Prediction of extubation failure in the paediatric cardiac ICU using machine learning and high-frequency physiologic data

Background: Cardiac intensivists frequently assess patient readiness to wean off mechanical ventilation with an extubation readiness trial despite it being no more effective than clinician judgement alone. We evaluated the utility of high-frequency physiologic data and machine learning for improving the prediction of extubation failure in children with cardiovascular disease. Methods: This was a retrospective analysis of clinical registry data and streamed physiologic extubation readiness trial data from one paediatric cardiac ICU (12/2016-3/2018). We analysed patients’ final extubation readiness trial. Machine learning methods (classification and regression tree, Boosting, Random Forest) were performed using clinical/demographic data, physiologic data, and both datasets. Extubation failure was defined as reintubation within 48 hrs. Classifier performance was assessed on prediction accuracy and area under the receiver operating characteristic curve. Results: Of 178 episodes, 11.2% (N = 20) failed extubation. Using clinical/demographic data, our machine learning methods identified variables such as age, weight, height, and ventilation duration as being important in predicting extubation failure. Best classifier performance with this data was Boosting (prediction accuracy: 0.88; area under the receiver operating characteristic curve: 0.74). Using physiologic data, our machine learning methods found oxygen saturation extremes and descriptors of dynamic compliance, central venous pressure, and heart/respiratory rate to be of importance. The best classifier in this setting was Random Forest (prediction accuracy: 0.89; area under the receiver operating characteristic curve: 0.75). Combining both datasets produced classifiers highlighting the importance of physiologic variables in determining extubation failure, though predictive performance was not improved. Conclusion: Physiologic variables not routinely scrutinised during extubation readiness trials were identified as potential extubation failure predictors. Larger analyses are necessary to investigate whether these markers can improve clinical decision-making.

Significant Physical and Exercise-Related Variables for Exercise-Centred Lifestyle: Big Data Analysis for Gynaecological Cancer Patients

This study investigated the characteristics of gynaecological cancers and is aimed at identifying significant risk variables using the National Health Insurance Sharing Service database to develop practical interventions for affected patients. Data regarding patients with uterine and ovarian cancer from the National Health Insurance Sharing Service database were collected and analysed using Student’s t -test, logistic regression, and receiver operating characteristic curve analyses. Student’s t -test analyses revealed that age, body mass index, blood pressure, and waist variables differed significantly among patients with uterine cancer. Gamma-glutamyl transpeptidase levels were higher in patients with ovarian cancer than in patients with uterine cancer. Physical fitness function tests reflected the status of patients with cancer. Moreover, physical disability was associated with an increased incidence of ovarian cancer. Intensive exercise for 20 min more than 1 time per week must be avoided to prevent uterine cancer. Receiver operating characteristic curve analyses showed that the optimal cutoff value for one-leg standing time, a prognostic and preventive factor in ovarian cancer, was 9.50 s (sensitivity, 94.9%; specificity, 96.9%). Controlling significant variables for each gynaecological cancer type in an individualised and optimised manner is recommended, including by maintenance of an adjusted exercise-centred lifestyle.

Posterior National Institutes of Health Stroke Scale Improves Prognostic Accuracy in Posterior Circulation Stroke

