Separate Set Research Articles

Prediction of Hypoglycemia From Continuous Glucose Monitoring in Insulin-Treated Patients With Type 2 Diabetes Using Transfer Learning on Type 1 Diabetes Data: A Deep Transfer Learning Approach.

Hypoglycemia is common in insulin-treated type 2 diabetes (T2D) patients, which can lead to decreased quality of life or premature death. Deep learning models offer promise of accurate predictions, but data scarcity poses a challenge. This study aims to develop a deep learning model utilizing transfer learning to predict hypoglycemia. Continuous glucose monitoring (CGM) data from 226 patients with type 1 diabetes (T1D) and 180 patients with T2D were utilized. Data were structured into one-hour samples and labeled as hypoglycemia or not depending on whether three consecutive CGM values were below 3.9 [mmol/L] (70 mg/dL) one hour after the sample. A convolutional neural network (CNN) was pre-trained with the T1D data set and subsequently fitted using a T2D data set, all while being optimized toward maximizing the area under the receiver operating characteristics curve (AUC) value, and it was externally validated on a separate T2D data set. The developed model was externally validated with 334 711 one-hour CGM samples, of which 15 695 (4.69%) were labeled as hypoglycemic. The model achieved an AUC of 0.941 and a positive predictive value of 40.49% at a specificity of 95% and a sensitivity of 69.16%. The transfer learned CNN model showed promising performance in predicting hypoglycemic episodes and with slightly better results than a non-transfer learned CNN model.

Robust hybrid algorithms for regularization and variable selection in QSAR studies

This study introduces a robust hybrid sparse learning approach for regularization and variable selection. This approach comprises two distinct steps. In the initial step, we segment the original dataset into separate training and test sets and standardize the training data using its mean and standard deviation. We then employ either the LASSO or sparse LTS algorithm to analyze the training set, facilitating the selection of variables with non-zero coefficients as essential features for the new dataset. Secondly, the new dataset is divided into training and test sets. The training set is further divided into k folds and evaluated using a combination of Random Forest, Ridge, Lasso, and Support Vector Regression machine learning algorithms. We introduce novel hybrid methods and juxtapose their performance against existing techniques. To validate the efficacy of our proposed methods, we conduct a comprehensive simulation study and apply them to a real-life QSAR analysis. The findings unequivocally demonstrate the superior performance of our proposed estimator, with particular distinction accorded to SLTS+LASSO. In summary, the twostep robust hybrid sparse learning approach offers an effective regularization and variable selection applicable to a wide spectrum of real-world problems.

Thrombomodulin Protects Against Cell-Free Hemoglobin Mediated Acute Kidney Injury in Sickle Cell Anemia

