Performance of a Large Language Model in Screening Citations.

Importance: Large language models (LLMs) are promising as tools for citation screening in systematic reviews. However, their applicability has not yet been determined. Objective: To evaluate the accuracy and efficiency of an LLM in title and abstract literature screening. Design, Setting, and Participants: This prospective diagnostic study used the data from the title and abstract screening process for 5 clinical questions (CQs) in the development of the Japanese Clinical Practice Guidelines for Management of Sepsis and Septic Shock. The LLM decided to include or exclude citations based on the inclusion and exclusion criteria in terms of patient, population, problem; intervention; comparison; and study design of the selected CQ and was compared with the conventional method for title and abstract screening. This study was conducted from January 7 to 15, 2024. Exposures: LLM (GPT-4 Turbo)-assisted citation screening or the conventional method. Main Outcomes and Measures: The sensitivity and specificity of the LLM-assisted screening process was calculated, and the full-text screening result using the conventional method was set as the reference standard in the primary analysis. Pooled sensitivity and specificity were also estimated, and screening times of the 2 methods were compared. Results: In the conventional citation screening process, 8 of 5634 publications in CQ 1, 4 of 3418 in CQ 2, 4 of 1038 in CQ 3, 17 of 4326 in CQ 4, and 8 of 2253 in CQ 5 were selected. In the primary analysis of 5 CQs, LLM-assisted citation screening demonstrated an integrated sensitivity of 0.75 (95% CI, 0.43 to 0.92) and specificity of 0.99 (95% CI, 0.99 to 0.99). Post hoc modifications to the command prompt improved the integrated sensitivity to 0.91 (95% CI, 0.77 to 0.97) without substantially compromising specificity (0.98 [95% CI, 0.96 to 0.99]). Additionally, LLM-assisted screening was associated with reduced time for processing 100 studies (1.3 minutes vs 17.2 minutes for conventional screening methods; mean difference, -15.25 minutes [95% CI, -17.70 to -12.79 minutes]). Conclusions and Relevance: In this prospective diagnostic study investigating the performance of LLM-assisted citation screening, the model demonstrated acceptable sensitivity and reasonably high specificity with reduced processing time. This novel method could potentially enhance efficiency and reduce workload in systematic reviews.

Machine learning algorithm for predicting 30-day mortality in patients receiving rapid response system activation: A retrospective nationwide cohort study.

This study investigated the accuracy of a machine learning algorithm for predicting mortality in patients receiving rapid response system (RRS) activation. This retrospective cohort study used data from the In-Hospital Emergency Registry in Japan, which collects nationwide data on patients receiving RRS activation. The missing values in the dataset were replaced using multiple imputations (mode imputation, BayseRidge sklearn. linear model, and K-nearest neighbor model), and the enrolled patients were randomly assigned to the training and test cohorts. We established prediction models for 30-day mortality using the following four types of machine learning classifiers: Light Gradient Boosting Machine (LightGBM), eXtreme Gradient Boosting, random forest, and neural network. Fifty-two variables (patient characteristics, details of RRS activation, reasons for RRS initiation, and hospital capacity) were used to construct the prediction algorithm. The primary outcome was the accuracy of the prediction model for 30-day mortality. Overall, the data from 4,997 patients across 34 hospitals were analyzed. The machine learning algorithms using LightGBM demonstrated the highest predictive value for 30-day mortality (area under the receiver operating characteristic curve, 0.860 [95 % confidence interval, 0.825-0.895]). The SHapley Additive exPlanations summary plot indicated that hospital capacity, site of incidence, code status, and abnormal vital signs within 24 h were important variables in the prediction model for 30-day mortality.

Incidence and mortality of community-acquired and nosocomial infections in Japan: a nationwide medical claims database study.