Background and Purpose: The National Institutes of Health Stroke Scale (NIHSS) underestimates clinical severity in posterior circulation stroke and patients presenting with low NIHSS may be considered ineligible for reperfusion therapies. This study aimed to develop a modified version of the NIHSS, the Posterior NIHSS (POST-NIHSS), to improve NIHSS prognostic accuracy for posterior circulation stroke patients with mild-moderate symptoms. Methods: Clinical data of consecutive posterior circulation stroke patients with mild-moderate symptoms (NIHSS <10), who were conservatively managed, were retrospectively analyzed from the Basilar Artery Treatment and Management registry. Clinical features were assessed within 24 hours of symptom onset; dysphagia was assessed by a speech therapist within 48 hours of symptom onset. Random forest classification algorithm and constrained optimization were used to develop the POST-NIHSS in the derivation cohort. The POST-NIHSS was then validated in a prospective cohort. Poor outcome was defined as modified Rankin Scale score ≥3 at 3 months. Results: We included 202 patients (mean [SD] age 63 [14] years, median NIHSS 3 [interquartile range, 1–5]) in the derivation cohort and 65 patients (mean [SD] age 63 [16] years, median NIHSS 2 [interquartile range, 1–4]) in the validation cohort. In the derivation cohort, age, NIHSS, abnormal cough, dysphagia and gait/truncal ataxia were ranked as the most important predictors of functional outcome. POST-NIHSS was calculated by adding 5 points for abnormal cough, 4 points for dysphagia, and 3 points for gait/truncal ataxia to the baseline NIHSS. In receiver operating characteristic analysis adjusted for age, POST-NIHSS area under receiver operating characteristic curve was 0.80 (95% CI, 0.73–0.87) versus NIHSS area under receiver operating characteristic curve, 0.73 (95% CI, 0.64–0.83), P =0.03. In the validation cohort, POST-NIHSS area under receiver operating characteristic curve was 0.82 (95% CI, 0.69–0.94) versus NIHSS area under receiver operating characteristic curve 0.73 (95% CI, 0.58–0.87), P =0.04. Conclusions: POST-NIHSS showed higher prognostic accuracy than NIHSS and may be useful to identify posterior circulation stroke patients with NIHSS <10 at higher risk of poor outcome.

Premio de Investigación SCHOT 2020: desarrollo y validación de un modelo multivariables de predicción de estadía hospitalaria en pacientes mayores de 65 años sometidos artroplastia total de cadera electiva en Chile utilizando aprendizaje de máquinas

Resumen Introducción La predicción de la estadía hospitalaria luego de una artroplastia total de cadera (ATC) electiva es crucial en la evaluación perioperatoria de los pacientes, con un rol determinante desde el punto de vista operacional y económico. Internacionalmente, se han empleado macrodatos (big data, en inglés) e inteligencia artificial para llevar a cabo evaluaciones pronósticas de este tipo. El objetivo del presente estudio es desarrollar y validar, con el empleo del aprendizaje de máquinas (machine learning, en inglés), una herramienta capaz de predecir la estadía hospitalaria de pacientes chilenos mayores de 65 años sometidos a ATC por artrosis. Material y Métodos Empleando los registros electrónicos de egresos hospitalarios anonimizados del Departamento de Estadísticas e Información de Salud (DEIS), se obtuvieron los datos de 8.970 egresos hospitalarios de pacientes sometidos a ATC por artrosis entre los años 2016 y 2018. En total, 15 variables disponibles en el DEIS, además del porcentaje de pobreza de la comuna de origen del paciente, fueron incluidos para predecir la probabilidad de que un paciente presentara una estadía acortada (< 3 días) o prolongada (> 3 días) luego de la cirugía. Utilizando técnicas de aprendizaje de máquinas, 8 algoritmos de predicción fueron entrenados con el 80% de la muestra. El 20% restante se empleó para validar las capacidades predictivas de los modelos creados a partir de los algoritmos. La métrica de optimización se evaluó y ordenó en un ranking utilizando el área bajo la curva de característica operativa del receptor (area under the receiver operating characteristic curve, AUC-ROC, en inglés), que corresponde a cuan bien un modelo puede distinguir entre dos grupos. Resultados El algoritmo XGBoost obtuvo el mejor desempeño, con una AUC-ROC promedio de 0,86 (desviación estándar [DE]: 0,0087). En segundo lugar, observamos que el algoritmo lineal de máquina de vector de soporte (support vector machine, SVM, en inglés) obtuvo una AUC-ROC de 0,85 (DE: 0,0086). La importancia relativa de las variables explicativas demostró que la región de residencia, el servicio de salud, el establecimiento de salud donde se operó el paciente, y la modalidad de atención son las variables que más determinan el tiempo de estadía de un paciente. Discusión El presente estudio desarrolló algoritmos de aprendizaje de máquinas basados en macrodatos chilenos de libre acceso, y logró desarrollar y validar una herramienta que demuestra una adecuada capacidad discriminatoria para predecir la probabilidad de estadía hospitalaria acortada versus prolongada en adultos mayores sometidos a ATC por artrosis. Conclusión Los algoritmos creados a traves del empleo del aprendizaje de máquinas permiten predecir la estadía hospitalaria en pacientes chilenos operado de artroplastia total de cadera electiva.

