Predictors Of Bacteremia Research Articles

P-850. External Validation of Multiple Bacteremia Prediction Models in a Large US Emergency Department

Abstract Background We previously proposed bacteremia prediction models that might save a quarter of blood cultures (BCs) while missing few bacteremia patients. In this study, we aimed to validate these models in the United States.Table 1:Discrimination, calibration, and clinical usefulness upon external validation for five bacteremia prediction models.Basic variables were age, systolic blood pressure (mmHg), heart rate (beats/min), respiratory rate (breaths/min), temperature (Celsius) and Glasgow coma scale. The basic model was tested in 46236 patients, the model with CRP was tested in 8393 patients, the model with PCT and the only PCT model was tested in 12864 patients and the model with both CRP and PCT was tested in 2317 patients. A sensitivity of 95% means that out of 100 patients with bacteremia, the model detects 95 patients and misses 5 patients. CI = confidence interval, BCs = blood cultures, CRP = C-reactive protein, PCT = Procalcitonin. Methods We validated five models using retrospectively collected data. We included patients ≥ 18 years old with one or more BCs drawn during their visit to University of Michigan Hospital, Ann Arbor Emergency Department, from October 2013 until October 2023. The models used basic information such as age and vital signs, with or without the addition of C-reactive protein (CRP), procalcitonin (PCT) or both. We created cohorts based on availability of the biomarkers CRP and PCT. Missing data (except biomarkers) was statistically imputed. Discrimination was assessed with the C-statistic and calibration using calibration plots. Clinical usefulness was evaluated by sensitivity, specificity, and reduced BCs. Results We studied 46,608 patients, of whom 372 were excluded because of suspected endocarditis or indwelling vascular catheter. Median age was 61 years, 52% were male. 3816 patients had bacteremia (8.3%), while 2674 (5.8%) patients had contaminated BCs. PCT was known for 12,864 patients, CRP for 8,393 patients and 2,317 patients had both PCT and CRP biomarkers available. We observed substantial improved performance for models containing PCT (Table 1). The best performing model had a C-statistic of 0.80 (95% CI 0.79-0.82), and adequate calibration (intercept -0.20 (95% CI -0.27 to -0.14), slope 0.79 (95% CI 0.75-0.84)). All models somewhat underestimated the risk of bacteremia in the lower risk ranges, leading to more missed bacteremia patients compared to the European cohorts used to develop the models. At a minimum sensitivity of 95%, the models were able to reduce 13-30% of BCs. Conclusion Recently proposed prediction models have satisfactory discrimination and calibration for US patients. However, almost all models would have missed more bacteremic episodes compared to the European development cohorts, underlining the importance of validation in other geographical settings. Procalcitonin showed substantial predictive performance for bacteremia and should be considered for any bacteremia prediction model. Disclosures Laraine Washer, MD, Ferring Pharmaceuticals: Advisor/Consultant

Procalcitonin for safe reduction of unnecessary blood cultures in the emergency department: Development and validation of a prediction model

Blood cultures (BCs) are commonly ordered in emergency departments (EDs), while a minority yields a relevant pathogen. Diagnostic stewardship is needed to safely reduce unnecessary BCs. We aimed to develop and validate a bacteremia prediction model for ED patients, with specific focus on the benefit of incorporating procalcitonin. We included adult patients with suspected bacteremia from a Dutch ED for a one-year period. We defined 23 candidate predictors for a "full model", of which nine were used for an automatable "basic model". Variations of both models with C-reactive protein and procalcitonin were constructed using LASSO regression, with bootstrapping for internal validation. External validation was done in an independent cohort of patients with confirmed infection from 71 Spanish EDs. We assessed discriminative performance using the C-statistic and calibration with calibration curves. Clinical usefulness was evaluated by sensitivity, specificity, saved BCs, and Net Benefit. Among 2111 patients in the derivation cohort (mean age 63 years, 46% male), 273 (13%) had bacteremia, versus 896 (20%) in the external cohort (n=4436). Adding procalcitonin substantially improved performance for all models. The basic model with procalcitonin showed most promise, with a C-statistic of 0.87 (0.86-0.88) upon external validation. At a 5% risk threshold, it showed a sensitivity of 99% and could have saved 29% of BCs while only missing 10 out of 896 (1.1%) bacteremia patients. Procalcitonin-based bacteremia prediction models can safely reduce unnecessary BCs at the ED. Further validation is needed across a broader range of healthcare settings.