BACKGROUND: It is important to determine the prevalence and prognosis of community-acquired infection (CAI) and nosocomial infection (NI) to develop treatment strategies and appropriate medical policies in aging society. METHODS: Patients hospitalized between January 2010 and December 2019, for whom culture tests were performed and antibiotics were administered, were selected using a national claims-based database. The annual trends in incidence and in-hospital mortality were calculated and evaluated by dividing the patients into four age groups. RESULTS: Of the 73,962,409 inpatients registered in the database, 9.7% and 4.7% had CAI and NI, respectively. These incidences tended to increase across the years in both the groups. Among the patients hospitalized with infectious diseases, there was a significant increase in patients aged ≥ 85 years (CAI: + 1.04%/year and NI: + 0.94%/year, P < 0.001), while there was a significant decrease in hospitalization of patients aged ≤ 64 years (CAI: -1.63%/year and NI: -0.94%/year, P < 0.001). In-hospital mortality was significantly higher in the NI than in the CAI group (CAI: 8.3%; NI: 14.5%, adjusted mean difference 4.7%). The NI group had higher organ support, medical cost per patient, and longer duration of hospital stay. A decreasing trend in mortality was observed in both the groups (CAI: -0.53%/year and NI: -0.72%/year, P < 0.001). CONCLUSION: The present analysis of a large Japanese claims database showed that NI is a significant burden on hospitalized patients in aging societies, emphasizing the need to address particularly on NI.

Claudin-2 upregulation enhances intestinal permeability, immune activation, dysbiosis, and mortality in sepsis.

Intestinal epithelial expression of the tight junction protein claudin-2, which forms paracellular cation and water channels, is precisely regulated during development and in disease. Here, we show that small intestinal epithelial claudin-2 expression is selectively upregulated in septic patients. Similar changes occurred in septic mice, where claudin-2 upregulation coincided with increased flux across the paracellular pore pathway. In order to define the significance of these changes, sepsis was induced in claudin-2 knockout (KO) and wild-type (WT) mice. Sepsis-induced increases in pore pathway permeability were prevented by claudin-2 KO. Moreover, claudin-2 deletion reduced interleukin-17 production and T cell activation and limited intestinal damage. These effects were associated with reduced numbers of neutrophils, macrophages, dendritic cells, and bacteria within the peritoneal fluid of septic claudin-2 KO mice. Most strikingly, claudin-2 deletion dramatically enhanced survival in sepsis. Finally, the microbial changes induced by sepsis were less pathogenic in claudin-2 KO mice as survival of healthy WT mice injected with cecal slurry collected from WT mice 24 h after sepsis was far worse than that of healthy WT mice injected with cecal slurry collected from claudin-2 KO mice 24 h after sepsis. Claudin-2 upregulation and increased pore pathway permeability are, therefore, key intermediates that contribute to development of dysbiosis, intestinal damage, inflammation, ineffective pathogen control, and increased mortality in sepsis. The striking impact of claudin-2 deletion on progression of the lethal cascade activated during sepsis suggests that claudin-2 may be an attractive therapeutic target in septic patients.

Epidemiology and patterns of empiric antimicrobial therapy practice in patients with community-onset sepsis using data from a Japanese nationwide medical claims database-the Japan Sepsis Alliance (JaSA) study group.

Objectives: We aimed to describe empiric antimicrobial options for patients with community-onset sepsis using nationwide real-world data from Japan. Methods: This retrospective cohort study used nationwide Japanese data from a medical reimbursement system database. Patients aged ≥20 years with both presumed infections and acute organ dysfunction who were admitted to hospitals from the outpatient department or emergency department between 2010 and 2017 were enrolled. We described the initial choices of antimicrobials for patients with sepsis stratified by intensive care unit (ICU) or ward. Results: There were 1,195,741 patients with community-onset sepsis; of these, 1,068,719 and 127,022 patients were admitted to the wards and ICU, respectively. Third-generation cephalosporins and carbapenem were most commonly used for patients with community-onset sepsis. We found that 1.7% and 6.0% of patients initially used antimicrobials for methicillin-resistant Staphylococcus aureus coverage in the wards and ICU, respectively. Although half of the patients initially used antipseudomonal agents, only a few patients used a combination of antipseudomonal agents. Moreover, few patients initially used a combination of antimicrobials to treat methicillin-resistant Staphylococcus aureus and Pseudomonas sp. Conclusion: Third-generation cephalosporins and carbapenem were most frequently used for patients with sepsis. A combination therapy of antimicrobials for drug-resistant bacteria coverage was rarely provided to these patients.

Association between hospital spending and in-hospital mortality of patients with sepsis based on a Japanese nationwide medical claims database study.

