An interpretable machine learning-based prediction model for risk of death for patients with ischemic stroke in intensive care unit / 南方医科大学学报

OBJECTIVE@#To construct an inherent interpretability machine learning model as an explainable boosting machine model (EBM) for predicting one-year risk of death in patients with severe ischemic stroke.@*METHODS@#We randomly divided the data of 2369 eligible patients with severe ischemic stroke in the MIMIC-Ⅳ(2.0) database, who were admitted in ICU in 2008 to 2019, into a training dataset (80%) and a test dataset (20%), and assessed the prognosis of the patients using the EBM model. The prediction performance of the model was evaluated by calculating the area under the receiver operating characteristic (AUC) curve. The calibration curve and Brier score were used to evaluate the degree of calibration of the model, and a decision curve was generated to assess the net clinical benefit.@*RESULTS@#The EBM model constructed in this study had good discrimination power, calibration and net benefit, with an AUC of 0.857 (95% CI: 0.831-0.887) for predicting prognosis of severe ischemic stroke. Calibration curve analysis showed that the standard curve of the EBM model was the closest to the ideal curve. Decision curve analysis showed that the model had the greatest net benefit rate at the prediction probability threshold of 0.10 to 0.80. The top 5 independent predictive variables based on the EBM model were age, SOFA score, mean heart rate, mechanical ventilation, and mean respiratory rate, whose significance scores ranged from 0.179 to 0.370.@*CONCLUSION@#This EBM model has a good performance for predicting the risk of death within one year in patients with severe ischemic stroke and allows clinicians to better understand the contributing factors of the patients' outcomes through the model interpretability.

Predictors and Outcome of in-Hospital Cardiac Arrest Patients in Emergency Department Suez Canal University Hospitals

Background: In-hospital cardiac arrest (IHCA) is defined as cessation of cardiac activity, confirmed by the absence of signs of circulation, in a hospitalized patient who had a pulse at the time of admission. The purpose of the present study was to record the definitive predictors of IHCA, focusing on the relation between cause and outcome as well as the influence of location on survival. Subjects and Methods: This prospective observational study (cross sectional) was carried out in Emergency Department at Suez Canal University Hospital and included 223 patients experiencing IHCA at the Emergency Department (ED). Results: Our study showed return of spontanous circulation (ROSC) rate of 27.4%, which is lower than those reported in other studies from the region. In our study, we found that the overall mean duration for comprehensive cardiopulmonary resuscitation (CPR) was 21 min (SD ± 10).We found that Pulse, RR, BP, Witnessed and advanced life support (ALS) interventions at time of event were significant positive predictors to ROSC with patients while age, modified early warning score (MEWS), Interval between collapse to start CPR and CPR duration were negative predictors to cognitive impairment with diabetic patients. Conclusions: IHCA can be predicted using different variable related to patients vital data, laboratories, radiological investigations and patient demographic data which helps in predicting and modifying the outcome in limited situations.

The comparison of apache II and apache IV score to predict mortality in intensive care unit in a tertiary care hospital

Background: The prognostication of critically ill patients, in a systematic way, based on definite objective data is an integral part of the quality of care in Intensive Care Unit (ICU). Acute physiology and chronic health evaluation (APACHE) scoring systems provide an objective means of mortality prediction in Intensive Care Unit (ICU). The aims of this study were to compare the performance of APACHE II and APACHE IV in predicting mortality in our intensive care unit (ICU).Methods: A prospective observational study was conducted in a 13 bedded intensive care unit (ICU) of a tertiary level teaching hospital. All the patients above the age of 12 years, irrespective of diagnosis managed in ICU for >24hours were enrolled. APACHE II and APACHE IV scores were calculated based on the worst values in the first 24hours of admission. All enrolled patients were followed up, and outcome was recorded as survivors or non survivors. Observed mortality rates were compared with predicted mortality rates for both the APACHE II and APACHE IV. Receiver operator characteristic curves (ROC) were used to compare accuracy of the two scores.Results: APACHE II score of the patients ranged from 1 to 32 and APACHE IV score of the patients ranged from 25 to 142. There was good correlation between APACHE II and APACHE IV scores with the spearman’s rho value of 0.776 (P<0.01). Discrimination for APACHE II and APACHE IV models were good with area under ROC curve of 0.805 and 0.832 respectively. APACHE IV was more accurate than APACHE II in this regard. The cut-off point with best Youden index for APACHE II was 17 and for APACHE IV were 72 respectively for predicting mortality.Conclusions: Discrimination was better for APACHE IV than APACHE II model however Calibration was better for APACHE II than APACHE IV model in present study. There was good correlation between the two models observed in present study.

