Receiver Operating Characteristic Area Under The Curve Research Articles

Nomogram Based on HRV for Predicting the Therapeutic Effects of Orthostatic Training in Children with Vasovagal Syncope

Background: This study intended to find out whether the parameters of heart rate variability (HRV) can predict the treatment efficacy of orthostatic training among pediatric cases of vasovagal syncope (VVS). Methods: Patients with VVS who underwent orthostatic training were retrospectively enrolled. Lasso and logistic regression were used to sift through variables and build the model, which is visualized using a nomogram. The model’s performance was evaluated through calibration plots, a receiver operating characteristic (ROC) curve, and decision curve analysis (DCA) for both datasets. Results: In total, 119 participants were included in the analysis, and 73 and 46 were assigned to the training and validation datasets, respectively. Five factors with nonzero coefficients were chosen based on lasso regression: age, the root means square of successive differences between normal sinus beats (rMSSD), standard deviation of the averages normal-to-normal intervals in all 5-min segments, minimum heart rate, and high frequency. Drawing from the logistic regression analysis results, the visual predictive model incorporated two variables, namely age and rMSSD. For the training dataset, the sensitivity was 0.686 and the specificity was 0.868 with an area under the curve (AUC) of 0.81 (95% CI, 0.71–0.91) for the ROC curve. For the validation dataset, the AUC of the ROC was 0.80 (95% CI, 0.66–0.93), while sensitivity and specificity were recorded at 0.625 and 0.909, respectively. In the calibration plots for both datasets, the predicted probabilities correlated well with the actual probabilities. According to the DCA, the visual predictive model gained a significant net benefit across a wide threshold range. Conclusions: Pediatric patients with VVS can benefit from orthostatic training using a visual predictive model comprising age and rMSSD.

Druggability of Pharmaceutical Compounds Using Lipinski Rules with Machine Learning

In the field of pharmaceutical research, identifying promising pharmaceutical compounds is a critical challenge. The observance of Lipinski's Rule of Five (RO5) is a fundamental criterion, but evaluating many compounds manually requires significant resources and time. However, the integration of computational techniques in drug discovery in its early stages has significantly transformed the pharmaceutical industry, enabling further efficient screening and selection of possible drug candidates. Therefore, this study explores RO5 using algorithms of Machine Learning (ML), offering a comprehensive method to predict the druggability of pharmaceutical compounds. The study developed, evaluated, and validated the performance metrics of multiple supervised machine learning models. The best model was used to build an application that can predict and classify potential drug candidates. The findings revealed promising capabilities across all models for drug classification. Among all the explored models, Random Forest (RF), Extreme Gradient Boost (XGBoost), and Decision Tree (DT) classifiers demonstrated exceptional performance, achieving near-perfect accuracy of 99.94%, 99.81% and 99.87% respectively. This highlights the robustness of ensemble learning methods in classifying compounds based on RO5 adherence. The comparative analysis of these models underscores the importance of considering balanced accuracy, precision, F1-score, recall, and Receiver Operating Characteristics-Area Under the Curve (ROC-AUC) score, interpretability, and computational efficiency when choosing between ML algorithms in drug discovery. The DrugCheckMaster application was subsequently developed using the most predictive model and is now available on Render (https://capstone-project-dc7w.onrender.com/).

A Machine Learning Model Based on Thyroid US Radiomics to Discriminate Between Benign and Malignant Nodules.

Background/Objectives: Thyroid nodules are a very common finding, mostly benign but sometimes malignant, and thus require accurate diagnosis. Ultrasound and fine needle biopsy are the most widely used and reliable diagnostic methods to date, but they are sometimes limited in addressing benign from malignant nodules, mainly with regard to ultrasound, by the operator's experience. Radiomics, quantitative feature extraction from medical images and machine learning offer promising avenues to improve diagnosis. The aim of this work was to develop a machine learning model based on thyroid ultrasound images to classify nodules into benign and malignant classes. Methods: For this purpose, images of ultrasonography from 142 subjects were collected. Among these subjects, 40 patients (28.2%) belonged to the class "malignant" and 102 patients (71.8%) belonged to the class "benign", according to histological diagnosis from fine-needle aspiration. This image set was used for the training, cross-validation and internal testing of three different machine learning models. A robust radiomic approach was applied, under the hypothesis that the radiomic feature could capture the disease heterogeneity among the two groups. Three models consisting of four ensembles of machine learning classifiers (random forests, support vector machines and k-nearest neighbor classifiers) were developed for the binary classification task of interest. The best performing model was then externally tested on a cohort of 21 new patients. Results: The best model (ensemble of random forest) showed Receiver Operating Characteristic-Area Under the Curve (ROC-AUC) (%) of 85 (majority vote), 83.7 ** (mean) [80.2-87.2], accuracy (%) of 83, 81.2 ** [77.1-85.2], sensitivity (%) of 70, 67.5 ** [64.3-70.7], specificity (%) of 88, 86.5 ** [82-91], positive predictive value (PPV) (%) of 70, 66.5 ** [57.9-75.1] and negative predictive value (NPV) (%) of 88, 87.1 ** [85.5-88.8] (* p < 0.05, ** p < 0.005) in the internal test cohort. It achieved an accuracy of 90.5%, a sensitivity of 100%, a specificity of 86.7%, a PPV of 75% and an NPV of 100% in the external testing cohort. Conclusions: The model constituted of four ensembles of random forest classifiers could identify all the malignant nodes and the consistent majority of benign in the external testing cohort.