Background: The effect of hospital spending on the mortality rate of patients with sepsis has not yet been fully elucidated. We hypothesized that hospitals that consume more medical resources would have lower mortality rates among patients with sepsis. Methods: This retrospective study used administrative data from 2010 to 2017. The enrolled hospitals were divided into quartiles based on average daily medical cost per sepsis case. The primary and secondary outcomes were the average in-hospital mortality rate of patients with sepsis and the effective cost per survivor among the enrolled hospitals, respectively. A multiple regression model was used to determine the significance of the differences among hospital categories to adjust for baseline imbalances. Results: Among 997 hospitals enrolled in this study, the crude in-hospital mortality rates were 15.7% and 13.2% in the lowest and highest quartiles of hospital spending, respectively. After adjusting for confounding factors, the highest hospital spending group demonstrated a significantly lower in-hospital mortality rate than the lowest hospital spending group (coefficient = -0.025, 95% confidence interval [CI] -0.034 to -0.015; p < 0.0001). Similarly, the highest hospital spending group was associated with a significantly higher effective cost per survivor than the lowest hospital spending group (coefficient = 77.7, 95% CI 73.1 to 82.3; p < 0.0001). In subgroup analyses, hospitals with a small or medium number of beds demonstrated a consistent pattern with the primary test, whereas those with a large number of beds or academic affiliations displayed no association. Conclusions: Using a nationwide Japanese medical claims database, this study indicated that hospitals with greater expenditures were associated with a superior survival rate and a higher effective cost per survivor in patients with sepsis than those with lower expenditures. In contrast, no correlations between hospital spending and mortality were observed in hospitals with a large number of beds or academic affiliations.

Machine-learning model for predicting oliguria in critically ill patients.

This retrospective cohort study aimed to develop and evaluate a machine-learning algorithm for predicting oliguria, a sign of acute kidney injury (AKI). To this end, electronic health record data from consecutive patients admitted to the intensive care unit (ICU) between 2010 and 2019 were used and oliguria was defined as a urine output of less than 0.5 mL/kg/h. Furthermore, a light-gradient boosting machine was used for model development. Among the 9,241 patients who participated in the study, the proportions of patients with urine output < 0.5 mL/kg/h for 6 h and with AKI during the ICU stay were 27.4% and 30.2%, respectively. The area under the curve (AUC) values provided by the prediction algorithm for the onset of oliguria at 6 h and 72 h using 28 clinically relevant variables were 0.964 (a 95% confidence interval (CI) of 0.963-0.965) and 0.916 (a 95% CI of 0.914-0.918), respectively. The Shapley additive explanation analysis for predicting oliguria at 6 h identified urine values, severity scores, serum creatinine, oxygen partial pressure, fibrinogen/fibrin degradation products, interleukin-6, and peripheral temperature as important variables. Thus, this study demonstrates that a machine-learning algorithm can accurately predict oliguria onset in ICU patients, suggesting the importance of oliguria in the early diagnosis and optimal management of AKI.

Efficiency and workload reduction of semi-automated citation screening software for creating clinical practice guidelines: a prospective observational study.

BACKGROUND: We evaluated the applicability of automated citation screening in developing clinical practice guidelines. METHODS: We prospectively compared the efficiency of citation screening between the conventional (Rayyan) and semi-automated (ASReview software) methods. We searched the literature for five clinical questions (CQs) in the development of the Japanese Clinical Practice Guidelines for the Management of Sepsis and Septic Shock. Objective measurements of the time required to complete citation screening were recorded. Following the first screening round, in the primary analysis, the sensitivity, specificity, positive predictive value, and overall screening time were calculated for both procedures using the semi-automated tool as index and the results of the conventional method as standard reference. In the secondary analysis, the same parameters were compared between the two procedures using the final list of included studies after the second screening session as standard reference. RESULTS: Among the five CQs after the first screening session, the highest and lowest sensitivity, specificity, and positive predictive values were 0.241 and 0.795; 0.991 and 1.000; and 0.482 and 0.929, respectively. In the secondary analysis, the highest sensitivity and specificity in the semi-automated citation screening were 1.000 and 0.997, respectively. The overall screening time per 100 studies was significantly shorter with semi-automated than with conventional citation screening. CONCLUSIONS: The potential advantages of the semi-automated method (shorter screening time and higher discriminatory rate for the final list of studies) warrant further validation. TRIAL REGISTRATION: This study was submitted to the University Hospital Medical Information Network Clinical Trial Registry (UMIN-CTR [UMIN000049366]).

