Area Under The ROC Curve Research Articles

Combi-Elastography versus Transient Elastography for Assessing the Histological Severity of Metabolic Dysfunction-Associated Steatotic Liver Disease.

Combi-elastography is a B-mode ultrasound-based method in which two elastography modalities are utilized simultaneously to assess metabolic dysfunction-associated steatotic liver disease (MASLD). However, the performance of combi-elastography for diagnosing metabolic dysfunction-associated steatohepatitis (MASH) and determining fibrosis severity is unclear. This study compared the diagnostic performances of combi-elastography and vibration-controlled transient elastography (VCTE) for identifying hepatic steatosis, fibrosis, and high-risk MASH. Participants who underwent combi-elastography, VCTE, and liver biopsy were selected from a prospective cohort of patients with clinically suspected MASLD. Combi-elastography-related parameters were acquired, and their performances were evaluated using area under the receiver-operating characteristic curve (AUROC) analysis. A total of 212 participants were included. The diagnostic performance for hepatic steatosis of the attenuation coefficient adjusted by covariates from combi-elastography was comparable to that of the controlled attenuation parameter measured by VCTE (AUROC, 0.85 vs 0.85; p=0.925). The performance of the combi-elastography-derived fibrosis index adjusted by covariates for diagnosing significant fibrosis was comparable to that of liver stiffness measured by VCTE (AUROC, 0.77 vs 0.80; p=0.573). The activity index from combi-elastography adjusted by covariates was equivalent to the FibroScan-aspartate aminotransferase score in diagnosing high-risk MASH among participants with MASLD (AUROC, 0.72 vs 0.74; p=0.792). The performance of combi-elastography is similar to that of VCTE when evaluating histology of MASLD.

Electronic Health Record Phenotyping of Pediatric Suicide-Related Emergency Department Visits.

Suicide is a leading cause of death among young people. Accurate detection of self-injurious thoughts and behaviors (SITB) underpins equity in youth suicide prevention. To compare methods of detecting SITB using structured electronic health information and measure algorithmic performance across demographics. This cross-sectional study used medical records among youths aged 6 to 17 years with at least 1 mental health-related emergency department (ED) visit in 2017 to 2019 to an academic health system in Southern California serving 787 000 unique individuals each year. Analyses were conducted between January and September 2023. Multiexpert electronic health record review ascertained the presence of SITB using the Columbia Classification Algorithm of Suicide Assessment. Random forest classifiers with nested cross-validation were developed using (1) International Statistical Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) codes for nonfatal suicide attempt and self-harm and chief concern and (2) all available structured data, including diagnoses, medications, and laboratory tests. Detection performance was assessed overall and stratified by age group, sex, and race and ethnicity. The sample comprised 2702 unique youths with an MH-related ED visit (1384 youths who identified as female [51.2%]; 131 Asian [4.8%], 266 Black [9.8%], 719 Hispanic [26.6%], 1319 White [48.8%], and 233 other race [8.6%]; median [IQR] age, 14 [12-16] years), including 898 children and 1804 adolescents. Approximately half of visits were related to SITB (1286 visits [47.6%]). Sensitivity of SITB detection using only codes and chief concern varied by age group and increased until age 15 years (6-9 years: 59.3% [95% CI, 48.5%-69.5%]; 10-12 years: 69.0% [95% CI, 63.8%-73.9%]; 13-15 years: 88.4% [95% CI, 85.1%-91.2%]; 16-17 years: 83.1% [95% CI, 79.1%-86.6%]), while specificity remained constant. The area under the receiver operating characteristic curve (AUROC) was lower among preadolescents (0.841 [95% CI, 0.815-0.867]) and male (0.869 [95% CI, 0.848-0.890]), Black (0.859 [95% CI, 0.813-0.905]), and Hispanic (0.861 [95% CI, 0.831-0.891]) youths compared with adolescents (0.925 [95% CI, 0.912-0.938]), female youths (0.923 [95% CI, 0.909-0.937]), and youths of other races and ethnicities (eg, White: 0.901 [95% CI, 0.884-0.918]). Augmented classification (ie, using all available structured data) outperformed classification with codes and chief concern alone (AUROC, 0.975 [95% CI, 0.968-0.980] vs 0.894 [95% CI, 0.882-0.905]; P < .001). In this study, diagnostic codes and chief concern underestimated SITB prevalence, particularly among minoritized youths. These results suggest that priority on algorithmic fairness in suicide prevention strategies must extend to accurate detection of youths with suicide-related emergencies.

Pediatric Early Warning Scores Before Rapid Response Poorly Predict Intensive Care Unit Transfers.

The Pediatric Early Warning Score (PEWS) is a clinical tool used to identify children at risk for clinical deterioration, but its utility remains debated, particularly in high-resource settings. Our objective with this study was to assess the predictive performance of the PEWS for unplanned PICU transfers after pediatric rapid response team (RRT) activation. A retrospective cohort study at a tertiary care academic children's hospital included all hospitalized patients up to 21 years old who had RRT activations between August 2021 and July 2022. Demographic and clinical data, the primary reason for RRT activation, and the modified Brighton PEWS were collected. The primary outcome was PICU transfer following RRT activation, and the secondary outcome was rapid escalation of care within 4 hours after RRT activation. Sensitivity, specificity, and area under the receiver operating characteristic curve (AUROC) were calculated at multiple PEWS thresholds. Of 297 RRT activations in 244 patients, 183 (63%) resulted in a PICU transfer, 75% of which were due to respiratory concerns. The PEWS was recorded in 89% of RRT activations within the preceding 4 hours. There was no significant difference in the PEWS between patients with or without PICU transfer or rapid escalation of care. The sensitivity, specificity, and AUROC of PEWS for predicting PICU transfer or rapid escalation of care were low (AUROC 0.495-0.613). PEWS within 4 hours before RRT activation was a poor predictor of PICU transfer or rapid escalation care. Further work is needed to develop a more sensitive and specific tool.

