Area Under ROC Curve Research Articles

Prevalence and predictors of help-seeking behavior among post-partum women with urinary incontinence in China and Indonesia: A cross-sectional survey based on Andersen Help-Seeking Model

IntroductionUrinary incontinence (UI) is a prevalent condition among post-partum women, and it can significantly affect their physical and psychosocial well-being. Therefore, it's crucial for post-partum women to discuss any UI symptoms they may be experiencing with their healthcare provider and seek appropriate treatment. PurposeTo investigate what leads post-partum women with UI to seek help in China and Indonesia based on the Andersen Help-Seeking Model. MethodsA cross-sectional digital survey among post-partum women (6 weeks to 1 year) was conducted from May to November 2021 in China and Indonesia. The survey contents included: (1) demographic characteristics, (2) social support, (3) health care needs, (4) capacity and resources. For analysis, descriptive statistics, independent sample t-tests and chi-square tests were used to determine the differences between help-seeking and non-help-seeking women with UI, and logistic regression analysis and the receiver operating characteristic curve (ROC) was used to determine the predictors of heal-seeking behavior in post-partum women with UI. ResultsThe prevalence of UI was 25 % (215/868) and 31 % (187/605) among post-partum women in China and Indonesia. Among post-partum women with UI, the help-seeking rate was 46 % (98/215) and 52 % (98/187) in China and Indonesia. Incontinence quality of Life, support from women of the family (Yes), knowing the available department for UI (Yes), current knowledge of UI and the current status of UI were independent risk factors for China (P<0.05). Support from husband (Yes), being asked about UI by a doctor (Yes) and the current status of UI were independent risk factors for Indonesia. The obtained area under ROC curve (AUC) for the model were 0.884 and 0.935 in China and Indonesia. ConclusionsThe prevalence of UI and the rate of seeking help for UI in Indonesia were higher than those in China. Social support, needs and the availability of resources and organizational support to assist patients in their help-seeking efforts, were the main predictors affecting help-seeking behavior among post-partum women with UI.

Diagnostic value of three-dimensional neuroretinal rim thickness for differentiation of superior segmental optic nerve hypoplasia

Little is known about the diagnostic utility of three-dimensional neuroretinal rim thickness (3D-NRRT) for differentiating patients with superior segmental optic nerve hypoplasia (SSOH) from normal-tension glaucoma (NTG). Since SSOH is defined by characteristic optic nerve head features, investigation of diagnostic usefulness of 3D-NRRT is necessary. In this cross-sectional study, 49 SSOH eyes, 52 NTG eyes, and 41 normal eyes were enrolled. Retinal nerve fiber layer thickness (RNFLT) and 3D-NRRT values, as obtained in the right-eye orientation by optical coherence tomography (OCT), were recorded. On RNFLT clock-hour comparison, the 11–3 clock-hour sectors were significantly thinner for SSOH than for NTG (all P < 0.01). As for 3D-NRRT, whereas the 1 and 2 sectors were significantly thinner for SSOH (P < 0.001, P = 0.004), the 6–11 sectors were significantly thinner for NTG (all P < 0.01). The area under receiver operating characteristic (AUROC) curves of the superior and nasal quadrants of RNFLT (0.838, 0.729) were significantly greater than those of 3D-NRRT (0.518, 0.588; P < 0.001, P = 0.043). However, the AUROCs of the inferior and temporal quadrants were significantly greater for 3D-NRRT (0.728, 0.760) than for RNFLT (0.527, 0.550; P = 0.008, P = 0.019). The appropriate use of 3D-NRRT can be useful in differentiating SSOH from NTG.

Development and validation of a combined hypoxia- and metabolism-related prognostic signature to predict clinical prognosis and immunotherapy responses in clear cell renal cell carcinoma.

Hypoxia and metabolism are closely correlated with the progression of cancer. We aimed to construct a combined hypoxia- and metabolism-related genes (HMRGs) prognostic signature to predict survival and immunotherapy responses in patients with clear cell renal cell carcinoma (ccRCC). The RNA-seq profiles and clinical data of ccRCC were acquired from the TCGA and the ArrayExpress (E-MTAB-1980) databases. Least absolute shrinkage and selection operator (LASSO) and univariate and multivariate Cox regression analyses were applied to establish a prognostic signature. The E-MTAB-1980 cohort was selected for validation. The effectiveness and reliability of the signature were further evaluated by Kaplan-Meier (K-M) survival and time-dependent receiver operating characteristic (ROC) curves. Further analyses, including functional enrichment, ssGSEA algorithm, CIBERSORT algorithm, and expression of immune checkpoints, were explored to investigate immune status and immunotherapy responses. We constructed a prognostic eight-gene signature with IRF6, TEK, PLCB2, ABCB1, TGFA, COL4A5, PLOD2, and TUBB6. Patients were divided into high-risk and low-risk groups based on the medium-risk score. The K-M analysis revealed that patients in the high-risk group had an apparently poor prognosis compared to those in the low-risk group in the TCGA (p < 0.001) and E-MTAB-1980 (p < 0.005). The area under ROC curve (AUC) of the prognostic signature was 0.8 at 1 year, 0.77 at 3 years, and 0.78 at 5 years in the TCGA, respectively, and was 0.82 at 1 year, 0.74 at 3 years, and 0.75 at 5 years in the E-MTAB-1980, respectively. Independent prognostic analysis confirmed the risk score as a separate prognostic factor in ccRCC patients (p < 0.001). The results of ssGSEA showed not only a high degree of immune cell infiltration but also high scores of immune-related functions in the high-risk group. The CIBERSORT analysis further confirmed that the abundance of immune cells was apparently different between the two risk groups. The risk score was significantly correlated with the expression of cytotoxic T lymphocyte-associated antigen-4 (CTLA4), lymphocyte-activation gene 3 (LAG3), and programmed cell death protein 1 (PD-1). The HMRGs signature could be used to predict clinical prognosis, evaluate the efficacy of immunotherapy, and guide personalized immunotherapy in ccRCC patients.