Thrombomodulin (TM) is a transmembrane protein that plays a pivotal role in maintaining microvascular health. Functions of TM include 1) binding thrombin and inhibiting its interaction with fibrinogen, 2) augmenting protein C function, and 3) down-regulating complement activation. Reduced TM function has been implicated in several thrombotic microangiopathies, including atypical hemolytic uremic syndrome and disseminated intravascular coagulation (DIC) (PMID 29866818). Previous in vitro studies have demonstrated that cell-free hemoglobin (Hb) mediates loss of TM from the endothelial surface which, in turn, is associated with a higher risk of kidney injury in patients with sickle cell disease (PMID 34929051). Infusion of intact, soluble TM may restore TM's function in the microvasculature, as observed in critically ill patients with DIC (PMID 17059423). In transgenic sickle mice (Townes model, Jackson Laboratory; Bar Harbor, Maine), we investigated whether 1) cell-free Hb reduces TM function leading to glomerular endothelial and kidney injury and 2) if infusion of intact, soluble TM (Asahi Kasei Pharma ©; Tokyo, Japan) can restore TM function and reduce the toxic effects of cell-free Hb in the kidney. Sickle mice (4 male, 4 females per condition) were challenged with cell-free Hb (0.24g/kg intravenous [iv]) or normal saline followed 2 hours later by either infusion of TM (5mg/kg subcutaneous, 1mg/kg iv) or normal saline. Blood and urine samples were collected 24 hours after the challenge and the mice were sacrificed for histopathologic evaluation. In a separate set of similar experiments, sickle mice (3 males, 3 females per condition) received iv microbubbles (Vevo MicroMarker TM Contrast) 24 hours after the challenge to assess renal cortical blood flow by contrast-enhanced ultrasound (Vevo2100 system). Mean ± standard error of mean values are provided. Comparisons between Hb-only and Hb+TM conditions were made using the Mann-Whitney test. Vasculopathy: Infusion of cell-free Hb resulted in decreased TM antithrombotic function within the glomerulus; infusion of TM 2 hours after the challenge restored these functions (fibrinogen relative fluorescent intensity [RFI]: control, 24.2 ± 0.9; Hb-only, 40.6 ± 1.7; Hb+TM: 26.8 ± 1.6) and anti-complement function (C3 RFI: control, 12.9 ± 0.4; Hb-only, 19.3 ± 1.4; Hb+TM, 14.1 ± 0.5) (P≤0.007). Circulating vascular injury biomarkers rose in the cell-free Hb challenged sickle mice and this was abated with TM rescue for p-selectin (control, 271 ± 25 ng/mL; Hb-only, 533 ± 47 ng/mL; Hb+TM, 356 ± 34 ng/mL), VEGF (control, 59 ± 4 pg/mL; Hb-only, 93 ± 7 pg/mL; Hb+TM 69 ± 3 pg/mL), and vWF (control: 91 ± 9 ng/mL, Hb-only: 299 ± 31 ng/mL, Hb+TM: 183 ± 22 ng/mL) (P≤0.01). Kidney Damage: Urine biomarkers of tubular injury increased after the cell-free Hb challenge and improved with TM (KIM-1: control, 317 ± 66 pg/day; Hb-only, 1201 ± 142 pg/day; Hb+TM, 638 ± 78 pg/day; P=0.003) (NGAL: control, 234 ± 34 ng/day; Hb-only, 1624 ± 145 ng/day; Hb+TM, 767 ± 118 ng/day; P=0.003). An elevation in serum creatinine concentration is used to define acute kidney injury. The cell-free Hb challenge led to an acute rise in serum creatinine while the TM rescued mice had serum creatinine concentrations that were significantly lower than the Hb-only challenged mice (P=0.03) and similar to the control conditions (Figure 1A). Preliminary testing of TM administered 24 hours after the cell-free Hb challenge has also led to kidney protective effects as assessed by serum creatinine, p-selectin, and vWF (P < 0.05). Cortical Blood Flow: Peak enhancement, a measure of relative blood volume, was reduced in the cell-free Hb challenged mice compared to control conditions by contrast-enhanced ultrasound. The TM-rescued mice had improved peak enhancement compared to the Hb-only treated mice (P=0.03) and similar to control conditions (Figure 1B). In conclusion, cell-free Hb reduces TM function resulting in increased vascular and kidney injury and decreased cortical blood flow. Infusion of soluble TM 2 hours after the cell-free Hb challenge restores TM function, protects against vascular and kidney injury, and preserves cortical blood flow. Our data highlights TM as a potential candidate therapy to preserve vascular function and abate kidney injury in patients with sickle cell disease during severe vaso-occlusive or hyperhemolytic episodes, when concentrations of cell-free Hb increase several fold.

Patient Experience with Signs, Symptoms, and Daily Impacts of Immune Thrombocytopenia