To Evaluate and Compare Ability of Two Prognostic Scoring Systems in Predicting the Mortality of Critically Ill Patients in Adult Intensive Care Unit

Introduction: Depression is a common psychiatric illness inthe elderly. It often co-exists with chronic neuropathic pain inold age group.Case report: We present a case report of an elderly patientwho was successfully treated with subanaesthetic intravenousinfusion for severe depression with suicidal intention andchronic neuropathic pain.Conclusion: Ketamine has been reported to be used indepression with suicidal features as well as refractory pain.Depression is a common psychiatric problem across the agegroups. It occurs due to neurochemical imbalance in the brainnamely dopamine, norepinephrine serotonin.

The Evaluation Of Pulp Score (PEPTIC Ulcer Perforation Score) As A Predictor Of Mortality Following Peptic Ulcer Perforation.

Introduction: Mortality and morbidity following perforated peptic ulcer (PPU) remains high, and mortality proportions of 25–30% have been reported in population–based studies.The aim of this study was to evaluate the efficacy of PULP SCORE in predicting 30 day mortality. Patients and methods: A series of 52 patients were enrolled in the study.Patients who underwent surgical treatment for perforated peptic ulcer were allotted points according to the PULP scoring system and stratified into high and low risk groups. All the data was prospectively analyzed. Observations and results:46 patients were in low risk and 6 patients were in high risk category. 5 patients were deceased in high risk group but none in low risk group. The PULP SCORE had a sensitivity of 83.33% and specificity of 97.83% in predicting mortality. In the ROC curve the AUC was 91.8%. 4 variables out of 8 variables in the score were found to be most important in predicting mortality. They were : 1. Treatment delay >24 hrs, 2. Shock on admission, 3. High ASA score, 4. Age >65 years. Conclusion:The prognostic predictors included in the PULP score can be readily identified prior to surgery, easy to use and feasible in emergency. The PULP score can assist in accurate and early identification of high–risk patients, and thus assist in risk stratification and triage of patients.

Factores clínicos y electrocardiográficos como predictores de mortalidad en pacientes en fase aguda de un primer evento cerebrovascular / Clinical and electrocardiographic factors as mortality predictors in patients in the acute phase of a first cerebrovascular event

