Prediction Of Sepsis Research Articles

Assessing the effects of data drift on the performance of machine learning models used in clinical sepsis prediction

BackgroundData drift can negatively impact the performance of machine learning algorithms (MLAs) that were trained on historical data. As such, MLAs should be continuously monitored and tuned to overcome the systematic changes that occur in the distribution of data. In this paper, we study the extent of data drift and provide insights about its characteristics for sepsis onset prediction. This study will help elucidate the nature of data drift for prediction of sepsis and similar diseases. This may aid with the development of more effective patient monitoring systems that can stratify risk for dynamic disease states in hospitals. MethodsWe devise a series of simulations that measure the effects of data drift in patients with sepsis, using electronic health records (EHR). We simulate multiple scenarios in which data drift may occur, namely the change in the distribution of the predictor variables (covariate shift), the change in the statistical relationship between the predictors and the target (concept shift), and the occurrence of a major healthcare event (major event) such as the COVID-19 pandemic. We measure the impact of data drift on model performances, identify the circumstances that necessitate model retraining, and compare the effects of different retraining methodologies and model architecture on the outcomes. We present the results for two different MLAs, eXtreme Gradient Boosting (XGB) and Recurrent Neural Network (RNN). ResultsOur results show that the properly retrained XGB models outperform the baseline models in all simulation scenarios, hence signifying the existence of data drift. In the major event scenario, the area under the receiver operating characteristic curve (AUROC) at the end of the simulation period is 0.811 for the baseline XGB model and 0.868 for the retrained XGB model. In the covariate shift scenario, the AUROC at the end of the simulation period for the baseline and retrained XGB models is 0.853 and 0.874 respectively. In the concept shift scenario and under the mixed labeling method, the retrained XGB models perform worse than the baseline model for most simulation steps. However, under the full relabeling method, the AUROC at the end of the simulation period for the baseline and retrained XGB models is 0.852 and 0.877 respectively. The results for the RNN models were mixed, suggesting that retraining based on a fixed network architecture may be inadequate for an RNN. We also present the results in the form of other performance metrics such as the ratio of observed to expected probabilities (calibration) and the normalized rate of positive predictive values (PPV) by prevalence, referred to as lift, at a sensitivity of 0.8. ConclusionOur simulations reveal that retraining periods of a couple of months or using several thousand patients are likely to be adequate to monitor machine learning models that predict sepsis. This indicates that a machine learning system for sepsis prediction will probably need less infrastructure for performance monitoring and retraining compared to other applications in which data drift is more frequent and continuous. Our results also show that in the event of a concept shift, a full overhaul of the sepsis prediction model may be necessary because it indicates a discrete change in the definition of sepsis labels, and mixing the labels for the sake of incremental training may not produce the desired results.

Role of Procalcitonin and C-reactive Protein as Predictors of Sepsis and in Managing Sepsis in Postoperative Patients: A Systematic Review.

Systemic inflammatory response syndrome (SIRS) and sepsis are inflammatory responsesto infection or trauma, causing symptoms and adverse outcomes such as organ shutdown and death. Different scoring systems can help in the diagnosis of SIRS and sepsis. Several biomarkers such as C-reactive protein (CRP), procalcitonin (PCT), and white blood cells (WBCs) can serve as predictors of sepsis. Surgery, trauma, and burns are the non-inflammatory causes of SIRS and sepsis. In postoperative patients, both inflammatory and non-inflammatory causes of immune response may co-exist. The role of inflammatory biomarkers in identifying sepsis development, deciding to use antibiotics, and discharging patients needs further exploration and clarity. We searched medical databases such as PubMed/Medline, PMC, ScienceDirect, Cochrane Library, and Google Scholar for relevant medical literature. The identified papers were screened, eligibility criteria were applied, and 15 research papers were identified. The finalized papers explored the roles of CRP and PCT in postoperative patients. Both CRP and PCT are raised in a postoperative patient, and then, gradually, the levels decrease. However, in case of an infection, these levels continue to rise and signify an infection, which may progress to sepsis. The cut-off values can guide decision-making about when to start antibiotics and discharge the patient. PCT was found to be more reliable in identifying the infection and preventing the development of sepsis. Further research is needed to identify the exact cut-off values that can help in decision-making.