Background Immune thrombocytopenia (ITP) is an autoimmune disorder characterized by increased risk of bleeding, as well as thromboembolic event. ITP as well as the currently available treatments for the disease can have a debilitating impact on patient's health-related quality of life (HRQoL). However, published evidence on symptoms experience by patients with ITP is limited. Objectives Explore the patient experience with signs, symptoms, and daily impacts of ITP and update the conceptual disease model for HRQoL in patients with ITP. Methods Patients aged ≥18 years diagnosed with primary ITP lasting >3 months were included in the study. Patients with any comorbidity where fatigue was a key symptom (e.g., anemia) were excluded. Concept elicitation semi-structured interviews were conducted with patients, which took approximately 40 minutes. One-on-one telephone interviews were conducted by an experienced and trained moderator. Patients were recruited in the study via third party recruitment agencies in the US. Interviews were audio-recorded, transcribed, and analyzed using MAXQDA. Multiple waves of interviews were conducted in separate sets of patients to ensure that saturation (when no new symptoms are reported in further rounds of interview) of responses is achieved; Wave 1 to 3 were carried out in 5 patients each, whereas Wave 4 was conducted in remaining 3 patients. Symptoms or impacts which were mentioned by ≥50% of patients and received an average bothersomeness rating of ≥5 on a 0-10 scale were considered salient. Results A total of 18 patients with ITP (mean [SD] age: 42.7 [14.7] years) were interviewed; 78% were female. Among included patients, median time since diagnosis was 4.2 (range: 0.3-24.2) years; all patients were diagnosed with primary ITP >12 months prior to the interview, i.e., had chronic ITP. Majority of patients had received ≥3 treatments for ITP (n=12, 66.7%), while 3 patients (16.7%) received 2 treatments, and 3 patients (16.7%) had 1 treatment for ITP. About 94% of patients (n=17) had received steroids previously. At the time of interview, 16 out of 18 patients were not receiving steroids, while 1 patient was on steroids (for 1 patient: unknown). Six key symptoms emerged as the most salient. Patients most frequently and spontaneously reported experiencing fatigue (n=17, 94%), followed by bruising (n=15, 83%) and petechiae (n=13, 72%), due to ITP. About half of the patients described symptoms related to sleep which included difficulty staying awake during the day, difficulty falling and staying asleep, and difficulty waking up in the morning. In addition, patients frequently reported bleeding problems (n=15, 83%), such as bleeding gums and nose (n=8, 44%), and bleedings that are difficult to stop (n=7, 39%). Saturation of signs and symptom concepts was achieved in Wave 2. Patients described different concepts of fatigue: physical fatigue, mental fatigue, or the emotional impact of fatigue. All patients who reported fatigue (n=17) experienced the physical aspect of it, which had a bigger impact on patients' lives. Thirteen patients reported only physical fatigue, while none of the patients experienced only mental fatigue. Fatigue was considered the most bothersome symptom by 10 out of 17 patients. Six key impacts emerged as the most salient: work, worry, and fear, limited physical or sporting activities, perceived need to be cautious to avoid getting hurt, and impact on friendships, family, and social life ( Figure 1). Saturation of signs and symptom concepts was achieved in Wave 3. The findings from the concept elicitation interviews were used to refine the conceptual model for ITP (Mathias et al. Health Qual Life Outcomes. 2008; Figure 2). Conclusions This qualitative study highlighted 6 key symptoms and 6 key impacts as the most salient among patients with chronic ITP. Fatigue was the most frequent and bothersome symptom, especially physical fatigue; other frequently reported symptoms were bruising, petechiae, and symptoms related to sleep (difficulty staying awake during the day, difficulty falling and staying asleep, and difficulty waking up in the morning). The most important impacts of ITP were work, worry, and fear, family, friends, and social life impacts, perceived need to be cautious, and limited physical or sporting activities. The findings obtained from the concept elicitation interviews were used to refine the conceptual model for ITP developed by Mathias et al 2008.

An Integrated Model of Customers’ Intention to Reuse Information Service: What’s New for Conversational Agents?

The application of conversational agents empowered by artificial intelligence (AI) is becoming more common among tourism and hospitality businesses. Past research on AI has mostly investigated embodied robots rather than disembodied conversational agents. This study investigates the determinants of customers’ intention to reuse chatbots through integrating three different perspectives: technology acceptance model, computer-mediated communication theories, and interpersonal communication theories. Using online self-reported data analyzed by hierarchical regression and multigroup structural equation modeling, the efficacy of different theoretical models in predicting chatbot adoption were compared. A separate data set was collected from instant messaging (IM) users to provide implications that further explain the distinction between disembodied conversational agents and human conversational partners. Results suggest that variables associated with TAM and interpersonal attraction exert greatest influence on customers’ trust and reuse intention toward conversational agents. Managerial implications are suggested for practitioners to improve chatbot and IM design.

Staphylococcus aureus activates NRLP3-dependent IL-1β secretion from human conjunctival goblet cells using α toxin and toll-like receptors 2 and 1.

We used cultured human conjunctival goblet cells to determine (i) whether the toxigenic S. aureus- induced activation of the epithelial goblet cells requires two signals to activate the NLRP3 inflammasome, (ii) if one signal is mediated by TLR1, TLR2, or TLR6, and (iii) if the S. aureus toxin α toxin is another signal for the activation of the inflammasome and secretion of mature IL-1β. Cultured cells were incubated with siRNA to knock down the different TLRs. After stimulation with toxigenic S. aureus RN6390, pro-IL-1β synthesis, caspase-1 activity, and mature IL-1β secretion were measured. In a separate set of experiments, the cells were incubated with toxigenic S. aureus RN6390 or mutant S. aureus ALC837 that does not express α toxin with or without exogenous α toxin. A gentamicin protection assay was used to determine if intracellular bacteria were active. We conclude that α toxin from toxigenic S. aureus triggers two separate mechanisms required for the activation of the NLRP3 inflammasome and secretion of mature IL-1β. In the first mechanism, α toxin secreted from internalized S. aureus produces a pore, allowing the internalized bacteria and associated pathogen-associated molecular patterns to interact with intracellular TLR2 and, to a lesser extent, TLR1. In the second mechanism, α toxin forms a pore in the plasma membrane, leading to an efflux of cytosolic K+ and influx of Ca2+. We conclude that α toxin by these two different mechanisms triggers the synthesis of pro-IL-1β and NLRP3 components, activation of capase-1, and secretion of mature IL-1β to defend against bacterial infection.

