Prediction Of Sepsis Research Articles

Association of low vitamin D level and full-term early-onset neonatal sepsis; a case-control study

BackgroundSepsis is one of the main causes of death in newborns worldwide. Vitamin D levels during fetal and neonatal periods have a significant role in the development of the immunological system. The study aims to evaluate the association between vitamin D levels and the risk of early-onset neonatal sepsis in full-term neonates in a developing country.MethodsThis case–control study was conducted at the Neonatal Intensive Care Units (NICUs) of Kasr Alainy Hospital, Cairo, Egypt. The study was composed of two groups; the sepsis group involved full-term neonates appropriate for gestational age with sepsis-related clinical signs. The control group included newborns with no signs of clinical/laboratory infection within 72 h of life. Blood samples were collected on admission during the first three days of life in both groups for the measurement of 25-hydroxyvitamin D levels, Complete Blood Count (CBC), C reactive protein (CRP), and blood culture.ResultsForty-five newborns with clinical and laboratory findings of early-onset neonatal sepsis within 72 h of life were enrolled, and the control group included forty-five newborns with no evidence of sepsis. Vitamin D levels in the sepsis group were significantly lower than in the control group. Apgar score at the first minute was significantly lower in the sepsis group. 57.8% of neonates with sepsis had positive blood cultures. There was a statistical difference between deficient, insufficient, and sufficient vitamin D levels regarding the duration of the NICU stay, which was longer in neonates with deficient vitamin D levels. CRP was significantly higher in neonates with deficient vitamin D levels. The area under the receiver operating characteristic curve for serum vitamin D in the prediction of neonatal sepsis was 0.76 at a cutoff < 19.7(ng/ml).ConclusionIn the current study, full-term newborns with EOS had considerably lower vitamin D levels than healthy controls. Through appropriate vitamin supplementation of the mothers during pregnancy, it could be possible to ensure adequate vitamin D levels for newborns. This may contribute to the reduction of the risk of EOS, together with the other well-known preventive measures (i.e. breastfeeding and intrapartum antibiotic prophylaxis).

A Letter about “Evaluation of Sepsis Prediction Models before Onset of Treatment”

A Letter about “Evaluation of Sepsis Prediction Models before Onset of Treatment”

Early prediction of sepsis-induced respiratory tract infection using a biomarker-based machine-learning algorithm

Early and differential diagnosis of sepsis is essential to avoid unnecessary antibiotic use and further reduce patient morbidity and mortality. Here, we aimed to identify predictors of sepsis and advance a machine-learning strategy to predict sepsis-induced respiratory tract infection (RTI). Patients with sepsis and RTI were selected via retrospective analysis, and essential population characteristics and laboratory parameters were recorded. To improve the performance of the primary model and avoid over-fitting, a recursive feature elimination with cross-validation (RFECV) strategy was used to screen the optimal subset of biomarkers and construct nine machine-learning models based on this subset; the average accuracy, precision, recall, and F1-score were used for evaluation of the models. We identified 430 patients with sepsis and 686 patients with RTI. A total of 39 features were collected, with 23 features identified for initial model construction. Using the RFECV algorithm, we found that the XGBoost classifier, which only needed to include seven biomarkers, demonstrated the best performance among all prediction models, with an average accuracy of 89.24 ± 2.28, while the Ridge classifier, which included 11 biomarkers, had an average accuracy of only 83.87 ± 4.69. The remaining models had prediction accuracies greater than 88%. We developed nine models for predicting sepsis using a strategy that combined RFECV with machine learning. Among these models, the XGBoost classifier, which included seven biomarkers, showed the best performance and highest accuracy for predicting sepsis and may be a promising tool for the timely identification of sepsis.

