Predicting Mortality in Severe Pneumonia Cases is Now Possible Thanks to Machine Learning Approach

By , in Health News Sci/Tech on . Tagged width:

Using a multivariate logistic regression model, a new pneumonia score known as the “Integrated CCI-APS” was established. The absence of ratings for accurately forecasting long-term mortality for fatal instances of pneumonia led researchers to attempt to build new pneumonia scores through the application of machine learning techniques. This score took into account a total of six different factors, including metastatic solid tumor, age, the Charlson Comorbidity Index, congestive heart failure, readmission, and the Acute Physiology Score III.

Patients diagnosed with pneumonia and admitted to the intensive care unit (ICU) will have their new scores used to forecast not just their 1-year mortality but also their death in the hospital.

Continue reading down below.

This study provided evidence that machine learning models have the potential to be useful in the construction of pneumonia scores. Several different machine learning methods, such as logistic regression, decision tree, random forest, multilayer perceptron, and XGBoost, were utilized. When it came to determining both 1-year mortality and death in the hospital in severe pneumonia patients, the integrated CCI-APS score performed much better than other scores that were available.

The research used data obtained via the MIMIC-IV and eICU records, drawing an analysis population from these two different kinds of information. Comparatively, the eICU system contained 13,760 patients, whereas the MIMIC-IV database only had 4,697 patients in it. The death rates at one year and in hospitals were the significant outcomes that were compared between the two datasets. Patients were separated into several subgroups according to whether or not they had been diagnosed with ventilator-associated pneumonia (VAP) or community-acquired pneumonia (CAP).

The area under the curve (AUC) values for predicting one-year death varied from 0.784 to 0.797, while the area under the curve (AUC) values for predicting hospital mortality fluctuated between 0.691 to 0.780. The ranges of accuracy that corresponded to the numbers included results such as 0.723 to 0.725 and 0.641 to 0.718.

In general, the findings of this study represent a potential advance toward more precisely forecasting the likelihood of death outcomes in patients with pneumonia and may lead to enhanced clinical decision-making in intensive care units.

Tiesha loves to share her passion for everything that’s beautiful in this world. Apart from writing on her beauty blog and running her own beauty channel on Youtube, she also enjoys traveling and photography. Tiesha covers various stories on the website.