Introducción: El accidente cerebrovascular (ACV) es la segunda causa de muerte y tercera causa de discapacidad en el mundo. Objetivos: Evaluar la asociación entre variables clínicas, electrocardiográficas, escalas neurológicas en pacientes con ACV como predictoras de mortalidad a 3 meses posteriores al egreso hospitalario. Materiales y métodos: Estudio de cohorte prospectivo con muestreo no probabilístico, en pacientes mayores de 18 años con primer ACV. Se evaluaron variables demográficos, clínicas, escalas neurológicas del Instituto Nacional de Salud (NIHSS) y canadiense (CNS), variabilidad de la frecuencia cardiaca (VFC) y del QT (QTV), dispersión del QT. Se determinó la mortalidad a los tres meses de seguimiento. Se realizó análisis bivariado y de regresión logística múltiple cuyo desenlace fue mortalidad a tres meses post egreso hospitalario, incluyendo variables con baja correlación (r< 0.4) y significancia estadística (p<0.05). Resultados: Se incluyeron 92 pacientes, 13 de los cuales fallecieron en la fase de tratamiento intrahospitalario. Se realizó seguimiento durante tres meses después del egreso hospitalario en 81 pacientes. La mortalidad total en tres meses de seguimiento fue del 21.7%(n=20). Se identificaron cinco variables predictoras de mortalidad en el modelo final: puntaje de escala NIHSS, frecuencia cardiaca media, VLF QT ≥36.311, LF/HF ≤1.019, valores extremos r-MSD (≥7.985o≤2.363) de VFC. La capacidad discriminatoria del modelo mediante el análisis del área bajo la curva fue de 0.95, con valores de sensibilidad y especificidad del 60% y 93% respectivamente. Conclusion: Altos puntajes de escala NIHSS, VLF-QT, frecuencia cardiaca media, así como valores bajos LF/HF y valores extremos r-MSD, fueron factores de riesgo independientes para mortalidad a los 90 días después de un primer ACV.

Introduction: Stroke is the second cause of death and third cause of disability worldwide. Objective: To assess association between clinical and electrocardiographic variables, neurological scales in stroke patients like predictors of mortality at three months after hospital discharge. Subjects and methods: Prospective cohort with nonprobabilistic sampling, in patients over 18 years with first stroke. Demographic and clinical variables, neurological scales (NIHSS, Canadian), heart rate (HRV) and QT variability (QTV), QT dispersion were evaluated. Mortality was determined during the 3 months follow up. Bivariate and multiple logistic regression analysis were performed with mortality at three months after discharge as outcome. Variables were included in the model if they have low correlation (r<0.4) and significant statistically p values (P< 0.05). Results: 92 patients were included in the study, 13 patients died during the intra-hospital stay, 81 were followed at 3 months after their hospital discharge. Total mortality in patients included at three months follow-up was 21.7 % (n=20). We identified five predictors of mortality in the final model: NIHSS score, mean heart rate, VLF QT ≥36,311, LF/HF ≤ 1,019, extreme values of r-MSD (≥ 7,985 or ≤ 2,363) of HRV. The area under the curve (AUC) of the model was 0,95 with sensitivity of 60% and specificity of 93%. Conclusions: High NIHSS scores, VLF-QT, mean heart rate, low values of LF/HF and high extreme values of r-MSD were independent risk factors for mortality at 90 days after a first stroke.

Validación de dos escalas de valor pronóstico en niños que ingresan a UCI / Validation of two prediction value scores in children admitted to the intensive care unit

Las escalas PIM (Índice de Mortalidad Pediátrica) y PELOD (Índice Pediátrico de Disfunción Orgánica) son sistemas de evaluación que permiten la estimación de la severidad de la enfermedad y el ajuste del riesgo de mortalidad en grupos heterogéneos de pacientes. El objetivo del presente trabajo fue el de validar las escalas PIM y PELOD en una Unidad de Cuidados Intensivos pediátrica (UCIP). Metodología. Fueron incluidos 97 niños con edad menor o igual a 12 años; las variables estudiadas fueron la mortalidad o sobrevida durante la estancia en UCI. PIM incluye 7 variables medidas durante la primera hora de admisión a UCI; PELOD incluye disfunción de seis sistemas orgánicos en 12 variables. Para estimar discriminación, se utilizó el área bajo la curva de rendimiento diagnóstico, y para evaluar calibración, la bondad de ajuste de Hosmer-Lemeshow. Resultados. Edad media 4,0 años (rango intercuartil 1,0-8,1); estancia 6,0 días; (rango 3,0 a 17,0); las principales causas de ingreso a UCIP fueron accidentes 30, sepsis 19, neurológicas 14. Desarrollaron disfunción orgánica múltiple 58 (59,8%) de 97. La mortalidad observada fue de 17,5%. La predicción de riesgo de mortalidad por PIM fue significativamente más alta en no sobrevivientes (0,48±0,35) que sobrevivientes (0,18±0,23; t test 3,40 p<0,003); calibración (p=0,025) y discriminación (área bajo la curva = 0,79 ± 0,057; p<0,001) de PIM fue buena. Conclusión: PIM es una medida válida de predicción de riesgo de mortalidad en UCIP en nuestro medio