Role of growth arrest-specific protein 6 in diagnosis and prognosis of sepsis in infants and children

Background Sepsis is a main cause of mortality in the pediatrics and neonates. Early goal-directed therapy is the cornerstone in sepsis management that is focuses on keeping organ perfusion and function normally. Diagnosis of pediatric sepsis in the emergency department is considered a significant challenge for clinicians for many reasons. These include a lack of fast and accurate diagnostic tools, an absence of simple and specific diagnostic criteria, and unclear symptoms during the early phase of illness. The best predictive tool in the future will likely be a combination of multiple biomarkers and physiomarkers. Growth arrest-specific protein 6 (Gas6) is an anti-apoptotic vitamin-K-dependent protein. It is secreted by endothelial cells and is important for the activation of endothelium during inflammation. Several studies hypothesize the use of Gas6 as a biomarker in the pathophysiology of sepsis, an early predictor of mortality, and were able to identify some life-threatening sepsis complications. Aim Evaluation of the diagnostic and prognostic values of plasma Gas6 in patients with sepsis, septic shock, and multiple organ dysfunction syndrome in infants and children. Patients and methods Case-control study was carried out from April 2019 to April 2020. 90 children were included and divided into two equal groups (cases and controls). The patients aged from 1 month to 18 years were diagnosed with sepsis, septic shock, or multiple organ dysfunction syndrome in the pediatric intensive care unit of Mansoura University Children’s Hospital. The level of plasma Gas6 was measured in both cases and controls and evaluated as regards its relation to diagnosis and outcome of the cases. Results The two groups were age and sex matched. The outcome of the cases group was: 16 cases are discharged, and 29 died. A significant elevation of plasma Gas6 level in the cases more than the controls (sensitivity= 84.4% and specificity= 66.7%). However, no statistically significant difference as regard the relation between plasma Gas6 level and the outcome of the cases (sensitivity= 55.2% and specificity= 62.7%). Conclusion Gas6 show statistically significant elevation in cases of sepsis, septic shock and multiorgan failure. Gas6 plasma level has high accuracy in prediction of sepsis and can be considered as a good negative marker of sepsis but not a reliable prognostic marker in septic patients.

Prevalence and predictors of multidrug-resistant bacteremia in liver cirrhosis.

Improved knowledge of local epidemiology and predicting risk factors of multidrug-resistant (MDR) bacteria are required to optimize the management of infections. This study examined local epidemiology and antibiotic resistance patterns of liver cirrhosis (LC) patients and evaluated the predictors of MDR bacteremia in Korea. This was a retrospective study including 140 LC patients diagnosed with bacteremia between January 2017 and December 2022. Local epidemiology and antibiotic resistance patterns and the determinants of MDR bacteremia were analyzed using logistic regression analysis. The most frequently isolated bacteria, from the bloodstream, were Escherichia coli (n = 45, 31.7%) and Klebsiella spp. (n = 35, 24.6%). Thirty-four isolates (23.9%) were MDR, and extended-spectrum beta-lactamase E. coli (52.9%) and methicillin-resistant Staphylococcus aureus (17.6%) were the most commonly isolated MDR bacteria. When Enterococcus spp. were cultured, the majority were MDR (MDR 83.3% vs. 16.7%, p = 0.003), particularly vancomycin-susceptible Enterococcus faecium. Antibiotics administration within 30 days and/or nosocomial infection was a significant predictor of MDR bacteremia (OR: 3.40, 95% CI: 1.24-9.27, p = 0.02). MDR bacteremia was not predicted by sepsis predictors, such as positive systemic inflammatory response syndrome (SIRS) or quick Sequential Organ Failure Assessment (qSOFA). More than 70% of strains that can be treated with a third-generation cephalosporin have been cultured. In cirrhotic patients, antibiotic administration within 30 days and/or nosocomial infection are predictors of MDR bacteremia; therefore, empirical administration of broad-spectrum antibiotics should be considered when these risk factors are present.

