Accuracy Research Articles

Does muscle strength predict working memory? A cross-sectional fNIRS study in older adults.

Previous research has primarily focused on the association between muscle strength and global cognitive function in older adults, while the connection between muscle strength and advanced cognitive function such as inhibition and working memory (WM) remains unclear. This study aimed to investigate the relationship among muscle strength, WM, and task-related cortex hemodynamics. We recruited eighty-one older adults. Muscle strength was measured using a grip and lower limb strength protocol. We measured the WM performance by using reaction time (RT) and accuracy (ACC) in the N-back task and the cortical hemodynamics of the prefrontal cortex (PFC) by functional near-infrared spectroscopy (fNIRS). We found positive correlations between grip strength (p < 0.05), 30-s sit-up (p < 0.05) and ACC, negative correlation between grip strength (p < 0.05) and RT. Furthermore, we observed positive correlations between grip strength and the level of oxygenated hemoglobin (HbO2) in dorsolateral prefrontal cortex, frontopolar area, ventrolateral prefrontal cortex (p < 0.05), and negative correlations between grip strength and the level of deoxygenated hemoglobin (Hb) in left dorsolateral prefrontal cortex, frontopolar area, left ventrolateral prefrontal cortex (p < 0.05). Additionally, we noticed positive correlations between RT and the level of Hb in left dorsolateral prefrontal cortex, right frontopolar area (p < 0.05), and negative correlations between RT and the level of HbO2 in left dorsolateral prefrontal cortex, frontopolar area (p < 0.05). However, the cortical hemodynamics did not mediate the relationship between muscle strength and WM performance (RT, ACC). The grip strength of older adults predicted WM in the cross-section study. The level of hemodynamics in PFC can serve as a predictor of WM.

Ethical Principles Do Not Support Mandatory Preanesthesia Pregnancy Screening Tests: A Narrative Review.

Respect for patient autonomy is a pillar of medical ethics, manifested predominantly through informed consent. Mandatory (routine) nonconsented preoperative urine pregnancy testing does not adequately respect patient autonomy, is potentially coercive, and has the potential to cause harm medically, psychologically, socially, and financially. Inaccuracies in pregnancy testing can result in false-positive and false-negative results, especially in early pregnancy. There is substantial scientific evidence that anesthesia is not harmful to the fetus, raising the question of whether pregnancy testing provides substantial benefit to the patient. Not performing a preanesthesia pregnancy test has not been associated with significant medicolegal consequences. We review the ethical implications of mandatory preanesthesia pregnancy testing in light of these facts.

Proteomic profiling identifies novel inflammation-related plasma proteins associated with ischemic stroke outcome

BackgroundThe inflammatory response to cerebral ischemia is complex; however, most clinical studies of stroke outcome focus on a few selected proteins. We, therefore, aimed to profile a broad range of inflammation-related proteins to: identify proteins associated with ischemic stroke outcome that are independent of established clinical predictors; identify proteins subsets for outcome prediction; and perform sex and etiological subtype stratified analyses.MethodsAcute-phase plasma levels of 65 inflammation-related proteins were measured in 534 ischemic stroke cases. Logistic regression was used to estimate associations to unfavorable 3-month functional outcome (modified Rankin Scale score > 2) and LASSO regressions to identify proteins with independent effects.ResultsTwenty proteins were associated with outcome in univariable models after correction for multiple testing (FDR < 0.05), and for 5 the association was independent of clinical variables, including stroke severity (TNFSF14 [LIGHT], OSM, SIRT2, STAMBP, and 4E-BP1). LASSO identified 9 proteins that could best separate favorable and unfavorable outcome with a predicted diagnostic accuracy (AUC) of 0.81; three associated with favorable (CCL25, TRAIL [TNFSF10], and Flt3L) and 6 with unfavorable outcome (CSF-1, EN-RAGE [S100A12], HGF, IL-6, OSM, and TNFSF14). Finally, we identified sex- and etiologic subtype-specific associations with the best discriminative ability achieved for cardioembolic, followed by cryptogenic stroke.ConclusionsWe identified candidate blood-based protein biomarkers for post-stroke functional outcome involved in, e.g., NLRP3 inflammasome regulation and signaling pathways, such as TNF, JAK/STAT, MAPK, and NF-κB. These proteins warrant further study for stroke outcome prediction as well as investigations into the putative causal role for stroke outcome.

