Machine Learning Can Predict Health Risks of Future Pregnancies

A machine learning approach can analyze placenta slides and inform more women of their health risks in future pregnancies, leading to lower healthcare costs and better outcomes, according to a study published in the American Journal of Pathology.

When a baby is born, doctors sometimes examine the placenta for features that might suggest health risks in any future pregnancies. Providers analyze placentas to look for a type of blood vessel lesion called decidual vasculopathy (DV). These indicate that the mother is at risk for preeclampsia, a complication that can be fatal to both the mother and baby in any future pregnancies.

Once detected, preeclampsia can be treated, so there is considerable benefit from identifying at-risk mothers before symptoms appear. However, although there are hundreds of blood vessels in a single slide, only one diseased vessel is needed to indicate risk. This makes examining the placenta a time-consuming process that must be performed by a specialist, so most placentas go unexamined after birth.

"Pathologists train for years to be able to find disease in these images, but there are so many pregnancies going through the hospital system that they don't have time to inspect every placenta," said Daniel Clymer, PhD, alumnus, Department of Mechanical Engineering, CMU.

Researchers trained a machine learning algorithm to recognize certain features in images of a thin slice of a placenta sample. The team showed the tool various images and indicated whether the placenta was diseased or healthy.

Because it’s difficult for a computer to look at a large picture and classify it, the team employed a novel approach through which the computer follows a series of steps to make the task more manageable. First, the computer detects all blood vessels in an image. Each blood vessel can then be considered individually, creating similar data packets for analysis.

Then, the computer can access each blood vessel and determine if it should be deemed diseased or healthy. At this phase, the algorithm also considers features of the pregnancy, such as gestational age, birth weight, and any conditions the mother might have. If there are any diseased blood vessels, then the picture is marked as diseased.

The tool achieved individual blood vessel classification rates of 94 percent sensitivity and 96 percent specificity, and an area under the curve of 0.99.

"Our algorithm helps pathologists know which images they should focus on by scanning an image, locating blood vessels, and finding patterns of the blood vessels that identify DV," said Clymer.

The team noted that the algorithm is meant to act as a companion tool for physicians, helping them quickly and accurately assess placenta slides for enhanced patient care.

"This algorithm isn't going to replace a pathologist anytime soon," Clymer explained. "The goal here is that this type of algorithm might be able to help speed up the process by flagging regions of the image where the pathologist should take a closer look."

The study demonstrates the importance of partnerships within the healthcare sector, the team said.

"This is a beautiful collaboration between engineering and medicine as each brings expertise to the table that, when combined, creates novel findings that can help so many individuals," said lead investigators Jonathan Cagan, PhD, and Philip LeDuc, PhD, professors of mechanical engineering at CMU, Pittsburgh.

The findings have significant implications for the use of artificial intelligence in healthcare.

"As healthcare increasingly embraces the role of artificial intelligence, it is important that doctors partner early on with computer scientists and engineers so that we can design and develop the right tools for the job to positively impact patient outcomes," noted co-author Liron Pantanowitz, MBBCh, formerly vice chair for pathology informatics at UPMC.

"This partnership between CMU and UPMC is a perfect example of what can be accomplished when this happens."