ROC AUC Research Articles

Machine learning models for gully erosion susceptibility assessment in the Tensift catchment, Haouz Plain, Morocco for sustainable development

Gully erosion is a widespread environmental danger, threatening global socio-economic stability and sustainable development. This study comprehensively applied seven machine learning (ML) models including SVM, KNN, RF, XGBoost, ANN, DT, and LR, and evaluated gully erosion susceptibility in the Tensift catchment and predict it within the Haouz plain, Morocco. To ensure the reliability of the findings, the study employed a robust combination of gully erosion inventory, sentinel images, and Digital Surface Model. Eighteen predictors, encompassing topographical, geomorphological, environmental, and hydrological factors, were selected after multicollinearity analyses. The gully erosion susceptibility of the study revealed that approximately 28.18% of the Tensift catchment is at a very high risk of erosion. Furthermore, 15.13% and 31.28% of the catchment are categorized as low and very low respectively. These findings extend to the Haouz plain, where 7.84% of the surface area are very highly risking erosion, while 18.25% and 55.18% are characterized as low and very low risk areas. To gauge the performance of the ML models, an array of metrics including specificity, precision, sensitivity, and accuracy were employed. The study highlights XGBoost and KNN as the most promising models, achieving AUC ROC values of 0.96 and 0.93 in the test phase. The remaining models namely RF (AUC ROC = 0.89), LR (AUC ROC = 0.80), SVM (AUC ROC = 0.81), DT (AUC ROC = 0.86), and ANN (AUC ROC = 0.78), also displayed commendable performance. The novelty of this research is its innovative approach to combat gully erosion through cutting edge ML models, offering practical solutions for watershed conservation, sustainable management, and the prevention of land degradation. These insights are invaluable for addressing the challenges posed by gully erosion within the region, and beyond its geographical boundaries and can be used for defining appropriate mitigation strategies at local to national scale.

A novel LDVP-based anomaly detection method for data streams

Recently, some progress has been made in anomaly detection for data streams. However, these methods still exhibit deficiencies in effectively balancing computational efficiency and detection accuracy. In this paper, we propose a novel anomaly detection method called Dynamic Window Anomaly Detection based on Local Density of Vector dot Product (DWAD-LDVP) for data streams. DWAD-LDVP uses an adaptive dynamic sliding window mechanism with a verification model to balance the relationship between computational efficiency and detection accuracy. By converting local density values measured by the vector dot product into outlier scores, DWAD-LDVP can improve the accuracy of anomaly detection compared to SOTA methods. Furthermore, an incremental computation module is adopted to process data streams efficiently, thereby reducing unnecessary computational overhead. Simulation experiments on synthetic data and real-world datasets validate the robustness and superior performance of the proposed DWAD-LDVP method. Specifically, the experimental results indicate that the proposed DWAD-LDVP achieves competitive performance measured by accuracy, precision, recall, F1 score, and ROC AUC, confirming its effectiveness in anomaly detection tasks.

Difference in mortality risk predicted by leukocyte and lymphocyte levels in COVID-19 patients infected with the Wild-type, Delta, and Omicron strains.

