1
|
Mehta S, Fernandez F, Villagran C, Matheus C, Ceschim M, Vieira D, Torres MA, Mazzini J, Quintero S, Pisana L, Nola F, Safie R, Munguia A, Krisciunas S, Sunkaraneni S. P1464Adoption of feedback to validate a machine learning model for single lead STEMI detection. Eur Heart J 2019. [DOI: 10.1093/eurheartj/ehz748.0229] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Indexed: 11/14/2022] Open
Abstract
Abstract
Background
We have explored the performance of a single lead EKG with Artificial Intelligence (AI) based algorithms in STEMI diagnosis, thus far lead V2 has yielded the best results. Anticipating the performance of the LUMENGT-AI model, we designed a feedback strategy with healthcare centers to expand the validation of our work.
Purpose
To create a pragmatic alternative to the existing gold standard, a 12-lead EKG, for STEMI diagnosis.
Methods
An observational, retrospective, case-control study. Sample: 2,543 exclusively STEMI (anterior, inferior and lateral wall) diagnosis, EKG records. Feedback: From healthcare centers, confirming STEMI diagnosis and location, was obtained (thrombolysis, primary Percutaneous Coronary Intervention (PCI), pharmaco invasive therapy or coronary artery bypass surgery). Records excluded other patient and medical information. Sample was derived from the private International Telemedical Systems (ITMS) database. LUMENGT-AI Algorithm was employed. Preprocessing: detection of QRS complexes using the wavelet system, segmentation of each EKG into individual heartbeats (53,667 total beats) with fixed window of 0.4s to the left and 0.9s to the right of main QRS; Classification: A 1-D convolutional neural network was implemented; “STEMI” and “Not-STEMI” classes were considered for each heartbeat per lead; individual probabilities were aggregated to generate the final label for each record. Training & Testing: 90% and 10% of the sample were used, respectively. Experiments: Intel PC i7 8750H processor at 2.21GHz, 16GB RAM, Windows 10 OS with NVidia GTX 1070 GPU, 8GB RAM.
Results
V2 was the most precise lead with an Accuracy of 93.6%, a Sensitivity of 89%, and a Specificity of 94.7%.
Conclusions
The strategic adoption of feedback from healthcare centers provided strong validation of our model. The results of AI-augmented, single lead EKG are encouraging. We anticipate that this approach will become a promising methodology in STEMI detection.
Collapse
Affiliation(s)
- S Mehta
- Lumen Foundation, Miami, United States of America
| | - F Fernandez
- Lumen Foundation, Miami, United States of America
| | - C Villagran
- Lumen Foundation, Miami, United States of America
| | - C Matheus
- Lumen Foundation, Miami, United States of America
| | - M Ceschim
- Lumen Foundation, Miami, United States of America
| | - D Vieira
- Lumen Foundation, Miami, United States of America
| | - M A Torres
- Lumen Foundation, Miami, United States of America
| | - J Mazzini
- Lumen Foundation, Miami, United States of America
| | - S Quintero
- Lumen Foundation, Miami, United States of America
| | - L Pisana
- Lumen Foundation, Miami, United States of America
| | - F Nola
- Lumen Foundation, Miami, United States of America
| | - R Safie
- Lumen Foundation, Miami, United States of America
| | - A Munguia
- Lumen Foundation, Miami, United States of America
| | - S Krisciunas
- Lumen Foundation, Miami, United States of America
| | | |
Collapse
|
2
|
Vieira D, Mehta S, Fernandez F, Villagran C, Frauenfelder A, Ceschim M, Matheus C, Torres MA, Mazzini J, Quintero S, Pisana L, Safie R, Nola F, Krisciunas S, Cecilio S. 3035Synergy of artificial intelligence and single lead EKG to detect and localize STEMI. Eur Heart J 2019. [DOI: 10.1093/eurheartj/ehz745.0003] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Indexed: 11/14/2022] Open
Abstract
Abstract
Background
The cumbersome, standard 12-lead electrocardiogram (EKG) challenges an efficient detection of ST-Elevation Myocardial Infarction (STEMI) in pre-hospital (ambulances) and hospital (portable devices) settings. We believe that our machine-learning algorithm embedded into a single lead EKG will be successful in acute care settings.
Purpose
To incorporate Artificial Intelligence-guided, single lead EKG interpretation, to facilitate easy and accurate STEMI detection in urgent situations.
Methods
This is an observational, retrospective, case-control study. A subset sample was generated from the International Telemedical Systems (ITMS) database that contains cardiologist annotated EKG records. Subset: A total of 2,542 exclusively confirmed STEMI diagnosis EKG records from enrolled healthcare centers in Mexico, Colombia, and Brazil; including specific ischemic heart wall (anterior, inferior, and lateral). Following discharge of treated patients, confirmation of STEMI diagnosis was obtained as feedback from healthcare centers. Records were anonymized EKG that excluded all medical information. Sample: A Standard 12 lead, 10-seconds length, 500Hz sampling frequency EKG was fed to the LUMENGT-AI STEMI detecting algorithm. Preprocessing: Detection of QRS complexes by wavelet system, segmentation of each EKG record into individual heartbeats (total dataset 27,152 beats) with fixed window of 0.4s to the left and 0.9s to the right of main QRS; Classification: A 1-D convolutional neural network was implemented, three classes were considered for individual heartbeats: “Anterior”, “Inferior” and “Lateral”, each corresponding to the heart wall affected. These individual probabilities were aggregated to generate the final label for each of the 12 leads. Training & Testing: 90% and 10% of the dataset was used respectively. Experiments: Intel PC i7 8750H processor at 2.21GHz, 16GB RAM, Windows 10 OS with a NVidia GTX 1070 GPU, 8GB RAM.
Results
Accuracy – Lead V2 (91.7%); Sensitivity Anterior wall – Lead V2 (97.4%); Sensitivity for Lateral wall – Lead I (10.0%); and Sensitivity for Inferior wall – Lead V2 (93.6%).
Conclusions
AI algorithms merged with a Single lead approach detect and localize STEMI within any setting. The V2 lead yields superior results for mapping of ischemic areas of the heart among the anterior and inferior walls. In contrast, diagnosis remains suboptimal for identifying the lateral wall. The usage of synergistic technologies facilitates easy, fast and early STEMI triage and management.
Collapse
Affiliation(s)
- D Vieira
- Lumen Foundation, Miami, United States of America
| | - S Mehta
- Lumen Foundation, Miami, United States of America
| | - F Fernandez
- Lumen Foundation, Miami, United States of America
| | - C Villagran
- Lumen Foundation, Miami, United States of America
| | | | - M Ceschim
- Lumen Foundation, Miami, United States of America
| | - C Matheus
- Lumen Foundation, Miami, United States of America
| | - M A Torres
- Lumen Foundation, Miami, United States of America
| | - J Mazzini
- Lumen Foundation, Miami, United States of America
| | - S Quintero
- Lumen Foundation, Miami, United States of America
| | - L Pisana
- Lumen Foundation, Miami, United States of America
| | - R Safie
- Lumen Foundation, Miami, United States of America
| | - F Nola
- Lumen Foundation, Miami, United States of America
| | - S Krisciunas
- Lumen Foundation, Miami, United States of America
| | - S Cecilio
- Lumen Foundation, Miami, United States of America
| |
Collapse
|