Early prediction of Sepsis: A comparative assessment on patients’ covariates

The early prediction of Sepsis is a highly important task regarding its impact on the mortality ratio for ICU patients. The earlier the Sepsis occurrence is foreseen, the sooner the patient can have the chance to be intervened and the patient can be prevented from getting major organ damage. However, it is a challenging task for healthcare professionals even though they have expert knowledge and standardized prediction scoring systems. Therefore the machine learning models that evaluate the patients’ regularly measured covariates as features, and make predictions by utilizing them have been acknowledged and have gathered attention to automatize and improve the early prediction process. However, the nature of the problem of early prediction of Sepsis requires patients to make an estimation based on past measurements, and reliable estimation cannot be made with the use of instant measurement results alone. To this end, additional features obtained by recent measurements have been proposed to be involved in the machine learning model’s estimation process by basing a specific dataset with a specific definition of the illness. This study broadly evaluates the previously proposed features to measure the effectiveness of their single and joint use. Unlike the previous research, this evaluation was done by multiple datasets, different types of additional features, and various machine learning methods for validation purposes. Experimental results have proved that the proposed machine learning model has a significant effect on early prediction, and the best improvement is obtained with the XGBoost classifier. However, it should be noted that quantity, especially the quality of the features used to feed the model has a critical effect on the model’s performance. Consequently, using additional features enhances the model’s performance dramatically.

Clinician perspectives on how situational context and augmented intelligence design features impact perceived usefulness of sepsis prediction scores embedded within a simulated electronic health record.

Obtain clinicians' perspectives on early warning scores (EWS) use within context of clinical cases. We developed cases mimicking sepsis situations. De-identified data, synthesized physician notes, and EWS representing deterioration risk were displayed in a simulated EHR for analysis. Twelve clinicians participated in semi-structured interviews to ascertain perspectives across four domains: (1) Familiarity with and understanding of artificial intelligence (AI), prediction models and risk scores; (2) Clinical reasoning processes; (3) Impression and response to EWS; and (4) Interface design. Transcripts were coded and analyzed using content and thematic analysis. Analysis revealed clinicians have experience but limited AI and prediction/risk modeling understanding. Case assessments were primarily based on clinical data. EWS went unmentioned during initial case analysis; although when prompted to comment on it, they discussed it in subsequent cases. Clinicians were unsure how to interpret or apply the EWS, and desired evidence on its derivation and validation. Design recommendations centered around EWS display in multi-patient lists for triage, and EWS trends within the patient record. Themes included a "Trust but Verify" approach to AI and early warning information, dichotomy that EWS is helpful for triage yet has disproportional signal-to-high noise ratio, and action driven by clinical judgment, not the EWS. Clinicians were unsure of how to apply EWS, acted on clinical data, desired score composition and validation information, and felt EWS was most useful when embedded in multi-patient views. Systems providing interactive visualization may facilitate EWS transparency and increase confidence in AI-generated information.

An interpretable model for sepsis prediction using multi-objective rule extraction

An interpretable model for sepsis prediction using multi-objective rule extraction

Sepsis and case fatality rates and associations with deprivation, ethnicity, and clinical characteristics: population-based case–control study with linked primary care and hospital data in England

PurposeSepsis is a life-threatening organ dysfunction caused by dysregulated host response to infection. The purpose of the study was to measure the associations of specific exposures (deprivation, ethnicity, and clinical characteristics) with incident sepsis and case fatality.MethodsTwo research databases in England were used including anonymized patient-level records from primary care linked to hospital admission, death certificate, and small-area deprivation. Sepsis cases aged 65–100 years were matched to up to six controls. Predictors for sepsis (including 60 clinical conditions) were evaluated using logistic and random forest models; case fatality rates were analyzed using logistic models.Results108,317 community-acquired sepsis cases were analyzed. Severe frailty was strongly associated with the risk of developing sepsis (crude odds ratio [OR] 14.93; 95% confidence interval [CI] 14.37–15.52). The quintile with most deprived patients showed an increased sepsis risk (crude OR 1.48; 95% CI 1.45–1.51) compared to least deprived quintile. Strong predictors for sepsis included antibiotic exposure in prior 2 months, being house bound, having cancer, learning disability, and diabetes mellitus. Severely frail patients had a case fatality rate of 42.0% compared to 24.0% in non-frail patients (adjusted OR 1.53; 95% CI 1.41–1.65). Sepsis cases with recent prior antibiotic exposure died less frequently compared to non-users (adjusted OR 0.7; 95% CI 0.72–0.76). Case fatality strongly decreased over calendar time.ConclusionGiven the variety of predictors and their level of associations for developing sepsis, there is a need for prediction models for risk of developing sepsis that can help to target preventative antibiotic therapy.