MV-STCNet: Breast cancer diagnosis using spatial and temporal dual-attention guided classification network based on multi-view ultrasound videos

Breast cancer is one of the most prevalent cancers among women worldwide. Breast ultrasound (US) video is a common means of clinical diagnosis and contains rich information about lesions. However, information from just a single view may lead to a higher rate of misdiagnosis. Also, the size, spatial locations, and time of occurrence of lesions vary widely between patients. Meanwhile, there is plentiful redundant information in the video. To address these issues, we propose a novel dual-branch classification model based on US videos. This model can extract lesion information in both the transverse and longitudinal planes and combine the diagnostic results of the two branches to diagnose a lesion. Additionally, we design a region-guided module (RGM) and time-guided module (TGM) to mitigate the impact of redundant information on the network. The TGM simulates radiologists’ temporal attention in the diagnosis process as a guiding message, helping the network focus on frames that are more beneficial to the diagnosis. The RGM considers the roughly located lesion region as a priori information to guide the network to focus on the lesion region, which is helpful in extracting and strengthening texture features for breast cancer classification. We validate the effectiveness of our model using our collected US video dataset, which comprises 1000 cases obtained from 888 patients. The results show that the accuracy (ACC), area under the ROC curve (AUC), and F1-score of the model are 86.40%, 92.08%, and 82.39%, respectively, and its performance is better than that of other state-of-the-art models.

Value of Molybdenum Target X-Ray and High-Frequency Color Doppler Flow Imaging in Early Diagnosis of Breast Carcinoma: A Comparative Analysis.

Breast carcinoma (BC) threatens the physical and mental health of women worldwide, and early diagnosis is important for improving patient outcomes and ensuring successful treatment. This research mainly aims to compare and analyze the value of molybdenum target X-ray and high-frequency color Doppler flow imaging (CDFI) in the early diagnosis of BC. First, 102 patients with suspected early-stage BC (ESBC) admitted to Henan Provincial People's Hospital were examined by molybdenum target X-ray and CDFI. Based on the pathological findings, the diagnostic efficiency data of the two diagnostic modalities such as positive detection rate (PDR), positive predictive value (PPV), negative predictive value (NPV), sensitivity (SEN), specificity (SPE), and accuracy (ACC), as well as imaging information like masses, microcalcifications (MCs), axillary lymph node (LN) metastases, and blood flow signal or vascular sign abnormalities were analyzed. CDFI contributed to higher PDR, PRV, NPV, SEN, and ACC than molybdenum target X-ray in ESBC diagnosis, but similar SPE. The combined diagnosis of molybdenum target X-ray plus CDFI contributed to even higher PDR, PRV, NPV, SEN, and ACC than molybdenum target X-ray alone and higher ACC than CDFI. Imaging inspection revealed that the number of cases of masses, axillary LN metastases, and abnormalities in blood flow signals or vascular signs detected by CDFI was significantly higher than that by molybdenum target X-ray, while the number of MCs was significantly lower. Molybdenum target X-ray plus CDFI is more effective in the diagnosis of ESBC and plays a complementary role in imaging examination, which can synergistically improve the diagnostic ACC of ESBC and is worthy of clinical promotion.

An Intrusion Detection Model using election-Based Feature Selection and K-NN

The Internet of Things (IoT) is a harmonized embedded object and sensor set. It is a target of intrusions, which leads to considering the security of the IoT environment. Considering their vital role in network security, intrusion detection systems (IDS) have received significant interest in the research community. IDS can be divided into anomaly intrusion detection systems (AIDS) and signature intrusion detection systems (SIDS). In this work, we integrate AIDS to secure IoT environments better and enhance the privacy of data circulated within sensors and actuators which interact with the real world. However, a network intrusion detection system (NIDS) mitigates security issues. We present in this paper an effective NIDS for IoT security relying on machine learning such as K-NN and feature selection. Specifically, we designed our model using K-Nearest Neighbors (K-NN) algorithm to boost the performance and enhance the IDS accuracy (ACC), with feature selection algorithms separately, such as principal component analysis (PCA), genetic algorithm (GA) and univariate statistical test. We elected every feature that two or three methods have selected. Hence, our election-based feature selection method lasted with five features where the model has shown promising results in ACC, false alarm rate (FAR), and detection rate (DR). To evaluate the performance of our model on imbalanced datasets, we ran our tests on the Bot-IoT dataset. We measured the model using Matthew's Correlation Coefficient (MCC) to have a better performance evaluation, where it scored up to 97%. The suggested model resulted in 99.99% ACC and considerably changed the training time from 21696 seconds for the entire dataset to 102 seconds for the five selected features. Our proposed model provides significant performances compared with previous models