Comparison of historical criterion and artificial intelligence in patients with left ventricular hypertrophy

Abstract Backgrounds Diagnostic criterion of left ventricular hypertrophy (LVH) on 12-leads electrocardiogram (ECG) were historically established. However, there was no study to evaluate the criterion comparing with machine learning methods. Purpose To verify the historical criterion of LVH on ECG, and to compare them with diagnosis by machine learning. Methods First, consecutive 60 patients with LVH were recruited, and one to one matching with age and sex to patients with normal cardiac function was performed. Finally, 120 patients (69.6 ± 12.6years, 38men per group) were enrolled. LVH was defined as at least one LV wall (septum, posterior wall, apex) showed thickness over 15mm on ultrasound echocardiography. No sinus rhythm, and wide QRS cases were excluded. The accuracy of historical criterion and ECG predictors were calculated by an assessment of whether a predictor was higher/lower than the cut-off value of receivor operating characteristics curve analysis (ROC). Ten machine learning methods were built using the significant predictors of logistic regression with 10-times cross validation. Results By logistic regression analysis, 77 significant predictors were extracted, and 22 predictors showed their area under ROC (AUROC) over 0.700. Among historical criterion, Cornell voltage showed high accuracy (0.783) and AUROC (0.808). Conversely, among AI methods, light gradient boosting machine demonstrated higher accuracy (0.843) and random forest method higher AUROC (0.882). Shapley additive explanation method (SHAP) demonstrated that V2/V2 S-wave amplitude and I/V5 T-wave amplitude played essential roles to build the AI models. Conclusions AI diagnosis on ECG for LVH showed powerful diagnostic performance comparing historical criterion.

Diagnostic performance of machine learning on 12-lead electrocardiogram for predicting multi-vessel coronary vasospastic angina

Abstract Background Multi-vessel coronary vasospastic angina (multi-VSA) is life-threatening disease. However, 12-leads electrocardiogram (ECG) in no chest symptom period was considered to have no predictive value for the multi-VSA. We tried to predict multi-VSA by machine learning (ML) on parameters of the 12-lead electrocardiogram. Purpose To predict multi-VSA by machine learning on ECG parameters in no chest pain period. Methods We recruited 227 consecutive sinus-rhythm patients (63.6±12.9years, 136men) who underwent acetylcholine-provocation test in coronary angiography (CAG). Multi-VSA was defined as spasm in at least 2 major branches which was provoked by ACh. ECG was recorded before CAG in no chest pain period. ML was performed on table data of ECG parameters (ex. ST level ant J point, R wave amplitude, and/or J wave in each lead) using several ensemble learning methods. Results 79 patients (35%) showed multi-VSA, and the breakdown of the multi-VSA was as follows: double vessel 44, triple vessel 35, including LAD 73, LCx 47, RCA73. Univariate logistic regression analysis extracted 23 significant but weak predictors, the highest area under receiver operating characteristics curve (AUROC) was 0.673 of Q wave amplitude in lead aVL. Conversely, ML demonstrated high diagnostic performance (AUROC of extra trees classifier: 0.817). Shapley additive explanation method in the extra trees showed male, QTc, J wave in lead II, and low amplitude of Q wave in lead I/aVL played essential roles to build the ML model. Conclusion Several parameters of 12-lead ECG in multi-VSA patients contains potential features of VSA, and their aggregation and ensemble learning can predict VSA with high diagnostic performance.

Thrombospondin 2 is a novel biomarker of essential hypertension and associated with nocturnal Na+ excretion and insulin resistance

ABSTRACT Background Thrombospondins (TSPs) play important roles in several cardiovascular diseases. However, the association between circulating (plasma) thrombospondin 2 (TSP2) and essential hypertension remains unclear. The present study was aimed to investigate the association of circulating TSP2 with blood pressure and nocturnal urine Na+ excretion and evaluate the predictive value of circulating TSP2 in subjects with hypertension. Methods and Results 603 newly diagnosed essential hypertensive subjects and 508 healthy subjects were preliminarily screened, 47 healthy subjects and 40 newly diagnosed essential hypertensive subjects without any chronic diseases were recruited. The results showed that the levels of circulating TSP2 were elevated in essential hypertensive subjects. The levels of TSP2 positively associated with systolic blood pressure (SBP), diastolic blood pressure (DBP), and other clinical parameters, including homeostasis model assessment of insulin resistance (HOMA-IR), brachial-ankle pulse wave velocity, and serum triglycerides, but negatively associated with nocturnal urine Na+ concentration and excretion and high-density lipoprotein cholesterol. Results of multiple linear regressions showed that HOMA-IR and nocturnal Na+ excretion were independent factors related to circulating TSP2. Mantel–Haenszel chi-square test displayed linear relationships between TSP2 and SBP (χ2 = 35.737) and DBP (χ2 = 26.652). The area under receiver operating characteristic curve (AUROC) of hypertension prediction was 0.901. Conclusion Our study suggests for the first time that the circulating levels of TSP2 may be a novel potential biomarker for essential hypertension. The association between TSP2 and blood pressure may be, at least in part, related to the regulation of renal Na+ excretion, insulin resistance, and/or endothelial function.