Federated learning with tensor networks: a quantum AI framework for healthcare

Abstract The healthcare industries frequently handle sensitive and proprietary data, and due to strict&amp;#xD;privacy regulations, they are often reluctant to share data directly. In today’s context, Federated&amp;#xD;Learning (FL) stands out as a crucial remedy, facilitating the rapid advancement of distributed machine&amp;#xD;learning while effectively managing critical concerns regarding data privacy and governance.&amp;#xD;The fusion of federated learning and quantum computing represents a groundbreaking interdisciplinary&amp;#xD;approach with immense potential to revolutionize various industries, from healthcare to&amp;#xD;finance. In this work, we propose a federated learning framework based on quantum tensor networks&amp;#xD;that takes advantage of the principles of many-body quantum physics. Currently, there are&amp;#xD;no known classical tensor networks implemented in federated settings. Furthermore, we investigated&amp;#xD;the effectiveness and feasibility of the proposed framework by conducting a differential privacy analysis&amp;#xD;to ensure the security of sensitive data across healthcare institutions. Experiments on popular&amp;#xD;medical image datasets show that the federated quantum tensor network model achieved a mean&amp;#xD;receiver-operator characteristic area under the curve (ROC-AUC) of 91-98%, outperforming several&amp;#xD;state-of-the-art federated learning methods while using fewer parameters. Experimental results&amp;#xD;demonstrate that the quantum federated global model, consisting of highly entangled tensor network&amp;#xD;structures, showed better generalization and robustness and achieved higher testing accuracy,&amp;#xD;surpassing the performance of locally trained clients under unbalanced data distributions among&amp;#xD;healthcare institutions.

A dynamic provenance graph-based detector for advanced persistent threats

Advanced Persistent Threats (APTs) pose a major cyber threat due to their stealthy, long-term nature and intricate complexity, making them particularly challenging to detect. Provenance graphs map interactions between system entities as directed, heterogeneous networks, offering rich semantic information valuable for threat identification. However, most existing approaches rely on static graph analysis, overlooking the critical dynamics of evolving threats. Dynamic graph analysis offers the potential to capture both temporal and structural insights, but current methods focus on single-perspective learning, failing to fully exploit the inherent relationships and evolving patterns within the data. To address this gap, we propose CGL-AD, a Contextualized Graph Learning APT Detector. CGL-AD leverages temporal graph learning to capture subtle temporal changes and overall structural transformations over time. It then integrates hash-based data stream frequency estimation techniques to identify local topological alterations, subsequently feeding these rich embeddings into a powerful sequence learning model. Experimental results on three widely used datasets: Streamspot, Camflow-apt and Shellshock datasets demonstrate that CGL-AD significantly outperforms existing methods. Specifically, CGL-AD outperforms the best baseline FLASH on these datasets by 1.5%, 3.7%, and 5.6% respectively in terms of Receiver Operating Characteristic -Area Under the Curve (ROC-AUC), effectively revealing hidden APT attack patterns in dynamic provenance graphs.

Near-Infrared Spectroscopy for Neonatal Sleep Classification.

Sleep, notably active sleep (AS) and quiet sleep (QS), plays a pivotal role in the brain development and gradual maturation of (pre) term infants. Monitoring their sleep patterns is imperative, as it can serve as a tool in promoting neurological maturation and well-being, particularly important in preterm infants who are at an increased risk of immature brain development. An accurate classification of neonatal sleep states can contribute to optimizing treatments for high-risk infants, with respiratory rate (RR) and heart rate (HR) serving as key components in sleep assessment systems for neonates. Recent studies have demonstrated the feasibility of extracting both RR and HR using near-infrared spectroscopy (NIRS) in neonates. This study introduces a comprehensive sleep classification approach leveraging high-frequency NIRS signals recorded at a sampling rate of 100 Hz from a cohort of nine preterm infants admitted to a neonatal intensive care unit. Eight distinct features were extracted from the raw NIRS signals, including HR, RR, motion-related parameters, and proxies for neural activity. These features served as inputs for a deep convolutional neural network (CNN) model designed for the classification of AS and QS sleep states. The performance of the proposed CNN model was evaluated using two cross-validation approaches: ten-fold cross-validation of data pooling and five-fold cross-validation, where each fold contains two independently recorded NIRS data. The accuracy, balanced accuracy, F1-score, Kappa, and AUC-ROC (Area Under the Curve of the Receiver Operating Characteristic) were employed to assess the classifier performance. In addition, comparative analyses against six benchmark classifiers, comprising K-Nearest Neighbors, Naive Bayes, Support Vector Machines, Random Forest (RF), AdaBoost, and XGBoost (XGB), were conducted. Our results reveal the CNN model's superior performance, achieving an average accuracy of 88%, a balanced accuracy of 94%, an F1-score of 91%, Kappa of 95%, and an AUC-ROC of 96% in data pooling cross-validation. Furthermore, in both cross-validation methods, RF and XGB demonstrated accuracy levels closely comparable to the CNN classifier. These findings underscore the feasibility of leveraging high-frequency NIRS data, coupled with NIRS-based HR and RR extraction, for assessing sleep states in neonates, even in an intensive care setting. The user-friendliness, portability, and reduced sensor complexity of the approach suggest its potential applications in various less-demanding settings. This research thus presents a promising avenue for advancing neonatal sleep assessment and its implications for infant health and development.