Machine learning methods for diagnostic disease prediction and treatment decisions in parvovirus B19 positive viral cardiomyopathy

Abstract Background Parvovirus B19 (B19V) is the most common type of virus found in endomyocardial biopsies (EMB) from patients with inflammatory heart disease. However, the detection of genomic B19V-DNA appears to have little clinical significance and is often considered an incidental finding. Recent analyses of a large cohort of B19V-positive patients revealed that the detection of RNA transcripts rather than DNA is of clinical importance, as B19V-RNA-positive patients have a significantly worse cardiovascular outcome. Here, we applied machine learning (ML) methods to assess the impact of B19V transcriptional activity on cardiac inflammation and patient clinical outcome. Methods and results A total of 305 patients with a known clinical course and a positive endomyocardial B19V-DNA test were included in this study. Follow-up measurements of the B19V-specific transcripts VP1/2 and NS1 revealed significantly worse clinical courses than in patients without B19V-RNA detection (P=0.019, hazard ratio (95% CI)=1.62 (1.10-2.57)). LVEF at follow-up was significantly lower in patients with detected B19V-RNA (P=0.007), and adverse events such as worsening LV function (P=0.003) or stable systolic dysfunction (P=0.013) were more frequent in this group. Importantly, analysis of inflammatory markers in EMB showed no difference in patients with and without RNA detection (P=0.432), suggesting that the deterioration in clinical outcome was related to B19V transcriptional activity rather than an increase in inflammation. The analysis of area under the receiver operating characteristic curve (AUROC) of B19V-RNA-negative patients revealed that the macrophage marker MAC-1 (AUROC=0.66, P=0.002) was the single most prognostically relevant parameter, while the T cell marker CD3 (AUROC=0.73, P&amp;lt;0.001) was identified as the most prognostically relevant marker in B19V-RNA-positive patients. The linear combination of several inflammatory markers increased the AUROC value to 0.67 in B19V-RNA-negative patients and to 0.74 in B19V-RNA-positive patients. To further increase prognostic accuracy, we trained ML models on features derived from EMB inflammatory markers. For the prediction of major adverse cardiac events within 60 months, the best ML-models achieved an AUROC of 0.87 in B19V-RNA-positive and of 0.86 in B19V-RNA-negative patients, while models that neglect the differentiation of patients with regard to the presence of B19V RNA transcripts only achieved an AUROC of up to 0.76. Conclusions These results show that B19V transcriptional activity is an independent risk marker for adverse cardiac events in patients with viral cardiomyopathy and thus confirm the clinical-therapeutic importance of myocardial biopsy. Furthermore, we present an ML-based algorithm that accurately indicates disease progression in B19V-positive cardiomyopathy patients and is thus helpful for appropriate therapy recommendation.

Interethnic validation of an artificial intelligence algorithm for prediction of atrial fibrillation using sinus rhythm electrocardiogram

Abstract Background Previous research has demonstrated acceptable diagnostic accuracy of artificial intelligence (AI)-enabled sinus rhythm (SR) electrocardiogram (ECG) interpretation for identifying paroxysmal atrial fibrillation (AF). However, interethnic validations of the AI algorithms have not been widely conducted. Purpose We aimed to develop our own AI model for the identification of paroxysmal AF based on SR ECGs in the Korean population and to validate its diagnostic performance in Brazilian citizens. Methods We trained a Transformer-based vision network on 90% of a dataset comprising 808,194 ECGs from 121,282 patients at Seoul National University Bundang Hospital (2003-2020). The remaining 10% of the dataset was used for internal validation. The model was trained to compute a risk score for paroxysmal AF or new-onset AF within 2 years. External validation was conducted using non-AF ECGs from the CODE 15% dataset provided by the Telehealth Network of Minas Gerais, Brazil. Results In internal validation, our AI model achieved an Area Under the Receiver Operating Characteristic Curve (AUROC) of 0.911 (95% CI: 0.902 - 0.921), with a sensitivity of 78.1% and a specificity of 89.0%. Subgroup analyses showed an AUROC of 0.893 (95% CI: 0.876 - 0.909) for patients in routine health checkups or outpatient settings, and 0.854 for patients with "Normal ECG" interpretations. In external validation with the CODE 15% dataset, the AI model exhibited an AUROC of 0.884 (95% CI: 0.869-0.900), which increased to 0.906 (95% CI: 0.893-0.919) when adjusted for age and sex (Figure 1). In the subset of patients with "Normal ECG" interpretations, the AUROC was 0.826 (95% CI: 0.769-0.883), increasing to 0.861 (95% CI: 0.814-0.908) after applying the same adjustments. Conclusions Our AI-powered SR ECG interpretation model demonstrated excellent diagnostic performance in predicting paroxysmal AF, with valid performance in the Brazilian population as well. This suggests that the model has potential for broad application across different ethnic groups.Figure 1

National Early Warning Score 2 to Predict ICU Admission of COVID-19 Patients in Bangladeshi Cohort

