Search for: All records

Background and Objectives: Sepsis is a leading cause of mortality in intensive care units (ICUs). The development of a robust prognostic model utilizing patients’ clinical data could significantly enhance clinicians’ ability to make informed treatment decisions, potentially improving outcomes for septic patients. This study aims to create a novel machine-learning framework for constructing prognostic tools capable of predicting patient survival or mortality outcome. Methods: A novel dataset is created using concatenated triples of static data, temporal data, and clinical outcomes to expand data size. This structured input trains five machine learning classifiers (KNN, Logistic Regression, SVM, RF, and XGBoost) with advanced feature engineering. Models are evaluated on an independent cohort using AUROC and a new metric, 𝛾, which incorporates the F1 score, to assess discriminative power and generalizability. Results: We developed five prognostic models using the concatenated triple dataset with 10 dynamic features from patient medical records. Our analysis shows that the Extreme Gradient Boosting (XGBoost) model (AUROC = 0.777, F1 score = 0.694) and the Random Forest (RF) model (AUROC = 0.769, F1 score = 0.647), when paired with an ensemble under-sampling strategy, outperform other models. The RF model improves AUROC by 6.66% and reduces overfitting by 54.96%, while the XGBoost model shows a 0.52% increase in AUROC and a 77.72% reduction in overfitting. These results highlight our framework’s ability to enhance predictive accuracy and generalizability, particularly in sepsis prognosis. Conclusion: This study presents a novel modeling framework for predicting treatment outcomes in septic patients, designed for small, imbalanced, and high-dimensional datasets. By using temporal feature encoding, advanced sampling, and dimension reduction techniques, our approach enhances standard classifier performance. The resulting models show improved accuracy with limited data, offering valuable prognostic tools for sepsis management. This framework demonstrates the potential of machine learning in small medical datasets. 

Abstract Supernova remnants are commonly considered to produce most of the Galactic cosmic rays via diffusive shock acceleration. However, many questions regarding the physical conditions at shock fronts, such as the magnetic-field morphology close to the particle acceleration sites, remain open. Here we report the detection of a localized polarization signal from some synchrotron X-ray emitting regions of Tycho’s supernova remnant made by the Imaging X-ray Polarimetry Explorer. The derived degree of polarization of the X-ray synchrotron emission is 9% ± 2% averaged over the whole remnant, and 12% ± 2% at the rim, higher than the value of polarization of 7%–8% observed in the radio band. In the west region, the degree of polarization is 23% ± 4%. The degree of X-ray polarization in Tycho is higher than for Cassiopeia A, suggesting a more ordered magnetic field or a larger maximum turbulence scale. The measured tangential direction of polarization corresponds to the radial magnetic field, and is consistent with that observed in the radio band. These results are compatible with the expectation of turbulence produced by an anisotropic cascade of a radial magnetic field near the shock, where we derive a magnetic-field amplification factor of 3.4 ± 0.3. The fact that this value is significantly smaller than those expected from acceleration models is indicative of highly anisotropic magnetic-field turbulence, or that the emitting electrons either favor regions of lower turbulence, or accumulate close to where the orientation of the magnetic field is preferentially radially oriented due to hydrodynamical instabilities. 

Abstract We report on a ∼5 σ detection of polarized 3–6 keV X-ray emission from the supernova remnant Cassiopeia A (Cas A) with the Imaging X-ray Polarimetry Explorer (IXPE). The overall polarization degree of 1.8% ± 0.3% is detected by summing over a large region, assuming circular symmetry for the polarization vectors. The measurements imply an average polarization degree for the synchrotron component of ∼2.5%, and close to 5% for the X-ray synchrotron-dominated forward shock region. These numbers are based on an assessment of the thermal and nonthermal radiation contributions, for which we used a detailed spatial-spectral model based on Chandra X-ray data. A pixel-by-pixel search for polarization provides a few tentative detections from discrete regions at the ∼ 3 σ confidence level. Given the number of pixels, the significance is insufficient to claim a detection for individual pixels, but implies considerable turbulence on scales smaller than the angular resolution. Cas A’s X-ray continuum emission is dominated by synchrotron radiation from regions within ≲10 17 cm of the forward and reverse shocks. We find that (i) the measured polarization angle corresponds to a radially oriented magnetic field, similar to what has been inferred from radio observations; (ii) the X-ray polarization degree is lower than in the radio band (∼5%). Since shock compression should impose a tangential magnetic-field structure, the IXPE results imply that magnetic fields are reoriented within ∼10 17 cm of the shock. If the magnetic-field alignment is due to locally enhanced acceleration near quasi-parallel shocks, the preferred X-ray polarization angle suggests a size of 3 × 10 16 cm for cells with radial magnetic fields.