Abstract 12177: Quantitative Flow Ratio is a Great Predictor of Microvascular Obstruction in ST-Segment Elevation Myocardial Infarction After Percutaneous Coronary Intervention

Backgrounds: Compare the predictive effects of ST-segment resolution (STR) and quantitative flow ratio (QFR) in the detection of MVO in STEMI patients after PCI. Research Questions: To investigate the efficacy of STR/QFR in the evaluation of MVO detected by CMR in STEMI after PCI. Methods: A prospective cohort study enrolled 208 consecutive patients with STEMI. The degree of STR post procedure and predischarge after reperfusion was calculated. QFR were measured according to the angiography, CMR was performed within median time 5 days after revascularization to determine MVO. Clinical factors were enrolled to predict the incidence of MVO. Results: The MVO/LV mass is 1.42±2.28 in 126 patients with MVO according to CMR, and the area under ROC curves (AUC) identified QFR is the best predictor (AUC: 0.784, P &lt;0.001). Factors including QFR, STR, TIMI flow, CK-MB, cTnI, WBC, and HsCRP showed good predictive effects for the incidence of MVO ( P &lt;0.05). Except for peak cTnI, QFR (%) was the best predictor (AUC: 0.784, P &lt;0.001) ( Figure 1 ). A nomogram for predicting MVO was done (c-index: 0.894) ( Figure 2 ). A significant correlation was observed between STR and various parameters ( P &lt;0.05), except for the correlation between predischarge STR and MVO/LV mass at 6 months. Predischarge STR has the best correlation with infarct size after the 6-month follow-up ( r =-0.509, P &lt;0.001) ( Figure 3 ). Regression analysis indicated that postoperative STR (β=-0.248, P =0.007) was a significant predictor of MVO/LV mass ( Figure 3 ). Figure 4 further analyzed heart function from in-hospital CMR between different postoperative STR levels, and from 6 months CMR between predischarge STR subgroups. Higher STR predicted less cardiac injury and more ideal index of cardiac remodeling. Conclusions: In STEMI patients after PPCI, QFR demonstrated better predictive effect of the incidence of MVO than STR. Predischarge STR is closely related to infarct size and ejection fraction after 6-month follow-up.

Abstract 13412: Diagnostic Performance of Machine Learning on 12-Lead Electrocardiogram for Predicting Multi-Vessel Coronary Vasospastic Angina

Introduction: Multi-vessel coronary vasospastic angina (multi-VSA) is life-threatening disease. We tried to predict multi-VSA by machine learning (ML) on 12-lead electrocardiogram (ECG). Hypothesis: Machine learning on 12-lead ECG has powerful diagnostic value for multi-VSA. Methods: We recruited 227 consecutive sinus-rhythm patients (63.6±12.9years, 136men) who underwent acetylcholine-provocation test in coronary angiography (CAG). Multi-VSA was defined as spasm in at least 2 major branches. ECG was recorded before CAG in no chest pain period. ML was performed on table data of ECG parameters using several ensemble learning methods. Results: 79 patients (35%) showed multi-VSA, and univariate logistic regression analysis extracted 23 significant but weak predictors, the highest area under receiver operating characteristics curve (AUROC) was 0.673. Conversely, ML demonstrated high diagnostic performance (AUROC of extra trees classifier: 0.817). Shapley additive explanation method showed male, QTc, J wave in lead II, and low amplitude of Q wave in lead I/aVL played essential roles to build the ML model. Conclusion: Several parameters of 12-lead ECG in multi-VSA patients contains potential features of VSA, and their aggregation and ensemble learning can predict VSA with high diagnostic performance.

Abstract 16426: A Risk Model to Predict Atrial Fibrillation in Diabetes Using Machine Learning: The ACCORD Study