Objective: To assess the predictive accuracy of the National Early Warning Score 2 (NEWS 2) for early prediction of clinical deterioration (ICU admission) of COVID-19- infected patients in Bangladesh. Methods: Data from 244 patients with COVID-19 infection admitted from July 2021- September 2021, to the Aichi Hospital Ltd, Bangladesh, were retrospectively reviewed. Consecutive adult patients with laboratory-confirmed COVID-19 initially admitted to non- ICU wards were included. Categorical and quantitative variables were expressed as numbers (percentages) and median (interquartile range, IQR). We evaluated the predictive performance of NEWS2 to predict ICU admission by comparing the area under the receiver operating characteristics curve (AUROC) at thresholds 5 and 7 and assessed its association with ICU admission by performing multivariate logistic analyses. STATA conducted all the statistical analyses. Results: Among the included 218 patients, 68 patients were transferred to ICU. The AUROC was 0.96 (Standard error 0.01, 95% confidence interval 0.93-0.98), revealing that NEWS2 at hospital admission was a good predictor of ICU admission. A NEWS2 threshold of 5 had higher sensitivity than a threshold of 7 (72.06% and 29.41%). A point of 7 also had high specificity (99.33% and 63.33%) and a high positive predictive value than a threshold of 5 (95.24% vs. 47.12%). The NEWS2 entries 5 and 7 were related to ICU admission found in multivariate logistic regression analysis. Conclusion: Our results suggest that NEWS2 at hospital admission is a good predictor for ICU admission of COVID-19 patients in Bangladesh. Screening patients at admission with this tool may identify patients at risk of clinical deterioration and help in better management. EWMCJ Vol. 12, No. 1&amp;2, January-July 2024: 60-65

AB-HFPEF score: A new and improved HFpEF prediction tool for the Indian patient

Abstract Background/Introduction Heart failure with preserved ejection fraction (HFpEF) remains underdiagnosed in low-middle income countries including India. The simple and validated H2FPEF score has largely replaced the diagnostic need for invasive right heart catheterization. The score has therapeutic implications by stratifying patients into low, intermediate and high-risk groups. The relevance of this score is yet to be ascertained in the Indian population. Purpose We hypothesized that the H2FPEF score may fall short in accurately predicting HFpEF in our population, thus warranting a new score. Methods We conducted a prospective observational study over one year at the out-patient department of a tertiary-care hospital in India. Patients presenting with clinical features of heart failure were screened and enrolled in the study. Demographic, clinical, laboratory and echocardiographic details were recorded by the principal investigator using an online data collection tool. Patients were stratified into ‘low risk’, ‘intermediate risk’ and ‘high risk’ probabilities, with respective cut-off values of 0-3, 4-6, and 7-9 using the point-based score, and &amp;lt;30%, 30-59%, and &amp;gt;60% via the continuous model of H2FPEF score. Pearson’s chi-square test identified significant risk factors that were purposed into a novel AB-HFPEF score. Cohen’s kappa correlation determined the degree of agreement in the risk stratifications of the H2FPEF and the AB-HFPEF score, using the continuous scoring model as the reference. Results Among the 189 patients screened, 137 were enrolled in the study. The baseline characteristics are recorded in Table 1. Pedal edema and uncontrolled hypertension were noteworthy risk factors. A novel AB-HFPEF score was subsequently designed and scored as follows: Age (&amp;gt; 60 years) - 1, BMI (&amp;gt; 30 kg/m2) - 2, uncontrolled Hypertension (&amp;gt;140/90 mmHg) - 1, Filling pressure (E/e' &amp;gt; 9) - 1, Pulmonary Artery Systolic Pressure (&amp;gt; 35 mmHg) - 1, pedal Edema - 1, atrial Fibrillation - 3. Receiver operating characteristics (ROC) curve analysis demonstrated an area under the ROC curve (AUC) of 0.85 (95%CI, 0.78 to 0.91) for ‘low risk’ strata and 0.87 (95%CI, 0.81 to 0.93) for ‘high risk’ strata. The optimal stratification was 0-2, 3-4, and 5-10 points for the ‘low risk’, ‘intermediate risk’, and ‘high risk’ respectively. There was a moderate agreement of the AB-HFPEF score with the reference score (κ = 0.48 (95% CI, 0.36 to 0.60), p &amp;lt; 0.001), while the point-based H2FPEF score had only a slight agreement (κ = 0.15 (95% CI, 0.05 to 0.25), p &amp;lt; 0.001), as illustrated in Figure 1. Conclusion(s) The novel AB-HFPEF score performs better risk stratification in Indian patients with HFpEF as compared to the point-based H2FPEF score. We hence propose the AB-HFPEF score as an instrument for tailored cardiology care in the Indian demographic, laying the cornerstone for larger validation and prognostication studies.Table 1:Baseline characteristicsFigure 1:Heat Map Illustration

Detection of ATTR cardiac amyloidosis using a novel artificial intelligence algorithm for wearable-adapted noisy single-lead electrocardiograms