The Optimal Cutoff Value of D-dimer Levels to Predict in Hospital Mortality in Severe Cases of Coronavirus Disease 2019

Introduction: Coronavirus disease 2019 (Covid-19) can cause coagulation parameters abnormalities such as an increase of D-dimer levels especially in severe cases. The purpose of this study is to determine the differences of D-dimer levels in severe cases of Covid-19 who survived and non-survived and determine the optimal cut-off value of D-dimer levels to predict in-hospital mortality. Method: Data were obtained from confirmed Covid-19 patients who were treated from June to September 2020. The Mann-Whitney U test was used to determine differences of D-dimer levels in surviving and non-surviving patients. The optimal cut-off value and area under the curve (AUC) of the D-dimer level in predicting mortality were obtained by the receiver operating characteristic curve (ROC) method. Results: A total of 80 patients were recruited in this study. Levels of D-dimer were significantly higher in non-surviving patients (median 3.346 mg/ml; minimum – maximum: 0.939 – 50.000 mg/ml) compared to surviving patients (median 1.201 mg/ml; minimum – maximum: 0.302 – 29.425 mg/ml), p = 0.012. D-dimer levels higher than 1.500 mg/ml are the optimal cut-off value for predicting mortality in severe cases of Covid-19 with a sensitivity of 80.0%; specificity of 64.3%; and area under the curve of 0.754 (95% CI 0.586 - 0.921; p = 0.010). Conclusions: D-dimer levels can be used as a predictor of mortality in severe cases of Covid-19.

The correlation between nutrition and frailty and the receiver operating characteristic curve of different nutritional indexes for frailty

Background Frailty is a kind of geriatric syndrome, which is very common in the elderly. Patients with malnutrition are at higher risk of frailty. This study explored the correlation between nutrition and frailty and compared the receiver operating characteristic curve of different nutritional indexes for frailty. Methods This cross-sectional study included 179 inpatients aged ≥65 years old. Frailty was measured using Fried Frailty Phenotype, handgrip strength was measured using JAMAR@Plus and the 4.57 m usual gait speed was measured using a stopwatch. Comprehensive nutritional assessment refers to the application of Mini Nutritional Assessment (MNA) to assess the nutritional status of patients. Results Compared with the non-frailty group, the upper arm circumference, calf circumference, hemoglobin, albumin, prealbumin, cholesterol and low density lipoprotein in the frailty group were lower (P < 0.05). Comprehensive nutritional assessment, whether as a categorical variable or a continuous variable, was significantly correlated with frailty (P < 0.05). Model1 showed that the risk of frailty in malnourished patients was 3.381 times higher than that in well nourished patients (P = 0.036). Model2 showed that the risk of frailty decreased by 13.8% for every 1 point increase in MNA score (P = 0.009). The area under the curves of albumin, prealbumin and hemoglobin was larger (AUC > 0.65), AUC was 0.718, 0.693 and 0.743, respectively. Conclusions Our results suggest that malnutrition is closely related to frailty. As for single nutritional indexes, albumin, prealbumin and hemoglobin were found to be associated with frailty. Further cohort studies are needed to verify their ability to screen for frailty.

