- Статьи
- Science and technology
- In concert with sounds: AI will customize children's hearing implants in minutes
In concert with sounds: AI will customize children's hearing implants in minutes
Russia has developed a hardware-software complex with a neural network that can customize a hearing implant for children in 10-15 minutes. Usually it takes three to five days, but with the new technology the child starts to hear well almost immediately. After this time, the child's hearing and speech skills can be fully developed. During this period, sessions with sign language therapists, speech therapists and psychologists are most effective.
AI for implant customization
The St. Petersburg Research Institute of Ear, Throat, Nose and Speech (St. Petersburg Research Institute of ENT) of the Ministry of Health of the Russian Federation has created a software and hardware complex (PAC) for the use of expert AI-based cochlear implant customization. A cochlear implant is an electronic device with the help of which deaf people can hear surrounding sounds and speech. The complex is equipped with the original domestic platform for artificial intelligence tasks Delta Sprut.
As medics told Izvestia, the adjustment of the cochlear implantation system is performed with the participation of several specialists, the most important being a sign language surgeon and a sign language teacher. Usually it lasts from several days to several months, depending on the age and preparedness of the child.
- The results of auditory-verbal rehabilitation depend directly on the correct setting of the speech processor parameters. Therefore, it is necessary to compensate for the hearing loss as much as possible and bring the perception of the deaf person closer to that of normal hearing people. This is achieved with the help of special equipment connected to the cochlear implantation system, which allows each child to customize the speech processor," Vladimir Dvoryanchikov, director of the St. Petersburg Research Institute of ENT, told Izvestia.
According to him, the parameters of speech processor tuning are individual and depend on the location of electrodes in the cochlea, individual sensitivity of the auditory nerve fibers and other features of the auditory system. In order to restore hearing, it is necessary to determine the optimal parameters of electrical stimulation for each individual electrode and the system as a whole (more than 50 parameters). For this purpose, different methods of research, consultations of doctors and classes of sign language teachers, adaptation together with parents are used in a usual case.
- The Institute has developed an artificial intelligence system that allows using objective research and computer technology to determine the main parameters of cochlear implantation system adjustment. All the experience accumulated by the Institute in this field was used to develop the AI. This technology makes it possible to carry out the necessary objective research of a deaf child's hearing within a few minutes and to predict the necessary parameters for the cochlear implant to work. Setting up the system with AI usually takes 10-15 minutes, - said Vladimir Dvoryanchikov.
After that, doctors continue to monitor the rehabilitation process and make minor adjustments to the settings if necessary, sign language teachers teach the child to hear and speak, said the doctor.
Neural networks in medicine
More than 1 thousand cochlear implantation surgeries are performed in the Russian Federation every year. These are high-tech interventions that require a large staff of highly qualified specialists and technological equipment.
The project was developed on the basis of the domestic ROSA Virtualization platform, which makes it possible to connect a large number of specialists.
- In many AI developments, video card memory is allocated to only one user. Our system allows rational resource allocation, so that several users can simultaneously work with the video memory of one video card, which significantly increases the efficiency of the system and speeds up the performance of tasks. Without such flexibility this process would be much more complicated and costly, - said Vyacheslav Kadomsky, Director for Strategic Development of ROSA IT STC.
When the entire system is running, first a test is conducted - nerve response telemetry. The results are loaded into the interface of the AI server, which predicts the recommended parameters of the implant settings. During the rehabilitation process, the child will be observed by specialists, and if necessary, minor adjustments can be made based on consultations.
- The most important thing is that the child will be able to hear the world as soon as possible. The experience of using a solution based on artificial intelligence allows us to achieve high accuracy of adjustment, - emphasized Vladimir Dvoryanchikov.
The development seriously saves time for specialists and allows children to get their hearing improvement medication faster, said Anton Averyanov, CEO of ST IT Group, expert of the TechNet NTI market. In childhood, when a child is still developing, every day of his or her knowledge of the world is very important, so saving as much as three to five days is essential.
Irina Kirichenko, Professor of the Department of Otorhinolaryngology of the Federal State Autonomous University of Russia (PFUR), noted that the use of neural networks in medical practice is an effective method of treating patients. However, if the new hardware-software complex will help to quickly adjust the implant, then further work without a specialized doctor is not possible, she explained.
- Setting up the implant is a technical component, but there is also a moment that cannot be excluded - it is communication with a sign language therapist, the moment of feedback. Doctors need not only to understand that the patient hears the sound, but he must also signal us a sign that he understands everything he is told,- said the doctor.
Traditional cochlear implant fitting relies on the patient's subjective sensations, which is especially problematic in children who cannot verbally describe their hearing experience. One of the possible solutions is to use objective methods to assess the auditory response, such as psychoacoustic tests or brain activity analysis, Dmitry Ravinsky, head of the ELVIS C project at Sensor-Tech Laboratory and an expert in the Neuronet STI market, told Izvestia.
- Artificial intelligence trained on such data, as well as on data from other patients, determines the quality of sound perception, captures the slightest changes in the response of the hearing system and offers more appropriate implant settings. This speeds up and improves the accuracy of device fitting, eliminates subjectivity and reduces the burden on audiologists," he said.
If the effectiveness of the technology is confirmed, it could become a new standard in the rehabilitation of cochlear implant patients, the expert emphasized.
- Some people are cautious about new technologies - this is normal when it comes to health. It is important that doctors explain the advantages of new technologies to parents of children. Confidence in such solutions is growing, especially when parents see real results - improvement of quality of life and reduction of time for rehabilitation, - stressed Maria Zhikhareva, the head of psychological and pedagogical direction of the charitable autonomous non-profit organization "Yaseneva Polyana".
According to the expert, the new approach will not only speed up the process, but also reduce the risk of errors that can occur during manual adjustment.
Переведено сервисом «Яндекс Переводчик»