Abstract Background Amyloid cardiomyopathy (ATTR-CM) is often underdiagnosed, with very few diagnosed before the onset of symptoms, leading to delayed use of disease-modifying therapies that could have substantial prognostic implications early in the disease course. In the US, the disease has a substantially higher prevalence among Black adults, who are further challenged by lower access to healthcare services and lower preventative care, further reducing the ability for clinician-led diagnosis. We sought to develop an approach for identifying ATTR-CM on noisy single-lead ECGs obtainable on portable and wearable ECG devices, allowing broad use in community screening. Purpose To develop a portable/wearable device-adapted AI-ECG algorithm capable of detecting ATTR-CM on noisy single-lead ECGs. Methods We identified patients with ATTR-CM from 2015 through June 2023 across 5 hospitals of a large U.S.-based hospital system using positive bone scintigraphy scans or pharmacotherapy with an approved transthyretin stabilizer. In the development cohort, we used lead I of 12 lead ECGs, commonly captured by portable and wearable ECG devices, to train a novel noise-adapted model explicitly designed to infer information against simulated real-world noises. For this, ECG signals were augmented using random Gaussian real-world noise within four frequency ranges at multiple signal-to-noise ratios. We tested the model in an independent set of cases and matched controls drawn from July through December 2023. Results The development cohort consisted of 1,011 ECGs from 234 patients (median age 79 years [IQR:70-85], 17% women), with 10:1 age- and sex-matched 10,110 controls with 10,110 ECGs (age 79 [IQR:70-85] years 17.7% female). In the independent test set of 139 ECGs from 47 patients (age 80 [75-86] years, 32% women) and matched controls (1390 ECGs and patients) from July through December 2023, the AUROC (area under the receiver operating characteristic curve) of the AI-ECG model for ATTR-CM was 0.90 [95%CI: 0.88-0.92] (A). The AUPRC was 0.44 [0.36-0.53]. The sensitivity and specificity of the model were 0.85 and 0.80. The positive predictive value ranged from 0.12 at a 3% prevalence level to 0.59 at a 25% prevalence (B). Conclusions A novel portable and wearable-devices adapted model demonstrates promising performance for detecting ATTR-CM on single-lead ECGs, representing a more accessible method for screening in community-dwelling adults.

Large language models to facilitate pregnancy prediction after invitro fertilization.

We evaluated the efficacy of large language models (LLMs), specifically, generative pre-trained transformer-4 (GPT-4), in predicting pregnancy following invitro fertilization (IVF) treatment and compared its accuracy with results from an original published study. Our findings revealed that GPT-4 can autonomously develop and refine advanced machine learning models for pregnancy prediction with minimal human intervention. The prediction accuracy was 0.79, and the area under the receiver operating characteristic curve (AUROC) was 0.89, exceeding or being at least equivalent to the metrics reported in the original study, that is, 0.78 for accuracy and 0.87 for AUROC. The results suggest that LLMs can facilitate data processing, optimize machine learning models in predicting IVF success rates, and provide data interpretation methods. This capacity can help bridge the knowledge gap between data scientists and medical personnel to solve the most pressing clinical challenges. However, more experiments on diverse and larger datasets are needed to validate and promote broader applications of LLMs in assisted reproduction.

Identifying heart failure dynamics using multi-point electrocardiograms and deep learning

Abstract Background Heart failure (HF) hospitalizations are related to poor survival outcomes, emphasizing the need for early detection and intervention. Recent studies have reported the potential of deep learning algorithms in identifying heart failure through electrocardiogram (ECG) with notable accuracy. However, evidence supporting their utility in the ongoing monitoring and follow-up of HF patients remains uncertain. Advanced diagnostic tools that can effectively track heart failure progression or improvement over time are needed. Purpose To address this challenge, we developed a deep learning model analyzing 12-lead ECG waveforms at two time points, aiming to detect HF deterioration or improvement, potentially revolutionizing HF management. Methods This study analyzed 32,045 ECGs from 6,892 unique adult patients (aged 18 years and older) visiting the Department of Cardiology at Kobe University Hospital. The participants were randomly assigned to training (4,284 patients), validation (1,034 patients), and test (1,034 patients) datasets. ElecTransformer was developed to classify HF status into deteriorated, improved, and no change classes based on ECG waveform signals at two different time points and to estimate brain natriuretic peptide (BNP) levels. Performance metrics, such as the area under the receiver operating characteristic curve (AUROC) and accuracy, were calculated, and attention mapping via gradient-weighted class activation mapping was utilized to interpret the model's decision-making ability. Results The patients had an average age of 65 years (±15.4 years) and BNP levels averaged 71.5 pg/mL (25.8 to 185.2 pg/mL). For HF status classification, ElecTransformer achieved an AUROC of 0.845 and an accuracy of 0.821, improving to an AUROC of 0.885 and an accuracy of 0.833 with baseline BNP integration. The ElecTransformer accurately estimated BNP levels from ECG data, evidenced by a correlation coefficient of 0.798. Averaged attention mapping showed that the QRS complex was the most important feature for determining model outputs with lesser effects seen for P waves and T waves variance Conclusions Harnessing a Transformer-based deep learning model to analyze multiple ECGs markedly improves the accuracy in detecting changes in heart failure status. This method offers a valuable non-invasive option for monitoring heart failure progression, potentially leading to better patient management and earlier intervention in heart failure treatment.Architecture and Performance of ModelsAveraged attention mapping

Machine learning approach for detection of valvular heart disease through digital cardiac auscultation: results from multi-center prospective cross-sectional study