Prognostic Accuracy of a Novel Prognostic Score Using a Combination of PCT, DDi, and Lactate Biomarkers in Patients With Sepsis: A Retrospective Cohort Study

Background A combination of multiple biomarkers will be more accurate in predicting the mortality of sepsis patients. Herein, we aimed to assess the ability to predict adverse outcomes of a novel scoring system using the combination of PCT, DDi, and lactate (PDLS) in patients with sepsis from the emergency department (ED) of a hospital. Methods The patients’ baseline characteristics, main laboratory data and outcome were collected from the patient's electronic medical record. A receiver operating characteristic curve (ROC) analysis determine the optimal cutoff points for biomarkers PCT, DDi and lactate and establish a PDLS system based on their cutoff points. ROC was used to compare the accuracy of PDLS to Sequential Organ Failure Assessment (SOFA) and Acute Physiology and Chronic Health Evaluation (APACHE) II scores in predicting short-term mortality in patients with sepsis. Results The analysis cohort included 1001 patients. 117 sepsis patients died in 28 days. An increase in PDLS was associated with higher mortality and adverse events including MV, VD, AICU, and CRRT. PDLS was an independent predictor of 28-day mortality, MV, VD, AICU, and CRRT. The Area Under the Receiver Operating Characteristic curve (AUROC) of PDLS (0.96; Cl=0.94-0.98) was significantly higher than that of SOFA (0.84; Cl=0.80-0.89) and APACHE II (0.84; Cl=0.79-0.88). Conclusion PDLS is an independent prognostic predictor of adverse clinical outcomes for sepsis patients and was superior to other prognostic scores, including SOFA and APACHE II.

Serum aberrant expression of miR-24-3p and its diagnostic value in Alzheimer’s disease

Aim: This study aimed to evaluate the effect of miR-24-3p in Alzheimer’s disease (AD). Materials & methods: A total of 198 participants were recruited in this study, including 104 AD patients and 94 healthy controls. Expression of miR-24-3p was detected using quantitative real-time PCR. Receiver-operating characteristic curve was used to assess the diagnostic value of miR-24-3p. In vitro AD model was established to evaluate the effect of miR-24-3p. The downstream target was detected by luciferase reporter gene assay. Results: Expression of miR-24-3p showed 1.6-fold increase in AD group compared with healthy controls, and a negative correlation of miR-24-3p with mini-mental state examination score was obtained. Receiver-operating characteristic curve showed satisfactory diagnostic accuracy. Downregulation of miR-24-3p promoted cell proliferation and inhibited cell apoptosis. KLF8 is a target gene of miR-24-3p. Conclusion: MiR-24-3p has a certain value in the diagnosis of AD and may be a potential biomarker.

Urosepsis After Percutaneous Nephrolithotomy (PCNL) - a New Prediction Rule and Scoring System

Objective:Development and validation of a scoring system to predict the risk of urosepsis after percutaneous nephrolithotomy.Methods:The risk factors associated with urosepsis following PCNL(Percutaneous Nephrolithotomy) were identified by meta-analysis. Based on the degree of association, different scores were assigned to these risk factors. Finally Risk assessment scoring system for urosepsis after percutaneous nephrolithotomy (PCNL) was established and validated using ROC (Receiver Operating Characteristic) curve. Results:Based on the degree of association, Women, age (≥60yrs), diabetes mellitus, blood routine (White blood cells≥10×109/L), Urinalysis (White blood cells≥+), Urine culture (Positive), stone size(≥2cm), staghorn stone, hydronephrosis (moderate-severe) were assigned 3, 2, 3, 2, 2, 2, 2, 3, 2 points respectively with a total score of 21 points. The area under the ROC(Receiver Operating Characteristic) curve was 0.913, at the cut-off point of 8.5, the sensitivity and specificity were 90% and 89.4% respectively. Conclusions:The PuRass scoring system could be a useful tool in predicting the risk of urosepsis after PCNL(Percutaneous Nephrolithotomy). Clinician should pay attention to patients with a score above 8.5 during the perioperative period.