A rapid and low-cost method for detection of nine kinds of vegetable oil adulteration based on 3-D fluorescence spectroscopy

The integrity of vegetable oil has become a focal concern within the global food industry, particularly in specific regions where economic interests drive authenticity issues. In this study, we used three-dimensional (3-D) fluorescence spectroscopy for the qualitative and quantitative analysis of adulteration in different vegetable oils, offering the advantages of rapid and low-cost detection. Nine types of adulterated vegetable oils were modeled using five machine learning methods: K-nearest neighbor (KNN), Random Forest (RF), Support Vector Machine (SVM), Partial Least Squares (PLS), and Convolutional Neural Network (CNN). In the qualitative analysis, the Rp2 and RMSEP values ranged from 0.89 to 0.99 and 0.01 to 0.05, respectively. For qualitatively analysis, a threshold of 5 % was established for detecting adulteration in soybean oil with palm oil and sesame oil with palm oil, achieving an accuracy (ACC) greater than 90 %. Meanwhile, a threshold of 10 % was effective for remaining seven adulterated oils, with an ACC exceeding 90 %. Moreover, an ACC rate of over 95 % was attained when the threshold was set at 15 %. The findings demonstrate that 3-D fluorescence spectroscopy presents a robust tool for detecting various adulterated oils, heralding a valuable contribution to food safety and quality control.

Artificial intelligence-based prediction of molecular and genetic markers for hepatitis C-related hepatocellular carcinoma.

Hepatocellular carcinoma (HCC) is the main cause of mortality from cancer globally. This paper intends to classify public gene expression data of patients with Hepatitis C virus-related HCC (HCV+HCC) and chronic HCV without HCC (HCV alone) through the XGboost approach and to identify key genes that may be responsible for HCC. The current research is a retrospective case-control study. Public data from 17 patients with HCV+HCC and 35 patients with HCV-alone samples were used in this study. An XGboost model was established for the classification by 10-fold cross-validation. Accuracy (AC), balanced accuracy (BAC), sensitivity, specificity, positive predictive value, negative predictive value, and F1 score were utilized for performance assessment. AC, BAC, sensitivity, specificity, positive predictive value, negative predictive value, and F1 scores from the XGboost model were 98.1, 97.1, 100, 94.1, 97.2, 100, and 98.6%, respectively. According to the variable importance values from the XGboost, the HAO2, TOMM20, GPC3, and PSMB4 genes can be considered potential biomarkers for HCV-related HCC. A machine learning-based prediction method discovered genes that potentially serve as biomarkers for HCV-related HCC. After clinical confirmation of the acquired genes in the following medical study, their therapeutic use can be established. Additionally, more detailed clinical works are needed to substantiate the significant conclusions in the current study.

End-to-end deep learning radiomics: development and validation of a novel attention-based aggregate convolutional neural network to distinguish breast diffuse large B-cell lymphoma from breast invasive ductal carcinoma.

