Diagnosing Heart Attacks with Artificial Intelligence
Time is of the essence when someone shows signs of an acute myocardial infarction (AMI), more commonly known as a heart attack. An electrocardiogram (ECG), the first line of diagnostic response in such cases, should be done within 10 minutes of a person being admitted to the hospital to limit heart muscle damage. However, the non-linearity and complexity of abnormal ECG signals can make it very difficult for emergency medical staff to detect heart attack characteristics quickly.
Taiwan-based EBM Technologies (EBM) noticed that some hospitals were turning to artificial intelligence (AI) to overcome these limitations and aid in interpreting ECG results. In many cases, AI was helping doctors to diagnose diseases that affect the heart. As a provider of medical information technology solutions, EBM has evolved its offerings to enable AI-assisted diagnosis. Its solutions include Picture Archiving and Communication Systems (PACS) and systems for storing electronic medical records. In November 2019, the company introduced its EBM AI Platform, a software platform for seamless data annotation, training, and advanced visualization and deployment of AI-based medical imaging applications.
For 30 years, EBM has provided its PACS via an on-premises model for hospitals and clinics. However, the technology landscape has shifted immensely since the company’s founding, and EBM wanted to remain at the forefront of innovation to help hospitals improve efficiency. In 2019, the company took the next step in its digital transformation by approaching Amazon Web Services (AWS) to discuss how AI models could be built and deployed on the cloud.
ECGs are often performed inside an ambulance before a patient arrives at the hospital. By sending ECG data to the cloud, medical personnel can receive an AI-assisted diagnosis before the patient even arrives. Although doctors still need to examine the ECG data, the AI-assisted diagnosis can help indicate if an incoming patient will require treatment for an AMI.
“By applying our AI model, doctors gain more confidence because they can respond quicker to emergencies when they receive ECG data on their mobile devices before the patient arrives,” explains William Pan, founder and chief executive officer at EBM Technologies. “Often, doctors do not have a second opinion to confer with on analyzing specific ECG data. But now they have an AI consultant to provide better insights so they can take a clear course of action.”
Halving the Time Taken to Build AI Models
EBM’s engagement with AWS began with weekly training sessions around building AI models using
Amazon SageMaker. EBM’s team of two data scientists then worked directly with AWS solutions architects to understand the nuances of constructing deep learning algorithms on the AWS Cloud over the course of four months.
EBM’s data team was pleasantly surprised at how straightforward the user interface is on Amazon SageMaker and how simple it is to deploy AI models across various server frameworks. Before this, the team had been deploying AI models on up to four different types of client devices configured to each client’s on-premises infrastructure. The constant need to log in to do work and then log out and shift to another device to continue work for another client is now a thing of the past.
Furthermore, EBM’s build process is faster too. AI models are compute-intensive, and it’s difficult to predict the amount of resources needed before a robust model is achieved. With AWS, EBM has the flexibility to add processing units as needed to support data-hungry AI models. “Building a prototype on the AWS Cloud takes just half the time it used to on premises, and we don’t even have to worry about our clients’ device requirements. Additionally, we don’t have to install edge locations for PACS anymore,” Pan says. It now takes just three months to build AI prototypes, compared to six months before moving to AWS.
Returning Diagnosis in 30 Seconds
The next phase of training for EBM is focused on integrating AI models with the AWS Cloud for deployment. The data team uses
Amazon Simple Storage Service (Amazon S3) to store ECG images that trigger
AWS Lambda serverless code, with
Amazon API Gateway to call the AI model. The output is then saved in an
Amazon DynamoDB database, and another AWS Lambda function is triggered to send the prediction to a group on the LINE social messaging app. Paramedics, hospital staff, and doctors are all members of this messaging group.
The first implementation of the AWS Cloud–based model has been running for two months in a hospital located in an offshore island in Taiwan, which was identified as a good candidate for this service because of the lack of onsite medical staff.
“We use this solution to help local medical staff stay alert of critical situations and shorten the rescue time for patients likely suffering from an AMI,” Pan says. Based on findings from its first implementation, the time taken to process ECG data and return an AI prediction in the ambulance is 30 seconds. This means that patients don’t need to wait for a diagnosis from an on-premises PACS upon arriving at the hospital. What’s more, the accuracy rate of the AWS Cloud-based AI model is currently at 98 percent.
Lower Backup Costs on the Cloud
While hospitals have traditionally been conservative in their use of technology and reluctant to dive into cloud computing because of data privacy concerns, many are now exploring options for storing patient data due to high backup costs associated with on-premises environments.
EBM is currently working to demonstrate to its clients that security can be improved on the AWS Cloud with low backup costs. For the average-sized hospital in Taiwan with 500 beds, the company estimates on-premises backups to cost about $30,000 per year, but with AWS that would go down to a few thousand dollars at most. “We believe that the AWS Cloud offers a highly secure cloud solution, and we see great growth potential in our current markets,” Pan says.
Eventually, the company hopes to fully migrate its on-premises PACS to the AWS Cloud. “Working with AWS has given us a way to reinvent our products. The AWS team has provided us with a lot of advice on how to move forward in the long term,” Pan says.
About EBM Technologies
EBM Technologies has been providing Picture Archiving Communication Systems (PACS) to hospitals and clinics in Taiwan, Thailand, Japan, and China for 30 years. It uses AI models to support doctors in diagnosing conditions such as heart attacks.
Benefits of AWS
- Gives doctors confidence in diagnosing heart attacks
- Returns AI predictions in 30 seconds
- Achieves a 98% accuracy rate for AI predictions
- Halves time required to build and deploy AI models
- Supports hospitals and clinics with digital transformation
AWS Services Used
Amazon SageMaker
Amazon SageMaker is a fully managed service that provides every developer and data scientist with the ability to build, train, and deploy machine learning (ML) models quickly. SageMaker removes the heavy lifting from each step of the machine learning process to make it easier to develop high quality models.
Learn more »
Amazon Simple Storage Service
Amazon Simple Storage Service (Amazon S3) is an object storage service that offers industry-leading scalability, data availability, security, and performance. This means customers of all sizes and industries can use it to store and protect any amount of data for a range of use cases, such as data lakes, websites, mobile applications, backup and restore, archive, enterprise applications, IoT devices, and big data analytics.
Learn more »
AWS Lambda
AWS Lambda lets you run code without provisioning or managing servers. You pay only for the compute time you consume. With Lambda, you can run code for virtually any type of application or backend service - all with zero administration.
Learn more »
Amazon DynamoDB
Amazon DynamoDB is a key-value and document database that delivers single-digit millisecond performance at any scale. It's a fully managed, multiregion, multimaster, durable database with built-in security, backup and restore, and in-memory caching for internet-scale applications.
Learn more »
Get Started
Companies of all sizes across all industries are transforming their businesses every day using AWS. Contact our experts and start your own AWS Cloud journey today.
