Examinando por Autor "del Castillo, Alberto"

Mostrando 1 - 2 de 2
  • Miniatura
    Ítem
    Hospital-wide sepsis detection: A machine learning model based on prospectively expert-validated cohort
    (MDPI, 2026-01-21) Borges-Sa, Marcio; Giglio, Andrés; Aranda, Maria; Socias, Antonia; del Castillo, Alberto; Pruenza, Cristina; Hernández, Gonzalo; Cerdá, Sofía; Socias, Lorenzo; Estrada, Victor; de la Rica, Roberto; Martin, Elisa; Martin-Loeches, Ignacio
    Background/Objectives: Sepsis detection remains challenging due to clinical heterogeneity and limitations of traditional scoring systems. This study developed and validated a hospital-wide machine learning model for sepsis detection using retrospectively developed data from prospectively expert-validated cases, aiming to improve diagnostic accuracy beyond conventional approaches. Methods: This retrospective cohort study analysed 218,715 hospital episodes (2014–2018) at a tertiary care centre. Sepsis cases (n = 11,864, 5.42%) were prospectively validated in real-time by a Multidisciplinary Sepsis Unit using modified Sepsis-2 criteria with organ dysfunction. The model integrated structured data (26.95%) and unstructured clinical notes (73.04%) extracted via natural language processing from 2829 variables, selecting 230 relevant predictors. Thirty models including random forests, support vector machines, neural networks, and gradient boosting were developed and evaluated. The dataset was randomly split (5/7 training, 2/7 testing) with preserved patient-level independence. Results: The BiAlert Sepsis model (random forest + Sepsis-2 ensemble) achieved an AUC-ROC of 0.95, sensitivity of 0.93, and specificity of 0.84, significantly outperforming traditional approaches. Compared to the best rule-based method (Sepsis-2 + qSOFA, AUC-ROC 0.90), BiAlert reduced false positives by 39.6% (13.10% vs. 21.70%, p < 0.01). Novel predictors included eosinopenia and hypoalbuminemia, while traditional variables (MAP, GCS, platelets) showed minimal univariate association. The model received European Medicines Agency approval as a medical device in June 2024. Conclusions: This hospital-wide machine learning model, trained on prospectively expert-validated cases and integrating extensive NLP-derived features, demonstrates superior sepsis detection performance compared to conventional scoring systems. External validation and prospective clinical impact studies are needed before widespread implementation.
  • Miniatura
    Ítem
    Prognostic Value of International Normalized Ratio and Thrombocytopenia in Early Risk Stratification of Septic Patients
    (MDPI, 2026-04-07) Tejada, Sofia; Giglio, Andres; Aranda, Maria; Socias, Antonia; del Castillo, Alberto; Mena, Joana; Franco, Sara; Ortega, Maria; Nieto, Yasmina; Borges-Sa, Marcio
    Background/Objective: Coagulopathy is a hallmark of sepsis, associated with poor outcomes. Although platelet count has commonly been used for risk stratification, its prognostic value has remained limited. This study compared the ability of the International Normalized Ratio (INR) and platelet count to predict in-hospital mortality in septic patients. Methods: A retrospective study was conducted including adult patients diagnosed with sepsis and admitted to Hospital Universitario Son Llàtzer (Spain) between 2006 and 2022. The INR and platelet count at diagnosis were categorized using clinical thresholds. The primary outcome was in-hospital mortality. Results: Among 6433 patients (60.6% females), mortality was 8.8%. Mortality increased from 6.3% (INR ≤ 1.2) to 20.2% (INR 2.0–3.0), slightly decreasing at INR > 3.0 (10.8%). The platelet count showed a weaker association, with the highest mortality observed at <50 × 109/L (24.6%). The combined markers identified a high-risk subgroup with 50% mortality (INR > 3.0 and platelet count < 50 × 109/L). In the full cohort, multivariable analysis confirmed the INR as an independent predictor of mortality (OR 2.183, p = 0.0002), whereas the platelet count was not significant. The model including the INR achieved an AUC 0.746, while adding the platelet count did not improve performance. Conclusions: the INR at diagnosis was a strong and independent predictor of in-hospital mortality, outperforming the platelet count. These findings could support the consideration of the INR in early risk stratification frameworks and highlight the need for prospective validation before integration into sepsis guidelines.