Apart from invasive pathological examination, there is no effective method to differentiate breast diffuse large B-cell lymphoma (DLBCL) from breast invasive ductal carcinoma (IDC). In this study, we aimed to develop and validate an effective deep learning radiomics model to discriminate between DLBCL and IDC. A total of 324 breast nodules from 236 patients with baseline 18F-fluorodeoxyglucose (18F-FDG) positron emission tomography/computed tomography (PET/CT) were retrospectively analyzed. After grouping breast DLBCL and breast IDC patients, external and internal datasets were divided according to the data collected by different centers. Preprocessing was then used to process the original PET/CT images and an attention-based aggregate convolutional neural network (AACNN) model was designed. The AACNN model was trained using patches of CT or PET tumor images and optimized with an improved loss function. The final ensemble predictive model was built using distance weight voting. Finally, the model performance was evaluated and statistically verified. A total of 249 breast nodules from Fudan University Shanghai Cancer Center (FUSCC) and 75 breast nodules from Shanghai Proton and Heavy Ion Center (SPHIC) were selected as internal and external datasets, respectively. On the internal testing, our method yielded an area under the curve (AUC), accuracy (ACC), sensitivity (SEN), specificity (SPE), positive predictive value (PPV), negative predictive value (NPV), and harmonic mean of precision and sensitivity (F1) of 0.886, 83.0%, 80.9%, 85.0%, 84.8%, 81.2%, and 0.828, respectively. Meanwhile on the external testing, the results were 0.788, 71.6%, 61.4%, 84.7%, 84.0%, 62.6%, and 0.709, respectively. Our study outlines a deep learning radiomics method which can automatically, noninvasively, and accurately differentiate breast DLBCL from breast IDC, which will be more in line with the needs and strategies of precision medicine, individualized diagnosis, and treatment.

Applying data mining techniques to predict vitamin D deficiency in diabetic patients.

Vitamin D is among the vitamins necessary for both adults' and children's health. It plays a significant role in calcium absorption, the immune system, cell proliferation and differentiation, bone protection, skeletal health, rickets, muscle health, heart health, disease pathogenesis and severity, glucose metabolism, glucose intolerance, varying insulin secretion, and diabetes. Because the 25-hydroxyvitamin D (25OHD) test, which is used to measure vitamin D is expensive and may not be covered in healthcare benefits in many countries, this study aims to predict vitamin D deficiency in diabetic patients. The prediction method is based on data mining techniques combined with feature selection by using historical electronic health records. The results were compared with a filter-based feature selection algorithm, namely relief-F. Non-valuable features were eliminated effectively with the relief-F feature selection method without any performance loss in classification. The performances of the methods were evaluated using classification accuracy (ACC), sensitivity, specificity, F1-score, precision, kappa results, and receiver operating characteristic (ROC) curves. The analyses have been conducted on a vitamin D dataset of diabetic patients and the results show that the highest classification accuracy of 97.044% was obtained for the support vector machines (SVM) model using radial kernel that contains 18 features.

Machine Learning Algorithms for COPD Patients Readmission Prediction

Abstract: Patients' readmission might be seen as a crucial aspect in lowering costs while maintaining high-quality patient care. As a result, anticipating and reducing readmission rates for patients will considerably enhance healthcare delivery. The goal of this research is to use machine learning algorithms to predict readmission of COPD (Chronic Obstructive Pulmonary Disease) patients. The major metrics for measuring models' prediction capability in each time frame were Area under Curve (AUC) and Accuracy (ACC). Then, the factors' relevance for each result was clearly recognized, and specified key variables were discriminated. With%91 ACC, our research had the best accuracy in predicting readmission

Overall Survival Prediction in Stereotactic Radiosurgery Patients with Glioblastoma Via a Deep-Learning Approach

Accurate and automated early survival prediction is critical for glioblastoma (GBM) patients as their poor prognosis requires timely treatment decision-making. We have developed a deep learning (DL)-based GBM overall survival (OS) prediction model based on a multi-institutional public dataset using only pre-operative basic structural multi-parametric magnetic resonance images (MRIs). The purpose of this study is to evaluate this DL-based OS prediction model with an institutional stereotactic radiosurgery (SRS) clinical trial dataset. The task of this study is to classify GBM patients into 3 OS classes: long-survivors (>15 months), mid-survivors (between 10 and 15 months) and short-survivors (< 10 months). The proposed OS prediction model is an ensemble of a ResNet-based classifier and a K-NN classifier. The ResNet-based classifier is trained in a Siamese fashion to explore inter-class differences. During testing, training sample features are implemented with a K-NN classifier to ensemble with the ResNet-based classifier. A public dataset from Medical Image Computing and Computer Assisted Intervention (MICCAI) Brain Tumor Segmentation (BraTS) challenge 2020 (235 patients) were used for model establishing and initial validation. Then the validated model was evaluated on 19 GBM patients from an institutional SRS clinical trial. Each data entry consists of pre-operative basic structural multi-parametric MRIs and survival days, as well as patient ages for BraTS data and basic clinical characteristics for institutional data. GBM sub-regions, including contrast-enhancing tumor, peri-tumoral edema, and necrotic/non-enhancing tumor core, were segmented in the multi-parametric MRIs by an in-house DL model for both datasets. The OS prediction model was trained on 90% of the segmented BraTS data and validated on the rest 10%, then further evaluated on the institutional data. The model performance was assessed by prediction accuracy (ACC) and the area under the curve (AUC). For this 3-class OS classification task, our DL-based prediction model achieved an ACC of 65.22% and an AUC of 0.81 on the BraTS dataset compared with the top-ranked result from the BraTS challenge 2020 (Rank 1st: ACC 61.7%), and an ACC of 52.63% and an AUC of 0.69 on the institutional dataset. Further analysis of the institutional dataset found that the predicted OS class had a statistically significant correlation with treatment volume (p = 0.012) and age (p = 0.006), which matches the analysis that the patients' ground truth OS class is statistical significantly correlated with treatment volume (p = 0.045). Our DL-based OS prediction model for GBM using basic structural multi-parametric pre-operative MRIs has demonstrated promising performance in both public and institutional dataset with minimal manual processing requirements. This OS prediction model can be potentially applied to assist timely clinical decision-making.