Automated digital technologies for supporting sepsis prediction in children: a scoping review protocol

IntroductionWhile there have been several literature reviews on the performance of digital sepsis prediction technologies and clinical decision-support algorithms for adults, there remains a knowledge gap in examining the development...

OnAI-Comp: An Online AI Experts Competing Framework for Early Sepsis Detection.

Sepsis is a major public concern due to its high mortality, morbidity, and financial cost. There are many existing works of early sepsis prediction using different machine learning models to mitigate the outcomes brought by sepsis. In the practical scenario, the dataset grows dynamically as new patients visit the hospital. Most existing models, being "offline" models and having used retrospective observational data, cannot be updated and improved dynamically using the new observational data. Incorporating the new data to improve the offline models requires retraining the model, which is very computationally expensive. To solve the challenge mentioned above, we propose an Online Artificial Intelligence Experts Competing Framework (OnAI-Comp) for early sepsis detection using an online learning algorithm called Multi-armed Bandit. We selected several machine learning models as the artificial intelligence experts and used average regret to evaluate the performance of our model. The experimental analysis demonstrated that our model would converge to the optimal strategy in the long run. Meanwhile, our model can provide clinically interpretable predictions using existing local interpretable model-agnostic explanation technologies, which can aid clinicians in making decisions and might improve the probability of survival.

Monocyte Distribution Width (MDW) as an Early Investigational Marker for the Diagnosis of Sepsis in an Emergency Department of a Tertiary Care Hospital in North India

Background Sepsis is an emergency state in response to an infectious process ultimately leadingto multiorgan dysfunction and death. There is an urgent need for sepsis detection methods, especially in emergency settings. To fill this gap, monocyte distribution width (MDW) was studied as an early indicator of sepsis. Aim To evaluateMDW as an early marker of sepsis. Material and methods This was a prospective observational study including critically ill adult patients who presented to the emergency department. MDW was measured using a DxH 900 Hematology Analyser (Beckman Coulter Inc., Miami, FL). Abnormal MDW (>20.0) was considered a predictor of sepsis. Results A total of 148 patients were included and categorizedaccording to the Sepsis-2 and Sepsis-3 criteria, as having sepsis (25.6%), sepsis with shock (21.6%), and non-sepsis (52.8%). In patients with sepsis with and without shock, MDW was 28.28 ± 9.20 and 28.02 ± 9.01, respectively, significantly higher than in patients without sepsis (p < 0.001). The diagnostic accuracy value of MDW testing for early sepsis detection was highly significant (0.74, p < 0.000). Conclusion MDW can be used as a marker for the early prediction of sepsis.

Identification of Molecular Subtypes and a Novel Prognostic Model of Sepsis Based on Ferroptosis-Associated Gene Signature.

Ferroptosis has recently been associated with immunological changes in sepsis. However, the clinical significance of ferroptosis-associated genes (FAGs) remains unknown. In this paper, a FAG-based prognostic model was constructed for sepsis patients using an integrated machine learning approach. The prognosis model was composed of 14 FAGs that classify the patients as high or low risk. Based on immunological study, it was found that the immune status differed between the high-risk and low-risk clusters. Cox regression analysis revealed that FAGs were independent risk factors for the overall survival of sepsis patients. ROC curves and nomograms revealed that the FAG-based model was robust for prognosis prediction. Lastly, NEDD4L was identified from the 14 FAGs as a potential hub gene for sepsis prediction.

A model for sepsis prediction after retrograde intrarenal surgery and the use of the preoperative/postoperative white blood cell ratio to predict progression from sepsis to septic shock.