Contactless calibration of microchanneled AFM cantilevers for fluidic force microscopy

AbstractAtomic force microscopy (AFM) is an analytical technique that is increasingly utilized to determine interaction forces on the colloidal and cellular level. Fluidic force microscopy, also called FluidFM, became a vital tool for biomedical applications. FluidFM combines AFM and nanofluidics by means of a microchanneled cantilever that bears an aperture instead of a tip at its end. Thereby, single colloids or cells can be aspirated and immobilized to the cantilever, for example, to determine adhesion forces. To allow for quantitative measurements, the so‐called (inverse) optical lever sensitivity (OLS and InvOLS, respectively) must be determined, which is typically done in a separate set of measurements on a hard, non‐deformable substrate. Here, we present a different approach that is entirely based on hydrodynamic principles and does make use of the internal microfluidic channel of a FluidFM‐cantilever and an external pressure control. Thereby, a contact‐free calibration of the (inverse) optical lever sensitivity (InvOLS) becomes possible in under a minute. A quantitative model based on the thrust equation, which is well‐known in avionics, and finite element simulations, is provided to describe the deflection of the cantilever as a function of the externally applied pressure. A comparison between the classical and the here‐presented hydrodynamic method demonstrates equal accuracy.

BOA: A CT-Based Body and Organ Analysis for Radiologists at the Point of Care.

The study aimed to develop the open-source body and organ analysis (BOA), a comprehensive computed tomography (CT) image segmentation algorithm with a focus on workflow integration. The BOA combines 2 segmentation algorithms: body composition analysis (BCA) and TotalSegmentator. The BCA was trained with the nnU-Net framework using a dataset including 300 CT examinations. The CTs were manually annotated with 11 semantic body regions: subcutaneous tissue, muscle, bone, abdominal cavity, thoracic cavity, glands, mediastinum, pericardium, breast implant, brain, and spinal cord. The models were trained using 5-fold cross-validation, and at inference time, an ensemble was used. Afterward, the segmentation efficiency was evaluated on a separate test set comprising 60 CT scans. In a postprocessing step, a tissue segmentation (muscle, subcutaneous adipose tissue, visceral adipose tissue, intermuscular adipose tissue, epicardial adipose tissue, and paracardial adipose tissue) is created by subclassifying the body regions. The BOA combines this algorithm and the open-source segmentation software TotalSegmentator to have an all-in-one comprehensive selection of segmentations. In addition, it integrates into clinical workflows as a DICOM node-triggered service using the open-source Orthanc research PACS (Picture Archiving and Communication System) server to make the automated segmentation algorithms available to clinicians. The BCA model's performance was evaluated using the Sørensen-Dice score. Finally, the segmentations from the 3 different tools (BCA, TotalSegmentator, and BOA) were compared by assessing the overall percentage of the segmented human body on a separate cohort of 150 whole-body CT scans. The results showed that the BCA outperformed the previous publication, achieving a higher Sørensen-Dice score for the previously existing classes, including subcutaneous tissue (0.971 vs 0.962), muscle (0.959 vs 0.933), abdominal cavity (0.983 vs 0.973), thoracic cavity (0.982 vs 0.965), bone (0.961 vs 0.942), and an overall good segmentation efficiency for newly introduced classes: brain (0.985), breast implant (0.943), glands (0.766), mediastinum (0.880), pericardium (0.964), and spinal cord (0.896). All in all, it achieved a 0.935 average Sørensen-Dice score, which is comparable to the one of the TotalSegmentator (0.94). The TotalSegmentator had a mean voxel body coverage of 31% ± 6%, whereas BCA had a coverage of 75% ± 6% and BOA achieved 93% ± 2%. The open-source BOA merges different segmentation algorithms with a focus on workflow integration through DICOM node integration, offering a comprehensive body segmentation in CT images with a high coverage of the body volume.

Enchondroma Detection from Hand Radiographs with an Interactive Deep Learning Segmentation Tool-A Feasibility Study.

Enchondromas are common benign bone tumors, usually presenting in the hand. They can cause symptoms such as swelling and pain but often go un-noticed. If the tumor expands, it can diminish the bone cortices and predispose the bone to fracture. Diagnosis is based on clinical investigation and radiographic imaging. Despite their typical appearance on radiographs, they can primarily be misdiagnosed or go totally unrecognized in the acute trauma setting. Earlier applications of deep learning models to image classification and pattern recognition suggest that this technique may also be utilized in detecting enchondroma in hand radiographs. We trained a deep learning model with 414 enchondroma radiographs to detect enchondroma from hand radiographs. A separate test set of 131 radiographs (47% with an enchondroma) was used to assess the performance of the trained deep learning model. Enchondroma annotation by three clinical experts served as our ground truth in assessing the deep learning model's performance. Our deep learning model detected 56 enchondromas from the 62 enchondroma radiographs. The area under receiver operator curve was 0.95. The F1 score for area statistical overlapping was 69.5%. Our deep learning model may be a useful tool for radiograph screening and raising suspicion of enchondroma.