New-onset organ dysfunction as a screening tool for the identification of sepsis and outcome prediction in dogs with systemic inflammation.

Sepsis in people is defined as a life-threatening organ dysfunction (OD) caused by a dysregulated host response to infection. In veterinary medicine, sepsis is still defined by the presence of systemic inflammation plus the evidence of infection. Based on recent veterinary studies, multiorgan dysfunction syndrome (MODS) has been associated with a worse outcome in sepsis. Thus, the screening for OD is warranted to identify the most critically ill patients. The aim of this study was to investigate the diagnostic value of new-onset OD for the prediction of sepsis and outcome in a population of critically ill dogs with systemic inflammation. Dogs admitted to the Emergency Room and/or the Intensive Care Unit with systemic inflammation, defined by a serum C-reactive protein concentration > 1.6 mg/dL, were retrospectively included. Enrolled dogs were categorized according to the presence of sepsis or non-infectious systemic inflammation. The presence of newly diagnosed OD was assessed based on criteria adapted from human literature and previously reported canine criteria. 275 dogs were included: 128 had sepsis and 147 had non-infectious systemic inflammation. The frequency of new-onset OD was not different between these groups. Only the presence of fluid-refractory hypotension was significantly associated with a diagnosis of sepsis (OR 10.51, 3.08-35.94; p < 0.0001). The frequency of at least two ODs was significantly higher in non-survivors compared to survivors, according to both the human and the veterinary criteria considered for the study (p = 0.0001 and p = 0.0004, respectively). Specifically, the presence of acute kidney injury, stupor or coma, prolonged Prothrombin Time and decreased Base Excess were associated with a higher risk of death in the multivariate binary logistic regression. In this population of critically ill dogs with systemic inflammation, the detection of newly diagnosed ODs was not able to predict sepsis diagnosis, other than the presence of fluid-refractory hypotension. However, given the strong prognostic significance associated with ODs, our results support the early screening for ODs in any severe inflammatory critical care condition to identify high-risk patients and optimize their management.

Clinical expert-assigned ground truth for sepsis prediction on admission to the ICU

Clinical expert-assigned ground truth for sepsis prediction on admission to the ICU

Application of Machine Learning and its use in determining Sepsis Prediction

Sepsis is a potentially life-threatening and serious condition that occurs when an infection spreads across the body and triggers a widespread inflammatory response. It has a significantly high death rate, especially for patients in the ICU. Early detection and treatment of sepsis is necessary. Machine learning models for sepsis detection can be trained on a variety of data sources, such as electronic health records (EHRs), vital signs, laboratory test results, and demographic information. Exploratory Data analysis was performed using different pre-processing techniques. To classify the disease, the Random Forest Classifier is used, and comparison also performed among various Classifiers like MLP, KNN, etc. Key Words: Sepsis Prediction, Random Forest Classifier, vital signs.

A tagging polymorphism in fat mass and obesity-associated (FTO) gene is associated with sepsis status in children.

Sepsis is one of the most common causes of death in patients admitted to intensive care units (ICUs). The development of sepsis is significantly influenced by genetic predisposition. In this study, we highlight a potential association between a variant of the fat mass and obesity-associated (FTO) gene and risk of sepsis in children and adolescents. We investigated a first-intron tagging FTO polymorphism (rs17817449) by comparing a severe condition (SC) group, comprising 598 paediatric patients (ages 0-19 years) admitted to an ICU with fever, systemic inflammatory response syndrome (SIRS), sepsis, severe sepsis, septic shock, or multiple organ dysfunction syndrome (MODS), with a control group consisting of 616 healthy young adults. We observed a lower prevalence (p < 0.01; OR = 0.59, 95% CI = 0.39-0.87) of the FTO TT genotype in febrile and SIRS patients compared to patients with severe illness. There was a borderline trend towards a lower prevalence of the FTO TT genotype in the control group compared to the SC group (p < 0.09, OR = 0.81, 95% CI = 0.62-1.06). Our findings suggest that rs17817449, a common FTO polymorphism, may be a predictor of sepsis in paediatric patients, and that higher body weight is protective against this clinical complication.

