Neural network recognizes pregnancy complications

Young scientists are training a neural network to predict pre-eclampsia in time. This is one of the most pressing problems of pregnancy complications. As experts say, it has no reliable early prognostic and diagnostic signs, effective measures of prevention and treatment. But it can cause premature labor.

This system will be useful for doctors in women's clinics, because it is at the stage of dispensary observation it is important to prescribe timely prevention. It is especially important for doctors from remote areas. By applying the method using artificial intelligence, they can timely recognize a high-risk patient and refer her to a perinatal center.

The system was created together with doctors by specialists from the IT park of Petrozavodsk State University. To train the artificial intelligence, they used 10 thousand anonymized records of pregnant women with a confirmed risk of premature birth. And from the database of women's clinics from all over Karelia: when building a forecast, experts took into account the living conditions in a particular region.

Thanks to the capabilities of artificial intelligence, which is far superior to traditional methods, as it processes huge amounts of data, it was possible to effectively identify patterns between various risk factors. At the same time, the predictions were more accurate and reliable compared to human assessments. Diagnostic automation also speeds up the work. The model is planned to be integrated into the existing medical information systems of the republic. It is already being tested in women's clinics.