Bloodstream infections (BSIs) are a severe public health threat due to their rapid progression into critical conditions like sepsis. This study presents a novel eXplainable Artificial Intelligence (XAI) framework to predict BSIs using historical electronic health records (EHRs). Leveraging a dataset from St. Olavs Hospital in Trondheim, Norway, encompassing 35,591 patients, the framework integrates demographic, laboratory, and comprehensive medical history data to classify patients into high-risk and low-risk BSI groups. By avoiding reliance on real-time clinical data, our model allows for enhanced scalability across various healthcare settings, including resource-limited environments. The XAI framework significantly outperformed traditional models, particularly with tree-based algorithms, demonstrating superior specificity and sensitivity in BSI prediction. This approach promises to optimize resource allocation and potentially reduce healthcare costs while providing interpretability for clinical decision-making, making it a valuable tool in hospital systems for early intervention and improved patient outcomes.
Infection in the immunocompromised patient is often challenging on multiple levels. It can be difficult to distinguish between manifestations of the underlying disease, infection or malignancy. Symptoms may be vague or even absent, deviations in the common inflammatory parameters discrete, imaging findings scarce and the causative microbe may be a true pathogen as well as opportunistic. Here, we report an immunosuppressed female in her late teens with a purulent meningitis due to Ureaplasma parvum—a very rare cause of infection in the central nervous system of adults. We wish to highlight the relevance of intracellular pathogens and the need to actively search for these microbes, especially when response to broad-spectrum antibiotic treatment is absent. Furthermore, we emphasise the need for adequate molecular microbial diagnostics in search of microbes that are difficult to identify by culture and where serology and antigen tests may be absent or unreliable due to immune suppression.
Background: Recent diagnostic methods have enabled the detection of often culture-negative pathogens, including anaerobic bacteria from the oral cavity. Characterising the microbial diversity and co-occurrence of bacteria in such infections is important for understanding the molecular pathophysiology in odontogenic brain abscesses. Case presentation: We describe a case of polymicrobial odontogenic brain abscess in a 59-year-old man of Vietnamese ethnicity with a documented increased risk of brain abscess due to Hereditary Haemorrhagic Telangiectasia (HHT). The microbiological diagnostic work-up included conventional culture, matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS), targeted 16S rDNA analysis using three broad-range group-specific PCR (polymerase chain reaction) assays and next-generation sequencing (NGS). A literature review was conducted, including database searches for each identified microorganism. Twelve anaerobic bacterial species were detected, among which Treponema medium, Capnocytophaga HMT-323 and Candidatus Saccharibacteria oral taxon 488 have not previously been reported in brain abscesses. In addition, we identified the extremely rare pathogens Arachnia propionica and Capnocytophaga ochracea. Conclusion: This is the first report of Ca. Saccharibacteria oral taxon 488 in a clinical sample and the first detection of any species from this phylum in a brain abscess, co-detected with A. propionica, consistent with its obligate epibiotic lifestyle. Our findings broaden the known microbial diversity associated with odontogenic brain abscesses and underscore the value of 16S rDNA NGS in characterising polymicrobial infections, particularly when fastidious or uncultivable organisms are involved.
Bloodstream infections (BSIs) represent a critical public health concern, primarily due to their rapid progression and severe implications such as sepsis and septic shock. This study introduces an innovative Explanable Artificial Intelligence (XAI) framework, leveraging historical electronic health records (EHRs) to enhance BSI prediction. Unlike traditional models that rely heavily on real-time clinical data, our XAI-based approach utilizes a comprehensive dataset incorporating demographic data, laboratory results, and full medical histories from St. Olavs Hospital, Trondheim, Norway, covering 35,591 patients between 2015 and 2020. We developed models to differentiate between high-risk and low-risk BSI cases effectively, optimizing healthcare resource allocation and potentially reducing healthcare costs. Our results demonstrate superior predictive accuracy, particularly the tree-based models, which significantly outperformed contemporary models in both specificity and sensitivity metrics. Author Summary In this research, we have developed a new tool that uses artificial intelligence to better predict bloodstream infections, which can lead to serious conditions like sepsis if not quickly identified and treated. It is the first of its kind framework that analyzes past health records and helps identify patients who are at high risk of infection more accurately than existing tools. Unlike existing tools our framework can be implemented at any stage of the patient trajectory and is the only framework to achieve good accuracy without the use of intimate patient features such as vital signs. This ability could enable doctors to prioritize care more pre-emptively, effectively, potentially saving lives and reducing unnecessary medical tests. Our approach is designed to be easily understood and used by both medical professionals and those with little technical expertise, making it a valuable addition to hospital systems.
Sepsis is characterized by profound immune dysregulation, including impaired innate immune responses and epigenetic reprogramming of monocytes. However, strategies to therapeutically restore immune function remain limited. Here, we identify a cell-penetrating peptide, P7-Pen, as a modulator of monocyte epigenetic state and inflammatory responsiveness. Using primary human monocytes and peripheral blood mononuclear cells (PBMCs) from healthy donors and sepsis patients, we demonstrate that P7-Pen enhances cytokine production in response to Toll-like receptor stimulation while having minimal effects under basal conditions. P7-Pen treatment increased global histone acetylation, particularly H3 acetylation, and prevented the development of endotoxin tolerance. Transcriptomic and functional analyses revealed restoration of inflammatory gene expression, including TNF, IL6, and IFNB1, in otherwise hyporesponsive cells. Mechanistically, we identified the lysine deacetylase ABHD14B as a direct binding partner of P7-Pen. Silencing of ABHD14B recapitulated the effects of P7-Pen, leading to enhanced histone acetylation and cytokine production. Importantly, P7-Pen selectively potentiated responses to TLR ligands without inducing basal hyperinflammation. Collectively, our findings identify ABHD14B-dependent epigenetic regulation as a key checkpoint in innate immune tolerance and establish P7-Pen as a novel tool to restore immune responsiveness in sepsis.
25 June 2026 in European Journal of Clinical Microbiology & Infectious Diseases
Purpose: To describe BC collection practices at a tertiary care university hospital in a prospective observational study and examine the association between blood volume and culture yield. Methods: Analysis of all blood cultures collected during a 6-week period in 2023 at St Olav’s University Hospital, Norway. Additionally, we analysed impact of antimicrobial treatment prior to sampling, sampling department, bottle type, sampling time and method, microorganisms isolated as well as patient age and sex. Results: Only 19.7% of BC bottles were filled with the recommended 8–10 mL of blood. The overall pathogen positivity rate of all BC bottles was 6.8%, while 13.3% of the patients had at least one positive BC bottle. Patients >60 years were twice as likely to have a positive blood culture as those aged 18–60 years (16.2% vs 8.1%). Recommended filling volumes were positively associated with recovery of microbes, whereas particularly overfilling >14 mL was associated with reduced microbial yield. Overall, bottles with no growth contained a significantly higher mean blood volume than bottles with growth. Conclusion: This study highlights the gap between established recommendations for BC collection and current practice at our hospital. Preanalytical optimisation of BC collection is needed to improve diagnostic yield in this seriously ill patient population. Overfilling of BC bottles was negatively associated with microbial yield and should be avoided.