A modified analysis protocol for the PhNR test.

Several studies have reported that glaucoma patients have abnormal photopic negative response (PhNR) results compared to reference control subjects. The International Society for Clinical Electrophysiology of Vision (ISCEV) released an extended protocol for PhNR (I-PhNR) in 2018. The purpose of this study was to compare the I-PhNR protocol to a similar protocol modified (M-PhNR) to enhance the performance of the method in detecting glaucomatous damage. Thirty subjects were enrolled in this study (12 glaucoma patients, 10 glaucoma suspects, 8 normal controls). PhNR tests were conducted with a Diagnosys E3 mobile system (Diagnosys LLC, Lowell, MA). I-PhNR tests utilized all parameters specified by the ISCEV requirement. M-PhNR tests used the same parameters as the ISCEV tests with the exceptions of a 5-45Hz bandpass filter and a novel, objective sweep-selection parameter. According to the ISCEV protocol, the PhNR relative to baseline (i.e., BT), a-wave and b-wave response amplitudes and BT/b-wave amplitude ratios were measured. Coefficients of variation, receiver operating characteristic (ROC) curves, and t-tests were used to assess the data from one randomly chosen eye per subject. The M-PhNR protocol resulted in a decrease in the intra-subject repeat test coefficient of variation and a decrease in the average inter-subject coefficient of variation for the glaucoma subjects. The ROC curves demonstrated an increase in the area under the curve (AUC) for the M-PhNR compared to the I-PhNR protocol. The sensitivity and specificity were also greater for the M-PhNR protocol. The M-PhNR protocol resulted in a decrease in intra-subject and inter-subject data variability which resulted in a significant increase in the ROC AUC, sensitivity, and specificity for glaucoma. Thus, the M-PhNR protocol shows promise as a better diagnostic tool than the I-PhNR protocol for detecting glaucoma.

Novel scoring system specific for patients treated with micro-axial percutaneous left ventricular assist device

Abstract Background Cardiogenic shock (CS) is characterized by high mortality. The impact of micro-axial percutaneous left ventricular assist devices (ma-pLVAD) remains unclear. Nevertheless, the utilization of these devices increased lately which binds considerable resources. Accurate prognosis assessment is crucial for physicians to identify patients who may benefit from the device upfront. Purpose We developed a novel scoring system predicting 30-day all-cause mortality specifically for CS patients treated with ma-pLVAD. Methods Data were collected from an ongoing Impella registry including more than 830 patients since 2014. Among them, 460 patients received a ma-pLVAD in CS. Based on 53 clinically relevant parameters univariate logistic regressions on 30-day-mortality were performed. Subsequently, parameters with a p-value &amp;lt; 0.1 were included in a multivariate logistic regression analysis (MLRA) using backwards selection by likelihood ratio. The remaining continuous parameters underwent a correlation and regression tree analysis (CART) to define the optimal thresholds and MLRA was performed on all the selected parameters for calculation of odds ratios. These were used for score building. All-cause mortality was assessed by Kaplan-Meier method. Area under the curve (AUC) based on receiver operating characteristic (ROC) was estimated to determine the predictive value of the novel score for 30-day-mortality. Validation was performed on a timely independent cohort of CS patients treated with ma-pLVAD. Results Patients were 68±1 years old with predominance of male gender (74.4% [n=342]). CS was mainly caused by acute myocardial infarction (AMI) (60% [n=276]). A cardio-pulmonary resuscitation before ma-pLVAD was performed in 49.0% [n=215]. Initial serum lactate was 8.6±0.3 mmol/l. All-cause mortality at 30 days was 65.1±2.2% (Fig. 1A). A model of 6 parameters resulted after first MLRA with following odds ratios provided in Tab. 1A. CART analysis revealed optimal thresholds for age with 68 a [Gini-coefficient (G)=0.1], for initial serum lactate with 6.5 mmol/l [G=0.042], thrombocytes with 210 Gpt/l [G=0.021], and mean arterial pressure (MAP) with 70 and 90 mmHg [G1=0.027; G2=0.008]. Results of MLRA on the 6 parameters including the determined thresholds are given in Tab. 1B. A 0 to 17 point risk score resulted by using the odds ratios (Tab. 1C). ROC analysis indicated strong predictive capability for 30-day mortality (ROC AUC [95% CI] 0.767 [0.722-0.811]; p&amp;lt;0.001, Fig. 1B), with discrimination details provided in Tab. 1D. In the validation cohort, the score demonstrated favorable performance (ROC AUC [95% CI] 0.793 [0.607-0.979]; p=0.016; n=24). Conclusion We developed an internally validated, clinically feasible scoring system to estimate 30-day-mortality of CS patients treated with ma-pLVAD. Further studies are mandatory for external validation.