The Pediatric Index of Mortality (PIM) and Pediatric Logistic Organ Dysfunction (PLOD) scale are scoring systems that allow assessment of the severity of illness and mortality risk adjustment in heterogeneous groups of patients. The aim of this study was to validate the accuracy and reliability of PIM and PELOD scoring in a pediatric Intensive Care Unit (ICU) Methods: 97 children under 12 years of age were included. Survival and mortality during the stay in the ICU were studied. PIM scale includes 7 parameters measured during the first hour of admission to the ICU; PELOD includes dysfunction of 6 organs and systems in 12 variables. The area under the curve was used to assess discrimination and calibration was assessed with the Hosmer-Lemeshow goodness of fit test. Results: The median patient age was 4,0 years (inter-quartile range 1,0-8,1), median length of stay was 6 days (range 3-17). Main causes for admission to the ICU were accidents 30, sepsis 19, neurological 14. Fifty eight patients (59,8%) developed multiple organic dysfunction. Observed mortality was 17,5%. Prediction of risk of mortality with PIM was significantly higher in non survivors (0,48 ± 0,35) than in survivors (0,18 ± 0,23); t test 3,40 p<0,003; calibration (p=0,025) and discrimination (area under the curve = 0,79±0,057; p<0,001) for PIM was good. Conclusions: PIM is a valid prediction index for mortality risk in pediatric ICU in our hospitals

Pre-growth mortality of Abies cilicica trees and mortality models performance.

In this study, we compared tree-growth rates (basal area increment) from recently dead and living Taurus fir (Abies cilicica Carr.) trees in the Kovada lake Forest of Isparta, Turkey. For each dead tree, tree-growth rates were analyzed for the presence of pre-death growth depressions in the study area (number of sample plots=11) in 2006. However, we compared both the magnitude and rate of growth prior to death to a control (living) group of trees. Basal area increment (BAI) averaged substantially less during the last 10 years before death than for control trees. Trees that died started diverging in growth, on average, 50-60 years before death. About 18% of trees that died had chronically slow growth, 46% had pronounced declines in growth, whereas 36% had good growth up to death. However, tree-ring-based growth patterns of dead and living Taurus fir trees were compared and used 12 mortality models that were derived using logistic regression from growth patterns of tree-ring series as predictor variables. The four models with the highest overall performance correctly classified 43.8-56.3% of all dead trees and 75.0-87.5% of all living trees, and they predicted 25.0-43.8% of all dead trees to die within 0-15 years prior to the actual year of death.

Validação de um escore de alerta precoce pré-admissão na unidade de terapia intensiva / Validation of an early warning score in pre-intensive care unit