Abstract Background Cardiac auscultation, an essential bedside diagnostic tool, often identifies significant valvular heart disease through murmurs. However, its accuracy is contingent upon the clinician's expertise, leading to a growing dependence on expensive imaging techniques. This study aimed to develop and validate a machine learning model for detecting valvular heart disease from cardiac auscultation. Methods The digital cardiac auscultation and associated echocardiograms were prospectively collected from three sites between May 1, 2022, and Dec 31, 2022. We trained, validated, and externally tested a machine learning based model to classify aortic stenosis, mitral stenosis, aortic regurgitation, mitral regurgitation, tricuspid regurgitation, oveall valvular heart disease from cardiac auscultation, age, sex, and comorbidity. We condcted training, and internal testing using datasets for eXtreme Gradient Boosting (XGBoost) machine learning model from hospital one and hospital two. The dataset from hospital three was employed for external testing. We evaluated the area under the receiver operating characteristic curve (AUROC), area under the precision recall curve (AUPRC), accuracy, sensitivity, specificity. Result We selected the cardiac auscultation and echocardiograms of a total of 4,321 patients, excluding those without valvular heart disease information, congenital heart disease, a history of valvular operation, and cardiac auscultation could not be interpreted. The diagnostic performance of heart murmurs for valvular heart disease across the three datasets generally demonstrated high sensitivity and PPV. Specifically, the sensitivity for aortic stenosis was 89.1%, mitral stenosis 88.8%, aortic regurgitation 88.5%, mitral regurgitation 89.8%, tricuspid regurgitation 90.1%, and for any valvular heart disease, it was 95.0%. However, the NPV was consistently lower, with values of 16.4%, 14.2%, 12.2%, 23.1%, 26.2%, and 65.3% for the respective conditions. In XGBoost machine learning model, the AUROC values for internal testing (N=1,403), along with their 95% confidence intervals, were as follows: aortic stenosis (0.93, 0.88–0.97), mitral stenosis (0.91, 0.81–0.97), aortic regurgitation (0.84, 0.79–0.90), mitral regurgitation (0.94, 0.90–0.98), tricuspid regurgitation (0.81, 0.75–0.87), and overall valvular heart disease (0.86, 0.82–0.89). For external testing (N=814), the AUROC values were: aortic stenosis (0.90, 0.83–0.96), mitral stenosis (0.88, 0.79–0.95), aortic regurgitation (0.94, 0.93–0.96), mitral regurgitation (0.93, 0.92–0.95), tricuspid regurgitation (0.89, 0.87–0.91), and overall valvular heart disease (0.92, 0.90–0.94). Conclusion The machine learning based model can accurately classify valvular heart diseases using information from digital cardiac auscultation.Figure 1.Study FlowFigure 2.AUC curve

Predicting physical fitness levels from resting ECG data: a machine learning approach in cardiovascular assessment

Abstract Background Exercise Stress Testing (EST), an inexpensive noninvasive modality, is widely used for preliminary cardiovascular evaluation and ongoing monitoring of cardiac patients. The MET (Metabolic Equivalent of Task) score from EST, indicating physical fitness, is associated with overall mortality risk and is extremely helpful for preoperative evaluation. Artificial Intelligence (AI) advances, particularly Deep Learning (DL) for resting ECG analysis, have significantly improved cardiac diagnostic accuracy. Objectives This study explores AI's ability to predict MET scores from resting ECGs, aiming to streamline fitness assessments and detect early changes in fitness levels. Methods The current analysis encompasses 45,518 consecutive EST evaluations from 31,532 individual patients from 2017 to 2023. MET scores were automatically computed using an established formula, while interpretations by certified cardiologists provided a benchmark for the AI model. Poor fitness was defined as being within the lowest quartile of MET scores, adjusted for sex and age. A deep learning framework was developed to process a 10-second snippet of resting phase ECG in raw waveform, predicting the likelihood of diminished fitness and thus acting as a binary classifier. The efficacy of this Poor Fitness AI model (PF-AI model) was assessed through metrics such as the area under the receiver operating characteristic curve (AUROC), along with sensitivity, specificity, and accuracy rates. Results The average age of the study group was 55±10 years, predominantly male (68.8%, n = 31,337). Among them, 1,152 were identified as having poor fitness. Applying the model to the entire cohort through 5-fold cross-validation yielded a sensitivity of 78%, a specificity of 81%, and an AUROC of 0.87 in predicting poor fitness. Conclusions Leveraging merely 10 seconds of resting ECG data, the PF-AI model proves to be a robust tool in predicting poor fitness levels, comparable to the outcomes of comprehensive stress testing.

DELINEATE-Regurgitation: deep learning for automated assessment of aortic, mitral, and tricuspid regurgitation from echocardiography