Impact bending strength of thermally-modified timber

Abstract The article offers a perspective on how thermal-modification affects the impact bending strength of five different wood species, an aspect that has not received as much attention as the well-studied static load behavior of thermally-modified timber (TMT). Since the TMTs are mainly employed as outdoor materials, where they may encounter impact forces, a comparative investigation into the flexibility and strength of these materials under impact is useful. This article evaluates different aspects of the TMT, such as deflection, strain in the impact region, the maximum force needed to initiate cracks, and the energy required for rupture. Wood planks from ash, beech, larch, oak, and spruce were thermally modified at 180 and 220 °C. They were cut into test specimens, while a separate set of unmodified specimens from each wood species served as the reference group. The specimens were subjected to an impact 3-point bending test, and an ultra-high-speed camera meticulously recorded the results. The images were processed by the digital image correlation (DIC) method to determine the deflection and strain distribution of the beams during the impact test. The deflection, maximum force, maximum longitudinal strain, and required work for rupture of each group were determined. The results showed that thermal-modification decreases the wood deflection and maximum longitudinal strain by approximately 50 %. In addition, the impact bending strength decreased by nearly 60 %. However, the impact bending strength did not exhibit a statistically significant decrease at 180 °C; in some cases, it even increased.

Anti-Clogging Effect of Continuous Hemodiafiltration With Blood Recirculation.

Pediatric dialysis requires low flow from the body, but greater flow is needed to prevent clogging. As a solution, we developed a new continuous hemodiafiltration system with blood recirculation (CHDF-R), which enables separate settings for blood flow from the body and to the hemofilter. We compared CHDF-R with conventional continuous hemodiafiltration (CHDF) of bovine plasma and blood by monitoring the transmembrane pressure (TMP) and observing the hemofilter membrane surface. When using bovine plasma, the postdialysis TMP with CHDF-R was significantly lower than with CHDF (median CHDF, 23.7; median CHDF-R, 18.1; p = 0.029). Likewise, when using bovine blood, the postdialysis TMP was also significantly lower with CHDF-R than with CHDF (median CHDF, 150; median CHDF-R, 100; p = 0.029). Moreover, the area of clogged membrane was significantly smaller with CHDF-R than with CHDF, and the inner membrane surface showed less material deposition with CHDF-R than CHDF. CHDF-R thus appears to suppress accumulation of clogging substances by producing higher shear stress within hollow fiber membranes.

The selective Rho Kinase 2 inhibitor Chroman-1, is a potent vasodilator but does not compromise left ventricular function in-vivo

Abstract Background In the field of cardioprotection, focus is turning to limiting microvascular obstruction (MVO) post myocardial infarction. The development of MVO carries a poor prognosis, which is independent of infarct burden [1]. Commonly used vasodilators in the treatment of MVO e.g., Ca2+ blockers, beta-blockers can induce negatively inotropic effects. This is a serious problem as hypotension and shock during acute MI are harmful. Rho kinase (ROCK) inhibitors induce vasodilation by inhibiting the phosphorylation of myosin light chain kinase (MLC2) in vascular smooth muscle (VSMC) [2]. ROCK1/2 inhibitors are known to cause transient negative inotropic effects [3]. However, the potential side-effects of the selective ROCK2i, Chroman-1 (Ch1) are unknown. Purpose We aimed to i) Confirm the expression of ROCK2 mRNA in aorta and coronary vasculature, ii) To investigate the vasodilatory properties of Ch1 iii) To confirm that Ch1 does not acutely impair left ventricular ejection fraction (LVEF%) and blood pressure in-vivo. Methods RNA scope in-situ hybridisation was performed with a fluorescent, multiplex assay for ROCK1/2 & VSMC mRNA in whole hearts, coronary vasculature and aorta from male SD rats. Thoracic aorta was dissected and 3mm segments suspended into a tissue bath. Vascular rings were constricted with 1µM PE and treated with [10-12]-[10-5]mM Ch1. In a separate set of in-vivo experiments, male SD rats 250-300g were anaesthetised with isoflurane, intubated and ventilated. LVEF% was measured in a closed chest model, following the addition of Ch1 (30-300µg/kg) or DMSO vehicle, administered i.p at 15-minute intervals. Mean arterial pressure (MAP, mmHg) and heart rate (BPM) were measured throughout. Results ROCK2 mRNA was localised to myocardium, aorta and coronary circulation VSMC. Figure 1 demonstrates that Ch1 induced dose dependent vasodilation of PE constricted aortic rings vs control (EC50 Ch1 95% CI [5.1e-008, 2.2e-007] n=6). In-vivo, Ch1 at all doses (30-300µg/kg) did not significantly affect LVEF% (p>0.05, n=5) nor average MAP at doses 30-100µg/kg (p>0.05, n=5). At the highest dose, i.e 300µg/kg, Ch1 decreased average MAP (mmHg) compared to control +/-SEM (76+/-7.1 vs 92+/-2.8, p=0.02, n=5), but not below 70mmHg threshold. Ch1 did not significantly affect mean heart rate at any dose (p>0.05, n=5). Conclusion The selective ROCK2 inhibitor Ch1, induces potent arterial vasodilation, without compromising LVEF% or haemodynamics. It is therefore, a suitable drug for further preclinical studies of MVO in ischaemia/reperfusion (I/R).Figure 1:Ch1 induces dose dependent vasodilation of aortic rings, **p<0.005, ***p<0.001