JUSTIFICATICA E OBJETIVOS: O prognóstico dos pacientes admitidos em unidade de terapia intensiva (UTI) tem relação com sua gravidade nos momentos que precedem a internação. O objetivo deste estudo foi avaliar a gravidade dos pacientes 12, 24 e 72h antes da admissão na UTI, bem como qual o parâmetro mais prevalente nesses pacientes e correlacionar o Modified Early Warning Score (MEWS) no pré-UTI com o desfecho (sobrevivência versus óbito), respectivo. MÉTODO: Análise retrospectiva de 65 pacientes, nas 72 horas que antecederam a admissão na UTI, no perío-do de julho a outubro de 2006. RESULTADOS: O APACHE II médio foi 22,2 ± 7,9 pontos, a mortalidade real de 54,6 por cento e a taxa de mortalidade padronizada foi 1,24. O MEWS médio foi 3,7 ± 0,2; 4,0 ± 0,2 e 5,1 ± 0,2 pontos, calculado 72, 48 e 24h antes da admissão na UTI, respectivamente. Registrou-se um percentual crescente de pacientes com MEWS > 3 pontos nas 72, 48 e 24h antes da admissão - 43,8 por cento, 59,4 por cento e 73,4 por cento, respectivamente. Dentre os parâmetros fisiológicos, a freqüência respiratória foi a que mais contribuiu para a pontuação do MEWS. A mortalidade foi maior entre os pacientes com MEWS > 3 pontos já 72 horas antes da admissão. Entre os pacientes que faleceram, verificou-se um aumento significativo no MEWS médio, 24 horas antes da admissão à UTI (em relação ao registrado, 72 horas antes), fato não identificado nos sobreviventes. CONCLUSÕES: O MEWS identificou com fidelidade a gravidade dos pacientes admitidos na UTI, sugerindo ser um escore confiável à aplicação nas instancias que precedem a UTI.

BACKGROUND AND OBJECTIVES: Prognosis of patients in the intensive care unit (ICU) has a relation with their severity just before admission. The Modified Early Warning Score (MEWS) was used to evaluate the severe condition of patients 12, 24 and 72 hours before admission in the ICU, assess the most prevalent parameters and correlate the MEWS before ICU with the outcome (survival versus death). METHODS: Retrospective analyses of 65 patients consecutively admitted to the ICU from July to October, 2006 evaluating the physiological parameters 72 hours prior to admission. RESULTS: APACHE II mean was 22.2 ± 7.9 points, mortality was 54.6 percent and standardized mortality ratio means was 1.24. MEWS means were 3.7 ± 0.2; 4.0 ± 0.2 and 5.1 ± 0.2 points, calculated 72, 48 and 24 hours previous to ICU admission, respectively. An increasing percentage of patients with MEWS > 3 points within 72, 48 and 24 hours before admission - 43.8 percent, 59.4 percent and 73.4 percent, respectively was recorded. Among the included physiological parameters respiratory rate contributed the most to the MEWS. Highest mortality was found in patients with MEWS > 3 points already found 72 hours before admission. Patients who died presented with a significant increase in the MEWS 24 hours prior to admission to the ICU (in relation to the MEWS recorded 72 hours before) but the situation was not identified in survivors. CONCLUSIONS: MEWS closely identified the severity of patients admitted to the ICU, suggesting that it can be a reliable score, useful in the situations preceding the ICU.

Comparing the Usefulness of the Initial Acute Physiologic and Chronic Health Evaluation (APACHE) II Score in the Emergency Department (ED) and the Mortality in Emergency Department Sepsis (MEDS) Score for Predicting the Prognosis of Septic Patients Admitt / 대한구급학회지

BACKGROUND: To determine the prognostic value of the initial APACHE II score in the ED compared with the classic APACHE II score in the ICU and to check the usefulness of the MEDS score together for more rapid risk stratification of septic patients admitted to the ICU via the ED. METHODS: We prospectively checked the initial APACHE II and MEDS scores of all the patients who had systemic inflammatory response syndrome in the ED and the classic APACHE II scores after admission to the ICU, as well 6 months later. We enrolled the only sepsis cases in the final diagnosis after reviewing the medical records. We evaluated the predictive abilities of the initial APACHE II and MEDS scores compared with the classic APACHE II score. RESULTS: During 6 months, 58 patients diagnosed with sepsis were enrolled. Twenty-four (41.4%) patients died within 28 days of admission and 34 patients survived. The mortality group had a significantly higher mean classic APACHE II score (19 +/- 6.7 vs. 15 +/- 5.0, p < 0.01) and a higher mean MEDS score (16.67 +/- 2.70 vs. 8.91 +/- 3.11, p < 0.01) than the survivor group. The initial APACHE II score at the ED was not significantly different between the two groups. ROC analysis showed the discriminative power of the MEDS score in predicting mortality was much better than the APACHE II score (areas under the curves of the APACHE II score in the ED and ICU, and the MEDS scores were 0.668, 0.807, and 0.967, respectively; p < 0.01). CONCLUSIONS: The initial APACHE II score in the ED did not predict mortality better than the classic APACHE II score. However, the MEDS score predicted the poor prognosis of septic patients more rapidly and accurately in the ED than the APACHE II model.