This study aimed to investigate the changing trends, level differences, and prognostic performance of the leukocyte and lymphocyte levels of patients infected with the Wild strains, Delta strains and Omicron strains to provide a reference for prognostic assessment. In the current study, we conducted a retrospective cross-sectional study to evaluate the changing trends, level differences, and prognostic performance of leukocyte and lymphocyte of different strains at admission and discharge may already exist in patients with coronavirus disease-2019 (COVID-19) infected with the Wild type, Delta, and Omicron strains. A retrospective cross-sectional study was conducted. We recruited and screened the 243 cases infected with the Wild-type strains in Wuhan, the 629 cases infected with the Delta and 116 cases infected strains with the Omicron strains in Xi'an. The leukocyte and lymphocyte levels were compared the cohort of Wild-type infection with the cohort of Delta and the Omicron. The changes in the levels of leukocytes and lymphocytes exhibit a completely opposite trend in patients with COVID-19 infected with the different strains. The lymphocyte level at admission and discharge in patients with COVID-19 infected with Omicron strains (area under curve [AUC] receiver operating characteristic curve [ROC] 72.8-90.2%, 82.8-97.2%) presented better performance compared patients with COVID-19 infected with Wild type strains (AUC ROC 60.9-80.7%, 82.3-97.2%) and Delta strains (AUC ROC 56.1-84.7%, 40.3-93.3%). Kaplan-Meier curves showed that the leukocyte levels above newly established cutoff values and the lymphocyte levels below newly established cutoff values had a significantly higher risk of in-hospital mortality in COVID-19 patients with Wild-type and Omicron strains (P < .01). The levels of leukocyte and lymphocyte at admission and discharge in patients with COVID-19 infected with the Wild type, Delta, and Omicron strains may be differences among strains, which indicates different death risks. Our research may help clinicians identify patients with a poor prognosis for severe acute respiratory syndrome coronavirus 2 infection.

Comparative analysis of machine learning classification algorithms for predicting olive anthracnose disease

&lt;p&gt;Olive anthracnose (OA) is the most damaging fungal disease of the olive tree worldwide. In the context of integrated pest management, the development of predictive models could be used for early diagnosis and control. In the current study, a dataset consisting of 58 cases, coming from 5 locations and 12 olive cultivars, was used to study the relationship between ΟΑ incidence (OAI) and 35 heterogeneous variables. These variables include orchard characteristics, olive fruit parameters, foliar and soil nutrients, soil parameters and soil texture classes. The Random Forest-Recursive Feature Elimination with Cross Validation (RF-RFECV) feature selection method identified Location, water content, P, Ca, Mg, exchangeable Mg, trace Zn, trace Cu as possible new indicators associated with OAI. The objective of this study was to investigate whether these variables have a predictive value for OAI. Six different machine learning classification algorithms, namely decision tree (DT), gradient boosting (GB), logistic regression (LR), random forest (RF), k-nearest neighbors (KNN) and support vector machine (SVM), were developed for predicting conditions leading to OAI &amp;gt; 0% and 10%. Grid search hyperparameter optimization was employed to optimize model parameters. The final models were evaluated in terms of several standard metrics, such as accuracy, sensitivity, specificity and ROC AUC score. Findings suggested that GB performance was superior compared to the other models for the prediction of the occurrence of OA disease (OAI &amp;gt; 0%) with an accuracy of 86.7%, a sensitivity of 100%, a specificity of 75% and a ROC-AUC score of 93%, while for the prediction of the spread of the disease (OAI &amp;gt; 10%), DT stood out with an accuracy of 86.7%, a sensitivity of 81.8%, a specificity of 100% and a ROC-AUC score of 91%.&lt;/p&gt;

Contribution of small airway inflammation to the development of COPD

BackgroundLittle attention has been paid to the pathophysiological changes in the natural history of chronic obstructive pulmonary disease (COPD). The destructions of the small airways were visualized on thoracic micro-computed tomography scan. We investigated whether small airway inflammation (SAI) was the risk for the development of COPD.MethodsA total of 1062 patients were enrolled and analyzed in the study. The partitioned airway inflammation was determined by exhaled nitric oxide (NO) of FnNO, FeNO50, FeNO200, and calculated CaNOdual. Both FeNO200 and CaNOdual were compared to detect the promising predictor for peripheral airway/alveolar inflammation in COPD. The correlation between exhaled NO and white cell classification was evaluated to determine the inflammation type during the development of COPD.ResultsExhaled NO levels (FnNO, FeNO50, FeNO200, and CaNOdual) were the highest in the COPD group compared with all other groups. Furthermore, compared with controls, exhaled NO levels (FeNO50, FeNO200, and CaNOdual) were also significantly higher in the emphysema, chronic bronchitis, and smoking groups. FeNO200 was found to be a promising predictor for peripheral airway/alveolar inflammation (area under the curve [AUC] of the receiver operating characteristic [ROC] curve, area under the curve [AUC] = 0.841) compared with CaNOdual (AUC ROC = 0.707) in COPD. FeNO200 was the main risk factor (adjusted odds ratio, 2.191; 95% CI, 1.797–2.671; p = 0.002) for the development of COPD. The blood eosinophil and basophil levels were correlated with FeNO50 and FeNO200.ConclusionThe complete airway inflammations were shown in COPD, whereas SAI was the main risk factor for the development of COPD, which might relate to eosinophil and basophil levels.