Epidemiology of sepsis in a Japanese administrative database.

Sepsis is the leading cause of death worldwide. Considering regional variations in the characteristics of patients with sepsis, a better understanding of the epidemiology in Japan will lead to further development of strategies for the prevention and treatment of sepsis. To investigate the epidemiology of sepsis, we conducted a systematic literature review of PubMed between 2003 and January 2023. Among the 78 studies using a Japanese administrative database, we included 20 that defined patients with sepsis as those with an infection and organ dysfunction. The mortality rate in patients with sepsis has decreased since 2010, reaching 18% in 2017. However, the proportion of inpatients with sepsis is increasing. A study comparing short-course (≤7 days) and long-course (≥8 days) antibiotic administration showed lower 28-day mortality in the short-course group. Six studies on the treatment of patients with septic shock reported that low-dose corticosteroids or polymyxin B hemoperfusion reduced mortality, whereas intravenous immunoglobulins had no such effect. Four studies investigating the effects of treatment in patients with sepsis-associated disseminated intravascular coagulation demonstrated that antithrombin may reduce mortality, whereas recombinant human soluble thrombomodulin does not. A descriptive study of medical costs for patients with sepsis showed that the effective cost per survivor decreased over an 8-year period from 2010 to 2017. Sepsis has a significant impact on public health, and is attracting attention as an ongoing issue. Further research to determine more appropriate prevention methods and treatment for sepsis should be a matter of priority.

CHRONIC ETHANOL USE WORSENS GUT PERMEABILITY AND ALTERS TIGHT JUNCTION EXPRESSION IN A MURINE SEPSIS MODEL.

ABSTRACT: Alcohol use disorder is associated with increased mortality in septic patients. Murine studies demonstrate that ethanol/sepsis is associated with changes in gut integrity. This study examined intestinal permeability after ethanol/sepsis and investigated mechanisms responsible for alterations in barrier function. Mice were randomized to drink either 20% ethanol or water for 12 weeks and then were subjected to either sham laparotomy or cecal ligation and puncture (CLP). Intestinal permeability was disproportionately increased in ethanol/septic mice via the pore, leak, and unrestricted pathways. Consistent with increased permeability in the leak pathway, jejunal myosin light chain (MLC) kinase (MLCK) expression and the ratio of phospho-MLC to total MLC were both increased in ethanol/CLP. Gut permeability was altered in MLCK -/- mice in water/CLP; however, permeability was not different between WT and MLCK -/- mice in ethanol/CLP. Similarly, jejunal IL-1ß levels were decreased while systemic IL-6 levels were increased in MLCK -/- mice in water/CLP but no differences were identified in ethanol/CLP. While we have previously shown that mortality is improved in MLCK -/- mice after water/CLP, mortality was significantly worse in MLCK -/- mice after ethanol/CLP. Consistent with an increase in the pore pathway, claudin 4 levels were also selectively decreased in ethanol/CLP WT mice. Furthermore, mRNA expression of jejunal TNF and IFN-γ were both significantly increased in ethanol/CLP. The frequency of CD4 + cells expressing TNF and IL-17A and the frequency of CD8 + cells expressing IFN-γ in Peyer's Patches were also increased in ethanol/CLP. Thus, there is an ethanol-specific worsening of gut barrier function after CLP that impacts all pathways of intestinal permeability, mediated, in part, via changes to the tight junction. Differences in the host response in the setting of chronic alcohol use may play a role in future precision medicine approaches toward the treatment of sepsis.

Machine learning algorithms for predicting days of high incidence for out-of-hospital cardiac arrest.