To study the predictors of sepsis and progression to septic shock after RIRS; to establish and validate predictive models accordingly. In total, 1220 patients were included in the study during. Eight hundred forty-eight patients were assigned to the development cohort and 372 to the validation cohort according to medical record. Univariate and multivariate logistic regression analyses were used to screen independent risk factors for post-RIRS (Retrograde intrarenal surgery) sepsis and progression to septic shock. Nomogram prediction models were established according to the related independent risk factors. Areas under the receiver operating characteristic curves, calibration plots, and DCA (Decision curve analysis) were used to estimate the discrimination, calibration and clinical usefulness of the prediction model, respectively. In the development cohort, sepsis occurred in 59 patients, 16 of whom developed septic shock. Multivariate logistic regression analyses showed that the independent risk factors for sepsis after RIRS were preoperative D-J stent implantation, hydronephrosis > 6.25 HU (Hounsfield units), AGR (Albumin/globulin ratio) < 1.95, hs-CRP/Alb (High-sensitivity C-reactive protein/albumin ratio) > 0.060, operating time > 67.5min, and urinary nitrite positivity. The preoperative/postoperative WBC ratio > 1.5 was an independent risk factor for progression from sepsis to septic shock. In the development cohort, the AUC (Area under curve) for predicting sepsis risk was 0.845, and the AUC for predicting septic shock risk was 0.896; in the validation cohort, the corresponding values were 0.896 and 0.974, respectively. In the development cohort, the calibration test P values in the sepsis and septic shock cohorts, respectively, were 0.921 and 0.817; in the validation cohort, these values were 0.882 and 0.859. DCA of the model in the sepsis and septic shock cohorts showed threshold probabilities of 10-90% in the development cohort and 10-50% and 10-20% in the validation cohort. These individualized nomogram prediction models can improve the early identification of patients at risk for developing sepsis after RIRS or progressing from sepsis to septic shock.

Predictors and Health Care Utilization of Sepsis Post-Ureteroscopy in a U.S.-Based Population: Results from the Endourological Society TOWER Collaborative.

Purpose: To investigate the incidence, predictive factors, and health care utilization of sepsis post-ureteroscopy (URS) in patients enrolled in commercial insurance plans. Materials and Methods: A retrospective claims analysis was conducted using the IBM® MarketScan® commercial database. Patients ≥18 years were included if they had URS between January 2015 and October 2019 and developed sepsis within 30 days of URS. Multivariate logistic regression was used to identify various clinical and demographic predictors of sepsis post-URS. All-cause health care utilization (i.e., inpatient admissions and intensive care unit [ICU] stays) and all-cause health care costs up to 1 month post-septic event were measured. Results: Among the 104,100 URS patients meeting the inclusion criteria, 5.5% developed sepsis. Patients with diabetes (odds ratio [OR] = 1.52; p < 0.0001), older age (age 55-64 vs 18-34; OR = 1.35; p < 0.0001), baseline sepsis (OR = 3.51; p < 0.0001), baseline inpatient visits (OR = 1.17; p = 0.0012), and higher Elixhauser comorbidity scores (OR = 1.09; p < 0.0001) had a significantly higher likelihood of developing sepsis post-URS. In septic patients, 94.8% required inpatient care and 35% were admitted to the ICU. Mean hospital stay for septic patients was 6.86 days. Average all-cause health care cost per patient at 1 month in the septic cohort was $49,625 vs $17,782 in the nonseptic cohort indicating an incremental all-cause cost of $31,843 (p < 0.0001). Conclusions: A total of 5.5% of commercially insured patients undergoing URS developed sepsis post-URS. Diabetes, older age, baseline sepsis, baseline inpatient visit, and higher comorbidity score were all found to be independent predictors of post-URS sepsis. Patients with sepsis post-URS had higher health care utilization and costs indicating that sepsis is both a significant clinical and economic event.

Performance effectiveness of vital parameter combinations for early warning of sepsis-an exhaustive study using machine learning.

ObjectiveTo carry out exhaustive data-driven computations for the performance of noninvasive vital signs heart rate (HR), respiratory rate (RR), peripheral oxygen saturation (SpO2), and temperature (Temp), considered both independently and in all possible combinations, for early detection of sepsis.Materials and methodsBy extracting features interpretable by clinicians, we applied Gradient Boosted Decision Tree machine learning on a dataset of 2630 patients to build 240 models. Validation was performed on a geographically distinct dataset. Relative to onset, predictions were clocked as per 16 pairs of monitoring intervals and prediction times, and the outcomes were ranked.ResultsThe combination of HR and Temp was found to be a minimal feature set yielding maximal predictability with area under receiver operating curve 0.94, sensitivity of 0.85, and specificity of 0.90. Whereas HR and RR each directly enhance prediction, the effects of SpO2 and Temp are significant only when combined with HR or RR. In benchmarking relative to standard methods Systemic Inflammatory Response Syndrome (SIRS), National Early Warning Score (NEWS), and quick-Sequential Organ Failure Assessment (qSOFA), Vital-SEP outperformed all 3 of them.ConclusionIt can be concluded that using intensive care unit data even 2 vital signs are adequate to predict sepsis upto 6 h in advance with promising accuracy comparable to standard scoring methods and other sepsis predictive tools reported in literature. Vital-SEP can be used for fast-track prediction especially in limited resource hospital settings where laboratory based hematologic or biochemical assays may be unavailable, inaccurate, or entail clinically inordinate delays. A prospective study is essential to determine the clinical impact of the proposed sepsis prediction model and evaluate other outcomes such as mortality and duration of hospital stay.

