Respiratory diseases make everyday life difficult in countless ways and affect a large number of people in Hungary. How can artificial intelligence help these patients, and how could a tool developed by an international research team led from Pécs revolutionize the future of healthcare?

Dr. Tamás Ágh, principal investigator

University of Pécs

One such condition is chronic obstructive pulmonary disease (COPD), which affects 10 percent of the population over 40 – roughly half a million people. Numerous studies aim to improve the quality of life of those affected. One project under the HU-rizon programme, a flagship initiative of the NRDI Office designed to support Hungarian-led international research collaborations, is led by researchers at the University of Pécs and uses artificial intelligence to develop a tool that could become an effective weapon against the disease. This promising project will also be presented at the second stop of the HU-rizon Roadshow in Pécs on 9 April. We asked Dr Tamás Ágh, the principal investigator, about the disease and how artificial intelligence can help patients. 

“Diagnosing COPD is already challenging, as most people do not seek medical attention for the early symptoms – coughing, shortness of breath – until their quality of life is significantly affected. For this reason, the diagnosis is usually made at a moderately advanced stage of the disease, most often following a sudden deterioration known as an acute exacerbation event. It is during such an event that patients are examined for the first time.” 

How does the technology you are developing help patients? 

This tool is designed to make life easier for patients who have already been diagnosed with the disease. COPD patients can experience periods of deterioration several times a year, which is why it is important to anticipate the likelihood of such an event occurring for a specific individual under given circumstances. This is exactly what the model we have developed aims to predict; we are working on it as part of an international collaboration funded under the HU-rizon programme. 

So, something that has so far been unpredictable can now be predicted? 

Practically speaking, yes. Our artificial intelligence model collects information from medical records, analyses prescription and dispensing data, and also takes environmental factors into account, such as weather and air pollution data. Research shows that these periods of deterioration occur more frequently in winter than in summer, and air pollution also increases the susceptibility to exacerbations. In a city with higher smog levels, a patient is more likely to experience a deterioration than in a rural area. Prescription and dispensing data are also very important, as a patient who has regularly failed to buy their prescribed medicine over the past six months is more likely to experience a deterioration, indicating that they may not be adhering properly to their prescribed medication regimen. 

How will this innovative model work in practice? 

It is clear that many factors influence the occurrence of such a deterioration, ranging from the patient’s age, sex, and weight to the severity of the disease, how they take their medications, and even the expected weather in the coming period. Our model assists general practitioners by collecting data generated during patient care and analysing it together with other influencing factors. From this analysis, the treating physician can clearly see the likelihood that the patient in front of them will experience a deterioration within the next 3–6 months. Physicians do not need to request, review, and process a huge amount of data – all they have to do is run the algorithm we developed on a computer, and the result is immediately displayed on the screen. 

This would be a significant advance in medical care. Can we expect similar tools for other diseases? 

Artificial intelligence, when combined with the large healthcare databases available, is a powerful tool in the hands of medical science. On this principle, of course, other similar tools can also be developed. In Hungary, the National Health Insurance Fund has been storing health data on individuals going back decades. No human could possibly review, process, and interpret such a vast amount of data, but artificial intelligence can perform this work, identifying the necessary patterns. The same approach could, for example, be applied to patients with hypertension – allowing the risk of a stroke to be predicted in advance. Thus, the model developed for COPD could, in the future, be adapted to other diseases as well. 

What is the current status of the project? 

This is a three-year research project that we launched in January 2025 under the HU-rizon International Research Excellence Collaboration Programme. The project is led by researchers at the University of Pécs, with international partners including the Medical University of Łódź in Poland and the University Medical Center Groningen in the Netherlands. The first year focused on mapping similar models previously developed by others, identifying the factors that could be useful for us, and assessing potential data sources. We are currently building version zero of the model, after which testing and validation will follow. We aim to test the completed model in practice, involving 100–150 patients in primary care practices to see how the technology performs in real-life conditions. Finally, we plan to develop a health-economic model to examine the potential health benefits and resource savings for the Hungarian healthcare system if this system were implemented nationwide. 

The prototype of this artificial intelligence-based system – capable of revolutionizing healthcare – may therefore be developed in Pécs. Register and attend the HU-rizon Roadshow event on 9 April, where Dr Tamás Ágh will provide a detailed overview of the key directions and achievements of their project and the research findings to date as part of the HU-rizon research being conducted at the University of Pécs. 

Further details are available on the HU-rizon Roadshow website at https://nkfih.gov.hu/hu-rizont-roadshow-pecs. Participation in the event is free of charge but requires registration.