Inter-patient ECG heartbeat classification for arrhythmia classification: a new approach of multi-layer perceptron with weight capsule and sequence-to-sequence combination.

Objective: The objective of this research is to construct a method to alleviate the problem of sample imbalance in classification, especially for arrhythmia classification. This approach can improve the performance of the model without using data enhancement. Methods: In this study, we have developed a new Multi-layer Perceptron (MLP) block and have used a Weight Capsule (WCapsule) network with MLP combined with sequence-to-sequence (Seq2Seq) network to classify arrhythmias. Our work is based on the MIT-BIH arrhythmia database, the original electrocardiogram (ECG) data is classified according to the criteria recommended by the American Association for Medical Instrumentation (AAMI). Also, our method's performance is further evaluated. Results: The proposed model is evaluated using the inter-patient paradigm. Our proposed method shows an accuracy (ACC) of 99.88% under sample imbalance. For Class N, sensitivity (SEN) is 99.79%, positive predictive value (PPV) is 99.90%, and specificity (SPEC) is 99.19%. For Class S, SEN is 97.66%, PPV is 96.14%, and SPEC is 99.85%. For Class V, SEN is 99.97%, PPV is 99.07%, and SPEC is 99.94%. For Class F, SEN is 97.94%, PPV is 98.70%, and SPEC is 99.99%. When using only half of the training sample, our method shows that the SEN of Class N and V is 0.97% and 5.27% higher than the traditional machine learning algorithm. Conclusion: The proposed method combines MLP, weight capsule network with Seq2seq network, effectively addresses the problem of sample imbalance in arrhythmia classification, and produces good performance. Our method also shows promising potential in less samples.

Self-supervised temporal autoencoder for egocentric action segmentation

Given an egocentric video, action temporal segmentation aims to temporally segment the video into basic units, each depicting an action. As the camera is constantly moving, some important objects may disappear in some consecutive frames and cause an abrupt change in the visual content. Recently works fail to deal with this condition in the absence of manually annotating abundant frames. In this study, we propose a temporal-aware clustering method for egocentric action temporal segmentation: a self-supervised temporal autoencoder (SSTAE). Instead of directly learning visual features, the SSTAE is implemented by encoding the preceding target frame and predicting subsequent frames in the temporal relationship domain, which takes into account the local temporal consistency. Our proposed algorithm is guided by the reconstruction and predicted losses. Consequently, local temporal contexts are naturally integrated into the feature representation, and a clustering step is performed. Experiments on three egocentric datasets demonstrate the our proposed approach outperforms the state-of-the-art methods by clustering Accuracy(ACC) 7.57%, Normalized Mutual Information(NMI) 8.17%, Adjusted Rand Index(ARI) 8.6%.

