An earlier diagnosis will allow doctors to perform corrective surgery and take other actions to help newborns with heart defects.
Almost half of all deaths caused by congenital heart defects occur in children who are less than 1 year old.
However, there may be a solution on the way to help these infants.
A new artificial intelligence (AI)-based technology may be able to diagnose heart problems more quickly and accurately than a medical professional to significantly improve the odds of survival.
According to the Centers for Disease Control and Prevention (CDC), congenital heart defects are the most common type of birth defect.
Dr. Kolawole Oyelese, perinatologist with Atlantic Maternal Fetal Medicine in New Jersey, told Healthline “While only about 1 percent of babies will be born with a congenital heart defect, almost 25 percent of those children will have a heart defect that requires surgery to correct it within the first year.”
The CDC reports that from 1999 to 2006, there were almost 42,000 deaths related to congenital heart defects in the United States. This means the defects were either the main cause of death or contributed to death in some way.
Over the 7-year period tracked by the CDC, congenital heart defects were listed as the main cause of death for 27,960 people.
A 2010 study found that 48 percent of the deaths due to these defects occurred before a child had reached their first birthday.
Oyelese says undetected congenital heart defects are a serious problem.
“Because when a baby has a serious heart defect, very often the outcome depends on an accurate diagnosis in utero or at time of birth,” Oyelese told Healthline.
He adds that babies with severe heart defects who are not diagnosed before birth could die in the first month, sometimes becoming severely ill while still in the maternity ward nursery.
“Sometimes, babies with undiagnosed heart disease will be discharged home, only to come back very sick days later, or even die at home,” said Oyelese.
Diagnosis of heart problems before a baby is born allows for prompt, lifesaving treatment.
Fetal diagnosis currently depends on observations by experienced medical professionals using ultrasound imaging.
Human error makes it unfortunately common for babies to be born without having had their heart problem diagnosed.
However, treating congenital heart defects within a week after birth is known to markedly improve the prognosis.
Therefore, many attempts have been made to develop a technology that makes rapid and accurate diagnosis possible.
Machine learning is a field of computer science that gives computer systems the ability to learn using statistical techniques.
This allows AI to progressively improve its performance on a specific task only using data, without the need of someone to actually modify its programming.
Machine learning can be used to allow a diagnostic system to detect disease faster and much more accurately than a human being.
But, this requires that the computer has lots of information on normal and abnormal subjects for the disease involved.
The problem is that heart defects in children are somewhat infrequent, so there isn’t enough information available to teach the AI.
Because of this, a diagnosis based on machine learning wasn’t accurate enough to be used clinically.
That is, until now.
A research group led by scientists from the RIKEN Center for Advanced Intelligence Project (AIP), collaborating with Fujitsu Ltd. and Showa University, decided to take on the challenge.
They have successfully developed a new machine learning technology that can accurately predict disease using relatively small and incomplete collections of data.
Typically, fetal heart experts determine if parts of the heart, such as valves or blood vessels, are in the correct positions by comparing normal and abnormal fetal heart images and relying on their professional experience.
The RIKEN researchers found a computer process that was similar to how humans worked called “object detection.” This allowed the AI to both distinguish position and classify multiple objects appearing in fetal heart images.
“This breakthrough was possible thanks to the accumulated discussions among the experts on machine learning and fetal heart diagnosis. RIKEN AIP has many AI experts and opportunities for collaboration, like this project. We hope that the system will go into widespread use by means of the successful cooperation among clinicians, academia, and the company,” said Masaaki Komatsu, a RIKEN AIP researcher who led the project in a press release.
The researchers say that their next step is to carry out clinical trials in university hospitals throughout Japan.
These trials will increase the number of fetal ultrasound images in the database, further improving the accuracy of the AI system.
The RIKEN team anticipates that once this AI system is implemented, its accuracy and speed will significantly reduce medical disparities due to human error between the various regions.
However, Oyelese doesn’t think AI will replace human professionals any time soon.
“AI has its limitations,” he noted. “While it may help make the diagnosis more accurate, it is still not a replacement for years of expertise, clinical experience, or training.”
Artificial intelligence is vastly improving the speed and accuracy of medical diagnosis.
Researchers at RIKEN have solved a problem that prevented the use of AI to rapidly diagnose congenital heart defects so treatment can be rendered as soon as possible.
This new AI system will help countless children who might have otherwise suffered health problems or even death due to an undiagnosed heart defect.