A Vision on User-Centered Implementation and Evaluation of Explainable AI for Predicting Hospital-Onset Bacteremia.

In recent years, artificial intelligence (AI) has gained momentum in many fields of daily live. In healthcare, AI can be used for diagnosing or predicting illnesses. However, explainable AI (XAI) is needed to ensure that users understand how the algorithm arrives at a decision. In our research project, machine learning methods are used for individual risk prediction of hospital-onset bacteremia (HOB). This paper presents a vision on a step-wise process for implementation and evaluation of user-centered XAI for risk prediction of HOB. An initial requirement analysis revealed first insights on the users' needs of explainability to use and trust such risk prediction applications. The findings were then used to propose step-wise process towards a user-centered evaluation.

The Impact of Information Relevancy and Interactivity on Intensivists' Trust in a Machine Learning-Based Bacteremia Prediction System: Simulation Study.

The exponential growth in computing power and the increasing digitization of information have substantially advanced the machine learning (ML) research field. However, ML algorithms are often considered "black boxes," and this fosters distrust. In medical domains, in which mistakes can result in fatal outcomes, practitioners may be especially reluctant to trust ML algorithms. The aim of this study is to explore the effect of user-interface design features on intensivists' trust in an ML-based clinical decision support system. A total of 47 physicians from critical care specialties were presented with 3 patient cases of bacteremia in the setting of an ML-based simulation system. Three conditions of the simulation were tested according to combinations of information relevancy and interactivity. Participants' trust in the system was assessed by their agreement with the system's prediction and a postexperiment questionnaire. Linear regression models were applied to measure the effects. Participants' agreement with the system's prediction did not differ according to the experimental conditions. However, in the postexperiment questionnaire, higher information relevancy ratings and interactivity ratings were associated with higher perceived trust in the system (P<.001 for both). The explicit visual presentation of the features of the ML algorithm on the user interface resulted in lower trust among the participants (P=.05). Information relevancy and interactivity features should be considered in the design of the user interface of ML-based clinical decision support systems to enhance intensivists' trust. This study sheds light on the connection between information relevancy, interactivity, and trust in human-ML interaction, specifically in the intensive care unit environment.

Utility of shaking chills as a diagnostic sign for bacteremia in adults: a systematic review and meta-analysis

BackgroundAccurate prediction of bacteremia is essential for guiding blood culture collection and optimal antibiotic treatment. Shaking chills, defined as a subjective chill sensation with objective body shivering, have been suggested as a potential predictor of bacteremia; however, conflicting findings exist. To address the evidence gap, we conducted a systematic review and meta-analysis of studies to assess the diagnostic accuracy of shaking chills for predicting bacteremia among adult patients.MethodsWe included studies reporting the diagnostic accuracy of shaking chills or chills for bacteremia. Adult patients with suspected bacteremia who underwent at least one set of blood cultures were included. Our main analysis focused on studies that assessed shaking chills. We searched these studies through CENTRAL, MEDLINE, Embase, the World Health Organization ICTRP Search Portal, and ClinicalTrials.gov. Study selection, data extraction, evaluation for risk of bias, and applicability using the QUADAS-2 tool were conducted by two independent investigators. We estimated a summary receiver operating characteristic curve and a summary point of sensitivity and specificity of the index tests, using a hierarchical model and the bivariate model, respectively.ResultsWe identified 19 studies with a total of 14,641 patients in which the accuracy of shaking chills was evaluated. The pooled sensitivity and specificity of shaking chills were 0.37 (95% confidence interval [CI], 0.29 to 0.45) and 0.87 (95% CI, 0.83 to 0.90), respectively. Most studies had a low risk of bias in the index test domain and a high risk of bias and a high applicability concern in the patient-selection domain.ConclusionsShaking chills are a highly specific but less sensitive predictor of bacteremia. Blood cultures and early initiation of antibiotics should be considered for patients with an episode of shaking chills; however, the absence of shaking chills must not lead to exclusion of bacteremia and early antibiotic treatment.

