AI may have an ‘eye’ on growing babies: Could predict premature birth as early as 31 weeks
About 10% of all newborns born in the United States in 2021 were preterm — meaning they were born before 37 weeks of gestation, according to the Centers for Disease Control and Prevention (CDC).
Premature birth also accounts for about 16% of infant deaths.
Now, researchers at Washington University in St. Louis, Missouri, are trying to improve these odds. Use of artificial intelligence.
They developed a deep learning model that could predict premature birth by analyzing the electrical activity in a woman’s uterus during pregnancy — then they tested the model in a study that appeared in the medical journal PLOS One. was published.
AI tool helps doctors make sense of turbulent patient data and identify diseases: ‘more meaningful’ conversations
“The main thing is that it is possible to take data before the 31st week and predict preterm birth up to the 37th week” – what surprised the researchers, RA Nehorai, PhD, professor of electrical engineering Washington University in St. Louistold Fox News Digital.
“The AI/Deep Learning automatically learned the most informative features from the data that are relevant to the prediction of preterm birth,” he added.
Additionally, research suggests that preterm birth is an abnormal physiological condition, not just a A pregnancy that ended earlyNehorai said.
During the study, researchers performed electrohysterograms (EHGs), which use electrodes on the abdomen to record electrical activity in the uterus.
They took recordings of these electrical currents from 159 pregnant women who were at least 26 weeks along and “trained” an AI model on that data.
Social media influence in medically induced coma after ruptured cerebral aneurysm emergency C-section
They combined this data with medical information such as the woman’s age and weight, fetal weight and any bleeding in the first or second trimester.
About 19% of women in the study gave birth prematurely. In theory, data from those women could be used as a benchmark to predict preterm birth.
“The advantage of our approach is that it’s cheap to make,” said Nehorai of the new research. “Our model was effective in prediction with short EHG recordings, which may make the model easier to use, more cost-effective in the clinical setting and possibly usable in the home setting.”
Looking ahead, the researchers believe the method should be adopted by hospitals and obstetricians as part of women’s routine pregnancy screening. This will then allow pregnant women to seek care and make lifestyle changes as needed to protect their baby’s health.
“Our work contributes to the goal of using EHG measuring devices to accurately predict preterm birth.”
“A device dedicated to implementing our method should be built for this purpose,” Nehorai said.
It’s hard to say how long it might take before this type of test becomes widely available, the researchers said.
“There are already some EHG measurement devices on the market – however, predicting preterm birth from EHG data remains challenging,” said Uri Goldztegen, a PhD candidate in the Department of Biomedical Engineering working under Professor Nehorai’s supervision at Washington. is.” University.
New AI ‘cancer chatbot’ provides 24/7 support to patients and families: ‘compassionate approach’
“Our work contributes to the goal of using EHG measuring devices to accurately predict preterm birth,” he told Fox News Digital.
EHG measurements typically take between 30 and 60 minutes, with additional time needed to set the device on the mother’s abdomen, Goldztegen noted.
“We have shown that predictions can be made based on short EHG measurements of less than five minutes, without greatly reducing prediction accuracy,” he told Fox News Digital. “This finding is important, because the long duration of EHG measurements is a significant limitation for its adoption in clinical settings.”
The ‘promise’ of deep learning – but caveats
Dr. Susie Lipinski, board-certified OB/GYN at Pediatrics Medical Group In Denver, ColoradoWas not involved in the study but shared his input on whether deep learning technology could help solve the problem of preterm births in the US.
“Being able to predict who’s at risk before they go into labor would be very beneficial,” Lipinski told Fox News Digital. “Using a deep learning model appears promising; however, this study has a relatively small number of patients, so it cannot be determined how applicable it is to a larger population.”
“Previous studies using AI has not shown much reliability, so it will need more studies and larger patient populations before we can start using this method,” he added.
Another potential limitation is that EHG measurements are used in very few locations, Dr.
“The standard in most hospitals and offices is to use a tocodynamometer, which measures pressure, not electricity,” he explained.
Ozampic, Vegovi and pregnancy risks: what you need to know about this issue
If EHG becomes the way to assess for preterm birth, hospitals, birth centers and offices will have to buy new equipment, which could delay adoption in low-resource areas like rural and inner cities, Lipinski said.
“The higher rate of preterm birth in this study than the national average also raises questions about applicability,” he told Fox News Digital. “No demographics were given about the patient population, so there’s no way to see how it reflects the population of the entire country.”
“Being able to predict who is at risk before labor presents itself.”
There is also the possibility of false positives, Lipinski explained.
“Although this method predicts better than our current methods, there are still many patients who will be identified as at risk who may not have a preterm birth,” he said. “This false positive result will cause a huge burden of stress on the patient, as well as increase its use Health care resources.”
If and when this becomes the new standard of care, Lipinski said, treatment for preterm labor will need to improve.
“Our issues with preterm birth are two-fold: we have poor prognosis, but also poor prevention options after 26 weeks,” she added.
The researchers share the main limitations of the study
According to Goldztegen, the study has two main limitations.
“First, we developed our work using about 160 samples from two public datasets,” he said. “Although this amount of data was sufficient for our initial investigation, developing and validating a medical product would require a much larger dataset.”
A second limitation comes from the nature of deep learning, which can produce accurate results but is usually hard to interpret, Goldztegen said.
“In other words, it’s challenging to understand how the algorithm makes predictions,” he explained.
Click here to sign up for our health newsletter
In a discussion of the findings in the Medical Journal, the authors noted that “although machine learning algorithms can contribute to improving healthcare and much research is progressing in this area, significant challenges remain.”
“Developing and validating a medical product would require a very large dataset.”
Among those challenges: Identifying the reasons behind the algorithm’s predictions can be difficult, the researchers wrote.
Click here to get the Fox News app
The researchers also stated, “In our case, although our predictions may influence pregnancy management, our predictions will need to be supplemented with additional clinical trials to determine which therapies reduce the risk of preterm birth.” is more likely to reduce and improve outcomes,” the researchers added.