Association of 5-times sit-to-stand test with muscle weakness and prognosis in older patients with cardiovascular disease

Abstract Background/Introduction In the elderly with cardiovascular disease (CVD), maintaining muscle strength and physical function is essential for preventing adverse outcomes like falls and hospitalizations. The decline in lower limb strength signals increased morbidity and mortality risks. The 5-times sit-to-stand test (5STS) is a practical method for assessing lower limb strength, yet its predictive value for muscle weakness and prognosis in older CVD patients is underexplored. Purpose This study aims to elucidate the 5STS's role in identifying muscle weakness and its impact on the prognosis of older patients with CVD, offering insights for targeted interventions. Methods We included 3585 patients aged ≥ 65 years with CVD (mean age, 75.8 ± 6.6 years, 64.5% male) who underwent assessment of the 5STS before discharge. Maximal quadriceps isometric strength (QIS) was evaluated as a marker of lower limb strength and expressed relative to body weight (% body weight: BW). The primary outcome was all-cause mortality. We used the area under the curve (AUC) of the receiver operating characteristic (ROC) curve for each sex to calculate the cut-off values of the 5STS for low QIS (male &amp;lt; 45%BW, female &amp;lt; 35%BW), which have been reported to be associated with prognosis. In addition, we performed Kaplan-Meier curves and Cox regression analyses in five groups defined by 5STS time quartile and 5STS inability to assess the prognostic significance of 5STS. Cox regression analysis was performed as a multivariate model including the following variables; age, sex, body mass index, acute coronary syndrome, heart failure, hypertension, diabetes, dyslipidemia, current smoker, prior myocardial infarction, prior heart failure, prior chronic obstructive pulmonary disease, prior dementia, albumin, hemoglobin, estimated glomerular filtration rate, C-reactive protein. Result Over a median follow-up of 2.23 years (interquartile range 0.87-4.52 years), 657 (18.3%) patients died. Kaplan-Meier analysis indicated that both slow 5STS performance and inability to execute the 5STS were significantly linked with poorer prognosis (log-rank p &amp;lt; 0.001, Figure 1). After comprehensive adjustment, the slowest quartile and inability to perform the 5STS were associated with significantly higher mortality rates compared to the fastest quartile (HR 1.34; 95% CI: 1.05-1.72; p = 0.021; HR: 2.28; 95% CI: 1.60-3.24; p &amp;lt; 0.001, Figure 2). The ROC AUC for 5STS in identifying low QIS was 0.75 (95% CI: 0.73-0.77) for males and 0.75 (95% CI: 0.72-0.77) for females, with cut-off times established at 9.71s for males and 9.48s for females. Conclusion The 5STS, assessable in diverse settings without specialized equipment and reflecting key clinical outcomes, could serve as a standard tool for evaluating lower limb function.

Evaluation of a Proteomics-Guided Protein Signature for Breast Cancer Detection in Breast Tissue.

The distinction between noncancerous and cancerous breast tissues is challenging in clinical settings, and discovering new proteomics-based biomarkers remains underexplored. Through a pilot proteomic study (discovery cohort), we aimed to identify a protein signature indicative of breast cancer for subsequent validation using six published proteomics/transcriptomics data sets (validation cohorts). Sequential window acquisition of all theoretical (SWATH)-based mass spectrometry revealed 370 differentially abundant proteins between noncancerous tissue and breast cancer. Protein-protein interaction-based networks and enrichment analyses revealed dysregulation in pathways associated with extracellular matrix organization, platelet degranulation, the innate immune system, and RNA metabolism in breast cancer. Through multivariate unsupervised analysis, we identified a four-protein signature (OGN, LUM, DCN, and COL14A1) capable of distinguishing breast cancer. This dysregulation pattern was consistently verified across diverse proteomics and transcriptomics data sets. Dysregulation magnitude was notably higher in poor-prognosis breast cancer subtypes like Basal-Like and HER2 compared to Luminal A. Diagnostic evaluation (receiver operating characteristic (ROC) curves) of the signature in distinguishing breast cancer from noncancerous tissue revealed area under the curve (AUC) ranging from 0.87 to 0.9 with predictive accuracy of 80% to 82%. Upon stratifying, to solely include the Basal-Like/Triple-Negative subtype, the ROC AUC increased to 0.922-0.959 with predictive accuracy of 84.2%-89%. These findings suggest a potential role for the identified signature in distinguishing cancerous from noncancerous breast tissue, offering insights into enhancing diagnostic accuracy.