Timing and prediction of secondary bacteremia in patients with COVID-19: A retrospective cohort study.

We aimed to aid the appropriate use of antimicrobial agents by determining the timing of secondary bacteremia and validating and updating clinical prediction models for bacteremia in patients with COVID-19. We performed a retrospective cohort study on all hospitalized patients diagnosed with COVID-19 who underwent blood culture tests from January 1, 2020, and September 30, 2021, at an urban teaching hospital in Japan. The primary outcome measure was secondary bacteremia in patients with COVID-19. Of the 507 patients hospitalized with COVID-19, 169 underwent blood culture tests. Eleven of them had secondary bacteremia. The majority of secondary bacteremia occurred on or later than the 9th day after symptom onset. Positive blood culture samples collected on day 9 or later after disease onset had an odds ratio of 22.4 (95% CI 2.76-181.2, p < 0.001) compared with those collected less than 9 days after onset. The area under the receiver operating characteristic curve of the modified Shapiro rule combined with blood culture collection on or after the 9th day from onset was 0.919 (95% CI, 0.843-0.995), and the net benefit was high according to the decision curve analysis. The timings of symptom onset and hospital admission may be valuable indicators for making a clinical decision to perform blood cultures in patients hospitalized with COVID-19.

Bacteriological Profile, Antimicrobial Susceptibility Patterns and Predictors of Bacteremia in Neonates with Clinical Sepsis at KCMC Hospital, Northern Tanzania

Bacteriological Profile, Antimicrobial Susceptibility Patterns and Predictors of Bacteremia in Neonates with Clinical Sepsis at KCMC Hospital, Northern Tanzania

Combination of machine learning algorithms with natural language processing may increase the probability of bacteremia detection in the emergency department: A retrospective, big-data analysis of 94,482 patients.

Prompt diagnosis of bacteremia in the emergency department (ED) is of utmost importance. Nevertheless, the average time to first clinical laboratory finding range from 1 to 3 days. Alongside a myriad of scoring systems for occult bacteremia prediction, efforts for applying artificial intelligence (AI) in this realm are still preliminary. In the current study we combined an AI algorithm with a Natural Language Processing (NLP) algorithm that would potentially increase the yield extracted from clinical ED data. This study involved adult patients who visited our emergency department and at least one blood culture was taken to rule out bacteremia. Using both tabular and free text data, we built an ensemble model that leverages XGBoost for structured data, and logistic regression (LR) on a word-analysis technique called bag-of-words (BOW) Term Frequency-Inverse Document Frequency (TF-IDF), for textual data. All algorithms were designed in order to predict the risk for bacteremia with ED patients whose blood cultures were sent to the laboratory. The study cohort comprised 94,482 individuals, of whom 52% were males. The prevalence of bacteremia in the entire cohort was 9.7%. The model trained on the tabular data yielded an area under the curve (AUC) of 73.7% for XGBoost, while the LR that was trained on the free text achieved an AUC of 71.3%. After checking a range of weights, the best combination was for 55% weight on the XGBoost prediction and 45% weight on the LR prediction. The final model prediction yielded an AUC of 75.6%. Harnessing artificial intelligence to the task of bacteremia surveillance in the ED settings by a combination of both free text and tabular data analysis improved predictive performance compared to using tabular data alone. We recommend that future AI applications based on our findings should be assimilated into the clinical routines of ED physicians.

A Simple Bacteremia Score for Predicting Bacteremia in Patients with Suspected Infection in the Emergency Department: A Cohort Study.

