Digital Oracle

Why would each person create a neurodwainer

Alexander Bulanov, Izvestia

Thanks to the rapid development of artificial intelligence, everyone will soon be able to acquire a digital double — a virtual analogue of the body. Personal mobile sensors will send medical data to the patient's personal account in real time, and a special program will process them. This will allow not only to monitor the physical condition of the user, but also to predict the risks of developing diseases. The project of Russian scientists to create a human neurodeveloper is the first step towards the emergence of a global system of such monitoring.

Medical card account

In the future, personal accounts may replace the usual medical records, in which information about the patient's health status and all medical procedures that he has undergone will be accumulated. The functions of such a system may consist not only in simple data storage, but also in their machine processing, involving a preliminary assessment of the patient's health without the participation of medical personnel.

One of the most interesting ways to solve such a problem is the implementation of the project of scientists of the Nizhny Novgorod State University named after N.I. Lobachevsky "Human Neurodeveloper". The development involves both additional collection of medical data and the use of artificial intelligence systems for their interpretation. The concept of the project has already been created, scientists are developing technologies for its implementation.

The hardware part of the complex will be represented by a small mobile device that includes medical sensors. They are designed for continuous monitoring of such important indicators as respiratory rate, blood pressure, cardiac activity characteristics (cardiogram), body temperature, and so on. The received data will be promptly transmitted over the network to a special server, where, in addition to these indicators, the full history of medical observations from the patient's medical record will have to be stored.

Personal consultation

At the next stage, information will be processed using a neural network agent (personal neural network) — it should give a prognosis of the patient's condition. In addition, the developers plan to teach the system to offer its user options for further actions, which, depending on the situation, can range from "you need a good night's sleep" to "you need an urgent consultation with a cardiologist" or "call an ambulance urgently!".

The implementation of such functionality will be possible thanks to a special combined machine learning system, which is embedded in the neurodeveloper.

— Within the framework of our system, each person will have their own neural network that deals with their patient and is trained on his indicators, — said Alexander Gorban, a leading scientist at the Laboratory of advanced methods of multidimensional data analysis of Lobachevsky Nizhny Novgorod State University, professor at the University of Leicester (Great Britain). — At the same time, all personal neural network agents will be able to interact with each other, exchanging experience. Thus, a collective adaptive mind will be created, consisting of a huge number of small artificial intelligence systems. They will be continuously trained both independently and with the help of agents working with other patients.

According to the creators of the neural twin, such an approach to training neural networks can be considered optimal. If the system assumed the training of each individual neural network solely on the data of its patient, then it simply would not have enough information to correctly make predictions.

— On the other hand, to use a giant neural network that would assess the condition of all patients at once, huge computing resources would be required, which would make the project too expensive and difficult from the point of view of technical implementation, — Alexander Gorban added. — The neural network agents we offer, on the contrary, allow us to save resources, since a small mobile device will be enough for each of them to work.

However, the uniqueness of the project may create some difficulties in its implementation. According to Valentin Malykh, a researcher at the laboratory of Neural Systems and Deep learning at MIPT, the introduction of a neural double will require a long preparatory period (most likely several months), during which it will be necessary to monitor physiological indicators in a sufficiently large group of people and carry out initial training of the system based on the data obtained.

— In addition, developers will inevitably face the issue of protecting biometric data, which is personal and should not fall into the wrong hands (their distribution is limited by law. — "Izvestia"), — the expert clarified. — However, if the system is highly resistant to hacking, this administrative barrier can be overcome.

Unload the doctor

The introduction of the new system is proposed to begin with people suffering from chronic diseases (diabetes, cardiovascular diseases), since their state of health causes particular concerns among doctors and requires constant monitoring. The development of the system will largely depend on the improvement of wearable sensors.

— Mobile biometric sensors are still only developing, today their applicability is limited, — said Maxim Dyachenko, director of XOresearch. — In particular, if a full-fledged multichannel electrocardiography is able to determine the pre-infarction state, then the possibilities of its analogues are most often reduced to the definition of various types of arrhythmias. However, even the analysis of this information can save a person from a number of unpleasant consequences, provided that they work correctly with data and use high-level neural network technology.

If we talk about the creation of a global monitoring system, then its appearance will mean the reform of the entire medical sphere, which will allow doctors to remove a significant part of the daily burden by redistributing it to neural network assistants. This will allow specialists to devote more time to patients. And it will improve the quality of medical care, which, thanks to the advent of automatic forecasts, will be provided more timely.

Portal "Eternal youth" http://vechnayamolodost.ru