Deep learning to predict long-term mortality in patients requiring 7 days of mechanical ventilation

A machine learning-based prediction of hospital mortality in mechanically ventilated ICU patients

BACKGROUND

Mechanical ventilation (MV) is vital for critically ill ICU patients but carries significant mortality risks. This study aims to develop a predictive model to estimate hospital mortality among MV patients, utilizing comprehensive health data to assist ICU physicians with early-stage alerts.

METHODS

We developed a Machine Learning (ML) framework to predict hospital mortality in ICU patients receiving MV. Using the MIMIC-III database, we identified 25,202 eligible patients through ICD-9 codes. We employed backward elimination and the Lasso method, selecting 32 features based on clinical insights and literature. Data preprocessing included eliminating columns with over 90% missing data and using mean imputation for the remaining missing values. To address class imbalance, we used the Synthetic Minority Over-sampling Technique (SMOTE). We evaluated several ML models, including CatBoost, XGBoost, Decision Tree, Random Forest, Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Logistic Regression, using a 70/30 train-test split. The CatBoost model was chosen for its superior performance in terms of accuracy, precision, recall, F1-score, AUROC metrics, and calibration plots.

RESULTS

The study involved a cohort of 25,202 patients on MV. The CatBoost model attained an AUROC of 0.862, an increase from an initial AUROC of 0.821, which was the best reported in the literature. It also demonstrated an accuracy of 0.789, an F1-score of 0.747, and better calibration, outperforming other models. These improvements are due to systematic feature selection and the robust gradient boosting architecture of CatBoost.

CONCLUSION

The preprocessing methodology significantly reduced the number of relevant features, simplifying computational processes, and identified critical features previously overlooked. Integrating these features and tuning the parameters, our model demonstrated strong generalization to unseen data. This highlights the potential of ML as a crucial tool in ICUs, enhancing resource allocation and providing more personalized interventions for MV patients.

Machine learning based prediction of prolonged duration of mechanical ventilation incorporating medication data

Rationale

Duration of mechanical ventilation is associated with adverse outcomes in critically ill patients and increased use of resources. The increasing complexity of medication regimens has been associated with increased mortality, length of stay, and fluid overload but has never been studied specifically in the setting of mechanical ventilation.

Objective

The purpose of this analysis was to develop prediction models for mechanical ventilation duration to test the hypothesis that incorporating medication data may improve model performance.

Methods

This was a retrospective cohort study of adults admitted to the ICU and undergoing mechanical ventilation for longer than 24 hours from October 2015 to October 2020. Patients were excluded if it was not their index ICU admission or if the patient was placed on comfort care in the first 24 hours of admission. Relevant patient characteristics including age, sex, body mass index, admission diagnosis, morbidities, vital signs measurements, severity of illness, medication regimen complexity as measured by the MRC-ICU, and medical treatments before intubation were collected. The primary outcome was area under the receiver operating characteristic (AUROC) of prediction models for prolonged mechanical ventilation (defined as greater than 5 days). Both logistic regression and supervised learning techniques including XGBoost, Random Forest, and Support Vector Machine were used to develop prediction models.

Results

The 318 patients [age 59.9 (SD 16.9), female 39.3%, medical 28.6%] had mean 24-hour MRC-ICU score of 21.3 (10.5), mean APACHE II score of 21.0 (5.4), mean SOFA score of 9.9 (3.3), and ICU mortality rate of 22.6% (n=72). The strongest performing logistic model was the base model with MRC-ICU added, with AUROC of 0.72, positive predictive value (PPV) of 0.83, and negative prediction value (NPV) of 0.92. The strongest overall model was Random Forest with an AUROC of 0.78, a PPV of 0.53, and NPV of 0.90. Feature importance analysis using support vector machine and Random Forest revealed severity of illness scores and medication related data were the most important predictors.

Conclusions

Medication regimen complexity is significantly associated with prolonged duration of mechanical ventilation in critically ill patients, and prediction models incorporating medication information showed modest improvement in this prediction.

Characteristics, ICU Interventions, and Clinical Outcomes of Patients With Palliative Care Triggers in a Mixed Community-Based Intensive Care Unit