Bacteremia is a life-threatening condition that has increased in prevalence over the past two decades. Prompt recognition of bacteremia is important; however, identification of bacteremia requires 1 to 2 days. This retrospective cohort study, conducted from 10 November 2014 to November 2019, among patients with suspected infection who visited the emergency department (ED), aimed to develop and validate a simple tool for predicting bacteremia. The study population was randomly divided into derivation and development cohorts. Predictors of bacteremia based on the literature and logistic regression were assessed. A weighted value was assigned to predictors to develop a prediction model for bacteremia using the derivation cohort; discrimination was then assessed using the area under the receiver operating characteristic curve (AUC). Among the 22,519 patients enrolled, 18,015 were assigned to the derivation group and 4504 to the validation group. Sixteen candidate variables were selected, and all sixteen were used as significant predictors of bacteremia (model 1). Among the sixteen variables, the top five with higher odds ratio, including procalcitonin, neutrophil-lymphocyte ratio (NLR), lactate level, platelet count, and body temperature, were used for the simple bacteremia score (model 2). The proportion of bacteremia increased according to the simple bacteremia score in both cohorts. The AUC for model 1 was 0.805 (95% confidence interval [CI] 0.785-0.824) and model 2 was 0.791 (95% CI 0.772-0.810). The simple bacteremia prediction score using only five variables demonstrated a comparable performance with the model including sixteen variables using all laboratory results and vital signs. This simple score is useful for predicting bacteremia-assisted clinical decisions.

Clinical usefulness of procalcitonin/albumin and blood urea nitrogen/albumin ratios for the early detection of bacteremia

Background: Early prediction of bacteremia is important because sepsis may develop if bacteremia is not treated in time. We aimed to determine whether blood tests are clinically useful in predicting gram-negative versus gram-positive bacteremia. Methods: We retrospectively analyzed the medical records of 16,569 patients who underwent blood culture testing among those who visited St. Mary’s Hospital of Uijeongbu, Catholic University from January 1, 2018 to December 31, 2018. After excluding 15,580 patients with insufficient data for analysis, the total study population was 989. Results: In the multivariable logistic regression analysis for gram-negative bacteremia, after adjusting for major variables, procalcitonin (odds ratio [OR] 1.02, P &lt; 0.001), high-sensitivity C-reactive protein (OR 1.03, P = 0.010), and albumin (OR 0.55, P = 0.010) were significant. In the same model, lactic acid (OR 1.08, P = 0.010), blood urea nitrogen (BUN) (OR 1.02, P = 0.010), and albumin (OR 0.50, P = 0.005) were significant for gram-positive bacteremia. We constructed blood indicators by combining blood tests that were significant in multivariable logistic regression analysis. Comparing the area under the curve (AUC) of the receiver operating characteristic curve, the AUC of the procalcitonin/albumin ratio in gramnegative bacteremia was significantly higher than that of procalcitonin (0.846 vs. 0.837, P = 0.005), and the AUC of the BUN/albumin ratio in gram-positive bacteremia was significantly higher than that of BUN (0.709 vs. 0.679, P = 0.007). Conclusion: The procalcitonin/albumin ratio is expected to be helpful in the early prediction of gram-negative bacteremia, whereas the BUN/albumin ratio in the early prediction of gram-positive bacteremia.

1031. Can We Safely Reduce Unnecessary Blood Cultures in Patients with Suspected Infection in the Emergency Department? The PredictED Study to Develop and Validate a Bacteremia Prediction Model

