Daily Sepsis Research Analysis
Three impactful studies advance sepsis science along complementary axes: a Nature Communications paper introduces a geometric deep learning framework (InfEHR) that improves phenotype inference for culture‑negative neonatal sepsis and post‑operative AKI from EHRs; a meta-analysis of RCTs shows continuous β‑lactam infusion modestly reduces hospital mortality and increases clinical cure; and mechanistic work implicates suPAR–RAGE signaling in sepsis‑associated AKI, with antibody pretreatment amelio
Summary
Three impactful studies advance sepsis science along complementary axes: a Nature Communications paper introduces a geometric deep learning framework (InfEHR) that improves phenotype inference for culture‑negative neonatal sepsis and post‑operative AKI from EHRs; a meta-analysis of RCTs shows continuous β‑lactam infusion modestly reduces hospital mortality and increases clinical cure; and mechanistic work implicates suPAR–RAGE signaling in sepsis‑associated AKI, with antibody pretreatment ameliorating injury in mice.
Research Themes
- Explainable/graph-based AI for sepsis phenotyping and diagnosis
- Pharmacokinetic/pharmacodynamic optimization of β-lactams in critical illness
- Mechanistic biomarkers and targets in sepsis-associated acute kidney injury (suPAR–RAGE)
Selected Articles
1. InfEHR: Clinical phenotype resolution through deep geometric learning on electronic health records.
InfEHR converts whole EHRs into temporal graphs and uses deep geometric learning to infer clinical likelihoods with minimal labeled data. In neonatal culture-negative sepsis and postoperative AKI, it markedly improved sensitivity over physician heuristics while maintaining high specificity. This demonstrates scalable, label-efficient sepsis phenotyping with potential for earlier recognition and risk stratification.
Impact: Introduces a novel, label-efficient geometric learning paradigm that enhances detection of low-prevalence sepsis phenotypes directly from routine EHRs, addressing a key bottleneck in clinical AI.
Clinical Implications: Hospitals could deploy InfEHR-like tools to flag culture-negative neonatal sepsis or predict postoperative AKI earlier, guiding timely diagnostics and preventive interventions without extensive manual labeling.
Key Findings
- Temporal-graph representation of EHRs enabled high-performance probabilistic inference with few labels.
- Sensitivity improved substantially for culture-negative neonatal sepsis (0.60 vs 0.04) and postoperative AKI (0.71 vs 0.20), with preserved specificity.
- Outperformed physician heuristics across two independent health systems, highlighting scalability.
Methodological Strengths
- Geometric deep learning on temporal EHR graphs reduces reliance on large labeled datasets.
- External testing across two health systems improves generalizability.
Limitations
- Exact cohort sizes and prospective clinical impact (e.g., time-to-treatment reduction) were not reported.
- Algorithm performance compared to comprehensive machine learning baselines beyond heuristics was not detailed.
Future Directions: Prospective, interventional studies to test whether InfEHR-triggered alerts improve time-sensitive outcomes; integration with explainability and clinician-in-the-loop workflows; broader validation across diseases and health systems.
Electronic health records contain multimodal data that can inform clinical decisions but are often unsuited for advanced machine learning analyses due to lack of labeled data. Here, we present InfEHR, a framework to automatically compute clinical likelihoods from whole electronic health records without requiring large volumes of labeled training data. InfEHR applies deep geometric learning through a procedure that converts whole electronic health records to temporal graphs that naturally capture phenotypic dynamics, leading to unbiased representations. Using only few labeled examples, InfEHR computes and automatically revises probabilities achieving highly performant inferences, especially in low-prevalence diseases. We test InfEHR using electronic health records from Mount Sinai Health System and UC Irvine Medical Center against physician-provided heuristics on neonatal culture-negative sepsis (3% prevalence) and postoperative acute kidney injury (21% prevalence). InfEHR demonstrated superior performance: for culture-negative sepsis (sensitivity: 0.60 vs. 0.04, specificity: 0.98 vs. 0.99) and post-operative acute kidney injury (sensitivity: 0.71 vs. 0.20, specificity: 0.93 vs. 0.98). Our study demonstrates the application of geometric deep learning in electronic health records for probabilistic inference in real-world clinical settings at scale.