To what extent is end-tidal carbon dioxide a predictor of sepsis?

Background: Sepsis is a life-threatening organ dysfunction caused by a dysregulated host response to infection. It is a major cause of death worldwide; 245 000 cases are reported in the UK annually with a mortality rate of 20.3%. Rapid diagnosis and rapid treatment of sepsis can significantly reduce mortality but sepsis can be difficult to diagnose. End-tidal carbon dioxide (EtCO2) is the measurement of expired CO2 using capnometry and waveform capnography. For CO2 to be exhaled, it must be metabolised and transported before being exhaled by effective ventilation; EtCO2 can therefore provide an indication of metabolism, circulation and ventilation. EtCO2 has already been shown to be an indicator of other metabolic acidosis conditions so this review aims to identify the usefulness of EtCO2 in identifying sepsis. Methods: A systematic literature search was conducted between March and April 2021 using the CINAHL Plus and MEDLINE databases. The results were screened and evaluated. Results: Of the 44 papers identified in the original search, seven were included in this review. Conclusion: This review suggests an EtCO2 of ≤25 mmHg (3.3 kPa) in patients with a suspected infection is diagnostic of sepsis and therefore could be used to increase the speed and accuracy of diagnosis and potentially reduce sepsis mortality. It also identifies gaps in research around UK practice and in comparing EtCO2 against UK sepsis guidelines and diagnostic tools such as the UK Sepsis Trust guidelines.

Evaluation of the effectiveness of an automated sepsis predictive tool on patient outcomes.

To evaluate the effectiveness of a multidisciplinary, hospital-wide program as part of an electronic sepsis alert tool. We used data from 15 hospitals about adult patients with severe sepsis or septic shock. Nine intervention hospitals implemented an Epic sepsis prediction tool, education, and standardized order sets (six control hospitals did not). A difference-in-difference approach evaluated their effectiveness: 1) pre-implementation period (January 1, 2016-November 15, 2018) and 2) implementation period (November 16, 2018-June 30, 2019). Outcomes included mortality, receipt of the SEP-1 bundle of care, broad spectrum antibiotic use, ICU stay, and length of stay of 6926 patients. The difference of 6.7 percentage points between the intervention and control groups in SEP-1 bundle completion was not statistically significant (p = 0.105). The increase over time for antibiotic administration ≤1 h of time zero was not larger for hospitals in the intervention group (11.7%) compared to the control-group (7.6%, p = 0.084). Differences among hospitals in both groups were not statistically different for mortality (p = 0.174), ICU stays (p = 0.174), and length of stay (p = 0.652) from pre- to implementation period. The intervention to facilitate timely sepsis care did not improve patient outcomes among those with severe sepsis or septic shock.

Biomarkers for the Prediction and Judgement of Sepsis and Sepsis Complications: A Step towards precision medicine?

Sepsis and septic shock are a major public health concern and are still associated with high rates of morbidity and mortality. Whilst there is growing understanding of different phenotypes and endotypes of sepsis, all too often treatment strategies still only employ a “one-size-fits-all” approach. Biomarkers offer a unique opportunity to close this gap to more precise treatment approaches by providing insight into clinically hidden, yet complex, pathophysiology, or by individualizing treatment pathways. Predicting and evaluating systemic inflammation, sepsis or septic shock are essential to improve outcomes for these patients. Besides opportunities to improve patient care, employing biomarkers offers a unique opportunity to improve clinical research in patients with sepsis. The high rate of negative clinical trials in this field may partly be explained by a high degree of heterogeneity in patient cohorts and a lack of understanding of specific endotypes or phenotypes. Moving forward, biomarkers can support the selection of more homogeneous cohorts, thereby potentially improving study conditions of clinical trials. This may finally pave the way to a precision medicine approach to sepsis, septic shock and complication of sepsis in the future.