Clinical Characteristics and Predictors of Sepsis in Otogenic Meningitis

Clinical Characteristics and Predictors of Sepsis in Otogenic Meningitis

Applying Machine Learning to Blood Count Data Predicts Sepsis with ICU Admission.

Timely diagnosis is crucial for sepsis treatment. Current machine learning (ML) models suffer from high complexity and limited applicability. We therefore created an ML model using only complete blood count (CBC) diagnostics. We collected non-intensive care unit (non-ICU) data from a German tertiary care centre (January 2014 to December 2021). Using patient age, sex, and CBC parameters (haemoglobin, platelets, mean corpuscular volume, white and red blood cells), we trained a boosted random forest, which predicts sepsis with ICU admission. Two external validations were conducted using data from another German tertiary care centre and the Medical Information Mart for Intensive Care IV database (MIMIC-IV). Using the subset of laboratory orders also including procalcitonin (PCT), an analogous model was trained with PCT as an additional feature. After exclusion, 1 381 358 laboratory requests (2016 from sepsis cases) were available. The CBC model shows an area under the receiver operating characteristic (AUROC) of 0.872 (95% CI, 0.857-0.887). External validations show AUROCs of 0.805 (95% CI, 0.787-0.824) for University Medicine Greifswald and 0.845 (95% CI, 0.837-0.852) for MIMIC-IV. The model including PCT revealed a significantly higher AUROC (0.857; 95% CI, 0.836-0.877) than PCT alone (0.790; 95% CI, 0.759-0.821; P < 0.001). Our results demonstrate that routine CBC results could significantly improve diagnosis of sepsis when combined with ML. The CBC model can facilitate early sepsis prediction in non-ICU patients with high robustness in external validations. Its implementation in clinical decision support systems has strong potential to provide an essential time advantage and increase patient safety.

Early Physician Gestalt Versus Usual Screening Tools for the Prediction of Sepsis in Critically Ill Emergency Patients

ObjectiveCompare physician gestalt to existing screening tools for the identification of sepsis in the initial minutes of presentation when time sensitive treatments must be initiated. MethodsThis prospective observational study conducted with consecutive encounter sampling took place in the emergency department of an academic, urban, safety net hospital between September 2020 and May 2022. The study population included critically ill emergency department patients excluding traumas, transfers, and self-evident diagnoses. Emergency physician gestalt was measured using a visual analog scale (VAS) from 0-100 at 15 and 60 minutes after patient arrival. The primary outcome was an explicit sepsis hospital discharge diagnosis. Clinical data were recorded up to 3 hours to compare Systemic Inflammatory Response Syndrome (SIRS), Sequential Organ Failure Assessment (SOFA), quick SOFA (qSOFA), Modified Early Warning Score (MEWS) and a logistic regression machine learning model utilizing Least Absolute Shrinkage and Selection Operator (LASSO) for variable selection. The screening tools were compared utilizing ROC analysis and calculation of AUC. Results2,484 patient-physician encounters involving 59 attending physicians were analyzed. 275 patients (11%) received an explicit sepsis discharge diagnosis. When limited to available data at 15 minutes, initial VAS (AUC 0.90; 95% CI 0.88-0.92) outperformed all tools including LASSO (0.84, 95% CI 0.82-0.87), qSOFA (0.67; 95% CI 0.64-0.71), SIRS (0.67; 95% 0.64-0.70), SOFA (0.67; 95% CI 0.63-0.70) and MEWS (0.66; 95% CI 0.64-0.69). Expanding to data available at 60 minutes did not meaningfully change results. ConclusionsAmong adults presenting to an emergency department with an undifferentiated critical illness, physician gestalt in the first 15 minutes of the encounter outperformed other screening methods in identifying sepsis.

Early warning scores for sepsis identification and prediction of in-hospital mortality in adults with sepsis: A systematic review and meta-analysis.