2. Continuous versus intermittent infusion of β-lactams in patients with sepsis and septic shock: a systematic review and meta-analysis.
Pooling 11 RCTs (9,166 patients), continuous β‑lactam infusion did not reduce overall or ICU mortality but was associated with lower hospital mortality, higher end‑of‑study survival, and higher clinical cure. Safety and length of stay outcomes were similar to intermittent dosing.
Impact: Synthesizes the highest-level evidence to date on β‑lactam infusion strategies in sepsis, supporting potential practice adjustments toward continuous infusion in select ICU settings.
Clinical Implications: Consider continuous infusion for time-dependent β‑lactams in septic adults to improve clinical cure and potentially reduce hospital mortality, while balancing logistical demands and patient-specific pharmacokinetics.
Key Findings
- Across 11 RCTs, continuous infusion did not change overall mortality (RR 0.94; 95% CI 0.88–1.01) or ICU mortality (RR 0.94; 95% CI 0.88–1.01).
- Continuous infusion reduced hospital mortality (RR 0.92; 95% CI 0.85–0.99) and increased clinical cure rate (RR 1.42; 95% CI 1.12–1.80).
- No significant differences in ICU/hospital length of stay or adverse events between strategies.
Methodological Strengths
- Pre-registered (PROSPERO) systematic review with RoB 2.0 and GRADE assessment.
- Large cumulative sample size and RCT-only inclusion.
Limitations
- Heterogeneity in antibiotic classes, infusion protocols, and co-interventions.
- Benefits did not extend to overall or ICU mortality; potential publication bias cannot be fully excluded.
Future Directions: Patient-level meta-analyses and pragmatic RCTs using therapeutic drug monitoring to individualize continuous infusion, focusing on pathogen MICs and augmented renal clearance.
OBJECTIVE: To assess whether continuous infusion of β-lactam antibiotics improves clinical outcomes compared to intermittent infusion in adult patients with sepsis or septic shock. METHODS: We conducted a systematic review and meta-analysis of randomized controlled trials comparing continuous versus intermittent β-lactam infusion. Databases searched included PubMed, Scopus, Web of Science, and Embase. Risk of bias was assessed using the RoB 2.0 tool, and the certainty of evidence was evaluated using GRADE. RESULTS: Eleven studies involving 9,166 patients were analyzed, comparing continuous versus intermittent infusion of β-lactams in sepsis or septic shock. There was no significant difference in overall mortality (RR 0.94; 95% CI: 0.88-1.01) or ICU mortality (RR 0.94; 95% CI: 0.88-1.01). Continuous infusion was associated with lower hospital mortality (RR 0.92; 95% CI: 0.85-0.99), higher survival at the end of the study (RR 1.04; 95% CI: 1.02-1.07), and higher clinical cure rate (RR 1.42; 95% CI: 1.12-1.80). No significant differences were observed in the length of stay in the ICU (MD 0.75 days; 95% CI: -1.17 to 2.68) or hospital stay (MD -2.51 days; 95% CI: -10.13 to 5.12), or in the adverse events (RR 0.82; 95% CI: 0.60-1.12). CONCLUSION: Continuous infusion of β-lactams could reduce hospital mortality and increase the clinical cure rate in critically ill patients, although its effect on overall mortality, hospital stay, and adverse events remains uncertain. PROSPERO number: CRD42024613938.
3. Soluble urokinase plasminogen activator receptor promotes endoplasmic reticulum stress and apoptosis susceptibility through RAGE in sepsis acute kidney injury.
Serum suPAR was elevated in ICU patients with AKI and mechanistically promoted ER stress and apoptosis in renal tubular cells via RAGE binding. In septic mice, uPAR monoclonal antibody pretreatment reduced ROS, ER stress, and kidney injury, nominating suPAR–RAGE as a biomarker–target axis in sepsis-associated AKI.