Introduction: Atrial fibrillation (AF) is the most common cardiac arrhythmia seen in clinical practice, which necessitates identifying factors that can precipitate this dysrhythmia. We utilized the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial to build an AF risk prediction model specifically in the type 2 diabetes (T2DM) population, using machine learning (ML) approaches, and comparing to the established CHARGE-AF risk model. Methods: We utilized a patient cohort from the ACCORD clinical trial without prior AF. Baseline characteristics were collected as potential risk factors and incident AF was defined from ECG at follow-up. A random forest (RF) binary classifier was trained with 27 variables, including age, race, sex, smoking, waist circumference, BMI, blood pressure (BP), history of heart disease, medications, cholesterol, renal function, and left ventricular hypertrophy by ECG. All ACCORD subjects had T2DM, therefore this variable was not included. The CHARGE-AF model consisted of a Cox proportional hazard model trained on the following variables: age, race, height, weight, systolic/diastolic BP, current smoking, heart failure, prior MI, and BP medication. Area under ROC curve (AUC) was measured with randomized, stratified, 5-fold cross validation with 80/20 splits. Imputation and resampling methods were used for missing data and class imbalance. Results: Out of 9307 patients in the ACCORD cohort, 175 developed AF during the study period of 6.26 years, with a mean time to AF event of 3.68 years. The RF model, with an AUC of 0.912, showed improved prediction of AF over the Cox model, which had an AUC of 0.784. Permutation feature importance revealed age, race, waist circumference, weekly alcohol consumption, sex, and GFR as the most predictive features of the RF model (Fig 1). Conclusions: The CHARGE-AF risk model performed modestly on the ACCORD cohort, while a RF model showed superior performance.

Diagnostic accuracy of ePOS score in predicting DNR labeling after ICU admission: A prospective observational study (ePOS-DNR)

BackgroundResuscitation can sometimes be futile and making a do-not-resuscitate (DNR) decision is in the best interest of the patient. The electronic poor outcome screening (ePOS) score was developed to predict 6-month poor outcomes of critically ill patients. We explored the diagnostic accuracy of the ePOS score in predicting DNR decisions in the intensive care unit (ICU). MethodsThis study was conducted at the ICU of a tertiary referral hospital in Saudi Arabia between March and May 2023. Prospectively, we calculated ePOS scores for all eligible consecutive admissions after 48 h in the ICU and recorded the DNR orders. The ability of the score to predict DNR was explored using logistic regression. Youden's ideal cut-off value was calculated using the DeLong method, and different diagnostic accuracy measures were generated with corresponding 95 % confidence intervals (CIs). ResultsWe enrolled 857 patients, 125 received a DNR order and 732 did not. The average ePOS score of DNR and non-DNR patients was 28.2±10.7 and 15.2±9.7, respectively. ePOS score, as a predictor of DNR order, had an area under receiver operator characteristic (AUROC) curve of 81.8 % (95% CI: 79.0 to 84.3, P <0.001). Youden's ideal cut-off value >17 was associated with a sensitivity of 87.2 (95% CI: 80.0 to 92.5, P <0.001), specificity of 63.9 (95% CI: 60.3 to 67.4, P <0.001), positive predictive value of 29.2 (95% CI: 24.6 to 33.8, P <0.001), negative predictive value of 96.7 (95% CI: 95.1 to 98.3, P <0.001), and diagnostic odds ratio 12.1 (95% CI: 7.0 to 20.8, P <0.001). ConclusionsIn this study, the ePOS score performed well as a diagnostic test for patients who will be labeled as DNR during their ICU stay. A cut-off score >17 may help guide clinical decisions to withhold or commence resuscitative measures.

Early and fair COVID-19 outcome risk assessment using robust feature selection

Personalized medicine plays an important role in treatment optimization for COVID-19 patient management. Early treatment in patients at high risk of severe complications is vital to prevent death and ventilator use. Predicting COVID-19 clinical outcomes using machine learning may provide a fast and data-driven solution for optimizing patient care by estimating the need for early treatment. In addition, it is essential to accurately predict risk across demographic groups, particularly those underrepresented in existing models. Unfortunately, there is a lack of studies demonstrating the equitable performance of machine learning models across patient demographics. To overcome this existing limitation, we generate a robust machine learning model to predict patient-specific risk of death or ventilator use in COVID-19 positive patients using features available at the time of diagnosis. We establish the value of our solution across patient demographics, including gender and race. In addition, we improve clinical trust in our automated predictions by generating interpretable patient clustering, patient-level clinical feature importance, and global clinical feature importance within our large real-world COVID-19 positive patient dataset. We achieved 89.38% area under receiver operating curve (AUROC) performance for severe outcomes prediction and our robust feature ranking approach identified the presence of dementia as a key indicator for worse patient outcomes. We also demonstrated that our deep-learning clustering approach outperforms traditional clustering in separating patients by severity of outcome based on mutual information performance. Finally, we developed an application for automated and fair patient risk assessment with minimal manual data entry using existing data exchange standards.

Machine Learning Provides Individualized Prediction of Outcomes after First Complete Remission in Adult AML Patients - Results from the Harmony Platform