The early warning scores (EWS), quick Sequential Organ Failure Assessment (qSOFA) and systemic inflammatory response syndrome (SIRS) criteria have been proposed as sepsis screening tools. This review aims to summarise and compare the performance of EWS with the qSOFA and SIRS criteria for predicting sepsis diagnosis and in-hospital mortality in patients with sepsis. A systematic review with meta-analysis. Seven databases were searched from January 1, 2016 until March 10, 2022. Study quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies 2 tool. Sensitivity, specificity, likelihood ratios and diagnostic odd ratios were pooled by using the bivariate random effects model. Overall performance was summarised by using the hierarchical summary receiver-operating characteristics curve. This paper adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses of Diagnostic Test Accuracy Studies (PRISMA-DTA) guidelines. Ten studies involving 52,474 subjects were included in the review. For predicting sepsis diagnosis, the pooled sensitivity of EWS (65%, 95% CI: 55, 75) was similar to SIRS ≥2 (70%, 95% CI: 49, 85) and higher than qSOFA ≥2 (37%, 95% CI: 20, 59). The pooled specificity of EWS (77%, 95% CI: 64, 86) was higher than SIRS ≥2 (62%, 95% CI: 41, 80) but lower than qSOFA ≥2 (94%, 95% CI: 86, 98). Results were similar for the secondary outcome of in-hospital mortality. Although no one scoring system had both high sensitivity and specificity, the EWS had at least equivalent values in most measures of diagnostic accuracy compared with SIRS or qSOFA. Healthcare systems in which EWS is already in place should consider whether there is any clinical benefit in adopting qSOFA or SIRS. This systematic review did not directly involve patient or public contribution to the manuscript.

Serum IgG and IgA level in children with acute lymphoblastic leukaemia and its relation with outcome in induction chemotherapy

Background: Immunosuppression in children with Acute Lymphoblastic Leukaemia (ALL) escalates vulnerability to severe infectious complications which leads to higher morbidity and mortality rates among patients. Immunoglobulins are important part of defence mechanism of the body contributing to the humoral immunity. Of all the immunoglobulins, IgG and IgA represents most common type of antibody in blood circulation and provides most long-lived antibody-based immunity against invading pathogens. In earlier years, studies have shown that, serum immunoglobulins are found to be decreased during chemotherapy and this reduction is related with severe sepsis and death in patients. Early identification of immune status can be much useful for assessment of severity and better risk stratification of the patients. Objective: Evaluation of serum IgG and IgA levels in children with ALL and its relationship with sepsis in induction phase of chemotherapy. Method: This prospective observational study was conducted in the Departments of Paediatric Haematology &amp; Oncology and Department of General Paediatrics, Bangabandhu Sheikh Mujib Medical University, Dhaka. During January 2022 to December 2022, a total 60 paediatric patients of newly diagnosed ALL were included as study population. As control, 30 patients were included in this study from General paediatrics and Paediatric Haematology &amp; Oncology, BSMMU to compare baseline immunoglobulin levels with the study group. Complete history and physical examination of all patients were undertaken, and all prior reports were collected. Serum IgG and IgA level assay was done before starting chemotherapy and at the end of induction chemotherapy in the study group. Patients diagnosed with ALL received chemotherapy modified UK-ALL 2003 protocol. History and data of any event of sepsis, along with the symptoms of complications were taken from all patients during the course of induction chemotherapy. Results: The mean age of the cases were 5.49±3.15 (mean ± SD) years, which ranged from 1.9 to 15 years. Among the cases, 41 patients (68.3%) were male and 19 (31.7%) females. Seven (11.7%) subjects expired during induction chemotherapy. In comparison to the control group the mean IgG level in the study group was lower at baseline and even lower at the end of induction and it was statistically significant (p&lt;0.05). The mean IgG level was low in patients with septic events in comparison with the patients with no septic events. Difference of mean of IgG between septic and aseptic patients was significant only at the end of induction chemotherapy (p&lt;0.05). The mean IgA level was low in study group in comparison with the control group and it was statistically significant (p&lt;0.05) only at the end of induction chemotherapy. The mean IgA level was low in patients with septic events in comparison with the patients with no septic events. Difference of mean of IgA between septic and aseptic patients was significant (p&lt;0.05). Based on the receiver-operator characteristic (ROC) curves of change of S. IgG, with a cut off value 50% decreased from baseline having 68.8% sensitivity and 29.7% specificity for prediction of sepsis. Considering change of S. IgA with a cut off value 50% decreased from baseline having 67.6% sensitivity and 27.5% specificity for prediction of sepsis. Conclusion: In our study, the mean IgG and IgA level of the cases were low in comparison to the control group. The mean IgG and IgA level of the patients were lower who had infectious complications, and it was more marked after induction chemotherapy. Therefore, evaluation of immunoglobulin profiles is recommended to be considered in patients with ALL in further studies.