Impact: Bridges biomarker discovery with mechanism and therapeutic modulation, revealing suPAR–RAGE signaling in septic AKI and preclinical antibody efficacy.
Clinical Implications: suPAR may aid early AKI risk stratification in sepsis, and the suPAR–RAGE pathway presents a therapeutic target warranting clinical translation of uPAR/RAGE-directed agents.
Key Findings
- Serum suPAR levels were significantly higher in ICU patients with AKI (n=124 cohort).
- suPAR induced ER stress, ROS, and apoptosis-related proteins in HK-2 cells and increased ER stress in mouse renal cortex.
- suPAR physically interacted with RAGE (co-immunoprecipitation, colocalization, docking), and uPAR mAb pretreatment reduced septic AKI in mice.
Methodological Strengths
- Translational design combining human cohort biomarker analysis with in vitro and in vivo mechanistic validation.
- Multiple orthogonal assays (co-IP, fluorescence colocalization, docking) support the suPAR–RAGE interaction.
Limitations
- Clinical cohort size is modest and observational; causality in humans remains inferential.
- Antibody efficacy was shown in pretreatment models; therapeutic timing and safety in established AKI require study.
Future Directions: Prospective sepsis cohorts to validate suPAR thresholds for AKI prediction; phase I/II trials of uPAR/RAGE-targeted agents with mechanistic biomarkers; exploration of combinatorial antioxidant/ER stress modulators.
BACKGROUND: Acute kidney injury (AKI) is a common complication among critically ill patients, associated with an increased risk of adverse outcomes. There is an urgent need for novel biomarkers to assist in the early detection and management of AKI. Soluble urokinase plasminogen activator receptor (suPAR) is an inflammation-related, immune-derived molecule implicated in the pathogenesis of several diseases, including kidney diseases. METHODS: We characterized the ability of serum suPAR levels to diagnose AKI in 124 patients admitted to the intensive care unit (ICU). Additionally, in vivo and in vitro experiments were performed to explore the underlying mechanisms between suPAR and the development of AKI. We stimulated HK-2 cells with suPAR to investigate its effects on HK-2 cells. Additionally, Methods such as receptor inhibitors, protein docking, and co—immunoprecipitation were used to study how suPAR acts on HK-2 cells. We further explored whether the uPAR monoclonal antibody could alleviate acute kidney injury in septic mice. RESULTS: We found that serum suPAR levels were significantly elevated in patients with AKI. In addition, total suPAR/uPAR was elevated in the renal cortex of AKI mice, and serum suPAR levels were also increased. In vitro cell experiments demonstrated that suPAR stimulation promoted endoplasmic reticulum stress (ER stress) and the expression of apoptosis—related proteins in HK-2 cells and increased intracellular reactive oxygen species (ROS). Consistently, mice injected intraperitoneally with recombinant suPAR also exhibited elevated ER stress in the renal cortex. Furthermore, we discovered that suPAR was immunoprecipitated with the receptor of advanced glycation end products (RAGE), and recombinant suPAR labeled with FITC was fluorescently colocalized with RAGE on HK-2 cells, indicating that RAGE was involved in the signal transduction of suPAR. Additionally, protein docking results showed that suPAR can form a protein–protein complex with RAGE through hydrogen bonds. Pretreatment with uPAR monoclonal antibody alleviated kidney injury in septic mice and reduced the levels of ROS, apoptosis, and endoplasmic reticulum stress in the kidneys of septic AKI mice. CONCLUSIONS: Our study demonstrates that high levels of suPAR are positively correlated with the occurrence of AKI. suPAR promotes endoplasmic reticulum stress and increases susceptibility to apoptosis in renal tubular epithelial cells. RAGE on the cell membrane can bind to suPAR, participating in the activation of suPAR-mediated endoplasmic reticulum stress pathways and the expression of apoptosis-related proteins. Pretreatment with uPAR monoclonal antibody alleviates acute kidney injury in septic mice. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s10020-025-01352-w.