Background: Acute myeloid leukemia (AML) is a heterogeneous disease with high disease-related mortality and allogeneic hematopoietic stem cell transplantation (allo-HSCT) remains the only curative option for a large proportion of patients. However, allo-HSCT has a significant transplant-related mortality (TRM), so a careful evaluation between risk of disease relapse and TRM is needed for the challenging decision of whom to transplant in first complete remission (CR1). The European LeukemiaNet (ELN) provides a 3-tier static risk classification which is already a useful guidance, although dynamic risk assessment using measurable residual disease (MRD) information should also be taken into account. However, the intermediate-risk group is characterized by a wide heterogeneity and there are also some favorable-risk patients who require allo-HSCT because of early relapses after CR1. Therefore, more precise prediction tools are warranted to provide additional information valuable for clinical decision-making. Aims: To develop and validate a model that provides individualized outcome estimations in adult AML patients achieving CR1 with intensive treatment approaches. Methods: From the HARMONY Platform including 7467 AML patients at the time of the analysis, we selected patients aged 18-70 years old who received intensive treatment, achieved CR and were not consolidated with allo-HSCT in CR1. Patients who were refractory to first line chemotherapy or who had a relapse-free survival (RFS) of less than two months after achieving CR were excluded. A Bayesian Additive Regression Trees (BART) nonparametric machine learning model was used to predict the individualized likelihood of RFS, cumulative incidence of relapse (CIR) and overall survival (OS) in 1863 AML patients who were also selected based on the availability of comprehensive mutational information by next-generation sequencing analysis. Genomic aberrations that were present in at least 10 patients were included in the model. In order to test the accuracy of the model, a 10-fold cross-validation was applied using the area under receiver operating curve (AUC) at predefined time points (1, 2, 3, 4 and 5 years). Subsequently, the model was validated using an external publicly available database (Tazi et al, Nat. Commun. 2022) which included 722 patients based on identical inclusion criteria. Results: The study population of 1863 adult AML patients in CR1 included 51.2% of males and median age was 50 years. Regarding ELN2022 classification, 52% of patients were classified as favorable, 30% as intermediate and 18% as adverse risk. Most frequently mutated genes were NPM1 (35.9%), DNMT3A (25.6%) and NRAS (21.6%). The HARMONY model was able to estimate individualized likelihood of RFS, CIR and OS for all patients, providing confidence intervals for all estimations. Moreover, we could predict outcomes with increased accuracy over ELN2022 at all predefined time points. At 5 years, the AUC value was significantly better for the HARMONY model than ELN2022 in predicting RFS (0.729 vs 0.679, p=0.003), CIR (0.729 vs 0.680, p=0.002) and OS (0.733 vs 0.675, p&amp;lt;0.001) ( figure 1A). Validation in an external independent AML cohort confirmed superiority of the HARMONY model in predicting RFS (5-year AUC 0.714 vs 0.609, p &amp;lt;0.001), CIR (0.713 vs 0.610, p &amp;lt;0.001) and OS (0.737 vs 0.638, p &amp;lt;0.001) as compared to the ELN2022 risk stratification ( figure 1B), despite the fact that there was a different distribution of risk categories (28% favorable, 45% intermediate and 27% adverse risk). Conclusions: Analysis of large well-defined AML cohorts allows the development of increasingly precise predictive models for clinically relevant scenarios. The HARMONY machine learning model provides individualized outcome prediction for patients aged 18-70 years in CR1 consolidated without allo-HSCT and might be used in the future for clinical decision-making. The model can be accessed online via an interactive web calculator () and, while it has been validated using a large independent AML cohort, future studies are required focusing on prospective validation. Furthermore, the incorporation of additional variables such as MRD and the inclusion of patients treated with targeted therapies and non-intensive approaches will provide more accurate risk predictions.

Telomere Length Testing of Lymphocytes Alone By Flow-FISH Is a Highly Sensitive and Specific Test to Screen for a Telomere Biology Disorder in a Cohort of Children and Very Young Adults with Bone Marrow Failure