Comparison of Predict Mortality Scoring Systems for Spontaneous Intracerebral Hemorrhage Patients / 성인간호학회지

PURPOSE: The purpose of this study was to evaluate and compare the predictive ability of three mortality scoring systems; Acute Physiology and Chronic Health Evaluation(APACHE) III, Simplified Acute Physiology Score(SAPS) II, and Mortality Probability Model(MPM) II in discriminating in-hospital mortality for intensive care unit(ICU) patients with spontaneous intracerebral hemorrhage. METHODS: Eighty-nine patients admitted to the ICU at a university hospital in Daejeon Korea were recruited for this study. Medical records of the subject were reviewed by a researcher from January 1, 2003 to March 31, 2004, retrospectively. Data were analyzed using SAS 8.1. General characteristic of the subjects were analyzed for frequency and percentage. RESULTS: The results of this study were summarized as follows. The values of the Hosmer-Lemeshow's goodness-of-fit test for the APACHE III, the SAPS II and the MPM II were chi-square H=4.3849 p=0.7345, chi-square H= 15.4491 p=0.0307, and chi-square H=0.3356 p=0.8455, respectively. Thus, The calibration of the MPM II found to be the best scoring system, followed by APACHE III. For ROC curve analysis, the areas under the curves of APACHE III, SAPS II, and MPM II were 0.934, 0.918 and 0.813, respectively. Thus, the discrimination of three scoring systems were satisfactory. For two-by-two decision matrices with a decision criterion of 0.5, the correct classification of three scoring systems were good. CONCLUSION: Both the APACHE III and the MPM II had an excellent power of mortality prediction and discrimination for spontaneous intracerebral hemorrhage patients in ICU.

Predictive Comparisons of Procalcitonin (PCT) Level, Arterial Ketone Body Ratio (AKBR), APACHE III Score and Multiple Organ Dysfunction Score (MODS) in Systemic Inflammatory Response Syndrome (SIRS)

Procalcitonin (PCT) is a newly introduced marker of systemic inflammation and bacterial infection. A marked increase in circulating PCT level in critically ill patients has been related with the severity of illness and poor survival. The goal of this study was to compare the prognostic power of PCT and three other parameters, the arterial ketone body ratio (AKBR), the acute physiology, age, chronic health evaluation (APACHE) III score and the multiple organ dysfunction score (MODS), in the differentiation between survivors and nonsurvivors of systemic inflammatory response syndrome (SIRS). The study was performed in 95 patients over 16 years of age who met the criteria of SIRS. PCT and AKBR were assayed in arterial blood samples. The APACHE III score and MODS were recorded after the first 24 hours of surgical ICU (SICU) admission and then daily for two weeks or until either discharge or death. The patients were divided into two groups, survivors (n=71) and nonsurvivors (n=24), in accordance with the ICU outcome. They were also divided into three groups according to the trend of PCT level: declining, increasing or no change. Significant differences between survivors and nonsurvivors were found in APACHE III score and MODS throughout the study period, but in PCT value only up to the 7th day and in AKBR only up to the 3rd day. PCT values of the three groups were not significantly different on the first day between survivors and nonsurvivors. Receiver operating characteristic (ROC) curves for prediction of mortality by PCT, AKBR, APACHE III score and MODS were 0.690, 0.320, 0.915 and 0.913, respectively, on the admission day. In conclusion, PCT could have some use as a mortality predictor in SIRS patients but was less reliable than APACHE III score or MODS.