Abstract Background Valvular heart diseases, including aortic regurgitation (AR), mitral regurgitation (MR), and tricuspid regurgitation (TR) are significant contributors to cardiovascular morbidity and mortality worldwide. Artificial intelligence (AI) may improve echocardiographic assessment, but no models to date have successfully evaluated valvular regurgitation. Purpose The goal of this study was to develop an AI system to improve the accuracy of echocardiographic AR, MR, and TR interpretation. Methods Complete transthoracic echocardiogram (TTE) studies performed at a single center were split into train (n=65,301), validation (14,018), and test sets (13,746). The Deep Learning for Echo Analysis, Tracking, and Evaluation of Valvular Regurgitation (DELINEATE-Regurgitation) system intakes entire TTE studies, identifies color Doppler clips showing each valve, and yields a study level classification of AR, MR, and TR on a 6-grade scale (none/trace, mild, mild-moderate, moderate, moderate-severe, severe) using the cardiologist interpretation as a reference standard. This system used a hybrid neural network, leveraging the spatiotemporal feature learning capabilities of convolutional neural networks with the sequential processing strengths of transformer networks. The end-to-end hybrid approach allows the model to learn nuanced representations of the data including features within and across color Doppler videos. Model accuracy was assessed using quadratically weighted Cohen’s kappa (k), area under the receiver operator characteristic curve (AUROC) for the detection of both "moderate or greater" and "severe" regurgitation, and AI-cardiologist agreement the 6-grade scale. Additionally, an analysis was conducted using a 4-grade regurgitation scale (none/trace, mild, moderate, and severe) by rolling up intermediate classes (mild-moderate and moderate-severe) into the next higher class for both cardiologist determinations and model predictions. Results The DELINEATE-Regurgitation system demonstrated a high accuracy in the classification of AR (k=0.857), MR (k=0.867), and TR (k=0.847) (Figure 1). DELINEATE predicted the same severity as cardiologists (AR 91.8%, MR 79.9%, TR 77.7%) or within ±1 grade accuracy (AR 99.5%, MR 98.6%, TR 98.6%). The models had excellent binary discrimination in the detection of moderate or greater (AR AUROC 99.4, MR 98.6, TR 98.1) and severe (AR 99.7, MR 99.3, TR 99.6) regurgitation (Figure 2). In the 4-grade scale, the model had exact agreement with cardiologists in 92.3% of AR, 82.4% of MR, and 80.6% of TR cases. Panel adjudication of AI-cardiologist disagreement and correlation with quantitative metrics at higher levels of regurgitation will be presented. Conclusions The DELINEATE-Regurgitation AI system demonstrates a high level of accuracy in the assessment of AR, MR, and TR using color Doppler TTE videos. Future work will focus on automating quantification and developing strategies for optimal clinical deployment.Figure 1 - Confusion MatricesFigure 2 - ROC Curves

Artificial intelligence applied to electrocardiographic images for scalable screening of cardiac amyloidosis

Abstract Background Transthyretin amyloid cardiomyopathy (ATTR-CM) remains largely under-recognized, under-diagnosed, and under-treated. We hypothesized that the myocardial remodeling seen in ATTR-CM may be detectable through artificial intelligence (AI) applied to 12-lead electrocardiographic (ECG) images. Purpose To develop an AI-ECG algorithm that can detect ATTR-CM from images of 12-lead electrocardiograms. Methods Across 5 hospitals of a large U.S.-based hospital system, we identified patients with ATTR-CM, defined by the presence of a positive bone scintigraphy scan or pharmacotherapy with an approved transthyretin stabilizer between 2015 and the first half of 2023. The development cohort consisted of 1,011 ECGs from 234 patients (age 79 [IQR:70-85], n=176 (17.4%) women), who were age- and sex-matched in a 10:1 ratio to 10,110 ECGs from 10,110 controls (age 79 [IQR:70-85] years, n=1,800 [17.7%] female). A convolutional neural network (CNN) pre-trained using a bio-contrastive pretext on ECGs before 2015 was fine-tuned for ATTR-CM using 5-fold cross-validation and subsequently tested in an independent set of cases (139 ECGs in 47 patients; age 80 [75-86] years, n=44 (31.7% women)) and matched controls (1390 ECGs and patients) from the second half of 2023. Results The AUROC (area under the receiver operating characteristic curve) of the AI-ECG model for discriminating ATTR-CM in the cross-validation set was 0.90 [95%CI: 0.89-0.91], and in the leave-out, temporally distinct dataset was 0.915 [95%CI: 0.89-0.94] (A), with a sensitivity of 0.85 [95%CI: 0.79-0.91] and specificity 0.80 [95%CI 0.78-0.82]. Representative saliency maps are shown in panel B. In simulations, the positive predictive value ranged from 0.12 [95%CI: 0.10-0.13] at a 3% prevalence level to 0.59 [95%CI: 0.56-0.62] at 25% prevalence. Conclusions We demonstrate that AI applied directly to ECG images represents a promising approach for the screening of ATTR-CM.

On the interpretability of the SVM model for predicting infant mortality in Bangladesh

BackgroundAlthough machine learning (ML) models are well-liked for their outperformance in prediction, greatly avoided due to the lack of intuition and explanation of their predictions. Interpretable ML is, therefore, an emerging research field that combines the performance and interpretability of ML models to create comprehensive solutions for complex decision-making analysis. Conversely, infant mortality is a global public health concern affecting health, social well-being, socio-economic development, and healthcare services. The study employs advanced interpretable ML techniques to anticipate and understand the factors affecting infant mortality in Bangladesh, overcoming the shortcomings of the conventional logistic regression (LR) model.MethodsBy utilizing the global surrogate model and local individual conditional expectation (ICE) interpretability technique, the interpretable support vector machine (SVM) has been used in this study to reveal significant characteristics of infant mortality using data from the Bangladesh Demographic and Health Survey (BDHS) 2017–18. To investigate intricate decision-making analysis of infant mortality, we adapted SVM and LR techniques with the hyperparameter tuning parameters. These models’ performances were initially assessed using the receiver operating characteristics (ROC) curve, run-time, and confusion matrix parameters with 100 permutations. Afterward, the SVM model’s model-agnostic explanation and the LR model’s interpretation were compared to enhance advanced comprehension for further insights.ResultsThe results of the 100 permutations demonstrated that the LR model (Average: accuracy = 0.9105, precision = NaN, sensitivity = 0, specificity = 1, F1-score = 0, area under the ROC curve (AUC) = 0.6780, run-time = 0.0832) outperformed the SVM model (Average: accuracy = 0.8470, precision = 0.1062, sensitivity = 0.0949, specificity = 0.9209, F1-score = 0.1000, AUC = 0.5632, run-time = 0.0254) in predicting infant mortality, but the LR model had a slower run-time and it was unable to predict any positive cases. The interpretation of LR analysis revealed that infant mortality rates decrease when mothers give birth after over two years, with higher educational attainment, overweight or obese mothers, working mothers, and families with polluted cooking fuel having lower rates. The local ICE interpretability technique, which depicts individual influences on the average likelihood of dying within the first birthday, explores the interpretable SVM model that mothers with normal BMIs, giving birth within two years, using less polluted cooking fuel, working mothers, and having male infant were more likely to experience infant death. The interpretable SVM model based on the global surrogate model also reveals that working mothers who used polluted cooking fuel at home and working women who used less polluted cooking fuel but had a longer period between pregnancies than two years would have higher infant death rates. Even among non-working mothers who used polluted cooking fuel and gave birth within two years of the preceding one, infant death rates were higher.ConclusionsThe interpretable SVM model reveals global interpretations help clinicians understand the entire conditional distribution, while local interpretations focus on specific instances, providing different insights into model behavior. Interpretable ML models aid policymakers, stakeholders, and families in understanding and preventing infant deaths by improving policy-making strategies and establishing effective family counseling services.