Predicting out-of-hospital cardiac arrest (OHCA) events might improve outcomes of OHCA patients. We hypothesized that machine learning algorithms using meteorological information would predict OHCA incidences. We used the Japanese population-based repository database of OHCA and weather information. The Tokyo data (2005-2012) was used as the training cohort and datasets of the top six populated prefectures (2013-2015) as the test. Eight various algorithms were evaluated to predict the high-incidence OHCA days, defined as the daily events exceeding 75% tile of our dataset, using meteorological and chronological values: temperature, humidity, air pressure, months, days, national holidays, the day before the holidays, the day after the holidays, and New Year's holidays. Additionally, we evaluated the contribution of each feature by Shapley Additive exPlanations (SHAP) values. The training cohort included 96,597 OHCA patients. The eXtreme Gradient Boosting (XGBoost) had the highest area under the receiver operating curve (AUROC) of 0.906 (95% confidence interval; 0.868-0.944). In the test cohorts, the XGBoost algorithms also had high AUROC (0.862-0.923). The SHAP values indicated that the "mean temperature on the previous day" impacted the most on the model. Algorithms using machine learning with meteorological and chronological information could predict OHCA events accurately.

Protocol to acquire time series data on adverse reactions following vaccination using a smartphone or web-based platform.

Data collection on adverse reactions in recipients after vaccination is vital to evaluate potential health issues, but health observation diaries are onerous for participants. Here, we present a protocol to collect time series information using a smartphone or web-based platform, thus eliminating the need for paperwork and data submission. We describe steps for setting up the platform using the Model-View-Controller web framework, uploading lists of recipients, sending notifications, and managing respondent data. For complete details on the use and execution of this protocol, please refer to Ikeda et al. (2022).1.

Mortality analysis among sepsis patients in and out of intensive care units using the Japanese nationwide medical claims database: a study by the Japan Sepsis Alliance study group.

BACKGROUND: A substantial number of sepsis patients require specialized care, including multidisciplinary care, close monitoring, and artificial organ support in the intensive care unit (ICU). However, the efficacy of ICU management on clinical outcomes remains insufficiently researched. Therefore, we tested the hypothesis that ICU admission would increase the survival rate among sepsis patients. METHODS: We conducted a retrospective study using the nationwide medical claims database of sepsis patients in Japan from 2010 to 2017 with propensity score matching to adjust for baseline imbalances. Patients aged over 20 years, with a combined diagnosis of presumed serious infection and organ failure, were included in this study. The primary outcome studied was the in-hospital mortality among non-ICU and ICU patients. In addition to propensity score matching, we performed a multivariable logistic regression analysis for the primary outcome. As the treatment policy was not extracted from the database, we performed sensitivity analyses to determine mortality differences in adults (20 ≤ age ≤ 64), independent patients, patients without malignant tumors, based on the assumption that treatment intensity is likely to increase in those population. RESULTS: Among 1,167,901 sepsis patients (974,289 in non-ICU and 193,612 in ICU settings), the unadjusted in-hospital mortality was 22.5% among non-ICU patients and 26.2% among ICU patients (3.7% [95% CI 3.5-3.9]). After propensity score matching, the in-hospital mortality was 29.2% among non-ICU patients and 25.8% among ICU patients ([Formula: see text] 3.4% [95% CI [Formula: see text] 3.7 to [Formula: see text] 3.1]). In-hospital mortality with a multivariable regression analysis ([Formula: see text] 5.0% [95% CI [Formula: see text] 5.2 to [Formula: see text] 4.8]) was comparable with the results of the propensity score matching analysis. In the sensitivity analyses, the mortality differences between non-ICU and ICU in adults, independent patients, and patients without malignant tumors were [Formula: see text] 2.7% [95% CI [Formula: see text] 3.3 to [Formula: see text] 2.2], [Formula: see text] 5.8% [95% CI [Formula: see text] 6.4 to [Formula: see text] 5.2], and [Formula: see text] 1.3% [95% CI [Formula: see text] 1.7 to [Formula: see text] 1.0], respectively. CONCLUSIONS: Herein, using the nationwide medical claims database, we demonstrated that ICU admission was potentially associated with decreasing in-hospital mortality among sepsis patients. Further investigations are warranted to validate these results and elucidate the mechanisms favoring ICU management on clinical outcomes.

Short- versus long-course antibiotic therapy for sepsis: a post hoc analysis of the nationwide cohort study.

