Development and validation of a nomogram to predict in-hospital cardiac arrest among patients admitted with acute coronary syndrome

Prediction model of in-hospital cardiac arrest using machine learning in the early phase of hospitalization

In hospitals, the deterioration of a patient's condition leading to death is often preceded by physiological abnormalities in the hours to days beforehand. Several risk-scoring systems have been developed to identify patients at risk of major adverse events; however, such systems often exhibit low sensitivity and specificity. To identify the risk factors associated with in-hospital cardiac arrest (IHCA), we conducted a retrospective cohort study at a tertiary medical center in Taiwan. Four machine learning algorithms were employed to identify the factors most predictive of IHCA. The support vector machine model was discovered to be the most effective at predicting IHCA. The ten most critical physiological parameters at 8 h prior to the event were pulse rate, age, white blood cell count, lymphocyte count, body temperature, body mass index, systolic and diastolic blood pressure, platelet count, and use of central nervous system-active medication. Using these parameters, we can enhance early warning and rapid response systems in our hospital, potentially reducing the incidence of IHCA in clinical practice.

Interpretable machine learning-based clinical prediction model for predicting lymph node metastasis in patients with intrahepatic cholangiocarcinoma

OBJECTIVE

Prediction of lymph node metastasis (LNM) for intrahepatic cholangiocarcinoma (ICC) is critical for the treatment regimen and prognosis. We aim to develop and validate machine learning (ML)-based predictive models for LNM in patients with ICC.

METHODS

A total of 345 patients with clinicopathological characteristics confirmed ICC from Jan 2007 to Jan 2019 were enrolled. The predictors of LNM were identified by the least absolute shrinkage and selection operator (LASSO) and logistic analysis. The selected variables were used for developing prediction models for LNM by six ML algorithms, including Logistic regression (LR), Gradient boosting machine (GBM), Extreme gradient boosting (XGB), Random Forest (RF), Decision tree (DT), Multilayer perceptron (MLP). We applied 10-fold cross validation as internal validation and calculated the average of the areas under the receiver operating characteristic (ROC) curve to measure the performance of all models. A feature selection approach was applied to identify importance of predictors in each model. The heat map was used to investigate the correlation of features. Finally, we established a web calculator using the best-performing model.

RESULTS

In multivariate logistic regression analysis, factors including alcoholic liver disease (ALD), smoking, boundary, diameter, and white blood cell (WBC) were identified as independent predictors for LNM in patients with ICC. In internal validation, the average values of AUC of six models ranged from 0.820 to 0.908. The XGB model was identified as the best model, the average AUC was 0.908. Finally, we established a web calculator by XGB model, which was useful for clinicians to calculate the likelihood of LNM.

CONCLUSION

The proposed ML-based predicted models had a good performance to predict LNM of patients with ICC. XGB performed best. A web calculator based on the ML algorithm showed promise in assisting clinicians to predict LNM and developed individualized medical plans.

Development of early prediction model of in-hospital cardiac arrest based on laboratory parameters

BACKGROUND

In-hospital cardiac arrest (IHCA) is an acute disease with a high fatality rate that burdens individuals, society, and the economy. This study aimed to develop a machine learning (ML) model using routine laboratory parameters to predict the risk of IHCA in rescue-treated patients.

METHODS

This retrospective cohort study examined all rescue-treated patients hospitalized at the First Medical Center of the PLA General Hospital in Beijing, China, from January 2016 to December 2020. Five machine learning algorithms, including support vector machine, random forest, extra trees classifier (ETC), decision tree, and logistic regression algorithms, were trained to develop models for predicting IHCA. We included blood counts, biochemical markers, and coagulation markers in the model development. We validated model performance using fivefold cross-validation and used the SHapley Additive exPlanation (SHAP) for model interpretation.

RESULTS

A total of 11,308 participants were included in the study, of which 7779 patients remained. Among these patients, 1796 (23.09%) cases of IHCA occurred. Among five machine learning models for predicting IHCA, the ETC algorithm exhibited better performance, with an AUC of 0.920, compared with the other four machine learning models in the fivefold cross-validation. The SHAP showed that the top ten factors accounting for cardiac arrest in rescue-treated patients are prothrombin activity, platelets, hemoglobin, N-terminal pro-brain natriuretic peptide, neutrophils, prothrombin time, serum albumin, sodium, activated partial thromboplastin time, and potassium.