Abstract Background Blood cultures are commonly used at emergency departments (EDs), while only 5-10% yields a relevant pathogen. We aimed to develop an automatable prediction model in ED patients with suspected bacteremia. This may reduce the use of blood cultures and prevent potential harm from false-positive blood cultures, while minimizing the risk of missing positive cultures. Methods In this observational study, we included consecutive adult patients who had a blood culture taken at the ED of the Haga Teaching Hospital, the Netherlands, for a one-year period. Demographics, laboratory, and outcome data were collected from electronic patient records. We defined 23 candidate predictors for our “full model”, of which nine were used for the "basic" model because they are readily automatable (Table 1). Regression analysis with the Least Absolute Shrinkage and Selection Operator was used to define the model. Bootstrapping was performed for internal validation, including imputation of missing values. We assessed discriminative performance using the C-statistic and calibration using the calibration intercept and slope. Clinical utility was assessed by sensitivity, specificity, negative and positive predictive values and decision curves. Results We included 2111 unique patients; mean age 63 years and 46% male. 272 patients had true positive blood cultures (13%); 79 patients (3.7%) had contaminated blood cultures. Our basic model included 8 predictors: age, systolic blood pressure, heart rate, breathing frequency, temperature, Glasgow coma score, C-reactive protein, procalcitonin. The corrected C-statistic was 0.81 (95% CI 0.79-0.83). The full model additionally included recent antibiotic use, indwelling vascular catheter, chills, and suspected infection site. The corrected C-statistic was 0.87 (95% CI 0.85-0.88). Calibration was good for both models. Sensitivity for bacteremia was 97.4% in the basic model, saving 18% of blood cultures (Table 2). The full model could save 26% of blood cultures while maintaining a sensitivity of 98.5% (Table 3).Figure 1.Decision curve analysis Conclusion The proposed bacteremia prediction models are designed for easy implementation in electronic patient records and can reduce unnecessary blood cultures. Further external validation is needed before implementation in clinical practice. Disclosures All Authors: No reported disclosures

Procalcitonin and Risk Prediction for Diagnosing Bacteremia in Hospitalized Patients: A Retrospective, National Observational Study.

Bacteremia is associated with significant morbidity and mortality. Timely, appropriate therapy may improve clinical outcomes, and therefore, determining which patients benefit from more comprehensive diagnostic strategies (i.e., direct specimen testing) could be of value. We performed an assessment of procalcitonin (PCT) and clinical characteristics in the discrimination of bacteremic hospitalizations. We analyzed 71,105 encounters and 14,846 visits of patients with bacteremia alongside 56,259 without an admission. The area under the receiver-operating characteristic (AUROC) curve for the prediction of bacteremia via procalcitonin was 0.782 (95% CI 0.779-0.787). The prediction modeling of clinical factors with or without PCT resulted in a similar performance to PCT alone. However, the clinically predicted risk of bacteremia stratified by PCT thresholds allowed the targeting of high-incidence bacteremia groups (e.g., ≥50% positivity). The combined use of PCT and clinical characteristics could be useful in diagnostic stewardship by targeting further advanced diagnostic testing in patients with a high predicted probability of bacteremia.

Proportion of Neutrophils in White Blood Cells as a Useful Marker for Predicting Bacteremic Acute Cholangitis

Objective A positive hemoculture in acute cholangitis is serious, but a blood culture result cannot be obtained at the initial diagnosis and so cannot be used for the severity assessment and decision-making concerning urgent/early biliary drainage. Accordingly, a predictor for bacteremia at the initial diagnosis of acute cholangitis would be particularly useful. We investigated the association between neutrophil proportions in white blood cell counts (%Neutro) and bacteremic acute cholangitis. Methods Of 166 patients with acute cholangitis who were diagnosed with the Tokyo guidelines 2018/2013 from April 2015 to March 2017, a total of 94 underwent blood culture assessments and were divided into those with a positive hemoculture (n=48) and a negative hemoculture (n=46) and then compared. A receiver operating characteristic curve analysis was used to evaluate the predictive ability of %Neutro and other inflammatory markers. Results The %Neutro values were significantly higher in the positive hemoculture group than in the negative hemoculture group (91.7% ±4.0% vs. 82.5% ±9.0%, P<0.0001). A cut-off %Neutro value of 89.7% was strongly associated with bacteremia (area under the curve 0.86, sensitivity 77.1%, specificity 80.4%). A %Neutro of ≥89.7% was a predictor of a positive hemoculture in univariate (P<0.0001) and multivariate analyses (P<0.001). Patients with a %Neutro ≥89.7% needed early biliary drainage more frequently than others (30/46, 65.2% vs. 18/48, 37.5%, p=0.0063). Conclusions %Neutro is an independent predictor of bacteremia in patients with acute cholangitis and may contribute to decision-making concerning early biliary drainage.