BACKGROUND: The appropriate duration of antibiotic treatment in patients with bacterial sepsis remains unclear. The purpose of this study was to evaluate the association of a shorter course of antibiotics on 28-day mortality in comparison with a longer course using a national database in Japan. METHODS: We conducted a post hoc analysis from the retrospective observational study of patients with sepsis using a Japanese claims database from 2010 to 2017. The patient dataset was divided into short-course (≤ 7 days) and long-course (≥ 8 days) groups according to the duration of initial antibiotic administration. Subsequently, propensity score matching was performed to adjust the baseline imbalance between the two groups. The primary outcome was 28-day mortality. The secondary outcomes were re-initiated antibiotics at 3 and 7 days, during hospitalization, administration period, antibiotic-free days, and medical cost. RESULTS: After propensity score matching, 448,146 pairs were analyzed. The 28-day mortality was significantly lower in the short-course group (hazard ratio, 0.94; 95% CI, 0.92-0.95; P < 0.001), while the occurrence of re-initiated antibiotics at 3 and 7 days and during hospitalization were significantly higher in the short-course group (P < 0.001). Antibiotic-free days (median [IQR]) were significantly shorter in the long-course group (21 days [17 days, 23 days] vs. 17 days [14 days, 19 days], P < 0.001), and short-course administration contributed to a decrease in medical costs (coefficient $-212, 95% CI; - 223 to - 201, P < 0.001). Subgroup analyses showed a significant decrease in the 28-day mortality of the patients in the short-course group in patients of male sex (hazard ratio: 0.91, 95% CI; 0.89-0.93), community-onset sepsis (hazard ratio; 0.95, 95% CI; 0.93-0.98), abdominal infection (hazard ratio; 0.92, 95% CI; 0.88-0.97) and heart infection (hazard ratio; 0.74, 95% CI; 0.61-0.90), while a significant increase was observed in patients with non-community-onset sepsis (hazard ratio; 1.09, 95% CI; 1.06-1.12). CONCLUSIONS: The 28-day mortality was significantly lower in the short-course group, even though there was a higher rate of re-initiated antibiotics in the short course.

Prehospital diagnostic algorithm for acute coronary syndrome using machine learning: a prospective observational study.

Rapid and precise prehospital recognition of acute coronary syndrome (ACS) is key to improving clinical outcomes. The aim of this study was to investigate a predictive power for predicting ACS using the machine learning-based prehospital algorithm. We conducted a multicenter observational prospective study that included 10 participating facilities in an urban area of Japan. The data from consecutive adult patients, identified by emergency medical service personnel with suspected ACS, were analyzed. In this study, we used nested cross-validation to evaluate the predictive performance of the model. The primary outcomes were binary classification models for ACS prediction based on the nine machine learning algorithms. The voting classifier model for ACS using 43 features had the highest area under the receiver operating curve (AUC) (0.861 [95% CI 0.775-0.832]) in the test score. After validating the accuracy of the model using the external cohort, we repeated the analysis with a limited number of selected features. The performance of the algorithms using 17 features remained high AUC (voting classifier, 0.864 [95% CI 0.830-0.898], support vector machine (radial basis function), 0.864 [95% CI 0.829-0.887]) in the test score. We found that the machine learning-based prehospital algorithms showed a high predictive power for predicting ACS.

Temporal trends of medical cost and cost-effectiveness in sepsis patients: a Japanese nationwide medical claims database.

BACKGROUND: Sepsis is the leading cause of death worldwide. Although the mortality of sepsis patients has been decreasing over the past decade, the trend of medical costs and cost-effectiveness for sepsis treatment remains insufficiently determined. METHODS: We conducted a retrospective study using the nationwide medical claims database of sepsis patients in Japan between 2010 and 2017. After selecting sepsis patients with a combined diagnosis of presumed serious infection and organ failure, patients over the age of 20 were included in this study. We investigated the annual trend of medical costs during the study period. The primary outcome was the annual trend of the effective cost per survivor, calculated from the gross medical cost and number of survivors per year. Subsequently, we performed subgroup and multiple regression analyses to evaluate the association between the annual trend and medical costs. RESULTS: Among 50,490,128 adult patients with claims, a total of 1,276,678 patients with sepsis were selected from the database. Yearly gross medical costs to treat sepsis gradually increased over the decade from $3.04 billion in 2010 to $4.38 billion in 2017, whereas the total medical cost per hospitalization declined (rate = - $1075/year, p < 0.0001). While the survival rate of sepsis patients improved during the study period, the effective cost per survivor significantly decreased (rate = - $1806/year [95% CI - $2432 to - $1179], p = 0.001). In the subgroup analysis, the trend of decreasing medical cost per hospitalization remained consistent among the subpopulation of age, sex, and site of infection. After adjusting for age, sex (male), number of chronic diseases, site of infection, intensive care unit (ICU) admission, surgery, and length of hospital stay, the admission year was significantly associated with reduced medical costs. CONCLUSIONS: We demonstrated an improvement in annual cost-effectiveness in patients with sepsis between 2010 and 2017. The annual trend of reduced costs was consistent after adjustment with the confounders altering hospital expenses.