Background Uncovering an underlying telomere biology disorder (TBD) in patients with bone marrow failure (BMF) has crucial implications for treatment and outcomes, including hematopoietic cell transplantation and malignancy risk. Telomere length (TL) testing of peripheral blood leukocytes by clinical Flow-FISH has become an important test in diagnosing TBDs. In seminal studies, Alter et al. compared age-adjusted TL lengths (&amp;lt; or ≥1 st percentile) in lymphocytes (lymphs), granulocytes (grans), and four lymph subsets in individuals with a known TBD and their unaffected relatives. They found that very low (VL) TL (&amp;lt;1 st percentile) in lymphs had a sensitivity (Sens) of 97%, specificity (Spec) of 91%, and positive predictive value (PPV) of 85% for a TBD diagnosis. VL TL in 3 or 4 lymph subsets had slightly better test performance with a Sens of 98%, Spec of 94%, and PPV of 89%. We hypothesized that test performance characteristics of flow FISH would be different if applied to a ‘real-world’ cohort, which could inform the use of TL in lymphs alone versus in the four lymph subsets for different clinical indications. Methods We performed a search of the electronic medical record to identify all individuals who had Flow-FISH performed at Texas Children's Hospital from 2007 - 2022 (Figure 1). We excluded those who had the 2-panel test (lymphs and grans) or family TL cascade testing. We collected the indication for and results of Flow-FISH, demographic and clinical characteristics, and genetic testing information via chart review. A composite score (Comp Score) was created to define TBD (Comp Score of 1 or 2) vs. Not TBD (Comp Score &amp;lt;1). Performance of lymph TL alone and TL in 3 or 4 lymph subsets in predicting TBD was evaluated using Sens, Spec, PPV, NPV, area under ROC curve (AUC), and Cohen's Kappa statistic (k). Results Our final cohort included 433 individuals who received the 6-panel test (lymphs, grans, and four lymph subsets) with a median age of 9.7 years, range 0.06-22.6. Top indications for Flow-FISH were severe aplastic anemia (SAA) (n=106), mild aplastic anemia (AA) (i.e., not fulfilling criteria for MAA or SAA) (103), cytopenias without known or documented bone marrow hypoplasia or dysplasia (70), moderate aplastic anemia (MAA) (37), non-therapy-related myelodysplastic syndrome (26), and an incidental TBD gene variant identified through comprehensive sequencing (17). Twenty-two subjects (5.1%) had a TBD, 17 (3.9%) of which had a pathogenic/likely pathogenic variant (P/LPV) in a known TBD gene (Figure 1). Seven of those with SAA (6.6%) had VL TL in lymphs, 5 (4.7%) of which also had VL TL in 3 or 4 of the lymph subsets, and only 1 (0.9%) had a TBD based on Comp Score. Importantly, none had VL TL in 3 or 4 of the lymph subsets without VL TL in lymphs. Five of those with MAA (13.5%) had VL TL in lymphs, all of which had VL TL in 3 or 4 lymph subsets. Three had a TBD based on Comp Score, and each had a PV in a TBD gene; a TBD variant was not identified in two despite exome sequencing and chromosomal microarray testing. Eighteen of those with mild AA (17.5%) had VL TL in lymphs, 15 (14.6%) also had VL TL in 3 or 4 lymph subsets, 9 (8.7%) had a TBD based on Comp Score, and 7 (6.8%) had a P/LPV in a TBD gene. The Sens of TL in lymphs &amp;lt; vs. ≥ 1 st percentile and TL &amp;lt; 1 st percentile in 3 or 4 vs. &amp;lt; 3 lymph subsets against TBD based on Comp Score in those with AA regardless of severity was 100%, and the Spec ranged from 90.4% (lymph TL testing alone in mild AA) to 96.2% (lymph subsets testing in SAA) (Table 1). The PPV for both test approaches was low (≤20%) in SAA due to the low prevalence of TBD in this population. PPV was between 50-60% in the other AA groups. The AUC for detecting a TBD in the various AA groups was comparably high for both test approaches (0.95-0.98). The two tests had a substantial to almost perfect agreement (k=0.82, p&amp;lt;0.001). Conclusions This single institutional cohort demonstrates that the prevalence of TBD in young patients presenting with SAA is remarkably low, and the Sens and Spec of TL testing in lymphs alone are sufficiently high to detect the rare TBD patient and prompt additional testing (e.g., genetics) in few patients due to false positive testing. The prevalence of TBD is higher in those with MAA or mild AA, yet both test approaches detect all TBD patients. Spec level in mild AA is marginally higher with lymph subset TL testing but, at ~94%, underscores the need for other clinical parameters (e.g., genetics) to establish a TBD diagnosis in a small percentage of patients.

Comparison of the effectiveness of delirium evaluation tools in intensive care patients: pre-deliric versions 1 and 2, E-pre-deliric and ICDSC.

This study's objective was to compare the effectiveness of the delirium prediction model (pre-deliric) and the early prediction model (E-pre-deliric) in delirium prediction in an intensive care unit (ICU) according to the Intensive Care Delirium Screening Checklist (ICDSC). Our aim was to determine these models' usability and cut-off values for ICU patients. We classified the studied patients based on their highest ICDSC scores (tested twice daily) during ICU hospitalization. ICDSC scores of 4 or higher indicated positive results for delirium, whereas a score of 0 represented a negative result. We recorded the patients' demographic and clinical details and characteristics and calculated their E-pre-deliric and pre-deliric version 1 and version 2 scores. To evaluate the effectiveness of the models, we used receiver operating characteristic (ROC) curve analysis. Two hundred fifty patients (55.6% males, mean age 60.6±18.7 years) participated in this study. Their mean Acute Physiology and Chronic Health Evaluation II (APACHE-II) score was 17.0±9.1. Delirium was more common in men, patients of older ages, those with high APACHE-II scores, those who had undergone urgent admissions, those with histories of trauma, those with high urea or creatinine values and those who had undergone sedation or mechanical ventilation. Compared to patients who did not develop delirium, those who did had longer ICU stays and hospital stays, as well as greater mortality risk. The cutoff values for the patients' pre-deliric version 1, pre-deliric version 2 and E-pre-deliric scores were 38% [area under ROC (AUROC)=1], 22% (AUROC=1) and 28% (AUROC=1), respectively. This study is the first to compare the pre-deliric and E-pre-deliric prediction models. These models' validity and reliability were acceptable. They were clinically useful, and we identified their cut-off values. These models provide options for early detection of delirium and are easily applicable in the ICU.

Echocardiography for Volume Assessment in Acute Myocardial Infarction.

