagnosed OCD patients and assessed HCs based on the Diagnostic and Statistical Manual for Mental Health Disorders, 5th edition (DSM-5) criteria. We measured the severity of OCD using the Yale-Brown Obsessive Compulsive Scale (Y-BOCS). Serum IL-1β and IL-6 levels were measured using ELISA kits following the appropriate methods. The results showed that serum IL-1β levels were significantly elevated in OCD patients compared to HCs (23.68±1.65 pg/ml vs. 15.75±1.02 pg/ml; p = 0.002). Similarly, OCD patients exhibited significantly higher serum IL-6 levels than HCs (44.97±0.73 pg/ml vs. 37.04±0.35 pg/ml; p<0.001). We observed both cytokines were positively correlated with the Y-BOCS scores in OCD patients (IL-1β: r = 0.380, p = 0.015; IL-6: r = 0.324, p = 0.026) which indicates their role in disease pathophysiology. These results suggest that serum IL-1β and IL-6 levels may be associated with the pathophysiology of OCD. Also, these cytokines levels in blood samples can serve as early risk assessment tools for the development of OCD. We recommend further studies in a large and homogeneous population to support these findings.

Machine-learning-based models assist the prediction of pulmonary embolism in autoimmune diseases: A retrospective, multicenter study.

Pulmonary embolism (PE) is a severe and acute cardiovascular syndrome with high mortality among patients with autoimmune inflammatory rheumatic diseases (AIIRDs). Accurate prediction and timely intervention play a pivotal role in enhancing survival rates. However, there is a notable scarcity of practical early prediction and risk assessment systems of PE in patients with AIIRD. In the training cohort, 60 AIIRD with PE cases and 180 age-, gender-, and disease-matched AIIRD non-PE cases were identified from 7254 AIIRD cases in Tongji Hospital from 2014 to 2022. Univariable logistic regression (LR) and least absolute shrinkage and selection operator (LASSO) were used to select the clinical features for further training with machine learning (ML) methods, including random forest (RF), support vector machines (SVM), neural network (NN), logistic regression (LR), gradient boosted decision tree (GBDT), classification and regression trees (CART), and C5.0 models. The performances of these models were subsequently validated using a multicenter validation cohort. In the training cohort, 24 and 13 clinical features were selected by univariable LR and LASSO strategies, respectively. The five ML models (RF, SVM, NN, LR, and GBDT) showed promising performances, with an area under the receiver operating characteristic (ROC) curve (AUC) of 0.962-1.000 in the training cohort and 0.969-0.999 in the validation cohort. CART and C5.0 models achieved AUCs of 0.850 and 0.932, respectively, in the training cohort. Using D-dimer as a pre-screening index, the refined C5.0 model achieved an AUC exceeding 0.948 in the training cohort and an AUC above 0.925 in the validation cohort. These results markedly outperformed the use of D-dimer levels alone. ML-based models are proven to be precise for predicting the onset of PE in patients with AIIRD exhibiting clinical suspicion of PE. Chictr.org.cn : ChiCTR2200059599.

Development of a deep learning-based automatic detection model for gastroesophageal varices using transient elastography

Backgroundand Purpose: This study investigates the application of deep learning in the early detection and risk assessment of gastroesophageal varices (GOV), focusing on non-invasive methods using transient elastography. MethodsThe study employs DenseNet201 and an enhanced DenseNet201-with-SSMV (Spleen Stiffness Measurement Value) models. These models are trained and tested on a dataset comprising clinical and transient elastography data from patients. The methodology includes preprocessing of data, model training, and validation. To compare the performance of these models, metrics such as accuracy, sensitivity, specificity, and area under the ROC curve are utilized. Additionally, a thorough cross-validation process is implemented to ensure the robustness and generalizability of the models. The study aims to establish the superiority of the DenseNet201-with-SSMV model over the standard DenseNet201 by demonstrating enhanced predictive performance in detecting gastroesophageal varices. ResultsThe DenseNet201-with-SSMV model demonstrated superior performance with a training accuracy of 94.25%, sensitivity of 91.70%, specificity of 96.19%, and AUC of 0.988 (95% CI: 0.982–0.994). In testing, it achieved an accuracy of 90.15%, sensitivity of 84.03%, specificity of 94.76%, and AUC of 0.946 (95% CI: 0.919–0.973), outperforming the DenseNet201 model. The study demonstrates the superior performance of the DenseNet201-with-SSMV model over the standard DenseNet201, particularly in accuracy and predictive capabilities. ConclusionThe integration of clinical indicators with deep learning models offers a promising, non-invasive approach for GOV detection, with potential implications for liver disease management. The study also identifies the need for larger datasets and exploration of additional clinical indicators to enhance model performance.

Integrating Deep Learning with Electronic Health Records for Early Glaucoma Detection: A Multi-Dimensional Machine Learning Approach.

Recent advancements in deep learning have significantly impacted ophthalmology, especially in glaucoma, a leading cause of irreversible blindness worldwide. In this study, we developed a reliable predictive model for glaucoma detection using deep learning models based on clinical data, social and behavior risk factor, and demographic data from 1652 participants, split evenly between 826 control subjects and 826 glaucoma patients. We extracted structural data from control and glaucoma patients' electronic health records (EHR). Three distinct machine learning classifiers, the Random Forest and Gradient Boosting algorithms, as well as the Sequential model from the Keras library of TensorFlow, were employed to conduct predictive analyses across our dataset. Key performance metrics such as accuracy, F1 score, precision, recall, and the area under the receiver operating characteristics curve (AUC) were computed to both train and optimize these models. The Random Forest model achieved an accuracy of 67.5%, with a ROC AUC of 0.67, outperforming the Gradient Boosting and Sequential models, which registered accuracies of 66.3% and 64.5%, respectively. Our results highlighted key predictive factors such as intraocular pressure, family history, and body mass index, substantiating their roles in glaucoma risk assessment. This study demonstrates the potential of utilizing readily available clinical, lifestyle, and demographic data from EHRs for glaucoma detection through deep learning models. While our model, using EHR data alone, has a lower accuracy compared to those incorporating imaging data, it still offers a promising avenue for early glaucoma risk assessment in primary care settings. The observed disparities in model performance and feature significance show the importance of tailoring detection strategies to individual patient characteristics, potentially leading to more effective and personalized glaucoma screening and intervention.