Left Ventricular Myocardial Dysfunction Evaluation in Thalassemia Patients Using Echocardiographic Radiomic Features and Machine Learning Algorithms

Heart failure caused by iron deposits in the myocardium is the primary cause of mortality in beta-thalassemia major patients. Cardiac magnetic resonance imaging (CMRI) T2* is the primary screening technique used to detect myocardial iron overload, but inherently bears some limitations. In this study, we aimed to differentiate beta-thalassemia major patients with myocardial iron overload from those without myocardial iron overload (detected by T2*CMRI) based on radiomic features extracted from echocardiography images and machine learning (ML) in patients with normal left ventricular ejection fraction (LVEF > 55%) in echocardiography. Out of 91 cases, 44 patients with thalassemia major with normal LVEF (> 55%) and T2* ≤ 20 ms and 47 people with LVEF > 55% and T2* > 20 ms as the control group were included in the study. Radiomic features were extracted for each end-systolic (ES) and end-diastolic (ED) image. Then, three feature selection (FS) methods and six different classifiers were used. The models were evaluated using various metrics, including the area under the ROC curve (AUC), accuracy (ACC), sensitivity (SEN), and specificity (SPE). Maximum relevance-minimum redundancy-eXtreme gradient boosting (MRMR-XGB) (AUC = 0.73, ACC = 0.73, SPE = 0.73, SEN = 0.73), ANOVA-MLP (AUC = 0.69, ACC = 0.69, SPE = 0.56, SEN = 0.83), and recursive feature elimination-K-nearest neighbors (RFE-KNN) (AUC = 0.65, ACC = 0.65, SPE = 0.64, SEN = 0.65) were the best models in ED, ES, and ED&ES datasets. Using radiomic features extracted from echocardiographic images and ML, it is feasible to predict cardiac problems caused by iron overload.

Volume-Assisted Estimation of Remnant Liver Function Based on Gd-EOB-DTPA Enhanced MR Relaxometry: A Prospective Observational Trial.

In the context of liver surgery, predicting postoperative liver dysfunction is essential. This study explored the potential of preoperative liver function assessment by MRI for predicting postoperative liver dysfunction and compared these results with the established indocyanine green (ICG) clearance test. This prospective study included patients undergoing liver resection with preoperative MRI planning. Liver function was quantified using T1 relaxometry and correlated with established liver function scores. The analysis revealed an improved model for predicting postoperative liver dysfunction, exhibiting an accuracy (ACC) of 0.79, surpassing the 0.70 of the preoperative ICG test, alongside a higher area under the curve (0.75). Notably, the proposed model also successfully predicted all cases of liver failure and showed potential in predicting liver synthesis dysfunction (ACC 0.78). This model showed promise in patient survival rates with a Hazard ratio of 0.87, underscoring its potential as a valuable tool for preoperative evaluation. The findings imply that MRI-based assessment of liver function can provide significant benefits in the early identification and management of patients at risk for postoperative liver dysfunction.

Different prioritization states of working memory representations affect visual searches: Evidence from an event-related potential study

Previous evidence has shown that the contents of working memory (WM) can bias visual selection. However, not much is known about how WM effects change when the WM representation is held in different prioritization states. Here, we investigated this problem using event-related potentials. Subjects maintained two colors in WM while performing a search task. One of the colors was retro-cued, indicating that it was 80 % likely to be the target of the memory test. During the search display, one of the distractors was a salient color singleton, and this singleton distractor could carry the same color as the cued WM representation, the uncued WM representation, or be irrelevant to the memory content. Behaviorally, the memory test performance was found to be better for the cued color than for the uncued color, and we observed lower search accuracy (ACC) and longer search reaction time (RT) when the singleton distractor matched the cued WM representation than when it matched an uncued or an irrelevant WM representation. Event-related potential (ERP) data showed that the P3 amplitude of cue-color distractor conditions was smaller than that of uncued-color distractor conditions and irrelevant-color distractor conditions. These findings clearly indicate that prioritizing an item for enhanced representational quality enables the item to bias attention to a greater extent.

Prediction of upcoming urinary tract infection after intracerebral hemorrhage: a machine learning approach based on statistics collected at multiple time points.