Prediction algorithm for ICU mortality and length of stay using machine learning.

Machine learning can predict outcomes and determine variables contributing to precise prediction, and can thus classify patients with different risk factors of outcomes. This study aimed to investigate the predictive accuracy for mortality and length of stay in intensive care unit (ICU) patients using machine learning, and to identify the variables contributing to the precise prediction or classification of patients. Patients (n = 12,747) admitted to the ICU at Chiba University Hospital were randomly assigned to the training and test cohorts. After learning using the variables on admission in the training cohort, the area under the curve (AUC) was analyzed in the test cohort to evaluate the predictive accuracy of the supervised machine learning classifiers, including random forest (RF) for outcomes (primary outcome, mortality; secondary outcome, length of ICU stay). The rank of the variables that contributed to the machine learning prediction was confirmed, and cluster analysis of the patients with risk factors of mortality was performed to identify the important variables associated with patient outcomes. Machine learning using RF revealed a high predictive value for mortality, with an AUC of 0.945 (95% confidence interval [CI] 0.922-0.977). In addition, RF showed high predictive value for short and long ICU stays, with AUCs of 0.881 (95% CI 0.876-0.908) and 0.889 (95% CI 0.849-0.936), respectively. Lactate dehydrogenase (LDH) was identified as a variable contributing to the precise prediction in machine learning for both mortality and length of ICU stay. LDH was also identified as a contributing variable to classify patients into sub-populations based on different risk factors of mortality. The machine learning algorithm could predict mortality and length of stay in ICU patients with high accuracy. LDH was identified as a contributing variable in mortality and length of ICU stay prediction and could be used to classify patients based on mortality risk.

Measurement of Intestinal Permeability During Sepsis.

Gut barrier function has been hypothesized to play a critical role in the pathophysiology of sepsis. Measuring intestinal permeability allows for a determination of barrier dysfunction under conditions of health and disease. Fluorescence-conjugated dyes such as fluorescein isothiocyanate-4 kDa dextran (FD4) have been commonly used for evaluating hyperpermeability. Here we describe a common method to measure gut permeability in vivo, following gavage with different sized dyes. In addition, we describe an ex vivo everted gut sac model that allows for discrimination of permeability by segmental geographic location along the intestine.

Tumor-Specific T Cells Exacerbate Mortality and Immune Dysregulation during Sepsis.

Sepsis induces significant immune dysregulation characterized by lymphocyte apoptosis and alterations in the cytokine milieu. Because cancer patients exhibit a 10-fold greater risk of developing sepsis compared with the general population, we aimed to understand how pre-existing malignancy alters sepsis-induced immune dysregulation. To address this question, we assessed the impact of tumor-specific CD8+ T cells on the immune response in a mouse model of cecal ligation and puncture (CLP)-induced sepsis. Tumor-bearing animals containing Thy1.1+ tumor-specific CD8+ T cells were subjected to CLP, and groups of animals received anti-Thy1.1 mAb to deplete tumor-specific CD8+ T cells or isotype control. Results indicated that depleting tumor-specific T cells significantly improved mortality from sepsis. The presence of tumor-specific CD8+ T cells resulted in increased expression of the 2B4 coinhibitory receptor and increased apoptosis of endogenous CD8+ T cells. Moreover, tumor-specific T cells were not reduced in number in the tumors during sepsis but did exhibit impaired IFN-γ production in the tumor, tumor draining lymph node, and spleen 24 h after CLP. Our research provides novel insight into the mechanisms by which pre-existing malignancy contributes to increased mortality during sepsis.