Abstract 13157: Multivariable ECG Machine Learning is Superior to QRS Duration and Traditional ECG Criteria for Detecting Left Ventricular Hypertrophy in Patients With Left Bundle Branch Block

Introduction: The utility of ECG to diagnose left ventricular hypertrophy (LVH) in patients with left bundle branch block (LBBB) is limited. We sought to compare the diagnostic yield of ECG using (i) QRS duration, (ii) published LVH criteria, and (iii) machine learning (ML) models to detect increased left ventricular mass indexed (LVMi) in the setting of LBBB. Hypothesis: ML is superior to QRS duration and LVH criteria to detect LVH among LBBB. Methods: 12-lead ECGs were processed to reconstruct orthogonal X, Y, Z leads using Kors’s matrix and obtain root-mean-squared (3D) ECG. R wave, S wave and overall amplitudes, voltage-time-integrals (VTIs), and other ECG features were extracted from all ECG leads. ML algorithms [logistic regression (LR), support vector classifier (SVC), decision trees (DT), random forest (RF), gradient boosted machine (GBM) and boosted trees (BT)] were trained to predict increased LVMi (women >95 g/m 2 , men >115 g/m 2 ) from ECG features on a training set of 2668 ECGs with typical LBBB and echocardiogram within 45 days before or after ECG. LVM was calculated using the ASE formula. We obtained ROC AUCs for prediction of increased LVMi by (i) QRS duration, (ii) published LVH criteria, and (iii) ML models in a separate validation set of adults with typical LBBB. Results: Among the validation set of 413 adults (53% women, age 73±12 yr) with LBBB, QRS duration alone had a higher AUC (women 0.657, men 0.703) for diagnosing increased LVMi compared to standard LVH criteria (Table). RF (women 0.688, men 0.738), GBM and DT models had better AUCs than QRS duration alone. Conclusions: In patients with LBBB, QRS duration ≥150 in women and ≥160 in men is a superior predictor of LVH than traditional voltage-based ECG criteria, but ML methods outperform all traditional ECG criteria and QRS duration.

Abstract 16427: Despite Machine Learning Models, QRS Duration Remains the Superior ECG Criterion for Predicting Left Ventricular Dilation in Patients With Left Bundle Branch Block

Introduction: The utility of ECG to diagnose left ventricular (LV) dilation in patients with left bundle branch block (LBBB) is not known. We sought to compare the diagnostic yield of ECG using (i) QRS duration, (ii) published LVH criteria, and (iii) machine learning (ML) models to detect increased left ventricular end diastolic volume indexed (LVEDVi) in the setting of LBBB. Hypothesis: ML is superior to QRS duration and LVH criteria to detect increased LVEDVi among LBBB. Methods: 12-lead ECGs were processed to reconstruct orthogonal X, Y, Z leads using Kors’s matrix and obtain root-mean-squared (3D) ECG. R wave, S wave and overall amplitudes, voltage-time-integrals (VTIs), and other ECG features were extracted from all ECG leads. ML algorithms [logistic regression (LR), support vector classifier (SVC), decision trees (DT), random forest (RF), gradient boosted machine (GBM) and boosted trees (BT)] were trained to predict increased LVEDVi (women >61 mL/m 2 , men >74 mL/m 2 ) from ECG features on a training set of 2668 ECGs with typical LBBB and echocardiogram within 45 days before or after ECG. LVEDV was measured using ASE biplane method of discs. We obtained ROC AUCs for prediction of increased LVEDVi by (i) QRS duration, (ii) published LVH criteria, and (iii) ML models in a separate validation set of adults with typical LBBB. Results: Among the validation set of 413 adults (53% women, age 73±12 yr) with LBBB, QRS duration alone had a higher AUC (women 0.668, men 0.699) for diagnosing increased LVMi compared to standard LVH criteria (Table). The best ML model (RF with overall AUC 0.694) did not substantially outperform QRS duration alone. Conclusions: In patients with LBBB, QRS duration ≥150 in women and ≥160 in men is a superior predictor of LV dilation than traditional ECG-based LVH criteria, with no additional value added by ML methods.