Transfer learning–based PET/CT three-dimensional convolutional neural network fusion of image and clinical information for prediction of EGFR mutation in lung adenocarcinoma

BackgroundTo introduce a three-dimensional convolutional neural network (3D CNN) leveraging transfer learning for fusing PET/CT images and clinical data to predict EGFR mutation status in lung adenocarcinoma (LADC).MethodsRetrospective data from 516 LADC patients, encompassing preoperative PET/CT images, clinical information, and EGFR mutation status, were divided into training (n = 404) and test sets (n = 112). Several deep learning models were developed utilizing transfer learning, involving CT-only and PET-only models. A dual-stream model fusing PET and CT and a three-stream transfer learning model (TS_TL) integrating clinical data were also developed. Image preprocessing includes semi-automatic segmentation, resampling, and image cropping. Considering the impact of class imbalance, the performance of the model was evaluated using ROC curves and AUC values.ResultsTS_TL model demonstrated promising performance in predicting the EGFR mutation status, with an AUC of 0.883 (95%CI = 0.849–0.917) in the training set and 0.730 (95%CI = 0.629–0.830) in the independent test set. Particularly in advanced LADC, the model achieved an AUC of 0.871 (95%CI = 0.823–0.919) in the training set and 0.760 (95%CI = 0.638–0.881) in the test set. The model identified distinct activation areas in solid or subsolid lesions associated with wild and mutant types. Additionally, the patterns captured by the model were significantly altered by effective tyrosine kinase inhibitors treatment, leading to notable changes in predicted mutation probabilities.ConclusionPET/CT deep learning model can act as a tool for predicting EGFR mutation in LADC. Additionally, it offers clinicians insights for treatment decisions through evaluations both before and after treatment.

Predictive Analysis of Endoscope Demand in Otolaryngology Outpatient Settings

Background: There has been a trend to transit reprocessing of flexible endoscopes from a high-level disinfectant (HLD) centralized manner to sterilization performed by nursing staff in some Ear, Nose, and Throat (ENT) clinics. In doing so, the clinic nursing staff are responsible for predicting and managing clinical demand for flexible endoscopes. The HLD disinfection process is time-consuming and requires specialized training and competency to be performed safely. Solely depending on human expertise for predicting the flexible endoscope demands is unreliable and produced a concern of an inadequate supply of devices available for diagnostic purposes. Method: The demand for flexible endoscopes for future patient visits has not been well studied but can be modeled based on patients’ historical information, provider, and other visit-related factors. Such factors are available to the clinic before the visit. Binary classifiers can be used to help inform the sterile processing department of reprocessing needs days or weeks earlier for each patient. Results: Among all our trained models, Logistic Regression reports an average AUC ROC score of 89% and accuracy of 80%. Conclusion: The proposed framework not only significantly reduces the reprocessing efforts in terms of time spent on communication, cleaning, scheduling, and transferring scopes, but also helps to improve patient safety by reducing the exposure risk to potential infections.

Using a Machine Learning Approach to Predict the Need for Elective Revision and Unplanned Surgery after Implant-based Breast Reconstruction.