Accurate prediction of urinary tract infection (UTI) following intracerebral hemorrhage (ICH) can significantly facilitate both timely medical interventions and therapeutic decisions in neurocritical care. Our study aimed to propose a machine learning method to predict an upcoming UTI by using multi-time-point statistics. A total of 110 patients were identified from a neuro-intensive care unit in this research. Laboratory test results at two time points were chosen: Lab 1 collected at the time of admission and Lab 2 collected at the time of 48 h after admission. Univariate analysis was performed to investigate if there were statistical differences between the UTI group and the non-UTI group. Machine learning models were built with various combinations of selected features and evaluated with accuracy (ACC), sensitivity, specificity, and area under the curve (AUC) values. Corticosteroid usage (p < 0.001) and daily urinary volume (p < 0.001) were statistically significant risk factors for UTI. Moreover, there were statistical differences in laboratory test results between the UTI group and the non-UTI group at the two time points, as suggested by the univariate analysis. Among the machine learning models, the one incorporating clinical information and the rate of change in laboratory parameters outperformed the others. This model achieved ACC = 0.773, sensitivity = 0.785, specificity = 0.762, and AUC = 0.868 during training and 0.682, 0.685, 0.673, and 0.751 in the model test, respectively. The combination of clinical information and multi-time-point laboratory data can effectively predict upcoming UTIs after ICH in neurocritical care.

Machine learning-based diagnosis and risk classification of coronary artery disease using myocardial perfusion imaging SPECT: A radiomics study

This study aimed to investigate the diagnostic performance of machine learning-based radiomics analysis to diagnose coronary artery disease status and risk from rest/stress Myocardial Perfusion Imaging (MPI) single-photon emission computed tomography (SPECT). A total of 395 patients suspicious of coronary artery disease who underwent 2-day stress-rest protocol MPI SPECT were enrolled in this study. The left ventricle myocardium, excluding the cardiac cavity, was manually delineated on rest and stress images to define a volume of interest. Added to clinical features (age, sex, family history, diabetes status, smoking, and ejection fraction), a total of 118 radiomics features, were extracted from rest and stress MPI SPECT images to establish different feature sets, including Rest-, Stress-, Delta-, and Combined-radiomics (all together) feature sets. The data were randomly divided into 80% and 20% subsets for training and testing, respectively. The performance of classifiers built from combinations of three feature selections, and nine machine learning algorithms was evaluated for two different diagnostic tasks, including 1) normal/abnormal (no CAD vs. CAD) classification, and 2) low-risk/high-risk CAD classification. Different metrics, including the area under the ROC curve (AUC), accuracy (ACC), sensitivity (SEN), and specificity (SPE), were reported for models’ evaluation. Overall, models built on the Stress feature set (compared to other feature sets), and models to diagnose the second task (compared to task 1 models) revealed better performance. The Stress-mRMR-KNN (feature set-feature selection-classifier) reached the highest performance for task 1 with AUC, ACC, SEN, and SPE equal to 0.61, 0.63, 0.64, and 0.6, respectively. The Stress-Boruta-GB model achieved the highest performance for task 2 with AUC, ACC, SEN, and SPE of 0.79, 0.76, 0.75, and 0.76, respectively. Diabetes status from the clinical feature family, and dependence count non-uniformity normalized, from the NGLDM family, which is representative of non-uniformity in the region of interest were the most frequently selected features from stress feature set for CAD risk classification. This study revealed promising results for CAD risk classification using machine learning models built on MPI SPECT radiomics. The proposed models are helpful to alleviate the labor-intensive MPI SPECT interpretation process regarding CAD status and can potentially expedite the diagnostic process.

Optimization of a Fabric Phase Sorptive Extraction protocol for the isolation of six bisphenols from juice pouches to be analysed by high performance liquid chromatography coupled with diode array detector

Fabric Phase Sorptive Extraction (FPSE) combined with high pressure liquid chromatography using to diode array detection (HPLC-DAD) was applied for the simultaneous determination of bisphenols (BPA, BPB, BPC, BPE, BPF, BPS) in juice pouches. The FPSE procedure was optimized with regards to the critical parameters that affect the performance of the method including the selection of the FPSE membrane type and size, adsorption time, extraction time, solvent volume desorption, magnetic stirring ratio, and salt addition. The FPSE membrane could be reused up to 14 times. The developed FPSE-HPLC-DAD method was validated in terms of linearity, sensitivity, accuracy andprecision. The limits of detection (LODs) were lower than 6.9 ng/mL, while the limits of quantification (LOQs) were lower than 21 ng/mL. The results obtained are satisfactory in terms of precision, accuracy and repeatability, with recoveries above 86% and CV values below 9.5%. The FPSE-HPLC-DAD method was successfully applied in the determination of six bisphenols in juice samples stored in pouches.