An Innovative Scoring System for Predicting Major Adverse Cardiac Events in Patients With Chest Pain Based on Machine Learning

Abstract

Chest pain is a common complaint in the emergency department, but this may prevent a diagnosis of major adverse cardiac events, a composite of all-cause mortality associated with cardiovascular-related illnesses. To determine potential predictors of major adverse cardiac events in Taiwan, a pilot study was performed, involving the data from 268 patients with major adverse cardiac events, which was by an artificial neural network method. Nine biomarkers were selected for identifying non-ST-elevation myocardial infarction from common chest pain patients. By using a machine learningbased feature selection technique, five biomarkers were chosen from a set of 37 candidate variables. A full and a reduced risk stratification model were built. The full model was based on the characteristics of both invasive (i.e., creatinine and troponin I) and non-invasive (i.e., age, coronary artery disease risk factors, and corrected QT interval) variables, and the reduced model was based only on non-invasive variable characteristics. The full model showed a sensitivity of 0.948 and a specificity of 0.546 when the cutoff was set at 2 points, with a missed major adverse cardiac events rate of 1.32%, a positive predictive value of 0.228, and a negative predictive value of 0.987. High performance was also obtained with the five major biomarkers in the predictor built by the machine learning algorithm. The full model had the highest performance, but the reduced model can be applied as a quick and reasonably performing diagnostic tool.

Applications of machine learning to undifferentiated chest pain in the emergency department: A systematic review

“…Prognostication studies varied in the timeframes considered. The longest time frame assessed was 90-day MACE by Wu et al [ 32 ] Wu et al used ML to select features for their risk stratification model, developing a full model that contained invasive (blood tests) variables, and a reduced model that only contained non-invasive variables. They also identify that in their data, QTc prolongation was a potentially novel predictor of MACE.…”

“…Conforti et al provided publicly available link to their dataset, however this link no longer works [ 42 ]. Wu et al stated that their dataset was available on reasonable request [ 32 ]. The code used for the ML models was not publicly available in any studies.…”

Understanding the bias in machine learning systems for cardiovascular disease risk assessment: The first of its kind review

“…3 (c). OBBM took 21% [71,78,86,90,92], LBBM took 8% [82,83], CUSIP took 8% [70,74], OBBM along with LBBM took 59% [72,73,[75][76][77]79,81,84,85,[87][88][89]91], OBBM, LBBM, and CUSIP all combined took 4% [69]. Fig.…”

“…ML-based CVD calculators were analyzed based on the outcome (ground truth) design, which was divided into five clusters (Fig. 2 (c)) such as death 17% [69][70][71][72], coronary artery calcification (CAC), which was 24% [73][74][75][76][77][78][79], myocardial infarction (MI) along with stroke and angina which was 29% [80][81][82][83][84][85][86], diabetes along with hypertension which was 12% [87][88][89] and the event-equivalent gold Fig. 1.…”

“…Fig. 2 (d) shows the contribution consisting of 77% binary [70][71][72]76,[78][79][80][81][82][83][84][85][86][87][88][89][90]92], 9% tertiary [75,91] and 14% quaternary [69,74,77]. Even though granularity does not hamper the performance of the ML system directly, it could introduce a bias during the monitoring of statins on CVD patients.…”