CONCLUSIONS

We developed a reliable machine learning-derived model that integrates readily available laboratory parameters to predict IHCA in patients treated with rescue therapy.

How to predict the death risk after an in-hospital cardiac arrest (IHCA) in intensive care unit? A retrospective double-centre cohort study from a tertiary hospital in China

OBJECTIVES

Our objective is to develop a prediction tool to predict the death after in-hospital cardiac arrest (IHCA).

DESIGN

We conducted a retrospective double-centre observational study of IHCA patients from January 2015 to December 2021. Data including prearrest diagnosis, clinical features of the IHCA and laboratory results after admission were collected and analysed. Logistic regression analysis was used for multivariate analyses to identify the risk factors for death. A nomogram was formulated and internally evaluated by the boot validation and the area under the curve (AUC). Performance of the nomogram was further accessed by Kaplan-Meier survival curves for patients who survived the initial IHCA.

SETTING

Intensive care unit, Tongji Hospital, China.

PARTICIPANTS

Adult patients (≥18 years) with IHCA after admission. Pregnant women, patients with 'do not resuscitation' order and patients treated with extracorporeal membrane oxygenation were excluded.

INTERVENTIONS

None.

PRIMARY AND SECONDARY OUTCOME MEASURES

The primary outcome was the death after IHCA.

RESULTS

Patients (n=561) were divided into two groups: non-sustained return of spontaneous circulation (ROSC) group (n=241) and sustained ROSC group (n=320). Significant differences were found in sex (p=0.006), cardiopulmonary resuscitation (CPR) duration (p<0.001), total duration of CPR (p=0.014), rearrest (p<0.001) and length of stay (p=0.004) between two groups. Multivariate analysis identified that rearrest, duration of CPR and length of stay were independently associated with death. The nomogram including these three factors was well validated using boot calibration plot and exhibited excellent discriminative ability (AUC 0.88, 95% CI 0.83 to 0.93). The tertiles of patients in sustained ROSC group stratified by anticipated probability of death revealed significantly different survival rate (p<0.001).

CONCLUSIONS

Our proposed nomogram based on these three factors is a simple, robust prediction model to accurately predict the death after IHCA.

Sex Differences in Platelet Reactivity in Patients With ST-Elevation Myocardial Infarction: A Sub-Analysis of the ON-TIME 3 Trial

Background: Fast and adequate platelet inhibition is one of the cornerstones in the treatment of patients with ST-elevation myocardial infarction (STEMI). The aim of this analysis is to examine sex differences in platelet inhibition in the acute treatment of STEMI patients.

Methods: Platelet reactivity units (PRU) and ticagrelor plasma concentrations of all patients in the ON-TIME 3 were compared according to sex. All patients were pre-treated with crushed ticagrelor, aspirin and heparin. Both univariable and multivariable analyses were performed.

Results: In this sub-analysis of the ON-TIME 3 trial, 195 STEMI patients, of which 58 female patients (29.7%) and 137 male patients (70.3%), were analyzed. PRU-values immediately post-PCI were not different in females [median 135 (IQR 47-228)] compared to males [160 (IQR 40-219), P = 0.92]. Ticagrelor plasma concentrations were higher in the females at the start of primary PCI [141 ng/mL (IQR 25-491) vs. 76 ng/mL (IQR 15-245), P = 0.049] and at 6 hours post-primary PCI [495 ng/mL (IQR 283-661) vs. 321 ng/mL (IQR 196-537), P = 0.001] compared to males. However, immediately post-primary PCI and at 1-hour post-primary PCI no significant differences in ticagrelor concentrations were seen between sexes. In multivariable analysis, sex was significantly associated with ticagrelor concentration (P = 0.04), but not with PRU (P = 0.93).

Conclusion: Effective platelet inhibition reached by crushed ticagrelor in STEMI patients was similar in both sexes. Females had similar or even higher ticagrelor plasma concentrations up to 6 hours post-primary PCI compared with males.