Two-stage implant-based reconstruction after mastectomy may require secondary revision procedures to treat complications, correct defects, and improve aesthetic outcomes. Patients should be counseled on the possibility of additional procedures during the initial visit, but the likelihood of requiring another procedure is dependent on many patient- and surgeon-specific factors. This study aims to identify patient-specific factors and surgical techniques associated with higher rates of secondary procedures and offer a machine learning model to compute individualized assessments for preoperative counseling. A training set of 209 patients (406 breasts) who underwent two-stage alloplastic reconstruction was created, with 45.57% of breasts (185 of 406) requiring revisional or unplanned surgery. On multivariate analysis, hypertension, no tobacco use, and textured expander use corresponded to lower odds of additional surgery. In contrast, higher initial tissue expander volume, vertical radial incision, and larger nipple-inframammary fold distance conferred higher odds of additional surgery. The neural network model trained on clinically significant variables achieved the highest collective performance metrics, with ROC AUC of 0.74, sensitivity of 84.2, specificity of 63.6, and accuracy of 62.1. The proposed machine learning model trained on a single surgeon's data offers a precise and reliable tool to assess an individual patient's risk of secondary procedures. Machine learning models enable physicians to tailor surgical planning and empower patients to make informed decisions aligned with their lifestyle and preferences. The utilization of this technology is especially applicable to plastic surgery, where outcomes are subject to a variety of patient-specific factors and surgeon practices, including threshold to perform secondary procedures.

Cardiologist-level interpretable knowledge-fused deep neural network for automatic arrhythmia diagnosis

BackgroundLong-term monitoring of Electrocardiogram (ECG) recordings is crucial to diagnose arrhythmias. Clinicians can find it challenging to diagnose arrhythmias, and this is a particular issue in more remote and underdeveloped areas. The development of digital ECG and AI methods could assist clinicians who need to diagnose arrhythmias outside of the hospital setting.MethodsWe constructed a large-scale Chinese ECG benchmark dataset using data from 272,753 patients collected from January 2017 to December 2021. The dataset contains ECG recordings from all common arrhythmias present in the Chinese population. Several experienced cardiologists from Shanghai First People’s Hospital labeled the dataset. We then developed a deep learning-based multi-label interpretable diagnostic model from the ECG recordings. We utilized Accuracy, F1 score and AUC-ROC to compare the performance of our model with that of the cardiologists, as well as with six comparison models, using testing and hidden data sets.ResultsThe results show that our approach achieves an F1 score of 83.51%, an average AUC ROC score of 0.977, and 93.74% mean accuracy for 6 common arrhythmias. Results from the hidden dataset demonstrate the performance of our approach exceeds that of cardiologists. Our approach also highlights the diagnostic process.ConclusionsOur diagnosis system has superior diagnostic performance over that of clinicians. It also has the potential to help clinicians rapidly identify abnormal regions on ECG recordings, thus improving efficiency and accuracy of clinical ECG diagnosis in China. This approach could therefore potentially improve the productivity of out-of-hospital ECG diagnosis and provides a promising prospect for telemedicine.

A radiomics strategy based on CT intra-tumoral and peritumoral regions for preoperative prediction of neoadjuvant chemoradiotherapy for esophageal cancer

ObjectiveDevelop a method for selecting esophageal cancer patients achieving pathological complete response with pre-neoadjuvant therapy chest-enhanced CT scans. MethodsTwo hundred and one patients from center 1 were enrolled, split into training and testing sets (7:3 ratio), with an external validation set of 30 patients from center 2. Radiomics features from intra-tumoral and peritumoral images were extracted and dimensionally reduced using Student's t-test and least absolute shrinkage and selection operator. Four machine learning classifiers were employed to build models, with the best-performing models selected based on accuracy and stability. ROC curves were utilized to determine the top prediction model, and its generalizability was evaluated on the external validation set. ResultsAmong 16 models, the integrated-XGBoost and integrated-random forest models performed the best, with average ROC AUCs of 0.906 and 0.918, respectively, and RSDs of 6.26 and 6.89 in the training set. In the testing set, AUCs were 0.845 and 0.871, showing no significant difference in ROC curves. External validation set AUCs for integrated-XGBoost and integrated-random forest models were 0.650 and 0.749. ConclusionIncorporating peritumoral radiomics features into the analysis enhances predictive performance for esophageal cancer patients undergoing neoadjuvant chemoradiotherapy, paving the way for improved treatment outcomes.