Predictors of bacteremia in febrile infants under 3 months old in the pediatric emergency department

IntroductionFever may serve as the primary indicator of underlying infection in children admitted to the pediatric emergency department (PED), especially in high-risk young infants. This study aimed to identify early clinical factors that could help predict bacteremia in young febrile infants.MethodsThe study included infants under 90 days of age who were admitted to the PED due to fever. Patients were divided into two groups based on the presence or absence of bacteremia and further divided into three age groups: (1) less than 30 days, (2) 30 to 59 days, and (3) 60 to 90 days. Several clinical and laboratory variables were analyzed, and logistic regression and receiver operating characteristic (ROC) analyses were used to identify potential risk factors associated with bacteremia in young febrile infants.ResultsA total of 498 febrile infants were included, of whom 6.4% were diagnosed with bacteremia. The bacteremia group had a higher body temperature (BT) at triage, especially in neonates, higher pulse rates at triage, longer fever subsidence time, longer hospital stays, higher neutrophil counts, and higher C-reactive protein (CRP) levels than those of the non-bacteremia group. ROC analysis showed that the best cut-off values for predicting bacteremia in infants with pyrexia were a BT of 38.7 °C, neutrophil count of 57.9%, and CRP concentration of 53.8 mg/L.ConclusionsA higher BT at triage, increased total neutrophil count, and elevated CRP levels may be useful for identifying bacteremia in young febrile infants admitted to the PED.

Development of an artificial intelligence bacteremia prediction model and evaluation of its impact on physician predictions focusing on uncertainty

Prediction of bacteremia is a clinically important but challenging task. An artificial intelligence (AI) model has the potential to facilitate early bacteremia prediction, aiding emergency department (ED) physicians in making timely decisions and reducing unnecessary medical costs. In this study, we developed and externally validated a Bayesian neural network-based AI bacteremia prediction model (AI-BPM). We also evaluated its impact on physician predictive performance considering both AI and physician uncertainties using historical patient data. A retrospective cohort of 15,362 adult patients with blood cultures performed in the ED was used to develop the AI-BPM. The AI-BPM used structured and unstructured text data acquired during the early stage of ED visit, and provided both the point estimate and 95% confidence interval (CI) of its predictions. High AI-BPM uncertainty was defined as when the predetermined bacteremia risk threshold (5%) was included in the 95% CI of the AI-BPM prediction, and low AI-BPM uncertainty was when it was not included. In the temporal validation dataset (N = 8,188), the AI-BPM achieved area under the receiver operating characteristic curve (AUC) of 0.754 (95% CI 0.737–0.771), sensitivity of 0.917 (95% CI 0.897–0.934), and specificity of 0.340 (95% CI 0.330–0.351). In the external validation dataset (N = 7,029), the AI-BPM’s AUC was 0.738 (95% CI 0.722–0.755), sensitivity was 0.927 (95% CI 0.909–0.942), and specificity was 0.319 (95% CI 0.307–0.330). The AUC of the post-AI physicians predictions (0.703, 95% CI 0.654–0.753) was significantly improved compared with that of the pre-AI predictions (0.639, 95% CI 0.585–0.693; p-value < 0.001) in the sampled dataset (N = 1,000). The AI-BPM especially improved the predictive performance of physicians in cases with high physician uncertainty (low subjective confidence) and low AI-BPM uncertainty. Our results suggest that the uncertainty of both the AI model and physicians should be considered for successful AI model implementation.

