Periodic Reporting for period 1 - OMI AI ECG Model (OMI AI ECG Model - application for more accurate heart attack diagnosis)
Reporting period: 2024-01-01 to 2024-10-31
Globally, 50M patients with chest pain present to emergency departments each year. For these patients, a 12-lead electrocardiogram (ECG) serves as a swift and readily available diagnostic test for identifying acute obstructive heart attacks (OMI), a life-threatening condition that requires prompt transfer to a cardiac catheterization laboratory.
The OMI AI ECG Model is an innovative AI-powered mobile application designed to improve the diagnosis of acute obstructive heart attacks, known as Occlusive Myocardial Infarction (OMI), using standard 12-lead electrocardiograms (ECGs). Traditional STEMI criteria, which are commonly used for ECG assessments, miss about 50% of OMI cases and have high false-positive rates, leading to delayed treatments and unnecessary invasive procedures. The OMI AI ECG Model addresses these shortcomings by being twice as sensitive as traditional methods and diagnosing OMI approximately 3 hours earlier.
This advanced AI model requires no additional hardware and is compatible with any ECG device. It leverages expert-trained artificial intelligence. Its interoperability ensures seamless integration with all healthcare provider systems, facilitating immediate implementation without disrupting existing workflows.
The OMI AI ECG Model's benefits are significant. It improves patient outcomes by reducing missed or delayed OMI diagnoses by 50%, potentially saving millions of lives. By decreasing unnecessary referrals and invasive procedures, it also contributes to substantial cost reductions in healthcare expenses. This optimization of workflow processes allows for more efficient patient management and better utilization of medical resources.
Clinical expertise and collaborative efforts support the development and validation of the OMI AI ECG Model. Conceived by pioneers of the OMI paradigm and supported by leading cardiologists, the model has demonstrated proven effectiveness.
The goal of this project is to complete the development of the AI model and its integration into the PMcardio for Organizations platform to conduct a prospective observational study and a randomized controlled trial (RCT) demonstrating safety and effectiveness. The project will also support regulatory approval and marketing activities to scale up and ensure the technology's implementation within healthcare systems.
The OMI AI ECG Model is an innovative AI-powered mobile application designed to improve the diagnosis of acute obstructive heart attacks, known as Occlusive Myocardial Infarction (OMI), using standard 12-lead electrocardiograms (ECGs). Traditional STEMI criteria, which are commonly used for ECG assessments, miss about 50% of OMI cases and have high false-positive rates, leading to delayed treatments and unnecessary invasive procedures. The OMI AI ECG Model addresses these shortcomings by being twice as sensitive as traditional methods and diagnosing OMI approximately 3 hours earlier.
This advanced AI model requires no additional hardware and is compatible with any ECG device. It leverages expert-trained artificial intelligence. Its interoperability ensures seamless integration with all healthcare provider systems, facilitating immediate implementation without disrupting existing workflows.
The OMI AI ECG Model's benefits are significant. It improves patient outcomes by reducing missed or delayed OMI diagnoses by 50%, potentially saving millions of lives. By decreasing unnecessary referrals and invasive procedures, it also contributes to substantial cost reductions in healthcare expenses. This optimization of workflow processes allows for more efficient patient management and better utilization of medical resources.
Clinical expertise and collaborative efforts support the development and validation of the OMI AI ECG Model. Conceived by pioneers of the OMI paradigm and supported by leading cardiologists, the model has demonstrated proven effectiveness.
The goal of this project is to complete the development of the AI model and its integration into the PMcardio for Organizations platform to conduct a prospective observational study and a randomized controlled trial (RCT) demonstrating safety and effectiveness. The project will also support regulatory approval and marketing activities to scale up and ensure the technology's implementation within healthcare systems.
With the grant funding, the OMI AI ECG Model has been further developed, demonstrating significantly improved sensitivity while maintaining a fixed specificity of 98%. The model also shows strong performance across various subgroups, including signal variations, different diagnoses, and morphological differences. It has been successfully integrated into the PMcardio for Organizations platform. An API interoperability module has been developed, enabling the PMcardio for Organizations platform to integrate seamlessly with other devices
A foundation for robust clinical evidence supporting the model’s safety and efficacy has been established with the submission of a prospective observational study protocol, and the first patients have already been enrolled.
To protect the intellectual property of the OMI AI ECG model, an IP management plan has been developed, and a trademark application for “Queen of Hearts” has been submitted.