Prediction of survival in out-of-hospital cardiac arrest: the updated Swedish cardiac arrest risk score (SCARS) model.

Out-of-hospital cardiac arrest (OHCA) is a major health concern worldwide. Although one-third of all patients achieve a return of spontaneous circulation and may undergo a difficult period in the intensive care unit, only 1 in 10 survive. This study aims to improve our previously developed machine learning model for early prognostication of survival in OHCA. We studied all cases registered in the Swedish Cardiopulmonary Resuscitation Registry during 2010 and 2020 (n = 55 615). We compared the predictive performance of extreme gradient boosting (XGB), light gradient boosting machine (LightGBM), logistic regression, CatBoost, random forest, and TabNet. For each framework, we developed models that optimized (i) a weighted F1 score to penalize models that yielded more false negatives and (ii) a precision-recall area under the curve (PR AUC). LightGBM assigned higher importance values to a larger set of variables, while XGB made predictions using fewer predictors. The area under the curve receiver operating characteristic (AUC ROC) scores for LightGBM were 0.958 (optimized for weighted F1) and 0.961 (optimized for a PR AUC), while for XGB, the scores were 0.958 and 0.960, respectively. The calibration plots showed a subtle underestimation of survival for LightGBM, contrasting with a mild overestimation for XGB models. In the crucial range of 0-10% likelihood of survival, the XGB model, optimized with the PR AUC, emerged as a clinically safe model. We improved our previous prediction model by creating a parsimonious model with an AUC ROC at 0.96, with excellent calibration and no apparent risk of underestimating survival in the critical probability range (0-10%). The model is available at www.gocares.se.

Friends Recommendation on Social Networks using the Bayesian Personalized Ranking-Matrix Factorization

In the digital landscape of social networking, the challenge of improving friend recommendation systems is pivotal for enhancing user interaction and fostering social connections. Addressing this challenge, the current study innovates by fusing Bayesian Personalized Ranking (BPR) with Matrix Factorization (MF), culminating in a novel BPR-MF model designed for the intricacies of social network relationships. The study harnesses a rich dataset from LastFM, comprising 27,806 interactions among 7,624 users, to analyze mutual follower patterns and augment the precision of friend recommendations. Through rigorous preprocessing and systematic evaluation of the BPR-MF model against different numbers of latent factors, the research uncovers that a configuration of 20 latent factors is most effective, achieving an RMSE of 0.156 and an AUC ROC of 0.800. This discovery addresses the critical problem of balancing computational complexity with prediction accuracy in recommendation models. It also demonstrates the necessity for a nuanced, data-driven approach to generate relevant social connections. The research sets a new direction for future studies aiming to capitalize on user interaction data to offer precise friend suggestions, all while upholding user privacy and avoiding reliance on personal data.

Comparison of iliopsoas tendonitis after hip resurfacing arthroplasty and total hip arthroplasty: A case-controlled investigation using a validated simulation.