Common computations formetacognition andmeta-metacognition.

Recent evidence shows that people have the meta-metacognitive ability to evaluate their metacognitive judgments of confidence. However, it is unclear whether meta-metacognitive judgments are made by a different system and rely on a separate set of computations compared to metacognitive judgments. To address this question, we asked participants (N = 36) to perform a perceptual decision-making task and provide (i) an object-level, Type-1 response about the identity of the stimulus; (ii) a metacognitive, Type-2 response (low/high) regarding their confidence in their Type-1 decision; and (iii) a meta-metacognitive, Type-3 response (low/high) regarding the quality of their Type-2 rating. We found strong evidence for the existence of Type-3, meta-metacognitive ability. In a separate condition, participants performed an identical task with only a Type-1 response followed by a Type-2 response given on a 4-point scale. We found that the two conditions produced equivalent results such that the combination of binary Type-2 and binary Type-3 responses acts similar to a 4-point Type-2 response. Critically, while Type-2 evaluations were subject to metacognitive noise, Type-3 judgments were made at no additional cost. These results suggest that it is unlikely that there is a distinction between Type-2 and Type-3 systems (metacognition and meta-metacognition) in perceptual decision-making and, instead, a single system can be flexibly adapted to produce both Type-2 and Type-3 evaluations recursively.

Implementation towards Blood Cancer Detection with Convolutional Neural Network

This Blood cancer is a life-threatening disease that requires early and accurate detection for effective treatment. In this project, we present a novel approach for blood cancer detection using a Convolutional Neural Network (CNN) model. The CNN model is trained on a dataset comprising cancer and normal blood cell images. Through extensive analysis and evaluation, we achieve a high level of accuracy in distinguishing between cancerous and normal blood cells. To evaluate the performance of our model, we conducted tests on a separate set of cancer and normal blood cell images. The accuracy of our model was determined by comparing the predicted labels with the ground truth labels. The results demonstrate that our CNN model achieves a commendable accuracy rate, making it a promising tool for blood cancer detection. Furthermore, we discuss the significance of our findings and their potential implications for early diagnosis and improved treatment outcomes. The robustness and reliability of our model contribute to its practical utility in clinical settings. By enabling early detection of blood cancer, our approach has the potential to positively impact patient outcomes and enhance the efficiency of treatment strategies. In conclusion, this project presents a novel approach for blood cancer detection using a CNN model. The results demonstrate the effectiveness of our model in accurately distinguishing between cancerous and normal blood cells. The proposed method holds promise for improving blood cancer diagnosis and ultimately contributing to better patient care.

222 High Throughput Screening of Bovine Mastitis from Raw Milk Using Maldi-Tof Ms and Machine Learning

Abstract To prevent pathogenic microorganisms associated with mastitis from entering the food supply, raw milk must be screened. This study investigated whether MALDI-TOF-MS combined with machine learning algorithms is a cost-effective method for screening dairy milk samples at high throughput. Raw milk from animals was analyzed using matrix-assisted laser desorption ionization mass spectrometry. Mass spectra were obtained for a training dataset of composite milk samples (n = 226) collected from a local dairy. The peaks from the mass spectra were then transported into a machine learning model, which detected non-obvious patterns that correspond with the mastitis condition. Finally, the diagnostic accuracy of the new model was evaluated using a separate set of milk samples (scoring set; n = 100). For optimal performance, decision trees, random forest models, gradient boosted trees, and neural network models were investigated. The results show that the gradient boosted decision tree model performs best for mastitis diagnosis, with simultaneous sensitivity and specificity greater than 0.8. The decision tree approach is conceptually straightforward and identifies candidate biomarker molecules indicative of mastitis. The combined MALDI TOF plus machine learning approach is effective for detecting mastitis in dairy cows and is likely a viable option for diagnosing a wide range of animal health issues. The MALDI TOF method is suitable for high-throughput screening because it only requires a microliter of inexpensive reagents.

