An AI system for forecasting of risks at pregnant women

2020: The announcement of an AI system for forecasting of risks at pregnant women

At the beginning of September, 2020 researchers from Carnegie-Mellon's University provided a technique of machine learning which allows to analyze samples of a placenta and to calculate risk for health of the woman at future pregnancies. A system is designed to help obstetricians-gynecologists who very need the forecast about possible complications of future women in labor. According to authors of the project, their development already began to be applied in clinical practice.

Among the most serious predictive signs researchers note damage of blood vessels of a placenta which carries the name of a detsidualny vaskulopatiya. Its presence assumes that young mother during pregnancy suffered from a preeklampsiya - a status which complicates 2-8% of pregnancies and can be fatal to mother and the child. If to detect a preeklampsiya at an early stage, then the patient can be cured before emergence of symptoms. But as inspection takes a lot of time and requires highly specialized skills, it will be out seldom.

The artificial intelligence predicting problems at pregnant women is provided

Researchers suggested to use AI to make assessment more available due to automatic search of the struck vessels on placenta micromedicines. The command trained the algorithm to reveal affected areas of vessels, having provided AI a huge number of images with placenta samples.

Pathologists for years train to find symptoms of a disease on these images, but in hospitals there are a lot of pregnant women that on inspections of each sample there is no time, - the researcher Daniel Clymer told. - Our AI helps doctors to reveal images with a possible vaskulopatiya

When testing the algorithm classified defeats more precisely, than professional pathologists. However researchers do not expect that a system will replace medical specialists. Instead of this AI should increase capacity of a research, reduce its cost and open to it broad access.^[1]

Notes