Background Errors caused by improper volume estimation increase acute mortality rates in acute myocardial infarction (AMI). We aimed to determine volume status in AMI patients using echocardiography and to correlate the findings with clinical outcomes. Methods This cross-sectional, single-center study was performed at a tertiary care center in India between August 2017 and September 2020 involving AMI patients. We performed echocardiography for all patients. Parameters such as left ventricle (LV) and atrium size, LV end-diastolic pressure, inferior vena cava (IVC) size and size variation, velocity stroke volume, and velocity time integral variation were measured. B-lines were recorded by scanning 32 regions on the anterior chest in the supine position using cardiac probes of echocardiography. Results A total of 184 patients were enrolled in the study with male predominance (82.1%). The mean age of patients was 58.2 ± 10.7 years. Dilated (>2.1 cm) and collapsible (<50%) IVC, and B-lines were significantly associated with heart failure (HF) (p<0.001; r=0.87 and p<0.001; r=0.74, respectively). The area under receiver operating characteristics (AUROC) curve to diagnose HF at a cut-off value of >10 for B-lines was 0.897 (0.842-0.951). AUROC curve for IVC size in diagnosing hypovolemia was 0.063 (0.000-0.130). Conclusions Volume status based on IVC size and B-lines detected byechocardiography hasa strongprognostic valuein AMIpatients and should be included in the routine assessment of thesepatients.

Esophageal cancer detection framework based on time series information from smear images

The gold standard for esophageal cancer diagnosis and treatment is the Thinprep Cytologic Test (TCT) of suspected sections. TCT refers to analyze the features of the lesion areas using tissue regions stained with hematoxylin and eosin. However, pathologists are currently subjected to high workloads and limited experiences. Moreover, automatic esophageal cancer detection methods typically screen the smears by focusing on suspicious lesions identified by cytotechnicians, cell misclassification often results in inaccurate and weak robust diagnoses, smear utilization rate is also reduced. Computer-assisted diagnosis also faces challenges such as information loss, non-quantitative analysis, and nonstandard diagnosing processes. To address these issues, this article proposes a deep learning based detection framework called the Esophageal Time Series Thinprep Cytologic Test (ETS-TCT) to provide a quantitative and robust screening of Whole Slide Imaging (WSI) for esophageal cancer. Our system is divided into three modules: Coarse–fine detection model, quantitative analysis model, and Fusion Long Short Term Memory-Support Vector Machine (FLSTM-SVM) model. These modules are organized to map the pathological information from cell-level to smear-level and offer interpretable predictions to pathologists. For each smear, we design the coarse–fine segmentation model for adaptively locating and classifying target cells, give mathematical descriptions for cell Deoxyribonucleic Acid (DNA) values using the quantitative analysis model, establish an early risk screening model for esophageal cancer using the optimized FLSTM-SVM network. To validate the methodology, we use two clinical datasets containing 580 samples, achieving the accuracy, sensitivity, specificity, and area under ROC (AUC) of 92.0%, 93.4%, 94.4%, and 0.913, respectively. In short, the framework provides robust and accurate diagnosis of esophageal cancer WSIs using the time series features. This screening can assist the pathologists in strengthening the diagnosis by robustly presenting the quantitative intermediate results as well as visual interpretations. Moreover, we clearly present and expatiate the concepts of time series features, highlighting the relationship between these features to physiological states of patients. Our results indicate the possibility of using the framework as a second system to the computer-assisted esophageal cancer screening.

Risk assessment of inferior alveolar nerve injury after wisdom tooth removal using 3D AI-driven models: A within-patient study

ObjectiveTo compare a three-dimensional (3D) artificial intelligence (AI)- driven model with panoramic radiography (PANO) and cone-beam computed tomography (CBCT) in assessing the risk of inferior alveolar nerve (IAN) injury after mandibular wisdom tooth (M3M) removal through a within-patient controlled trial. MethodsFrom a database of 6,010 patients undergoing M3M surgery, 25 patients met the inclusion criteria of bilateral M3M removal with postoperative unilateral IAN injury. In this within-patient controlled trial, preoperative PANO and CBCT images were available, while 3D-AI models of the mandibular canal and teeth were generated from the CBCT images using the Virtual Patient Creator AI platform (Relu BV, Leuven, Belgium). Five examiners, who were blinded to surgical outcomes, assessed the imaging modalities and assigned scores indicating the risk level of IAN injury (high, medium, or low risk). Sensitivity, specificity, and area under receiver operating curve (AUC) for IAN risk assessment were calculated for each imaging modality. ResultsFor IAN injury risk assessment after M3M removal, sensitivity was 0.87 for 3D-AI, 0.89 for CBCT versus 0.73 for PANO. Furthermore, the AUC and specificity values were 0.63 and 0.39 for 3D-AI, 0.58 and 0.28 for CBCT, and 0.57 and 0.41 for PANO, respectively. There was no statistically significant difference (p>0.05) among the imaging modalities for any diagnostic parameters. ConclusionThis within-patient controlled trial study revealed that risk assessment for IAN injury after M3M removal was rather similar for 3D-AI, PANO, and CBCT, with a sensitivity for injury prediction reaching up to 0.87 for 3D-AI and 0.89 for CBCT. Clinical significanceThis within-patient trial is pioneering in exploring the application of 3D AI-driven models for assessing IAN injury risk after M3M removal. The present results indicate that AI-powered 3D models based on CBCT might facilitate IAN risk assessment of M3M removal.

Screening of potential biomarkers for polycystic ovary syndrome and identification of expression and immune characteristics.