A comparative study of early stage Alzheimer’s disease classification using various transfer learning CNN frameworks

ABSTRACT The current research explores the improvements in predictive performance and computational efficiency that machine learning and deep learning methods have made over time. Specifically, the application of transfer learning concepts within Convolutional Neural Networks (CNNs) has proved useful for diagnosing and classifying the various stages of Alzheimer’s disease. Using base architectures such as Xception, InceptionResNetV2, DenseNet201, InceptionV3, ResNet50, and MobileNetV2, this study extends these models by adding batch normalization (BN), dropout, and dense layers. These enhancements improve the model’s effectiveness and precision in addressing the specified medical issue. The proposed model is rigorously validated and evaluated using publicly available Kaggle MRI Alzheimer’s data consisting of 1280 testing images and 5120 patient training images. For comprehensive performance evaluation, precision, recall, F1-score, and accuracy metrics are utilized. The findings indicate that the Xception method is the most promising of those considered. Without employing five K-fold techniques, this model obtains a 99% accuracy and 0.135 loss score. In addition, integrating five K-fold methods enhances the accuracy to 99.68% while decreasing the loss score to 0.120. The research further included the evaluation of the Receiver Operating Characteristic Area Under the Curve (ROC-AUC) for various classes and models. As a result, our model may detect and diagnose Alzheimer’s disease quickly and accurately.

An effective ensemble electricity theft detection algorithm for smart grid

AbstractSeveral machine learning and deep learning algorithms have been presented to detect the criminal behaviours in a smart grid environment in recent studies because of many successful results. However, most learning algorithms for the electricity theft detection have their pros and cons; hence, a critical research issue nowadays has been how to develop an effective detection algorithm that leverages the strengths of different learning algorithms. To demonstrate the performance of such an integrated detection model, the algorithm proposed first builds on deep neural networks, a meta‐learner for determining the weights of detection models for the construction of an ensemble detection algorithm and then uses a promising metaheuristic algorithm named search economics to optimise the hyperparameters of the meta‐learner. Experimental results show that the proposed algorithm is able to find better results and outperforms all the other state‐of‐the‐art detection algorithms for electricity theft detection compared in terms of the accuracy, F1‐score, area under the curve of precision‐recall (AUC‐PR), and area under the curve of receiver operating characteristic (AUC‐ROC). Since the results show that the meta‐learner of the proposed algorithm can improve the accuracy of deep learning algorithms, the authors expect that it will be used in other deep learning‐based applications.

Simple, Effective and Validated. VTE CASE Risk Assessment Score for Venous Thromboembolism in Metastatic Germ Cell Tumour Patients Before First-Line Chemotherapy.

Venous thromboembolism (VTE) may jeopardise excellent treatment results of germ cell tumours (GCT). We previously constructed a VTE risk score for GCT patients qualified for first-line chemotherapy (CTH), including vein compression, clinical stage (CS) and haemoglobin concentration. Validating our score in a separate cohort and establishing the cut-off point for the score. Re-assessing the numerical score in the training cohort. We retrospectively analysed a new cohort of GCT patients staged IS-IIIC. Area under the curve of receiver-operating characteristic (AUC-ROC) was calculated for the developed score, Khorana Risk Score (KRS) and Padua Prediction Score (PPS). AUC-ROC of the integer score was calculated for the training cohort. Cut-off point was established by Youden's and Liu's indices. Among 336 eligible patients in the validation cohort, VTE occurred in 41 (12.2%). AUC-ROC for our score, KRS and PPS were 0.818 (95% confidence interval (CI): 0.746-0.891), 0.608 (0.529-0.688) and 0.634 (0.547-0.720), respectively, p < 0.001. The optimal cut-off point for a low/high risk was 6 (≤ 6 vs. ≥ 7). In the training cohort, 369 patients had complete data on vein compression. AUC-ROC for our score, KRS and PPS were 0.819 (95% CI: 0.758-0.879), 0.710 (0.637-0.782) and 0.725 (0.651-0.800), p ≤ 0.001 and 0.015, respectively. Positive and negative predictive values were 30.8% and 96.5%, respectively. Our VTE risk score is a handy tool for GCT patients before first-line CTH for metastatic disease. Outperforming KRS and PPS, it has a good discriminatory value, especially for identifying low-risk patients.

Establishment and validation of a prediction model for gastric cancer with perineural invasion based on preoperative inflammatory markers.