Prediction of Pediatric Sepsis Using a Deep Encoding Network with Cross Features

Prediction of Pediatric Sepsis Using a Deep Encoding Network with Cross Features

Predictor Prognosis of Pediatric Septic Shock : Literature review

Introduction: The unpredictability of body response to organ dysfunction needs an effective tool to predict the prognosis of shock septic. Early recognition and treatment of septic shock improved prognosis and reduced mortality, especially in pediatrics. This review aimed to identify the predictor of mortality in pediatric septic shock. Methods: We performed a literature review of the predictor of mortality in pediatric septic shock conducted between 2015 and 2020 in ProQuest, Google Scholar, PubMed, and Science Direct. We used keywords (predictor or predictive) and (septic shock or septic), and (prognostic or prognosis) and (pediatric or children). The study selection was using the Preferred Reporting Items for Systematic Review and Meta-Analysis PRISMA framework. Results: 944 articles identified in ProQuest, 720 articles in Science Direct, 339 articles in Google Scholar, and 67 in Pubmed. Equally, the total articles were 2,070 articles, and there were 414 duplicates. After review of the complete texts was performed for 35 potential studies. In the full-text review, we excluded review articles (n = 3), different populations (n=8), and of poor quality (n = 20). Eventually, four papers were reviewed in this study. We found PELOD, PELOD-2, PIM, PIM 2, PIM 3, PMODS, PRISM, PRISM-III, PRISM-IV, and pSOFA as a predictor of sepsis in pediatrics. Conclusion: In conclusion, pSOFA is a more accurate screening result for estimating the risk of death by being 10 times more sensitive and specific. However, adding biomarkers to pSOFA will improve the accuracy of the predictor prognosis of pediatric sepsis.

Artificial and human intelligence for early identification of neonatal sepsis.

Artificial intelligence may have a role in the early detection of sepsis in neonates. Machine learning can identify patterns that predict high or increasing risk for clinical deterioration from a sepsis-like illness. In developing this potential addition to NICU care, careful consideration should be given to the data and methods used to develop, validate, and evaluate prediction models. When an AI system alerts clinicians to a change in a patient's condition that warrants a bedside evaluation, human intelligence and experience come into play to determine an appropriate course of action: evaluate and treat or wait and watch closely. With intelligently developed, validated, and implemented AI sepsis systems, both clinicians and patients stand to benefit. IMPACT: This narrative review highlights the application of AI in neonatal sepsis prediction. It describes issues in clinical prediction model development specific to this population. This article reviews the methods, considerations, and literature on neonatal sepsis model development and validation. Challenges of AI technology and potential barriers to using sepsis AI systems in the NICU are discussed.

Preoperative serum brain-derived neurotrophic factor as a predictive biomarker for sepsis after living-donor liver transplantation.

Although the survival rate after living-donor liver transplantation (LDLT) is improving, sepsis still limits the prognosis. Immune dysfunction and sarcopenia are often observed in LDLT patients, and increase susceptibility to infection. Brain-derived neurotrophic factor (BDNF) is a myokine produced by immune cells and skeletal muscle. We aimed to determine whether serum BDNF could be a feasible biomarker for sepsis of LDLT patients. We measured serum samples from 124 patients who underwent LDLT and 9 healthy volunteers for BDNF. We examined its correlation with incidence rate of sepsis. To clarify the source of BDNF, we examined its expression in lymphocytes, skeletal muscle cells, and hepatocytes. Patients who experienced sepsis showed worse short-term survival. Preoperative serum BDNF was lower in LDLT patients compared with healthy volunteers, and was also lower in Child-Pugh C compared with Child-Pugh A or B. Serum BDNF was inversely correlated with Model for End-Stage Liver Disease and controlling nutritional status (CONUT) scores, but had a weak positive correlation with skeletal muscle mass index (SMI). Multivariate analysis revealed that serum BDNF was independently associated with sepsis. Preoperative serum BDNF was a better predictor of sepsis in LDLT patients than CONUT score or SMI. Serum BDNF was positively correlated with lymphocyte counts, especially T cells. In vitro, T cells and skeletal muscle cells produced BDNF. Preoperative serum BDNF could be a predictive biomarker for sepsis after LDLT, by reflecting the systemic condition including hepatic function, nutritional status, and immune status.

Algorithmic fairness in computational medicine.