Iliopsoas tendonitis, typically caused by impingement with the acetabular cup, occurs in up to 18% of patients after total hip arthroplasty (THA) and up to 30% of patients after hip resurfacing arthroplasty (HRA). We have developed a simulation for detecting iliopsoas impingement and validated it in a previous study of THA patients. However, due to the difference in incidence between HRA and THA, this study had two aims. First, to validate the simulation in a cohort of HRA patients and, second, to comparethe results of the HRA and THA patients to understand any differences in their etiology. We conducted a retrospective search in an experienced surgeon's database for HRA patients with iliopsoas tendonitisand control patients without iliopsoas tendonitis, resulting in two cohorts of 12 patients. Using CT scans, 3D models of the each patient's prosthetic and bony anatomy were generated, landmarked, and simulated. Regarding validation of the simulation for HRA patients, impingement significantly predicted the probability of iliopsoas tendonitis in logistic regression models and the simulation had a sensitivity of 83%, specificity of 100%, and an AUC ROC curve of 0.95. Unexpectedly, the HRA cohort exhibited less impingement than the THA cohort. Our novel simulation has now been demonstrated to detect iliopsoas impingement and differentiate between the symptomatic and asymptomatic cohorts in investigations of THA and HRA patients. This tool has the potential to be used preoperatively, to guide decisions about optimal cup placement, and postoperatively, to assist in the diagnosis of iliopsoas tendonitis.

Comprehensive Classification of Iris Flower Species: A Machine Learning Approach

This study employs robust machine learning techniques to comprehensively assess the classification of Iris flower species. This study investigates the effectiveness of several machine learning algorithms in reliably classifying Iris flower species by utilizing a dataset that includes crucial morphological attributes such as sepal length, sepal width, petal length, and petal width. The algorithms under consideration are Decision Tree, Random Forest, Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and XGBoost. Every algorithm has its own distinct methodology for classification, where Decision Trees offer clear interpretability and Random Forest and XGBoost offer strong and complex ensemble techniques. The primary aim of this study is to assess and contrast different algorithms, considering not only their classification accuracy but also significant performance metrics including precision, recall, F1-score, ROC AUC, and specificity. This research provides valuable insights into the capabilities and constraints of each methodology when implemented on a meticulously organized and defined botanical dataset. It is expected that the results of this study will contribute significantly to the fields of artificial intelligence and botanical taxonomy, highlighting the capacity of these methods to accurately identify and categorize plant species.

Sample entropy correlates with intraventricular hemorrhage and mortality in premature infants early in life.

Mortality and intraventricular hemorrhage (IVH) are common adverse outcomes in preterm infants and are challenging to predict clinically. Sample entropy (SE), a measure of heart rate variability (HRV), has shown predictive power for sepsis and other morbidities in neonates. We evaluated associations between SE and mortality and IVH in the first week of life. Participants were 389 infants born before 32 weeks of gestation for whom bedside monitor data were available. A total of 29 infants had IVH grade 3 or 4 and 31 infants died within 2 weeks of life. SE was calculated with the PhysioNet open-source benchmark. Logistic regressions assessed associations between SE and IVH and/or mortality with and without common clinical covariates over various hour of life (HOL) censor points. Lower SE was associated with mortality by 4 HOL, but higher SE was very strongly associated with IVH and mortality at 24-96 HOL. Bootstrap testing confirmed SE significantly improved prediction using clinical variables at 96 HOL. SE is a significant predictor of IVH and mortality in premature infants. Given IVH typically occurs in the first 24-72 HOL, affected infants may initially have low SE followed by a sustained period of high SE. SE correlates with IVH and mortality in preterm infants early in life. SE combined with clinical factors yielded ROC AUCs well above 0.8 and significantly outperformed the clinical model at 96 h of life. Previous studies had not shown predictive power over clinical models. First study using the PhysioNet Cardiovascular Toolbox benchmark in young infants. Relative to the generally accepted timing of IVH in premature infants, we saw lower SE before or around the time of hemorrhage and a sustained period of higher SE after. Higher SE after acute events has not been reported previously.

