PPubMed4 Sep 2020Cited 48×

Improving preoperative risk-of-death prediction in surgery congenital heart defects using artificial intelligence model: A pilot study Improving preoperative risk-of-death prediction in surgery congenital heart defects

João Chang Junior, Fábio Binuesa, Luiz Fernando Caneo, Aida Luiza Ribeiro Turquetto, Elisandra Cristina Trevisan Calvo Arita, Aline Cristina Barbosa, Alfredo Manoel da Silva Fernandes, Evelinda Marramon Trindade, Fábio Biscegli Jatene, Paul-Eric Dossou, Marcelo Biscegli Jatene, Jaishankar Raman

Abstract

Congenital heart defects (CHD) are structural problems that arise in the formation of the heart or major blood vessels, with a significant impact on morbidity, mortality and health costs in children and adults. Defects vary in severity, from tiny holes between chambers that are resolved naturally or malformations that may require multiple surgical procedures, being a major cause of perinatal and infant mortality . Reported birth estimates for patients with congenital heart disease vary widely among studies worldwide. The estimate incidence of 9 per 1,000 live births is generally accepted, thus

A figure from Improving preoperative risk-of-death prediction in surgery congenital heart defects using artificial intelligence model: A pilot study Improving preoperative risk-of-death prediction in surgery congenital heart defects — fig. from the paper

§ The Valyu brief

The paper, in five minutes

Read from the full text via Valyu Contents

AI models predict preoperative mortality in congenital heart surgery with up to 92% sensitivity, using Random Forest as top performer.

What’s new

i.Developed and validated a preoperative mortality prediction model (CgntSCORE) for congenital heart disease surgery.
ii.Compared six widely cited AI algorithms and identified Random Forest as the top performer.
iii.Created a web application allowing clinicians to calculate individual mortality risk on browsers or smartphones.
iv.Identified key predictors such as previous ICU admission, diagnostic group, patient height, and arterial oxygen saturation, accounting for 67.8% of variable importance.
v.Presented the first individual CHD mortality prognosis using artificial intelligence.

How they did it

The study used data from 2,240 consecutive congenital heart disease patients treated at InCor between 2014 and 2018, extracting 84 variables. Six machine‑learning algorithms (MLP, Random Forest, Extra Trees, Stochastic Gradient Boosting, AdaBoost, Bagged Decision Trees) were trained and evaluated using 10‑fold cross‑validation with under‑ and over‑sampling to address class imbalance. Model performance was assessed with accuracy, sensitivity, specificity, precision, and ROC AUC metrics.

What they found

Random Forest achieved an AUC of 0.902 (90.2%) and a sensitivity (recall) of 92.2%.
Random Forest accuracy was 80.8% with precision of 54.6%.
Multilayer Perceptron obtained the highest accuracy (90.2%) and specificity (98.5%) but low sensitivity (20.8%) and AUC (84.6%).
Bagged Decision Trees showed specificity of 90.8% and sensitivity of 70.6%.
The top predictors (previous ICU admission, diagnostic group, patient height, arterial oxygen saturation, hypoplastic left heart syndrome, BMI) contributed 67.8% of importance in the Random Forest model.

Caveats

✕The dataset was highly unbalanced (1:9 mortality to survival ratio).
✕Only InCor’s single‑center data were used, limiting external generalizability.
✕The study is a pilot with a limited sample size of 2,240 patients.
✕Some algorithms showed low sensitivity, indicating difficulty identifying all high‑risk patients.

§ The Valyu brief

Reading the full paper and taking notes. This takes a few seconds…

§ Ask this paper

Ask a question about this paper

Valyu reads the full text and answers from what the paper actually says.

§ Related work in other archives

Improving preoperative risk-of-death prediction in surgery congenital heart defects using artificial intelligence model: A pilot study Improving preoperative risk-of-death prediction in surgery congenital heart defects

The paper, in five minutes

Ask a question about this paper

Similar papers in all four archives

Similar papers in all four archives

The paper, in five minutes