Researchers at UC Berkeley have used artificial intelligence to discover a previously overlooked warning signal in routine electrocardiograms that could help identify people at risk of sudden cardiac death more accurately than current methods, according to a study published in Nature.
More than 300,000 people in the United States die from sudden cardiac arrest each year, a condition distinct from a heart attack that occurs when the heart's electrical system suddenly fails. While implanted defibrillators can prevent some deaths, doctors have struggled to determine which patients need them most.
Current screening relies heavily on measuring how much blood the heart pumps with each contraction, a method that can identify a group with an annual sudden cardiac death rate of 4.6%. The AI system, trained on more than 440,000 electrocardiograms from Sweden and tested on patient records from the United States and Taiwan, identified a group with an annual death rate of 7%. It also recognised a broader population at elevated risk, including thousands of people who would not normally raise concerns under existing medical guidelines.
“Medical decisions are really hard, and I think that's why AI is so exciting for me,” said Ziad Obermeyer, associate professor at UC Berkeley's School of Public Health and lead author. “We can not only make better decisions, but also start to understand what's actually going on with these patients before their heart stops.”
Because the AI works with ordinary EKG images already routinely collected in hospitals, the technology could potentially improve screening without requiring a new type of medical test. The researchers aim to introduce the algorithm into hospital EKG databases; under the proposed approach, physicians would contact patients whose EKGs receive a high-risk assessment and offer them the opportunity to wear a patch for extended monitoring.
“Good AI starts with good data,” Obermeyer said. “Unfortunately, data like the ones we used for this study are incredibly hard to access. It's a big part of why there's so little clinical AI in use today.”
Assembling the datasets took approximately 10 years and involved contributions from Dandelion Health and Nightingale Open Science, two organisations Obermeyer helped establish to advance medical research through better data access.
Roughly 90% of people who experience sudden cardiac arrest outside a hospital die within minutes. “There are a lot of lives being lost from people who are dropping dead of sudden cardiac death that are preventable if we just had better AI tools to find these things,” he said.
Analysis
Why This Matters
- Could help prevent thousands of sudden cardiac deaths by identifying at-risk patients who are currently missed by standard screening
- Might reduce unnecessary implantations of expensive defibrillators, which never deliver a shock in about two-thirds of cases
- Opens a new avenue for understanding the electrical disturbances that cause sudden cardiac arrest
Background
Sudden cardiac arrest is a fatal emergency where the heart's electrical system fails, causing it to stop pumping blood. It is different from a heart attack, which involves blocked blood flow. Current risk assessment relies on ejection fraction, a measure of how much blood the heart pumps per beat, but this misses many at-risk individuals and leads to many defibrillator implantations that are never needed.
What to Watch
- Whether the algorithm is adopted in hospital systems in Sweden, Taiwan, and the United States for routine screening
- Plans to offer patients the ability to analyse their own EKG risk via a website Obermeyer has created
- Further research into the biological processes behind the electrical patterns the AI has learned to recognise
Checked
12 claims in this story checked against 1 source: 11 supported.
Show each claim
- SupportedAI detected a hidden heart signal that could predict sudden cardiac death.Latest Science News -- ScienceDaily: “researchers at UC Berkeley have identified a previously overlooked warning signal in routine heart scans that could make it easier to recognize people at risk before a fatal event occurs.”
- Partly supportedThe algorithm was trained on 440,000 EKGs.Latest Science News -- ScienceDaily: “The discovery was made using artificial intelligence trained on more than 440,000 electrocardiograms, or EKGs, collected in Sweden.”
- SupportedThe AI identified a group with an annual death rate of 7%, outperforming standard screening methods.Latest Science News -- ScienceDaily: “Using these traditional measurements, doctors can identify a group of patients with an annual sudden cardiac death rate of 4.6%. The AI system, however, identified a group with an annual death rate of 7%, demonstrating a greater ability to distinguish patients facing serious danger.”
- SupportedMore than 300,000 people in the United States die from sudden cardiac arrest each year.Latest Science News -- ScienceDaily: “More than 300,000 people in the United States die from sudden cardiac arrest each year.”
- SupportedCurrent screening measures how much blood the heart pumps with each contraction, and identifies a group with a 4.6% annual sudden cardiac death rate.Latest Science News -- ScienceDaily: “Current screening methods largely focus on how much blood the heart pumps with each contraction, a measurement used to assess whether someone might benefit from an implanted defibrillator. Using these traditional measurements, doctors can identify a group of patients with an annual sudden cardiac death rate of 4.6%.”
- SupportedThe AI was trained on Swedish EKGs and tested on patient records from the United States and Taiwan.Latest Science News -- ScienceDaily: “The team then evaluated the technology using thousands of additional patient records from the United States and Taiwan”
- SupportedObermeyer said, “Medical decisions are really hard, and I think that's why AI is so exciting for me,” and, “We can not only make better decisions, but also start to understand what's actually going on with these patients before their heart stops.”Latest Science News -- ScienceDaily: “"Medical decisions are really hard, and I think that's why AI is so exciting for me," said Ziad Obermeyer, an associate professor at UC Berkeley's School of Public Health and the study's lead author. "We can not only make better decisions, but also start to understand what's actually going on with these patients before their heart stops."”
- SupportedObermeyer said, “Good AI starts with good data,” and that the data used for the study are incredibly hard to access.Latest Science News -- ScienceDaily: “"Good AI starts with good data," said Obermeyer, who is also part of the joint UCSF-UC Berkeley Computational Precision Health program. "Unfortunately, data like the ones we used for this study are incredibly hard to access. It's a big part of why there's so little clinical AI in use today."”
- SupportedAssembling the datasets took approximately 10 years and involved contributions from Dandelion Health and Nightingale Open Science, which Obermeyer helped establish.Latest Science News -- ScienceDaily: “According to Obermeyer, assembling the datasets took approximately 10 years and involved contributions from two organizations he helped establish, Dandelion Health and Nightingale Open Science.”
- SupportedThe researchers aim to add the algorithm to hospital EKG databases, then have physicians offer high-risk patients a patch for extended monitoring.Latest Science News -- ScienceDaily: “The goal is to examine existing heart scans for signs of elevated risk and identify patients who may benefit from additional monitoring. Under the proposed approach, physicians would contact patients whose EKGs receive a high-risk assessment and offer them the opportunity to wear a patch”
- SupportedRoughly 90% of people who experience sudden cardiac arrest outside a hospital die within minutes.Latest Science News -- ScienceDaily: “roughly 90% of people who experience sudden cardiac arrest outside a hospital die within minutes”
- SupportedObermeyer said many deaths from sudden cardiac death could be prevented with better AI tools to find at-risk people.Latest Science News -- ScienceDaily: “"There are a lot of lives being lost from people who are dropping dead of sudden cardiac death that are preventable if we just had better AI tools to find these things."”