Association between inflammatory burden index and coronary slow flow phenomenon in patients with chest pain and no obstructive coronary arteries

BackgroundThe inflammatory burden index (IBI), a novel inflammation-based indicator, to is associated with the presence and prognosis of various diseases. However, few studies have focused on exploring the relationship between IBI and the coronary slow flow phenomenon (CSFP). In this study, we aimed to investigate the predictive value of IBI for CSFP in patients with chest pain and no obstructive coronary artery disease.MethodsA total of 1126 individuals with chest pain and no obstructive coronary arteries were consecutively included in this study. 71 patients developed CSFP were included in the CSFP group. A 1:2 age- and sex-matched patient with normal blood flow and angiographically proven normal coronary arteries was selected as the control group (n = 142). Plasma C-reactive protein (CRP), neutrophil, and lymphocyte counts were measured to determine the value of IBI.ResultsThe IBI were significantly higher in the CSFP group than in the controls (21.1 ± 6.5 vs. 14.5 ± 6.4, P < 0.001). The IBI increasedelevated with the increase of the numbers of vessels affected by CSFP. Multivariate logistic regression analysis revealed that IBI and body mass index (BMI) were independent predictors of CSFP. Receiver operating characteristic (ROC) curve analysis showed that when IBI was > 15.74, the sensitivity and specificity were 77.5% and 67.6%, respectively, and the area under the ROC curve (AUC) was 0.799 (95% CI: 0.737-0.862, P<0.001).ConclusionThe IBI may be an independent predictor of CSFP in patients with chest pain and normal coronary arteries. The IBI could improve the predictive value of CSFP compared with the indicators alone.

Proteomic and serological markers for diagnosing cardia gastric cancer and precursor lesions in a Chinese population

Cardia gastric cancer (CGC) is prevalent in East Asia, and noninvasive, cost-effective screening methods are needed. This study investigated the diagnostic value of serum pepsinogen (PG), gastrin-17 (G-17), Helicobacter pylori (H. pylori) antibodies, and proteomic profiling for CGC and precancerous lesions. We conducted a case-control study involving biopsy-confirmed patients with CGC (n = 60), low-grade intraepithelial neoplasia (CLGD, n = 60), high-grade intraepithelial neoplasia (CHGD, n = 64), and healthy controls (n = 120) matched for age and sex from high-incidence areas in China. Serological markers including PGI, PGII, G-17, and H. pylori were measured using ELISA and Western blot, while plasma protein markers were assessed using Olink® technology. The VSOLassoBag algorithm and nine machine learning (ML) algorithms were employed to identify crucial features and construct predictive models. Various evaluation metrics, including the area under the receiver-operating-characteristic curve (AUC), were utilized to compare predictive performance. Elevated PGII levels, decreased PGR, and H. pylori infection were significantly associated with an increased risk of CGC and precancerous lesions (P for trend < 0.05). The eXtreme Gradient Boosting (XGBoost) model performed best in discriminative ability among the 9 ML models. Following feature reduction based on predictive performance, a final explainable XGBoost model was developed, incorporating five protein biomarkers (CDHR2, ICAM4, PTPRM, CDC27, and FLT1). This model exhibited excellent performance in distinguishing individuals with CGC and precancerous lesions from healthy controls (AUC = 0.931 for CGC, 0.867 for CHGD, and 0.763 for CLGD), surpassing the traditional serological marker-based model. This study underscores the diagnostic potential of serological markers and proteomic profiling in the detection of CGC. Further validation and exploration of combined biomarker approaches are warranted to enhance early diagnosis and improve outcomes in high-risk populations.

Impact of nutrition-related laboratory tests on mortality of patients who are critically ill using artificial intelligence: A focus on trace elements, vitamins, and cholesterol.

