Prediction of Sepsis in the Intensive Care Unit With Minimal Electronic Health Record Data: A Machine Learning Approach
Thomas DesautelsJacob CalvertJana HoffmanMelissa JayYaniv KeremLisa ShiehDavid ShimabukuroUli K. ChettipallyMitchell D. FeldmanChris W. BartonDavid J. WalesRitankar Das
443
Citation
21
Reference
10
Related Paper
Citation Trend
Abstract:
Sepsis is one of the leading causes of mortality in hospitalized patients. Despite this fact, a reliable means of predicting sepsis onset remains elusive. Early and accurate sepsis onset predictions could allow more aggressive and targeted therapy while maintaining antimicrobial stewardship. Existing detection methods suffer from low performance and often require time-consuming laboratory test results.To study and validate a sepsis prediction method, InSight, for the new Sepsis-3 definitions in retrospective data, make predictions using a minimal set of variables from within the electronic health record data, compare the performance of this approach with existing scoring systems, and investigate the effects of data sparsity on InSight performance.We apply InSight, a machine learning classification system that uses multivariable combinations of easily obtained patient data (vitals, peripheral capillary oxygen saturation, Glasgow Coma Score, and age), to predict sepsis using the retrospective Multiparameter Intelligent Monitoring in Intensive Care (MIMIC)-III dataset, restricted to intensive care unit (ICU) patients aged 15 years or more. Following the Sepsis-3 definitions of the sepsis syndrome, we compare the classification performance of InSight versus quick sequential organ failure assessment (qSOFA), modified early warning score (MEWS), systemic inflammatory response syndrome (SIRS), simplified acute physiology score (SAPS) II, and sequential organ failure assessment (SOFA) to determine whether or not patients will become septic at a fixed period of time before onset. We also test the robustness of the InSight system to random deletion of individual input observations.In a test dataset with 11.3% sepsis prevalence, InSight produced superior classification performance compared with the alternative scores as measured by area under the receiver operating characteristic curves (AUROC) and area under precision-recall curves (APR). In detection of sepsis onset, InSight attains AUROC = 0.880 (SD 0.006) at onset time and APR = 0.595 (SD 0.016), both of which are superior to the performance attained by SIRS (AUROC: 0.609; APR: 0.160), qSOFA (AUROC: 0.772; APR: 0.277), and MEWS (AUROC: 0.803; APR: 0.327) computed concurrently, as well as SAPS II (AUROC: 0.700; APR: 0.225) and SOFA (AUROC: 0.725; APR: 0.284) computed at admission (P<.001 for all comparisons). Similar results are observed for 1-4 hours preceding sepsis onset. In experiments where approximately 60% of input data are deleted at random, InSight attains an AUROC of 0.781 (SD 0.013) and APR of 0.401 (SD 0.015) at sepsis onset time. Even with 60% of data missing, InSight remains superior to the corresponding SIRS scores (AUROC and APR, P<.001), qSOFA scores (P=.0095; P<.001) and superior to SOFA and SAPS II computed at admission (AUROC and APR, P<.001), where all of these comparison scores (except InSight) are computed without data deletion.Despite using little more than vitals, InSight is an effective tool for predicting sepsis onset and performs well even with randomly missing data.Keywords:
Mews
Electronic health record
Medical record
Objective:To compare the ability of early warning score(EWS) and modified early warning score(MEWS)to predict the mortality of admissions from emergency department.Methods:Randomly select 409 patients who were the emergency admissions in West China Hospital of Sichuan University.Collected the vital signs and the general state.Use the EWS and MEWS to value the patients,and use the receiver operating characteristics curve(ROC) to analyze the discrimination of the score and the risk of death.Results:The area under the curve of EWS is 0.849±0.132,the best cut-off value is 4 score.The area under the curve is 0.876±0.124,and the best cut-off value is 5 score.Conclusion:The MEWS have a better ability to identify the risk of emergency admissions' mortality,but both needs more improvement.