Development and derivation of bacteremia prediction model in patients with hepatobiliary infection

IntroductionHepatobiliary infections are common in the emergency department (ED), and the mortality rate for this condition is high. A suitable bacteremia prediction model would support prompt identification of bacteremia and appropriate management of hepatobiliary infections in the ED. Therefore, we attempted to produce a bacteremia prediction model with both internal and external validation for hepatobiliary infections in the ED. MethodsPatients with hepatobiliary infection were extracted from retrospective cohort databases of two tertiary hospitals from January 2018 to December 2019 and from January 2016 to December 2019, respectively. Independent risk factors were determined using multivariable logistic regression in a developmental cohort. We assigned a weighted value to predictive factors and developed a prediction model, which was validated both internally and externally. We assessed discrimination using the area under the receiver operating characteristics curve (AUC). ResultsOne hospital cohort of 1568 patients was randomly divided into a developmental group of 927 patients (60%) and an internal validation group of 641 patients (40%), and 736 people from the other hospital cohort were used for external validation. Bacteremia rates were 20.5%, 18.1%, and 23.1% in the developmental, internal, and external validation cohorts, respectively. Nine significant factors were used for predicting bacteremia, including age, three vital signs, and five laboratory tests. After applying our bacteremia prediction rule to the validation cohort, 56.5% and 53.8% of the internal and external validation groups were classified as low-risk bacteremia groups (bacteremia rates: 8.6% and 13.9%, respectively). The AUCs were 0.727 (95% confidence interval [CI]: 0.686–0.767), 0.730 (95% CI: 0.679–0.781), and 0.715 (95% CI: 0.672–0.758) for the developmental, internal, and external validation cohorts, respectively. The sensitivity and specificity for internal validation/external validation was 73.2%/67.6% and 63.0%/60.2%, respectively. ConclusionA bacteremia prediction model for hepatobiliary infection might be useful to predict the risk of bacteremia. It might also reduce the need for blood culture in low-risk patients.

Real-time artificial intelligence system for bacteremia prediction in adult febrile emergency department patients

BackgroundArtificial intelligence (AI) holds significant potential to be a valuable tool in healthcare. However, its application for predicting bacteremia among adult febrile patients in the emergency department (ED) remains unclear. Therefore, we conducted a study to provide clarity on this issue. MethodsAdult febrile ED patients with blood cultures at Chi Mei Medical Center were divided into derivation (January 2017 to June 2019) and validation groups (July 2019 to December 2020). The derivation group was utilized to develop AI models using twenty-one feature variables and five algorithms to predict bacteremia. The performance of these models was compared with qSOFA score. The AI model with the highest area under the receiver operating characteristics curve (AUC) was chosen to implement the AI prediction system and tested on the validation group. ResultsThe study included 5,647 febrile patients. In the derivation group, there were 3,369 patients with a mean age of 61.4 years, and 50.7% were female, including 508 (13.8%) with bacteremia. The model with the best AUC was built using the random forest algorithm (0.761), followed by logistic regression (0.755). All five models demonstrated better AUC than the qSOFA score (0.560). The random forest model was adopted to build a real-time AI prediction system integrated into the hospital information system, and the AUC achieved 0.709 in the validation group. ConclusionThe AI model shows promise to predict bacteremia in adult febrile ED patients; however, further external validation in different hospitals and populations is necessary to verify its effectiveness.

Machine learning of cell population data, complete blood count, and differential count parameters for early prediction of bacteremia among adult patients with suspected bacterial infections and blood culture sampling in emergency departments.

Bacteremia is a life-threatening complication of infectious diseases. Bacteremia can be predicted using machine learning (ML) models, but these models have not utilized cell population data (CPD). The derivation cohort from emergency department (ED) of China Medical University Hospital (CMUH) was used to develop the model and was prospectively validated in the same hospital. External validation was performed using cohorts from ED of Wei-Gong Memorial Hospital (WMH) and Tainan Municipal An-Nan Hospital (ANH). Adult patients who underwent complete blood count (CBC), differential count (DC), and blood culture tests were enrolled in the present study. The ML model was developed using CBC, DC, and CPD to predict bacteremia from positive blood cultures obtained within 4h before or after the acquisition of CBC/DC blood samples. This study included 20,636 patients from CMUH, 664 from WMH, and 1622 patients from ANH. Another 3143 patients were included in the prospective validation cohort of CMUH. The CatBoost model achieved an area under the receiver operating characteristic curve of 0.844 in the derivation cross-validation, 0.812 in the prospective validation, 0.844 in the WMH external validation, and 0.847 in the ANH external validation. The most valuable predictors of bacteremia in the CatBoost model were the mean conductivity of lymphocytes, nucleated red blood cell count, mean conductivity of monocytes, and neutrophil-to-lymphocyte ratio. ML model that incorporated CBC, DC, and CPD showed excellent performance in predicting bacteremia among adult patients with suspected bacterial infections and blood culture sampling in emergency departments.

