WEB DEMO: This webpage is a demo of the final developed web available for the clinician

MACRO: Machine learning models for early prognosis prediction in cardiogenic shock

10 patients were simulated and can be loaded. The patient presents both the admission features and the 24-hour features.

Two machine learning models are developed for predicting 30-day outcomes after cardiogenic shock in acute myocardial infarction on unseen data.

The Admission model provides a fast prediction using only variables available at admission time, like patient age, weight, comorbidities, ECG results, and clinical shock characteristics.

The Full model produces a more accurate prognosis by using laboratory values like lactate, creatine, and glucose, among others.

To use the models, first upload the generated data file with the patient information. After, you will need to choose a model to generate a prediction.

Make sure the type of features selected matches the model, otherwise you will face an error.

The model will return a Risk probability, that represents how likely it is for the patient to expire within 30 days. By clicking on the Patient ID number, a plot showing how the model arrives at that prediction is presented

Model output will be "Risk" if the Risk probability is above 70%, "No Risk" if the Risk probability is below 30%, and "Uncertain" otherwise