Polycystic ovary syndrome (PCOS) seriously affects the fertility and health of women of childbearing age. We look forward to finding potential biomarkers for PCOS that can aid clinical diagnosis. We acquired PCOS and normal granulosa cell (GC) expression profiles from the Gene Expression Omnibus (GEO) database. After data preprocessing, differentially expressed genes (DEGs) were screened by limma package, and Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis and Gene Set Enrichment Analysis (GSEA) were performed. Recursive feature elimination (RFE) algorithm and the least absolute shrinkage and selection operator (LASSO) Cox regression analysis were used to acquire feature genes as potential biomarkers. Time-dependent receiver operator characteristic curve (ROC curve) and Confusion matrix were used to verify the classification performance of biomarkers. Then, the expression characteristics of biomarkers in PCOS and normal cells were analyzed, and the insulin resistance (IR) score of samples was computed by ssGSEA. Immune characterization of biomarkers was evaluated using MCP counter and single sample gene set enrichment analysis (ssGSEA). Finally, the correlation between biomarkers and the scores of each pathway was assessed. We acquired 93 DEGs, and the enrichment results indicated that most of DEGs in PCOS group were significantly enriched in immune-related biological pathways. Further screening results indicated that JDP2 and HMOX1 were potential biomarkers. The area under ROC curve (AUC) value and Confusion matrix of the two biomarkers were ideal when separated and combined. In the combination, the training set AUC = 0.929 and the test set AUC = 0.917 indicated good diagnostic performance of the two biomarkers. Both biomarkers were highly expressed in the PCOS group, and both biomarkers, which should be suppressed in the preovulation phase, were elevated in PCOS tissues. The IR score of PCOS group was higher, and the expression of JDP2 and HMOX1 showed a significant positive correlation with IR score. Most immune cell scores and immune infiltration results were significantly higher in PCOS. Comprehensive analysis indicated that the two biomarkers had strong correlation with immune-related pathways. We acquired two potential biomarkers, JDP2 and HMOX1. We found that they were highly expressed in the PCOS and had a strong positive correlation with immune-related pathways.

Combined conventional ultrasonography with clinicopathological features to predict axillary status after neoadjuvant therapy for breast cancer: A case-control study.

This study aimed to evaluate the value of a model combining conventional ultrasonography and clinicopathologic features for predicting axillary status after neoadjuvant therapy in breast cancer. This retrospective study included 329 patients with lymph node-positive who underwent neoadjuvant systemic treatment (NST) from June 2019 to March 2022. Ultrasound and clinicopathological characteristics of breast lesions and axillary lymph nodes were analyzed before and after NST. The diagnostic efficacy of ultrasound, clinicopathological characteristics, and combined model were evaluated using multivariate logistic regression and receiver operator characteristic curve (ROC) analyses. The area under ROC (AUC) for the ability of the combined model to predict the axillary pathological complete response (pCR) after NST was 0.882, that diagnostic effectiveness was significantly better than that of the clinicopathological model (AUC of 0.807) and the ultrasound feature model (AUC of 0.795). In addition, eight features were screened as independent predictors of axillary pCR, including clinical N stage, ERBB2 status, Ki-67, and after NST the maximum diameter reduction rate and margins of breast lesions, the short diameter, cortical thickness, and fatty hilum of lymph nodes. The combined model constructed from ultrasound and clinicopathological features for predicting axillary pCR has favorable diagnostic results, which allowed more accurate identification of BC patients who had received axillary pCR after NST. A combined model incorporated ultrasound and clinicopathological characteristics of breast lesions and axillary lymph nodes demonstrated favorable performance in evaluating axillary pCR preoperatively and non-invasively.

Application of machine learning models in predicting insomnia severity: an integrative approach with constitution of traditional Chinese medicine.

This study sought to explore the utility of machine learning models in predicting insomnia severity based on Traditional Chinese Medicine (TCM) constitution classifications, with an aim to discuss the potential applications of such models in the treatment and prevention of insomnia. We analyzed a dataset of 165 insomnia patients from the Shanghai Minhang District Integrated Traditional Chinese and Western Medicine Hospital. TCM constitution was assessed using a standardized Constitution in Chinese Medicine (CCM) scale. Sleep quality, or insomnia severity, was evaluated using the Spiegel Sleep Questionnaire (SSQ). Machine learning models, including Random Forest Classifier (RFC), Support Vector Classifier (SVC), and K-Nearest Neighbors (KNN), were utilized. These models were optimized using Grid Search algorithm and were trained and tested on stratified patient data, with the TCM constitution classifications serving as primary predictors. The RFC outperformed others, achieving a weighted average accuracy, precision, recall, and F1-score of 0.91, 0.94, 0.92, and 0.92 respectively, it also effectively classified the severity of insomnia with high area under receiver operating characteristic curve (AUC-ROC) values. Feature importance analysis demonstrated the Damp-heat constitution as the most influential predictor, followed by Yang-deficiency, Qi-depression, Qi-deficiency, and Blood-stasis constitutions. The results demonstrate the potent utility of machine learning, specifically RFC, coupled with TCM constitution classifications in predicting insomnia severity. Notably, the constitution classifications such as Damp-heat and Yang-deficiency emerged as crucial determinants, emphasizing its potential in guiding targeted insomnia treatments. This approach enables the development of more personalized and efficient interventions, thereby enhancing patient outcomes.