Assessment of Sepsis Risk at Admission to the Emergency Department: Clinical Interpretable Prediction Model.

This study aims to develop an interpretable prediction model based on explainable artificial intelligence to predict bacterial sepsis and discover important biomarkers. A total of 1572 adult patients, 560 of whom were sepsis positive and 1012 of whom were negative, who were admitted to the emergency department with suspicion of sepsis, were examined. We investigated the performance characteristics of sepsis biomarkers alone and in combination for confirmed sepsis diagnosis using Sepsis-3 criteria. Three different tree-based algorithms-Extreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), Adaptive Boosting (AdaBoost)-were used for sepsis prediction, and after examining comprehensive performance metrics, descriptions of the optimal model were obtained with the SHAP method. The XGBoost model achieved accuracy of 0.898 (0.868-0.929) and area under the ROC curve (AUC) of 0.940 (0.898-0.980) with a 95% confidence interval. The five biomarkers for predicting sepsis were age, respiratory rate, oxygen saturation, procalcitonin, and positive blood culture. SHAP results revealed that older age, higher respiratory rate, procalcitonin, neutrophil-lymphocyte count ratio, C-reactive protein, plaque, leukocyte particle concentration, as well as lower oxygen saturation, systolic blood pressure, and hemoglobin levels increased the risk of sepsis. As a result, the Explainable Artificial Intelligence (XAI)-based prediction model can guide clinicians in the early diagnosis and treatment of sepsis, providing more effective sepsis management and potentially reducing mortality rates and medical costs.

Sepsis incidence, suspicion, prediction and mortality in emergency medical services: a cohort study related to the current international sepsis guideline

PurposeSepsis suspicion by Emergency Medical Services (EMS) is associated with improved patient outcomes. This study assessed sepsis incidence and recognition by EMS and analyzed which of the screening tools recommended by the Surviving Sepsis Campaign best facilitates sepsis prediction.MethodsRetrospective cohort study of claims data from health insurances (n = 221,429 EMS cases), and paramedics’ and emergency physicians’ EMS documentation (n = 110,419); analyzed outcomes were: sepsis incidence and case fatality compared to stroke and myocardial infarction, the extent of documentation for screening-relevant variables and sepsis suspicion, tools’ intersections for screening positive in identical EMS cases and their predictive ability for an inpatient sepsis diagnosis.ResultsIncidence of sepsis (1.6%) was similar to myocardial infarction (2.6%) and stroke (2.7%); however, 30-day case fatality rate was almost threefold higher (31.7% vs. 13.4%; 11.8%). Complete vital sign documentation was achieved in 8.2% of all cases. Paramedics never, emergency physicians rarely (0.1%) documented a sepsis suspicion, respectively septic shock. NEWS2 had the highest sensitivity (73.1%; Specificity:81.6%) compared to qSOFA (23.1%; Sp:96.6%), SIRS (28.2%; Sp:94.3%) and MEWS (48.7%; Sp:88.1%). Depending on the tool, 3.7% to 19.4% of all cases screened positive; only 0.8% in all tools simultaneously.ConclusionIncidence and mortality underline the need for better sepsis awareness, documentation of vital signs and use of screening tools. Guidelines may omit MEWS and SIRS as recommendations for prehospital providers since they were inferior in all accuracy measures. Though no tool performed ideally, NEWS2 qualifies as the best tool to predict the highest proportion of septic patients and to rule out cases that are likely non-septic.