SummaryMachine learning models are increasingly adopted for facilitating clinical decision-making. However, recent research has shown that machine learning techniques may result in potential biases when making decisions for people in different subgroups, which can lead to detrimental effects on the health and well-being of specific demographic groups such as vulnerable ethnic minorities. This problem, termed algorithmic bias, has been extensively studied in theoretical machine learning recently. However, the impact of algorithmic bias on medicine and methods to mitigate this bias remain topics of active discussion. This paper presents a comprehensive review of algorithmic fairness in the context of computational medicine, which aims at improving medicine with computational approaches. Specifically, we overview the different types of algorithmic bias, fairness quantification metrics, and bias mitigation methods, and summarize popular software libraries and tools for bias evaluation and mitigation, with the goal of providing reference and insights to researchers and practitioners in computational medicine.

An application based on bioinformatics and machine learning for risk prediction of sepsis at first clinical presentation using transcriptomic data.

Background: Linking genotypic changes to phenotypic traits based on machine learning methods has various challenges. In this study, we developed a workflow based on bioinformatics and machine learning methods using transcriptomic data for sepsis obtained at the first clinical presentation for predicting the risk of sepsis. By combining bioinformatics with machine learning methods, we have attempted to overcome current challenges in predicting disease risk using transcriptomic data. Methods: High-throughput sequencing transcriptomic data processing and gene annotation were performed using R software. Machine learning models were constructed, and model performance was evaluated by machine learning methods in Python. The models were visualized and interpreted using the Shapley Additive explanation (SHAP) method. Results: Based on the preset parameters and using recursive feature elimination implemented via machine learning, the top 10 optimal genes were screened for the establishment of the machine learning models. In a comparison of model performance, CatBoost was selected as the optimal model. We explored the significance of each gene in the model and the interaction between each gene through SHAP analysis. Conclusion: The combination of CatBoost and SHAP may serve as the best-performing machine learning model for predicting transcriptomic and sepsis risks. The workflow outlined may provide a new approach and direction in exploring the mechanisms associated with genes and sepsis risk.

Machine Learning in Healthcare: Two Case Studies

Machine learning (ML) is making significant inroads into the field of medicine. It can be used as a preventative measure by predicting a patient’s diagnosis and introducing early treatment to prevent adverse outcomes or lessen their impact. However, despite many demonstrated advantages of machine learning tools in health-care, their performance assessment remains partial at best. In particular, human interactions with machine learning tools in clinical settings remain poorly researched. This review examined machine learning tools in two important areas, sepsis diagnosis and suicide prediction. However, our exploration into the use of machine learning in sepsis and suicide prediction turned up no thorough human factors analyses of provider interactions with their machine learning tools, suggesting a critical research gap waiting to be filled.

Pentraxin 3 (PTX3) as a Predictor of Severity of Sepsis in Patients Admitted to an Intensive Care Unit: A Cross-Sectional Study From North India.

Background: Sepsis is a common clinical syndrome in critical patients in the medical intensive care unit. Many scoring systems and biomarkers are introduced to detect unfavorable outcomes in sepsis patients. This study aims to identify pentraxin 3 (PTX3) as a predictor of sepsis in patients who are critically ill and admitted to intensive care units.Materials and methods: This prospective observational survey purposively included 100 patients with sepsis identified by the Surviving Sepsis Campaign guidelines in the medical intensive care unit at one of the apex care centers in North India. Socio-demographic and clinical profiles were collected using a structured and validated checklist. Simple and multi-linear regression analyses were used to determine PTX3 as a predictor of sepsis.Results: A total of 100 patients were prospectively observed. Among them, 61% were males, and 39% were females, with a mean age of 50.78 (±13.53) years. From nine potential predictors, lactate (95% CI: 1.048-1.890, B: 1.469, p < 0.001), procalcitonin (95% CI: 0.136-0.270, B: 0.203, p < 0.001), and SOFA (Sequential Organ Failure Assessment) scores (95% CI: 0.112-0.450, B: 0.281, p = 0.001) significantly predict the changes in PTX3 level (R-square: 0.842, adjusted R-square: 0.826) in patients.Conclusions: PTX3 was found to correlate with the severity of sepsis as SOFA score and other markers like lactate, procalcitonin, and APACHE-II (Acute Physiology and Chronic Health Evaluation II) score.