Mews
Warning signs
Cite
Citations (0)
Mews
Triage
Cite
Citations (132)
Objective: To investigate the value of modified early warning score in elderly emergency patients. Method:216 consecutive elderly emergency patients in emergency ward and rescue room were scored with MEWS, 61 elderly patients scored with MEWS from emergency department to intensive care unit and were reviewed. Results:The study shows that increasing MEWS score was associated with worse outcome. The score of death group is significantly higher than survival group(P0.01). Not ICU admission is significantly lower than ICU admission(P0.05). Whereas There is no significantly difference in MEWS score between cardiac arrest in 6 hs and death after 6 hs(P0.05). Conclusion:The MEWS is a useful and appropriate risk-management tool. It can evaluate severity of elderly emergency patients and discriminate the emergency potential severity patients.
Mews
Cite
Citations (0)
Objective To study the prognostic value of modified early warning score (MEWS) and Acute Physiology and Chronic Health Evaluation (APACHE Ⅱ) score in emergency ICU elderly patients.Methods 113 elderly patients were collected in emergency ICU,after admission to calculate the MEWS score and APACHE Ⅱ score,the MEWS score and APACHE Ⅱ score difference were compared between death group and survival group.the differences were compared in MEWS score ≤ 3,4 ～6,7 ～ 9,≥ 10 scores groups respectively;analysis of the MEWS score and APACHE Ⅱ the relevance score.the correlation analysis of the MEWS score and APACHE Ⅱ score was given.Results The MEWS(7.91 ± 2.42) point and APACHE Ⅱ (21.9 ± 4.18) point in death group were greater than that in survival group [(4.51 ± 2.14) point,(19.53 ± 4.37) point] (t =7.49,2.70,all P ＜ 0.01).With MEWS score increasing,the fatality rate rose(P ＜0.01).the MEWS score and APACHE Ⅱ score was correlated(r =0.617,P ＜ 0.01).Conclusion The MEWS score was a simple and quick scoring system,as well as APACHE Ⅱ,it can predict the prognosis of elderly patients in emergency department,it has important application value.
Key words:
Modified early warning score ; Health status indicators; Aged; Prognosis
Mews
Case fatality rate
Cite
Citations (0)
Mews
Cite
Citations (0)
The 2016 definitions of sepsis included the quick Sepsis-related Organ Failure Assessment (qSOFA) score to identify high-risk patients outside the intensive care unit (ICU).We sought to compare qSOFA with other commonly used early warning scores.All admitted patients who first met the criteria for suspicion of infection in the emergency department (ED) or hospital wards from November 2008 until January 2016 were included. The qSOFA, Systemic Inflammatory Response Syndrome (SIRS), Modified Early Warning Score (MEWS), and the National Early Warning Score (NEWS) were compared for predicting death and ICU transfer.Of the 30,677 included patients, 1,649 (5.4%) died and 7,385 (24%) experienced the composite outcome (death or ICU transfer). Sixty percent (n = 18,523) first met the suspicion criteria in the ED. Discrimination for in-hospital mortality was highest for NEWS (area under the curve [AUC], 0.77; 95% confidence interval [CI], 0.76-0.79), followed by MEWS (AUC, 0.73; 95% CI, 0.71-0.74), qSOFA (AUC, 0.69; 95% CI, 0.67-0.70), and SIRS (AUC, 0.65; 95% CI, 0.63-0.66) (P < 0.01 for all pairwise comparisons). Using the highest non-ICU score of patients, ≥2 SIRS had a sensitivity of 91% and specificity of 13% for the composite outcome compared with 54% and 67% for qSOFA ≥2, 59% and 70% for MEWS ≥5, and 67% and 66% for NEWS ≥8, respectively. Most patients met ≥2 SIRS criteria 17 hours before the combined outcome compared with 5 hours for ≥2 and 17 hours for ≥1 qSOFA criteria.Commonly used early warning scores are more accurate than the qSOFA score for predicting death and ICU transfer in non-ICU patients. These results suggest that the qSOFA score should not replace general early warning scores when risk-stratifying patients with suspected infection.
