hedayatbehnam / primace

Predicting First-Year Survival after Percutaneous Coronary Interventions: A Machine Learning-Based ShinyApp Web Application in R
https://primace.aikadeh.com
Other
2 stars 0 forks source link
artificial-intelligence cardiology cardiovascular devops github-actions machine-learning mlops r shiny shiny-apps shiny-r shinyapps shinydashboard

PRIMACE Prediction Tool

PRIMACE

Continuous-Deployment Docker-Build-Push
R-CMD-Check S3-Bucket-Connection
Data-Modification-Check Webapp-Availability

GitHub issues GitHub pull requests License: MIT

Overview

PRIMACE is a shinyapp web application based on R language shiny package which can be used to predict major cardiovascular events (MACE) in first year following Primary Percutaneous Coronary Intervention (PPCI)
PPCI: Primary Percutaneous Coronary Intervention
MACE: Major Adverse Cardiovascular Events

Background

We published a related article with title of "Effects of opium use on one-year major adverse cardiovascular events (MACE) in the patients with ST-segment elevation MI undergoing primary PCI: a propensity score matched - machine learning based study" in BMC Complementary Medicine and Therapies journal.
ML models were as follows:

  1. Survival Random Forest
  2. Survival Extreme Gradient Boosting (xgboost)

Link to Online PRIMACE

PRIMACE
PRIMACE Prediction Tool

Link to Original Article

PRIMACE
Original Article in BMC Journal

Instructions

Preparing your dataset

Because preprocessing and training steps were conducted on our dataset, to use the saved models, you should rename your dataset variable to those similar to our study dataset.
The acceptable variables names are provided in Prediction Tool tab under Variables Name box.

Uploading dataset

After renaming variables, you can upload your dataset to the api in Upload File box. Acceptable file formats are .rds, .csv, .sav and .xlsx. If you do not upload a dataset, by default, a new test test with know target variable has been provided to the app to be used for prediction.

Selecting models and prediction

In Upload File section, you can select among two models (at the moment just random survival forest is available) we trained in our study. After selecting a model, push the Predict... button to initiates prediction on the dataset. It would take a little time to complete prediction process.

Dataset with known target variable

If your dataset have a columns named First_MACE_bin as target event variable and Time_to_MACE as time of event or censor, the app would peforms prediction and then assesses its prediction performance by different performance measures.

Dataset with unknown target variable

If your dataset does not have columsn named First_MACE_bin as target event variable and Time_to_MACE as time of event or censor, the app would just performs prediction, then a table of prediction of observations is provided. It would be applicable for prediction of patients at the time of MDCT.

Using Prediction of Your Patient

In Manual Prediction tab of Prediction Tool submenu, you can input your patient's features, then click Manual Prediction button to see the result of app prediction.