Assisted design of data science pipelines

When designing data science (DS) pipelines, end-users can get overwhelmed by the large and growing set of available data preprocessing and modeling techniques. Intelligent discovery assistants (IDAs) and automated machine learning (AutoML) solutions aim to facilitate end-users by (semi-)automating the process. However, they are expensive to compute and yield limited applicability for a wide range of real-world use cases and application domains. This is due to (a) their need to execute thousands of pipelines to get the optimal one, (b) their limited support of DS tasks, e.g., supervised classification or regression only, and a small, static set of available data preprocessing and ML algorithms; and (c) their restriction to quantifiable evaluation processes and metrics, e.g., tenfold cross-validation using the ROC AUC score for classification. To overcome these limitations, we propose a human-in-the-loop approach for the assisteddesignofdatasciencepipelines using previously executed pipelines. Based on a user query, i.e., data and a DS task, our framework outputs a ranked list of pipeline candidates from which the user can choose to execute or modify in real time. To recommend pipelines, it first identifies relevant datasets and pipelines utilizing efficient similarity search. It then ranks the candidate pipelines using multi-objective sorting and takes user interactions into account to improve suggestions over time. In our experimental evaluation, the proposed framework significantly outperforms the state-of-the-art IDA tool and achieves similar predictive performance with state-of-the-art long-running AutoML solutions while being real-time, generic to any evaluation processes and DS tasks, and extensible to new operators.

Duplex Vertical-Flow Rapid Tests for Point-of-Care Detection of Anti-dsDNA and Anti-Nuclear Autoantibodies.

The goal of this study is to develop a rapid diagnostic test for rheumatic disease and systemic lupus erythematosus (SLE) screening. A novel rapid vertical flow assay (VFA) was engineered and used to assay anti-nuclear (ANA) and anti-dsDNA (αDNA) autoantibodies from systemic lupus erythematosus (SLE) patients and healthy controls (HCs). Observer scores and absolute signal intensities from the VFA were validated via ELISA. The rapid point-of-care VFA test that was engineered demonstrated a limit of detection of 0.5 IU/mL for ANA and αDNA autoantibodies in human plasma with an inter-operator CV of 19% for ANA and 12% for αDNA. Storage stability was verified over a three-month period. When testing anti-dsDNA and ANA levels in SLE and HC serum samples, the duplex VFA revealed 95% sensitivity, 72% specificity and an 84% ROC AUC value in discriminating disease groups, comparable to the gold standard, ELISA. The rapid αDNA/ANA duplex VFA can potentially be used in primary care clinics for evaluating patients or at-risk subjects for rheumatic diseases and for planning follow-up testing. Given its low cost, ease, and rapid turnaround, it can also be used to assess SLE prevalence estimates.

Is the Validity of Logistic Regression Models Developed with a National Hospital Database Inferior to Models Developed from Clinical Databases to Analyze Surgical Lung Cancers?

In national hospital databases, certain prognostic factors cannot be taken into account. The main objective was to estimate the performance of two models based on two databases: the Epithor clinical database and the French hospital database. For each of the two databases, we randomly sampled a training dataset with 70% of the data and a validation dataset with 30%. The performance of the models was assessed with the Brier score, the area under the receiver operating characteristic (AUC ROC) curve and the calibration of the model. For Epithor and the hospital database, the training dataset included 10,516 patients (with resp. 227 (2.16%) and 283 (2.7%) deaths) and the validation dataset included 4507 patients (with resp. 93 (2%) and 119 (2.64%) deaths). A total of 15 predictors were selected in the models (including FEV1, body mass index, ASA score and TNM stage for Epithor). The Brier score values were similar in the models of the two databases. For validation data, the AUC ROC curve was 0.73 [0.68-0.78] for Epithor and 0.8 [0.76-0.84] for the hospital database. The slope of the calibration plot was less than 1 for the two databases. This work showed that the performance of a model developed from a national hospital database is nearly as good as a performance obtained with Epithor, but it lacks crucial clinical variables such as FEV1, ASA score, or TNM stage.