Mews
Cite
Citations (589)
Modified early warning scoring (MEWS) uses abnormalities in routine observations to identify patients at risk of critical illness. Nurses recorded scores at or above the medical response score of 3 on a hospital clinical information system during the first year of introducing MEWS to 10 wards in a university hospital. A total of 619 triggers were recorded in 365 patients. Fifty-nine required intensive care unit (ICU)/high dependency unit (HDU) care; 71 died. Survival was significantly worse for initial scores >4 (35/104 patients died) than for scores 3-4 (P<0.004). Multivariant analysis showed age (P<0.001) and trigger score (P<0.001) but not ward specialty (P=0.1) predicted death. Mean ages of survivors and non-survivors were 64 years (SD 18) and 74 years (SD 17), respectively. Addition of a score for age did not significantly increase the area under a receiver operator characteristic curve for the predictive value of MEWS scores. The study shows that increasing MEWS score is associated with worse outcome across a range of specialties and that nursing staff will use a patient information system to audit MEWS scores.
Mews
Specialty
Cite
Citations (49)
Abstract The commonly used Modified Early Warning score (MEWS) may poorly predict deterioration in COVID-19 patients. Therefore, an adjusted MEWS for COVID-19 patients (CEWS) was constructed. CEWS exceeded MEWS at all time points and impending death or intensive care unit (ICU) admission was strongly correlated with a persistently high CEWS.
Mews
Cite
Citations (3)
Objective To investigate the evaluation of modified early warning score( MEWS) on the condition and prognosis of patients in rescue rooms of departments of internal medicine.Methods A total of 268 medical patients admitted to this hospital from March to April 2013 were divided,according to MEWS,into groups A( with scores≥5,n = 97),B( with scores of 0-4,n = 171).The rates of ICU admission,mortality were compared between 2 groups; the whereabouts and 30 d outcomes of patients with different MEWS total scores were recorded,and the correlation between MEWS total score and ICU ad-mission rate analyzed.Results In group B,3 patients were admitted to ICU,1 died; in group A,61,9,respectively; the rates of ICU admission,mortality were higher in group A than in group B( P 0.05).Ninety-eight patients had MEWS total scores of 0-2( 1 admitted to ICU),73 had 3-4( 2 to ICU),72 had 5-6( 39 to ICU),13 had 7-8( 13 to ICU),12 had≥9( 9 to ICU).By linear correlation analysis,MEWS total score was positively correlated with ICU admission( r = 0.951,P = 0.013).Conclusion MEWS,having some abilities to predict patients' condition and prognosis,is worth widely applying.
Mews
Cite
Citations (0)
It is necessary to identify critical patients requiring hospitalization early due to the rapid increase in the number of COVID-19 cases.This study aims to evaluate the effectiveness of scoring systems such as emergency department triage early warning score (TREWS) and modified early warning score (MEWS) in predicting mortality in COVID-19 patients.In this retrospective cohort study, PCR positive patients evaluated for COVID-19 and decided to be hospitalized were evaluated. During the first evaluation, MEWS and TREWS scores of the patients were calculated. Intensive care needs as well as 24-h and 28-day mortality rates were evaluated.A total of 339 patients were included in the study. While 30 (8.8%) patients were hospitalized in the intensive care unit, 4 (1.2%) died in the emergency. The number of patients who died within 28 days was found to be 57 (16.8%). In 24-h mortality, the median MEWS value was found to be 7 (IQR 25-75) while the TREWS value was 11.5 (IQR 25-75). In the ROC analysis made for the diagnostic value of 28-day mortality of MEWS and TREWS scores, the area under the curve (AUC) for the MEWS score was found to be 0.833 (95% CI 0.777-0.888, p < 0.001) while it was identified as 0.823 (95% CI 0.764-0.882, p < 0.001) for the TREWS.MEWS and TREWS calculated at emergency services are effective in predicting 28-day mortality in patients requiring hospitalization due to COVID-19.
Mews
Triage
Cite
Citations (29)