AI in the service of
children's health

The answer is YES!

Kid-AID is a project aimed at creating tools based on artificial intelligence (AI) algorithms to assist healthcare professionals in making decisions regarding the treatment of children in emergencies.

Imagine a scenario: It’s a Friday evening. Lucy, an 8-month-old, starts to develop a high fever. She has a temperature of 39°C, is fussy, refuses to eat, and can’t sleep. Concerned, her parents take her to the Emergency Room. The on-call doctor lacks experience in assessing infants and, as a precaution, asks for a consulting specialist. They have to wait another 3 hours. By the time Lucy is seen by a specialist, it is already 11:30 PM. The baby is crying, and it’s hard to assess whether she is seriously ill or just tired. As a precaution, she is admitted to the hospital, taking the last available bed. Shortly afterward, another child in much worse condition arrives at the hospital. The search for space in another hospital begins…

Such situations occur every day, in every city, and not only in Poland. Why? Because assessing the health condition of infants and young children is a great challenge. It takes experience and practice to be confident in decision-making.

Medical University of Wrocław, in collaboration with Animativ and thanks to cooperation with St. Jadwiga Hospital in Trzebnica and Specialist Hospital in Wałbrzych, is leading a pioneering project that utilizes artificial intelligence (AI) to assist healthcare professionals in making critical decisions regarding the health of young patients.

Every day, thousands of children arrive at emergency departments with various symptoms. Studies show that only 4-6% of these cases require immediate medical intervention, and only 20% of children need hospitalization.

This means that many hospitalizations could be avoided if doctors had precise tools to assess a child's health condition.

Experienced pediatricians intuitively recognize the "ill appearance" of a seriously ill child, based on signals such as:

• the child's activity,
• breathing,
• skin tone,
• the child's behavior and responses.

Unfortunately, many of these signals are subjective and depend on the doctor’s experience. As a result, many cases are hospitalized even though they do not require immediate medical intervention.

A team led by Kamila Ludwikowska, MD, and Prof. Leszek Szenborn from the Department and Clinic of Pediatrics and Infectious Diseases at the Medical University of Wrocław is developing an AI-powered application. The goal of this app is not to replace the doctor, but to assist in assessing a child’s condition in difficult cases.

The application will:

• analyze short video recordings of children,
• use AI algorithms to identify critical symptoms such as breathing disturbances, movement pattern abnormalities, skin tone changes, and weakened responses,
• provide suggestions on whether immediate medical assistance, hospital observation, or further diagnostics are necessary.

As Kamila Ludwikowska, MD, emphasizes, AI "will not make a diagnosis," but it can successfully aid doctors in making assessments, particularly in areas where there is limited access to experienced staff or advanced diagnostic equipment.

Creating a Knowledge Database
The database will include:

• video recordings of young patients in varying health conditions,
• information on the course of the illness in children, such as symptoms, test results, medical history, and interventions taken.

The recordings, which will be made at the clinic in Wrocław and at the hospitals in Wałbrzych and Trzebnica, will allow algorithms to "learn" from real cases. As a result, the application will gain the ability to recognize subtle signs of a child's severe health condition.

In addition to supporting medical decision-making, the application has immense educational potential. Young doctors, medical students, and paramedics will be able to access the clinical case repository to gain experience in assessing children's health conditions.

Medical staff training focuses on the presence of specific symptoms and abnormalities. However, infants and young children display a wide range of behaviors that often fall outside these standard guidelines. To recognize when "something is off," it is essential to observe numerous examples of normal child behavior patterns. The recordings and histories of young patients collected for the Kid-AID project can serve as numerous examples, helping to understand the diversity of behaviors and symptoms in children both in health and illness. This will allow healthcare professionals to develop the skill of appropriately assessing a child's health condition more quickly. Additionally, they will have the opportunity to observe and learn about rare clinical cases of serious diseases in a much more comprehensive and accurate way than textbook descriptions provide.

As Prof. Szenborn emphasizes, this is especially important in times when, thanks to vaccinations, children are less likely to suffer from infectious diseases like meningococcal sepsis. While this is a beneficial change for population health, it also makes it more challenging for young doctors to gain practical knowledge in this area.

The project is being carried out in collaboration with Animativ Sp. z o.o., a company specializing in digital-medical solutions. This collaboration ensures the integration of medical knowledge with modern technologies.