White matter diffusion estimates in obsessive-compulsive disorder across 1653 individuals: machine learning findings from the ENIGMA OCD Working Group

White matter pathways, typically studied with diffusion tensor imaging (DTI), have been implicated in the neurobiology of obsessive-compulsive disorder (OCD). However, due to limited sample sizes and the predominance of single-site studies, the generalizability of OCD classification based on diffusion white matter estimates remains unclear. Here, we tested classification accuracy using the largest OCD DTI dataset to date, involving 1336 adult participants (690 OCD patients and 646 healthy controls) and 317 pediatric participants (175 OCD patients and 142 healthy controls) from 18 international sites within the ENIGMA OCD Working Group. We used an automatic machine learning pipeline (with feature engineering and selection, and model optimization) and examined the cross-site generalizability of the OCD classification models using leave-one-site-out cross-validation. Our models showed low-to-moderate accuracy in classifying (1) “OCD vs. healthy controls” (Adults, receiver operator characteristic-area under the curve = 57.19 ± 3.47 in the replication set; Children, 59.8 ± 7.39), (2) “unmedicated OCD vs. healthy controls” (Adults, 62.67 ± 3.84; Children, 48.51 ± 10.14), and (3) “medicated OCD vs. unmedicated OCD” (Adults, 76.72 ± 3.97; Children, 72.45 ± 8.87). There was significant site variability in model performance (cross-validated ROC AUC ranges 51.6–79.1 in adults; 35.9–63.2 in children). Machine learning interpretation showed that diffusivity measures of the corpus callosum, internal capsule, and posterior thalamic radiation contributed to the classification of OCD from HC. The classification performance appeared greater than the model trained on grey matter morphometry in the prior ENIGMA OCD study (our study includes subsamples from the morphometry study). Taken together, this study points to the meaningful multivariate patterns of white matter features relevant to the neurobiology of OCD, but with low-to-moderate classification accuracy. The OCD classification performance may be constrained by site variability and medication effects on the white matter integrity, indicating room for improvement for future research.

Comprehensive Classification of Fetal Health Using Cardiotocogram Data Based on Machine Learning

In the realm of obstetrics, the evaluation of fetal health remains a paramount yet challenging endeavor. Traditional approaches, such as electronic fetal monitoring (EFM), despite their widespread adoption, continue to grapple with uncertainties regarding their impact on neonatal outcomes and the reduction of emergency cesarean deliveries. This ambiguity is compounded by a prevailing confusion within the obstetric community about interpreting fetal heart rate patterns, often leading to inconsistent and subjective assessments. Addressing these complexities, our study presents an innovative machine learning-based techniques for the comprehensive classification of fetal health using cardiotocogram (CTG) data, offering a more objective and nuanced alternative to conventional methods. The core of our proposed solution is a novel model employing a sophisticated ensemble of machine learning classifiers, including Multi-Support Vector Machine (Multi-SVM), Decision Tree, Random Forest with Hyperparameter Tuning, XGBoost, and Neural Networks. This model is unique in its application, processing datasets in four different forms: raw datasets, datasets processed with MinMaxScaler, datasets subjected to feature selection using SelectKBest, and a combination of MinMaxScaler processing and SelectKBest feature selection. Such meticulous preprocessing, encompassing normalization and feature selection, is pivotal in ensuring equitable contribution from each feature, thereby optimizing the model's learning process and predictive accuracy. The effectiveness of our model is rigorously evaluated using a dataset comprising 2126 individual records from CTG exams, classified by specialist obstetricians into three types: Normal, Suspect, and Pathological. These records are exhaustively analyzed using various metrics, including Accuracy, Precision, Recall, F1-Score, ROC AUC, and Confusion Matrix. Among the classifiers, XGBoost emerged as the most proficient, consistently outperforming others across multiple metrics. This indicates its superior ability to accurately identify and categorize the different states of fetal health. Our findings thus underscore the significant promise of machine learning in revolutionizing fetal health monitoring, offering a more reliable, objective, and comprehensive method for assessing fetal well-being, with profound implications for prenatal care and clinical decision-making.