To communicate the benefits of the OMI AI ECG Model and its impact on patient outcomes, as well as to share preliminary project results, communication activities have been conducted according to the dissemination and communication plan. A risk management plan was also implemented to mitigate potential risks, and, to date, none have materialized. Additionally, a data management plan has been created and regularly updated to ensure secure data handling throughout the project lifecycle.
An Ethics Advisor was appointed at the start of the project, and their initial report confirms that the ethical framework and measures taken by Powerful Medical are satisfactory.
As outlined in Section 2 of D5.3 First interim implementation report, all deliverables due in the first reporting period (January 1, 2024, to October 31, 2024) have been successfully completed. For further details, please refer to the respective reports.
The investment component will be accessed later on to further help cover costs for mass commercialisation and adoption of the solution - mainly engaging key opinion leaders, validating strategies for reimbursement, participating in conferences and exhibitions, and post-market surveillance. It will also cover the long-term RCTs necessary for guideline implementation.
A foundation for robust clinical evidence supporting the model’s safety and efficacy has been established with the submission of a prospective observational study protocol, and the first patients have already been enrolled.
To protect the intellectual property of the OMI AI ECG model, an IP management plan has been developed, and a trademark application for “Queen of Hearts” has been submitted.
To communicate the benefits of the OMI AI ECG Model and its impact on patient outcomes, as well as to share preliminary project results, communication activities have been conducted according to the dissemination and communication plan. A risk management plan was also implemented to mitigate potential risks, and, to date, none have materialized. Additionally, a data management plan has been created and regularly updated to ensure secure data handling throughout the project lifecycle.
An Ethics Advisor was appointed at the start of the project, and their initial report confirms that the ethical framework and measures taken by Powerful Medical are satisfactory.
As outlined in Section 2 of D5.3 First interim implementation report, all deliverables due in the first reporting period (January 1, 2024, to October 31, 2024) have been successfully completed. For further details, please refer to the respective reports.
The investment component will be accessed later on to further help cover costs for mass commercialisation and adoption of the solution - mainly engaging key opinion leaders, validating strategies for reimbursement, participating in conferences and exhibitions, and post-market surveillance. It will also cover the long-term RCTs necessary for guideline implementation.
The OMI AI ECG Model is an externally and internationally validated AI model that uses image recognition and can interpret OMI from 12-lead ECGs from any device. Upon taking a picture of any ECG, AI technology detects OMI accurately and recommends personalised patient management. It is scalable, compatible with all ECG vendors, and can be deployed immediately into any existing emergency care setting using smartphones. OMI AI ECG Model harnesses the knowledge from thousands of previous patients, overcoming the barrier of learning complex ECG criteria. This solution is faster and more precise than current state-of-the-art criteria, cutting the number of missed or delayed acute obstructive heart attacks by 50%, ultimately saving millions and reducing heart failure incidence.
Using the OMI AI ECG Model is easy. After scanning the 12-lead ECG from any vendor, the user is able to diagnose OMI accurately using just a smartphone. After adding other clinical information about the patient into the algorithm, the user can get the best treatment and referral options from the latest clinical guidelines. The initial version of the model was trained on 18,000 ECGs from 10,000 patients. To substantially improve its accuracy in detecting OMI, the dataset was expanded during the first Reporting Period of the Project to over 152,000 ECGs from nearly 40,000 patients, more than tripling the ECG count and doubling the patient sample compared to the original dataset.
Since the project start, we conducted over 400 model training sessions, resulting in more than 2,000 different AI models from which we selected the best-performing AI model. This model demonstrated improved performance, with significantly increased sensitivity while maintaining a fixed specificity of 98%.
Using the OMI AI ECG Model is easy. After scanning the 12-lead ECG from any vendor, the user is able to diagnose OMI accurately using just a smartphone. After adding other clinical information about the patient into the algorithm, the user can get the best treatment and referral options from the latest clinical guidelines. The initial version of the model was trained on 18,000 ECGs from 10,000 patients. To substantially improve its accuracy in detecting OMI, the dataset was expanded during the first Reporting Period of the Project to over 152,000 ECGs from nearly 40,000 patients, more than tripling the ECG count and doubling the patient sample compared to the original dataset.
Since the project start, we conducted over 400 model training sessions, resulting in more than 2,000 different AI models from which we selected the best-performing AI model. This model demonstrated improved performance, with significantly increased sensitivity while maintaining a fixed specificity of 98%.