Parents can independently diagnose speech disorders in children using AI

The percentage of children with speech impairments is growing from year to year. In 1970-1980, only 25% of children had violations.

Today this figure is significantly higher - 58% - andcontinues to grow. According to international statistics, speech disorders are present in 17.5% of children under the age of 7. And the sooner they are diagnosed, the more effective the treatment will be.

“For almost 20 years of the company's existence,has accumulated extensive expertise in the field of neurology. And today we can help pediatric neurologists and parents not only in the treatment of delays in speech development through the production of effective and safe drugs. The creation of an automatic system for the diagnosis of speech defects in children, which reduces the time of their initial detection, is an innovative response to existing problems. And, despite the fact that the solution was launched relatively recently, we regularly receive positive feedback from parents and doctors, which testifies to its real benefits, ”emphasizes Marina Rykova, director of the marketing department at Geropharm.

"Spectrogram" is designed for children 3-6 years old andtheir parents is a chatbot with the function of sending audio messages. The main advantage is a playful approach that does not traumatize the child. Testing takes place in the format of entertainment - the bot invites the child to answer a few questions with the help of a parent. All exercises were developed by certified speech therapists and adapted to the created interface.

“Every second student has problems withpronunciation that could have been eliminated at an earlier age. But not all parents “reach” a specialist, writing off speech defects as childhood. Unfortunately, when the problem is already obvious to others, it is much more difficult to fix it, ”says speech therapist Natalya Sabanova.

The widget is powered by machine learning. Initially, the accumulation of "knowledge" by the program took place with the active participation of invited speech therapists: users sent audio recordings, and doctors checked them for speech defects and loaded the results into the system, thus teaching it.

Now human participation in the diagnostic processreduced to a minimum. When a user uploads a voice audio file to the system, the neural network checks it for speech disorders and sends the result to a speech therapist, who decides on the need for consultation. Then the test result is sent to the parent in the personal account, and also uploaded to the database. In the near future, Geropharm plans to fully automate the program.

To date, several hundred children have already been tested, and the diagnostic accuracy is more than 80%, provided that the recording is of high quality.

You can get tested on the PRO.MOZG project website.

Read also

On the threatening Earth asteroid Apophis noticed a dangerous phenomenon. What's happening?

Scientists have found out why children are the most dangerous carriers of COVID-19

What the Parker Solar Probe discovered when it flew as close to the Sun as possible