Gastric cancer (GC) is a prevalent malignant tumor of the digestive system, characterized by a poor prognosis and high recurrence rate. Perineural invasion (PNI), the neoplastic infiltration of nerves, is a significant predictor of survival outcome in GC. Accurate preoperative identification of PNI could facilitate patient stratification and optimal preoperative treatment. We therefore established and validated a preoperative risk assessment model for GC patients with PNI. We collected data from 1,195 patients who underwent surgical resection at our hospital between October 2020 and December 2023, with PNI confirmed by pathological examination. We gathered laboratory data, including blood cell count, blood type, coagulation index, biochemical indexes, and tumor markers. Eligible patients were randomly divided into a training set and a testing set at a ratio of 7:3. The important risk factors of PNI were evaluated by random forest package in RStudio. Receiver operating characteristic-area under the curve (ROC-AUC) analysis was used to evaluate the discriminatory ability of the factors for PNI. Univariate and multivariate logistic regression analyses were utilized to verity independent risk factors for patients with PNI, and the logistic regression model and nomogram were constructed based on the results. Calibration curve and decision curve analysis (DCA) were conducted to assess the predictive model. Finally, we verified the prediction equation model using the testing set. In the training set, 416 GC patients were pathologically diagnosed with PNI. The top 5 important risk factors for PNI were identified as carcinoembryonic antigen (CEA), fibrinogen-to-lymphocyte ratio (FLR), D-dimer, platelet-to-lymphocyte ratio (PLR), and carbohydrate antigen 19-9 (CA19-9), with optimal cut-off values of 3.89 ng/mL, 2.08, 0.24 mg/L, 122.37, and 14.85 U/mL, respectively. Multivariate logistic regression analysis confirmed that CEA, FLR, D-dimer, PLR, CA19-9, and CA72-4 as independent risk factors for PNI (P<0.05). We formulated the following predictive equation: Logit(P) = -1.211 + 0.695 × CEA + 0.546 × FLR + 0.686 × D-dimer + 0.653 × PLR + 0.515 × CA19-9 + 0.518 × CA72-4 (χ2=105.675, P<0.001). The model demonstrated an ROC-AUC value of 0.719 [95% confidence interval (CI): 0.681-0.757] in the training set, with a sensitivity of 68.51% and a specificity of 67.60%. The ROC-AUC value was 0.791 (95% CI: 0.750-0.831) in the testing set (sensitivity: 69.57%, specificity: 56.41%). Calibration curve and DCA confirmed that the model has good discrimination and accuracy. We successfully established and validated a prediction model for GC patients with PNI based on hematological indicators, hoping that this model can provide an adjunctive tool for predicting PNI in clinical work.

Beyond the distance between juridic and scientific decision-making process: An evidence-based prediction algorithm for professional liability assessment in healthcare-associated infections

BackgroundHAIs (Healthcare-Acquired-Infections) have been recently the subject of judgment n. 6386 pronounced on 3rd March 2023 by the Italian Supreme Court. This sentence provided three criteria to determine whether a health facility is responsible for the patient contracting a nosocomial infection, i.e. time criterion, topographical criterion and clinical criterion. Accordingly, the healthcare facility is obliged to prove the fulfillment of a series of preventive hygiene measures specifically detailed by the legislator. Herein, the positive predictive value of these criteria (“juridic criteria”) in the identification of professional liability for nosocomial infections was evaluated in comparison with clinical criteria reviewed by Infectious Disease specialists (“Infectious-Disease criteria”, i.e. presence of a Multidrug Resistant Organism (MDRO); development of surgical site infection; inadequate antibiotic therapy; inadequate disinfection). MethodsTwo retrospective cohorts were compared from the Portal of Telematic Services of the Ministry of Justice; 51 patients were extrapolated from Italian judgments concerning claims for Gram-negative nosocomial infections in the three-year period 2020–2022. On the other side, from the electronic database of University Hospital of Bari we extracted 349 patients affected by Gram-negative infections in the same timespan. Both “juridic” criteria and “Infectious-Disease” criteria were then applied to the full cohort after stratification for cohort of origin and after stratification for nosocomial or non-nosocomial infections. Predictive value of criteria was evaluated through receiver operating characteristic (ROC) curves and area under the curve (AUC). ResultsOverall, the incidence of definite nosocomial infections (according to final judgement or clinical records discharge letter) was 84 % in juridic cohort and 46 % in “real-world” series.Data suggested that the presence of all three juridic criteria [ROC AUC = 0.944 (95%CI = 0.924–0.963)] or the four clinical criteria [ROC AUC = 0.948 (95%CI = 0.928–0.969)] predicted well a case of nosocomial infection with professional liability.Moreover, by summarizing both criteria in a single classification system, the generated ROC curve (was the one with the highest AUC [0.9488 (95%CI = 0.928–0.969)].Accordingly, further tests were performed, evaluating the predictive value of one juridic criterium plus at one of more Infectious-Disease criteria. Interestingly, the ROCs curves demonstrated that the presence of at least 1 juridic criteria plus at least 2 Infectious Disease criteria reached a predictive value comparable to 2 or 3 juridic criteria. ConclusionsThe results highlight the efficiency of new criteria laid down in the judgment of the Italian Supreme Court to attribute liability for nosocomial infection despite the disputed distance between juridic and scientific decision-making process. In addition, the use of a combined score combining “juridic” and “Infectious-Disease” criteria provides a high-quality tool to be used by technical consultants to make up for lack of clinical documentation by passing judgments concerning litigation about professional liability in case of nosocomial infections. This sheds light on the possibility to face worldwide judicial inquiries with scientific rigor.