PSI-4 Soybean-Supplemented Nursing Ewes Showed Better Performance and Health Status Than Corn-Supplemented Ewes Stocked in Woodlands

Abstract Previous woodland-grazing studies recommended nutritional supplementation to young, growing goats and sheep as well as lactating does. However, information is limited on the role of supplement type to improve the performance and health status of nursing ewes raised in woodlands. This study tested a hypothesis that nursing ewes would remain healthy and perform equally well either with corn or soybean supplements. The study objective was to evaluate the performance and blood parameters of the nursing ewes when stocked in woodlands with corn or soybean supplements. Eighteen Katahdin-St. Croix cross-nursing ewes were divided into two groups. Each group was stocked in a separate set of woodland plots (3 plots/group) and rotated among those plots throughout the study period along with ad libitum hay from May to August 2022. Group-1 ewes were supplemented with whole corn and Group-2 ewes with whole soybean (0.5% of body weight). Animal performance data [body weight, FAMACHA score, and body condition score (BCS)] were measured on day 1, fortnightly during the study, and at the end of the study. For assessing the health status, blood samples were collected on days -1, 45, and 80 and analyzed for hematological and biochemical parameters using ProCyte-Dx Analyzer and Catalyst One (Table 1). Animal performance and blood parameter data were analyzed using the GLM procedure in SAS 9.4. Soybean-supplemented nursing ewes had greater body weight (22%, P < 0.05), and better FAMACHA score (13%, P < 0.01) and BCS (8%, P < 0.001) in comparison to corn-supplemented nursing ewes. Similarly, blood urea nitrogen, alkaline phosphate, cholesterol, and calcium concentration were greater in soybean-supplemented vs. corn-supplemented ewes (P < 0.05). Results suggest that soybean is a better supplement than corn for enhancing the performance and health status of nursing ewes grazing in woodlands.

A comprehensive strategy for phase detection of high entropy alloys: Machine learning and deep learning approaches

High Entropy Alloys (HEAs) are a novel category of materials with unique physical and mechanical properties, which makes them a promising candidate for various engineering applications. However, the accurate identification of different phases present in HEAs, controlling its superb properties, is still a challenging task. To address this issue, this study proposes a comprehensive approach based on machine learning (ML) and deep learning (DL) to detect the phases present in HEAs. A deep neural network (DNN) was successfully developed that uses unique feature values of a number of HEAs to accurately predict different phases present in the alloy system. The DNN architecture consists of multiple layers that allow it to learn and extract the relevant features from the data, and then classify them into their respective phases. To train and validate the DNN model, a large dataset of HEAs (with known phases and their corresponding feature values) was collected from the literature. This dataset was then used to train the DNN model using a supervised learning approach. The DNN model was also validated using a separate set of HEAs with unknown phases to test its accuracy and reliability. Comparison with known results demonstrated that the DNN model can accurately predict the phases present in HEAs with high accuracy. After implementing all ML and DL algorithms, a voting ensemble (containing six best performing algorithms) was constructed. The final accuracy was obtained as high as 84% which is quite reasonable for classification problem. Further, it was also observed how significant each feature is in determining the phase(s) of HEAs. The proposed approach offers a promising method for phase detection in HEAs, which can lead to a deeper understanding of their properties and facilitate the design of new materials for various applications.

A software defect prediction method based on learnable three-line hybrid feature fusion

Software defect prediction (SDP) plays a crucial role in ensuring the security and quality of software systems. However, it faces challenges posed by high-dimensional features present in software defect datasets and the limited effectiveness of traditional nonlinear dimensionality reduction methods in extracting essential feature information. To address these issues, we propose a novel approach called learnable three-line hybrid feature fusion (LTHFFA), which incorporates the principle of three-line hybrid breeding into feature fusion for the first time. In this method, three distinct dimensionality reduction techniques are initially employed to obtain three separate sets of features. Subsequently, a learnable weight factor feature fusion method is proposed to facilitate automatically learn and dynamically update of feature weights. By integrating the three feature sets based on the principle of three-line hybrid breeding, we derive learnable three-line hybrid fusion features. These features are then utilized in the context of software defect prediction. Experimental results demonstrate the superior performance of LTHFFA compared to nine other dimensionality reduction methods across seventeen publicly available software defect datasets. LTHFFA exhibits the ability to effectively integrate multiple feature sets, reduce feature redundancy, and enhance predictive accuracy. Moreover, statistical analysis using Friedman ranking and Holm's post-hoc test confirms the significant advantage of LTHFFA over alternative dimensionality reduction methods.