Sepsis biomarkers for early diagnosis of bacteremia in emergency department

Introduction: We compared the diagnostic values of individual and composite biomarkers used in the prediction of bacteremia in adult emergency department patients. Methodology: First-hour blood levels of C- reactive protein, procalcitonin, interleukin-6, lactate, lipopolysaccharide-binding protein and white blood cell count were collected from a 30-person control group and 47 adult patients. Patients included in this study were admitted to the emergency department on suspicion of sepsis. We categorized patients according to presence/absence of sepsis and bacteremia. Our control group was categorized as S-B -, septic patients with bacteremia were S+B+, and septic patients without bacteremia were S+B-. Results: All biomarkers showed a statistically significant elevation when S+B- and S+B+ groups were compared with the S-B-. When S+B+ group was compared with the S+B- group only procalcitonin and lactate levels had statistically significant elevation (p &lt; 0.005). Regression analysis demonstrated that lactate and procalcitonin were independently associated with having bacteremia in the state of sepsis and Hosmer-Lemeshow score was 0.772. The areas under the curve (AUC) values of biomarkers procalcitonin, lactate, C-reactive protein, combined 1 (procalcitonin+ lactate), and combined 2 (procalcitonin + lactate + C-reactive protein) were 0.773, 0.744, 0.523, 0.806, and 0.829 respectively. Conclusions: Combination of tests such as combined 1 or combined 2 were highly predictive of bacteremia in adult septic patients. Combined 2 demonstrated the best predictive performance and could be utilized as a tool to assist diagnosis of bacteremia before culture results are available.

Nausea Predicts Bacteremia in Immunocompetent Patients with Pneumococcal Community-Acquired Pneumonia: Secondary Data Analysis from a Prospective Cohort.

In pneumococcal community-acquired pneumonia (CAP), bacteremia is associated with increased mortality, but initial clinical severity scores frequently fail to identify bacteremic patients at risk. We have previously shown that gastrointestinal symptoms are common among patients admitted to the hospital with pneumococcal bacteremia. The aim of this study was to examine gastrointestinal symptoms and inflammatory responses in bacteremic and non-bacteremic pneumococcal CAP in a prospective cohort of immunocompromised and immunocompetent patients hospitalized with CAP. Logistic regression analysis was used to estimate the predictive value of gastrointestinal symptoms for pneumococcal bacteremia in patients with CAP. The Mann-Whitney test was used to compare inflammatory responses in patients with bacteremic vs. non-bacteremic pneumococcal CAP. Eighty-one patients with pneumococcal CAP were included, of whom 21 (26%) had bacteremia. Immunocompetent patients with pneumococcal CAP had an odds ratio of 16.5 (95% CI 3.0-90.9, p = 0.001) for bacteremia if nausea was present, whereas no such association was found in the immunocompromised patients (OR 0.22, 95% CI 0.02-2.05, p = 0.18). The serum levels of C-reactive protein, procalcitonin and interleukin 6 were significantly higher in the patients with bacteremic pneumococcal CAP compared to non-bacteremic pneumococcal CAP patients (p < 0.001, p = 0.005, and p = 0.019, respectively). In immunocompetent patients hospitalized with pneumococcal CAP, nausea may be a predictor of bacteremia. Bacteremic pneumococcal CAP patients display an increased inflammatory response compared to non-bacteremic pneumococcal CAP patients.