Evaluating Observer Reliability and Diagnostic Accuracy of CT-LEFAT Criteria for Post-Treatment Head and Neck Lymphedema: A Prospective Blinded Comparative Analysis of Oncologist Human Inter-Rater Performance.

Radiation-associated lymphedema and fibrosis (LEF) is a significant toxicity following radiation therapy (RT) for head and neck cancer (HNC) patients. Recently, the CT Lymphedema and Fibrosis Assessment Tool (CT-LEFAT) was developed to standardize LEF diagnosis through fat stranding visualized on CT. This study aims to evaluate the inter-observer reliability and diagnostic accuracy of the CT-LEFAT criteria. This study retrospectively evaluated 26 HNC patients treated with RT that received a minimum of two contrast-enhanced CT scans. Qualitative review was conducted by five physician raters to assess the fat stranding observed on CT according to the CT-LEFAT criteria. Fleiss' kappa analysis was used to assess the inter- and intra-rater reliability, and Receiver Operating Characteristic (ROC) Area Under the Curve (AUC) analysis was used to evaluate diagnostic accuracy. The inter-rater reliability across the six CT-LEFAT regions generally indicated a slight to fair agreement across all raters (0.04 ≤ kappa ≤ 0.36). Intra-observer agreement was generally fair to moderate (overall kappa=0.44). The ROC AUC analysis varied based on aggregation method used (0.60 ≤ average AUC ≤ 0.70). This specific use-case evaluating CT-LEFAT criteria displays limited performance. This suggests that additional materials, such as further training, refinement of imaging methods, or other processes may be required before achieving clinically-ready diagnostic performance of LEF diagnosis.

In vitro and In silico Antibacterial Evaluation of N-Methyl-2-phenylmaleimides

Background: Novel antibiotics are needed to stem the rise of antimicrobial resistance. N-Methyl-2-phenylmaleimide (NMP) compounds previously synthesised by our research group are structural analogues of 2,3,5-substituted perhydropyrrolo[3,4-d]isoxazole-4,6-diones found by others to have antibacterial activity. Objectives: This study aims to explain the significance of NMPs and their antibacterial activity. The antibacterial activity of the NMPs was determined against Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, and Pseudomonas aeruginosa. The partition coefficient of the NMPs and a pharmacophore model were used to explain their antibacterial activity. Methods: The Kirby Bauer Disc diffusion method was used to screen the NMPs for activity, while the broth microdilution method was used to determine the minimum inhibitory concentration (MIC) and minimum bactericidal concentration (MBC) of the active NMPs. Using the in vitro antibacterial activity of 2,3,5-substituted perhydropyrrolo[3,4-d]isoxazole-4,6-diones, a common feature pharmacophore model was constructed and validated. The rank score, fit value, enrichment factor (EF20%), and receiver operating characteristic area under the curve (ROC-AUC) were used as validation metrics. Results: The NMPs were only active against S. aureus, with compound 3 (4 µg/ml) being the most active. The majority of NMPs were bacteriostatic. A common feature pharmacophore model was validated (rank score: 120.5; fit value: 4; EF20%: 4.3; ROC-AUC: 0.9 ± 0.03) and showed that three hydrogen bond acceptors and a ring aromatic region are important for activity. Comparing the partition coefficient of the NMPs to their MIC a statistically significant correlation was found. Conclusion: NMPs can be used as lead compounds in future studies. The validated pharmacophore model and partition coefficient can be used to develop more active compounds.

MIDF-DMAP: Multimodal information dynamic fusion for drug molecule activity prediction

BackgroundComputer-aided drug development can alleviate limitations in drug research and development processes, such as long cycles, high costs, and limited targeting. Deep learning models can be used to predict the activity of drug compound molecules. However, existing models still need to adequately address semantic redundancy, conflicts, and noise between different data modalities. It results in low accuracy in predicting molecular activity and many false positive results that need further resolution. ObjectiveThis study aims to design a parallel deep learning model for processing multimodal molecular information fusion, dynamically adjusting the weights of different modal features during the fusion process. This approach ensures that high-quality modal information is more significant in molecular activity prediction, improving the model’s predictive accuracy. MethodsFirst, three modal features of drug molecules are extracted: one-dimensional fingerprints, two-dimensional topological structures, and three-dimensional geometric structures. Then, a dynamic weighting mechanism is proposed to fuse these three modal features organically. Finally, the fused features are input into the model classifier to obtain molecular activity prediction results. ResultsExperimental results show that the proposed model improved the Receiver Operating Characteristic-Area Under the Curve (ROC-AUC) value by 29.96% compared to the Weave model and 26.38% compared to the GC model on the BBBP dataset. Compared to the GraphMVP model and the multi-channel substructure graph model MSGG, the ROC-AUC value increased by 20.44% and 15.80%, respectively. This method also achieved good results on imbalanced samples and small molecule datasets. ConclusionThis study provides an effective multimodal dynamic fusion method for virtual drug screening, enriching computer-aided drug development theory and improving drug research and development efficiency.

