Novel ECG features and machine learning to optimize culprit lesion detection in patients with suspected acute coronary syndrome

Bouzid, Zeineb; Faramand, Ziad; Gregg, Richard E.; Helman, Stephanie; Martin‐Gill, Christian; Saba, Samir; Callaway, Clifton W.; Sejdić, Ervin; Al‐Zaiti, Salah S.

doi:10.1016/j.jelectrocard.2021.07.012

Cited by 13 publications

(7 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…After excluding 4,072 papers on title and abstract screening, we identified 278 studies for full-text screening, of which 106 studies were included for data extraction and subsequent analysis. 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 , 36 , 37 , 38 , 39 , 40 , 41 , 42 , 43 , 44 , 45 , 46 , 47 , 48 , 49 , 50 , 51 , 52 , 53 , 54 , 55 , 56 , 57 , 58 , 59 , 60 , 61 , 62 , 63 , 64 , 65 , 66 , 67 , 68 , 69 , 70 , 71 , 72 , 73 ,…”

Section: Resultsmentioning

confidence: 99%

Artificial intelligence and machine learning in prehospital emergency care: A scoping review

Chee,

Huang

et al. 2023

iScience

View full text Add to dashboard Cite

Section: Resultsmentioning

confidence: 99%

Artificial intelligence and machine learning in prehospital emergency care: A scoping review

Chee,

Huang

et al. 2023

iScience

View full text Add to dashboard Cite

“…AI has several advantages over traditional methods in PEC settings. It can effectively analyse and interpret high-dimensional data, such as EHR data, images, and ECG signals (18,24,45). AI can also integrate multimodal data (126) and model nonlinear relationships.…”

Section: Discussionmentioning

confidence: 99%

Artificial Intelligence and Machine Learning in Prehospital Emergency Care: A Systematic Scoping Review

Chee

Huang

et al. 2023

Preprint

View full text Add to dashboard Cite

Introduction: The literature on the use of AI in prehospital emergency care (PEC) settings is scattered and diverse, making it difficult to understand the current state of the field. In this scoping review, we aim to provide a descriptive analysis of the current literature and to visualise and identify knowledge and methodological gaps using an evidence map. Methods: We conducted a scoping review from inception until 14 December 2021 on MEDLINE, Embase, Scopus, IEEE Xplore, ACM Digital Library, and Cochrane Central Register of Controlled Trials (CENTRAL). We included peer-reviewed, original studies that applied AI to prehospital data, including applications for cardiopulmonary resuscitation (CPR), automated external defibrillation (AED), out-of-hospital cardiac arrest, and emergency medical service (EMS) infrastructure like stations and ambulances. Results: The search yielded 4350 articles, of which 106 met the inclusion criteria. Most studies were retrospective (n=88, 83.0%), with only one (0.9%) randomised controlled trial. Studies were mostly internally validated (n=96, 90.6%), and only ten studies (9.4%) reported on calibration metrics. While the most studied AI applications were Triage/Prognostication (n=52, 49.1%) and CPR/AED optimisation (n=26, 24.5%), a few studies reported unique use cases of AI such as patient-trial matching for research and Internet-of-Things (IoT) wearables for continuous monitoring. Out of 49 studies that identified a comparator, 39 reported AI performance superior to either clinicians or non-AI status quo algorithms. The minority of studies utilised multimodal inputs (n=37, 34.9%), with few models using text (n=8), audio (n=5), images (n=1), or videos (n=0) as inputs. Conclusion: AI in PEC is a growing field and several promising use cases have been reported, including prognostication, demand prediction, resource optimisation, and IoT continuous monitoring systems. Prospective, externally validated studies are needed before applications can progress beyond the proof-of-concept stage to real-world clinical settings.

show abstract

“…Such measurements rely on knowledge-driven markers (like QRS, ST-segment elevation, T-wave changes) re ecting the clinical knowledge of heart activity, and can be then used to de ne criteria and rules for a diagnostic evaluation by physicians. In addition to human evaluation, in the last years ECG features, which can vary in number and type depending on the program employed, have been used to feed ML methods for tabular data to derive a diagnostic model [4], [5].…”

Section: Introductionmentioning

confidence: 99%

Machine learning approaches for ECG-based models: discrimination and calibration for atrial fibrillation prediction

Baj

Gandin

Scagnetto

et al. 2023

Preprint

View full text Add to dashboard Cite

Background Machine learning (ML) methods to build prediction models starting from electrocardiogram (ECG) signals are an emerging research field. The aim of the present study is to investigate the performances of two ML approaches based on ECGs for the prediction of new-onset atrial fibrillation (AF), in terms of discrimination, calibration and sample size dependence. Methods We trained two models to predict new-onset AF: a convolutional neural network (CNN), that takes as input the raw ECG signals, and an eXtreme Gradient Boosting model (XGB), that uses the signal’s extracted features. A penalized logistic regression model (LR) was used as a benchmark. Discrimination was evaluated with the area under the ROC curve, while calibration with the integrated calibration index. We investigated the dependence of models’ performances on the sample size and on class imbalance corrections introduced with random under-sampling. Results CNN's discrimination was the most affected by the sample size, outperforming XGB and LR only above a certain threshold. Calibration showed only a small dependence on the sample size for all the models considered. Balancing the training set with random undersampling did not improve discrimination in any of the models. Instead, the main effect of imbalance corrections was to worsen the models’ calibration. The sample size emerged as a fundamental point for developing the CNN model, especially in terms of discrimination. The effect of the sample size on the other two models was weaker. Imbalance corrections led to poorly calibrated models, for all the approaches considered, reducing the clinical utility of the models. Conclusions Our results suggest that the choice of approach in the analysis of ECG should be based on the amount of data available, preferring more standard models for small datasets. Moreover, imbalance correction methods should be avoided when developing clinical prediction models, where calibration is crucial.

show abstract

Novel ECG features and machine learning to optimize culprit lesion detection in patients with suspected acute coronary syndrome

Cited by 13 publications

References 17 publications

Artificial intelligence and machine learning in prehospital emergency care: A scoping review

Artificial intelligence and machine learning in prehospital emergency care: A scoping review

Artificial Intelligence and Machine Learning in Prehospital Emergency Care: A Systematic Scoping Review

Machine learning approaches for ECG-based models: discrimination and calibration for atrial fibrillation prediction

Contact Info

Product

Resources

About