This study aimed to understand the collective impact of trace elements, vitamins, cholesterol, and prealbumin on patient outcomes in the intensive care unit (ICU) using an advanced artificial intelligence (AI) model for mortality prediction. Data from ICU patients (December 2016 to December 2021), including serum levels of trace elements, vitamins, cholesterol, and prealbumin, were retrospectively analyzed using AI models. Models employed included category boosting (CatBoost), extreme gradient boosting (XGBoost), light gradient boosting machine (LGBM), and multilayer perceptron (MLP). Performance was evaluated using area under the receiver operating characteristic curve (AUROC), accuracy, precision, recall, and F1-score. The performance was evaluated using 10-fold crossvalidation. The SHapley Additive exPlanations (SHAP) method provided interpretability. CatBoost emerged as the top-performing individual AI model with an AUROC of 0.756, closely followed by LGBM, MLP, and XGBoost. Furthermore, the ensemble model combining these four models achieved the highest AUROC of 0.776 and more balanced metrics, outperforming all models. SHAP analysis indicated significant influences of prealbumin, Acute Physiology and Chronic Health Evaluation II score, and age on predictions. Notably, the ratios of selenium to age and low-density lipoprotein to total cholesterol also had a notable impact on the models' output. The study underscores the critical role of nutrition-related parameters in ICU patient outcomes. Advanced AI models, particularly in an ensemble approach, demonstrated improved predictive accuracy. SHAP analysis offered insights into specific factors influencing patient survival, highlighting the need for broader consideration of these biomarkers in critical care management.

Deep learning-based super-resolution and denoising algorithm improves reliability of dynamic contrast-enhanced MRI in diffuse glioma

Dynamic contrast-enhanced MRI (DCE-MRI) is increasingly used to non-invasively image blood-brain barrier leakage, yet its clinical utility has been hampered by issues such as noise and partial volume artifacts. In this retrospective study involving 306 adult patients with diffuse glioma, we applied deep learning-based super-resolution and denoising (DLSD) techniques to enhance the signal-to-noise ratio (SNR) and resolution of DCE-MRI. Quantitative analysis comparing standard DCE-MRI (std-DCE) and DL-enhanced DCE-MRI (DL-DCE) revealed that DL-DCE achieved significantly higher SNR and contrast-to-noise ratio (CNR) compared to std-DCE (SNR, 52.09 vs 27.21; CNR, 9.40 vs 4.71; P < 0.001 for all). Diagnostic performance assessed by the area under the receiver operating characteristic curve (AUROC) showed improved differentiation of WHO grades based on a pharmacokinetic parameter \\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\hbox {V}_{{e}}$$\\end{document} (AUC, 0.88 vs 0.83, P = 0.02), while remaining comparable to std-DCE in other parameters. Analysis of arterial input function (AIF) reliability demonstrated that \\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\hbox {AIF}_{{DL}}$$\\end{document} exhibited superior agreement compared to \\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\hbox {AIF}_{{std}}$$\\end{document}, as indicated by mostly higher intraclass correlation coefficients (Time to peak, 0.79 vs 0.43, P < 0.001). In conclusion, DLSD significantly enhances both the image quality and reliability of DCE-MRI in patients with diffuse glioma, while maintaining or improving diagnostic performance.

Predicting neonatal mortality prior to discharge from hospital in prenatally diagnosed left congenital diaphragmatic hernia.

To evaluate the association of standardized prenatal imaging parameters and immediate neonatal variables with mortality prior to discharge in infants with isolated left congenital diaphragmatic hernia (LCDH), and to compare the performance of ultrasound- and magnetic resonance imaging (MRI)-based severity grading for the prediction of neonatal mortality. This was a retrospective study of infants with prenatally diagnosed isolated LCDH referred to a single tertiary center between 2008 and 2020. Fetuses with right or bilateral congenital diaphragmatic hernia, additional major structural anomaly or known genetic condition, as well as cases that underwent fetal intervention or declined postnatal intervention, were excluded. Ultrasound and MRI images were reviewed retrospectively. Univariable and multivariable analyses were performed, incorporating prenatal and immediate neonatal factors to analyze the association with neonatal mortality prior to discharge, and a prediction calculator was generated. The performance of ultrasound and that of MRI for the prediction of neonatal mortality were compared. Of 253 pregnancies with fetal CDH, 104 met the inclusion criteria, of whom 77 (74%) neonates survived to discharge. Seventy-five fetuses underwent both prenatal ultrasound and MRI. On multivariable analysis, observed/expected (o/e) lung-to-head ratio and o/e total fetal lung volume were associated independently with neonatal death (adjusted odds ratio, 0.89 (95% CI, 0.83-0.95) and 0.90 (95% CI, 0.84-0.97), respectively), whereas liver position was not. There was no significant difference in predictive performance between using ultrasound and MRI together (area under the receiver-operating-characteristics curve (AUC), 0.85 (95% CI, 0.76-0.93)) compared with using ultrasound alone (AUC, 0.81 (95% CI, 0.72-0.90); P = 0.19). The addition of neonatal parameters (gestational age at birth and small-for-gestational age) did not improve model performance (AUC, 0.87 (95% CI, 0.80-0.95)) compared with the combined ultrasound and MRI model (P = 0.22). There was poor agreement between severity assessment on ultrasound and MRI (Cohen's κ, 0.19). Most discrepancies were seen among cases deemed to be non-severe on ultrasound and severe on MRI, and outcomes were more consistent with MRI-based prognostication. In fetuses with prenatally diagnosed isolated LCDH, mortality prediction using standardized ultrasound and MRI measurements performed reasonably well. In cases classified as non-severe on ultrasound, MRI is recommended, as it may provide more accurate prognostication and assist in the determination of candidacy for fetal intervention. © 2024 The Author(s). Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.