Impact of Bacterial Etiology on Procalcitonin, C-reactive Protein and Hematological Parameters: Evaluating Mean Platelet Volume for Differentiating Gram-Negative and Gram-Positive Bacteria in Odontogenic Versus Non-odontogenic Head and Neck Abscesses.

Introduction Head and neck abscesses, which can originate from odontogenic or non-odontogenic sources, pose significant diagnostic challenges due to their diverse bacterial etiologies. This study aims to investigate the impact of bacterial etiology on procalcitonin (PCT), C-reactive protein (CRP), and various hematological parameters, and to assess the diagnostic performance of mean platelet volume (MPV) in differentiating between Gram-negativebacteria (GNB) and Gram-positive bacteria (GPB) in adults with odontogenic and non-odontogenic head and neck abscesses. Materials and methods Our retrospective analysis of a prospective study comprised 80 patients: 50 individuals (56% men, average age 41.6±18.18 years) with odontogenic and 30 patients (66.7% men, average age 44.53±15.49 years) with non-odontogenic head and neck abscesses during the period from July 2021 to June 2022. White blood cell count (WBC); neutrophil (Neu) and lymphocyte (Ly) count; MPV, and platelet count (PLT) were derived from the results of complete blood count. MPV/PLT (MPI) was calculated by dividing MPV by PLT. CRP levels (mg/l) were quantified via immunoturbidimetric analysis utilizing latex-enhanced particlesand PCT levels (ng/ml) by latex-enhanced immunoturbidimetric assay. Results In 25 (31.3%) of all 80 patients, no microorganisms were isolated (sterile cultures); in 28 (35%) resident microflora were isolated; in seven (8.8%) GNB were isolated; and in 17 (21.3%) GPB were isolated. CRP and Neu were significantly higher in patients with odontogenic abscesses compared to non-odontogenic ones. PLT and PCT were lower in patients with odontogenic abscesses vs those with non-odontogenic abscesses. Additionally, according to bacterial type, MPV, MPI and PCT were significantly higher in GPB compared to GNB. WBC, Neu and PLT were higher in patients with GNB vs GPB. Significant correlations were found between MPV andLy, and between MPV and Neu, regardless of the abscess origin or etiological factor. MPI exhibited an area under the curve of the receiver operating characteristic (AUC-ROC)=0.776, MPV of 0.541, and PCT of 0.568in distinguishing patients with GPB from GNB. A cut-off value of 0.029 was derived for MPI (70.6% sensitivity and 80% specificity). Conclusions This study highlights the impact of bacterial etiology on inflammatory and hematological markers in head and neck abscesses. Odontogenic abscesses showed higher CRP and Neu, indicating a stronger inflammatory response, while non-odontogenic abscesses had higher PLT, Ly, and PCT. MPI proved to be a more effective diagnostic marker (cut-off value of 0.029) than MPV or PCT for distinguishing between GPB and GNB, suggesting its valuable role in clinical practice for accurate and timely diagnosis.

Raman spectroscopy combined with machine learning and chemometrics analyses as a tool for identification atherosclerotic carotid stenosis from serum

Atherosclerosis carotid stenosis (ACS) is one of the main causes of stroke. Unfortunately, the highest number of people go to the doctor with an advanced disease or as a result of a stroke, because carotid atherosclerosis does not cause obvious symptoms. Therefore, it is important to find a diagnostic method to detect the disease during routine tests (using blood or serum). Consequently, in this article, Raman spectroscopy was tested as a potential diagnostic method. Indeed, Raman spectra of serum collected from ACS and control patients showed decrease of Raman peak around 1520 cm−1 and increase of peak around 3050 cm−1 in people with ACS. Moreover in people with ACS shift of peaks originating from amides II, I and lipids vibrations were noticed in comparison with control group. Interestingly, decision tree algorithm showed that peaks at 1656 cm−1 and 2957 cm−1 could be a spectroscopy markers of atherosclerotic carotid stenosis. Continuing, Principal Component Analysis (PCA) clearly showed distinguishing between serum collected from ACS and control patients, while machine learning algorithms showed high value of accuracy, sensitivity and selectivity (more than 90 %). Finally, value of area under the curve of Receiver Operating Characteristic (AUC-ROC) showed value of 0.81 for Raman range between 800 cm−1 and 1800 cm−1 and 0.86 for 2800 cm−1–3000 cm−1 range. Obtained results clearly showed possibility of Raman spectroscopy in detection of ACS from serum.