Objective: Integration of palliative care initiatives in the intensive care unit (ICU) benefit patients and improve outcomes. Palliative care triggers (PCTs) is a screening tool that aides in stratifying patients who would benefit most from an early palliative care approach. There is no consensus on PCT selection or best timing for implementation. We evaluated the clinical characteristics, ICU and palliative care interventions, and clinical outcomes of critically ill patients with PCT in a community-based mixed ICU. Methods: This retrospective study was conducted in a 44-bed adult, mixed ICU in a 407-bed community-based teaching hospital in Florida. Eleven PCTs were used as a screening tool during multidisciplinary rounds (MDRs). Patients were analyzed based on presence or absence of PCT as well as having met high (>2) versus low (<2) PCT. Data collected included patient demographics, ICU resource utilization and clinical outcomes. We considered a two-sided P value of less than .05 to indicate statistical significance with a 95% confidence interval. Results: Of 388 ICU patients, 189 (48.7%) met at least 1 PCT and 199 (51.3%) did not. The trigger group had higher Acute Physiology and Chronic Evaluation (APACHE) and Sequential Organ Failure Assessment (SOFA) scores within 24 h of ICU admission. The most common PCTs identified were ICU length of stay greater than 7 days or readmission to ICU, terminal prognosis and assisting family in transitioning goals of care. There were statistically significant differences in ICU resource utilization, palliative care interventions, and overall worse clinical outcomes in the trigger-detected group. Similar findings were seen in the cohort with high PCT (>2). Conclusions: Our study supports the implementation of a tailored 11-item palliative care screening tool to effectively identify ICU patients with high ICU and palliative care interventions and worse clinical outcomes.

The use of natural language processing in palliative care research: A scoping review

BACKGROUND

Natural language processing has been increasingly used in palliative care research over the last 5 years for its versatility and accuracy.

AIM

To evaluate and characterize natural language processing use in palliative care research, including the most commonly used natural language processing software and computational methods, data sources, trends in natural language processing use over time, and palliative care topics addressed.

DESIGN

A scoping review using the framework by Arksey and O'Malley and the updated recommendations proposed by Levac et al. was conducted.

SOURCES

PubMed, Web of Science, Embase, Scopus, and IEEE Xplore databases were searched for palliative care studies that utilized natural language processing tools. Data on study characteristics and natural language processing instruments used were collected and relevant palliative care topics were identified.

RESULTS

197 relevant references were identified. Of these, 82 were included after full-text review. Studies were published in 48 different journals from 2007 to 2022. The average sample size was 21,541 (median 435). Thirty-two different natural language processing software and 33 machine-learning methods were identified. Nine main sources for data processing and 15 main palliative care topics across the included studies were identified. The most frequent topic was mortality and prognosis prediction. We also identified a trend where natural language processing was frequently used in analyzing clinical serious illness conversations extracted from audio recordings.

CONCLUSIONS

We found 82 papers on palliative care using natural language processing methods for a wide-range of topics and sources of data that could expand the use of this methodology. We encourage researchers to consider incorporating this cutting-edge research methodology in future studies to improve published palliative care data.

Establishment of ICU Mortality Risk Prediction Models with Machine Learning Algorithm Using MIMIC-IV Database

Objective: The mortality rate of critically ill patients in ICUs is relatively high. In order to evaluate patients’ mortality risk, different scoring systems are used to help clinicians assess prognosis in ICUs, such as the Acute Physiology and Chronic Health Evaluation III (APACHE III) and the Logistic Organ Dysfunction Score (LODS). In this research, we aimed to establish and compare multiple machine learning models with physiology subscores of APACHE III—namely, the Acute Physiology Score III (APS III)—and LODS scoring systems in order to obtain better performance for ICU mortality prediction. Methods: A total number of 67,748 patients from the Medical Information Database for Intensive Care (MIMIC-IV) were enrolled, including 7055 deceased patients, and the same number of surviving patients were selected by the random downsampling technique, for a total of 14,110 patients included in the study. The enrolled patients were randomly divided into a training dataset (n = 9877) and a validation dataset (n = 4233). Fivefold cross-validation and grid search procedures were used to find and evaluate the best hyperparameters in different machine learning models. Taking the subscores of LODS and the physiology subscores that are part of the APACHE III scoring systems as input variables, four machine learning methods of XGBoost, logistic regression, support vector machine, and decision tree were used to establish ICU mortality prediction models, with AUCs as metrics. AUCs, specificity, sensitivity, positive predictive value, negative predictive value, and calibration curves were used to find the best model. Results: For the prediction of mortality risk in ICU patients, the AUC of the XGBoost model was 0.918 (95%CI, 0.915–0.922), and the AUCs of logistic regression, SVM, and decision tree were 0.872 (95%CI, 0.867–0.877), 0.872 (95%CI, 0.867–0.877), and 0.852 (95%CI, 0.847–0.857), respectively. The calibration curves of logistic regression and support vector machine performed better than the other two models in the ranges 0–40% and 70%–100%, respectively, while XGBoost performed better in the range of 40–70%. Conclusions: The mortality risk of ICU patients can be better predicted by the characteristics of the Acute Physiology Score III and the Logistic Organ Dysfunction Score with XGBoost in terms of ROC curve, sensitivity, and specificity. The XGBoost model could assist clinicians in judging in-hospital outcome of critically ill patients, especially in patients with a more uncertain survival outcome.