On October 28th, the BIRD Incubator hosted an insightful panel discussion as part of the AI2MED project, titled Building the Healthcare of Tomorrow. Experts from healthcare and AI sectors gathered to explore the role of artificial intelligence in reshaping diagnostics, treatment personalization, and patient monitoring. The AI2MED project’s mission is to close the skills gap between healthcare providers and AI applications, accelerating the adoption of AI for more accurate, efficient, and accessible healthcare.
How AI works and why it is different in Healthcare
Data scientist and senior researcher Mateo Sokač provided an in depth look at how AI functions and its distinct application in healthcare. He illustrated that AI models often work as “black boxes”, delivering effective results without always providing transparent insights into their processes. He explained that this lack of transparency poses a unique challenge in healthcare, where understanding why an AI recommendation is made can be just as crucial as the recommendation itself.
AI models trained on vast datasets may offer general patterns, but individual variability, especially in genetics or disease progression, demands cautious application. As Sokač observed, “AI might suggest a treatment based on millions of data points, yet the decision must always pass through a doctor who validates its relevance to the individual patient.” He underscored that while AI can handle complex computations, there is no substitute for a doctor’s understanding of the patient’s unique needs, especially in high stakes fields like oncology and genetics.
The challenge of responsibility in AI
Ognjen Bagatin, Owner of the Entrepreneur Media Platform, Director at Poliklinika Sinteza, and Co-founder at Futura Partners, highlighted both the potential and limitations of AI in healthcare. He emphasized that while AI tools, such as those used in radiology, are valuable for enhancing diagnostic precision and streamlining processes, they cannot replace human oversight. Bagatin underscored that AI’s role is fundamentally that of an aid, a tool designed to assist healthcare professionals and improve organizational efficiency.
However, he stressed the importance of responsibility: “AI cannot assume responsibility, it is ultimately humans who must interpret and act on AI’s recommendations.” Bagatin suggested that while AI contributes significantly to handling complex tasks, true value comes from its integration with human expertise, ensuring that medical professionals guide the decision-making process.
Real world impact of AI
A standout example of AI’s practical application in healthcare is the SENDD (System for Early Neurological Deviation Detection) project. Tomislav Strgar, Head of the AI department at TIS Grupa, detailed how SENDD was designed to aid early neurological assessments for infants. This system allows parents to record and submit several video clips of their child’s movements from the comfort of their home, eliminating the need for frequent in person visits to healthcare facilities. These video recordings are analyzed by AI models, and any flagged abnormalities prompt a review by doctors who make the final assessment and decision.
SENDD is based on the “General Movement Assessment” method, a long established approach for early detection of neurological deviations. The project was launched as an IRI (Innovation and Research) initiative by TIS Grupa in collaboration with Poliklinika Sabol. This initiative aimed to streamline the assessment process, making it more accessible for families while preserving the quality of medical evaluation. Strgar highlighted that the SENDD system allows for early identification of potential developmental issues, facilitating timely intervention.
Adoption of AI among medical professionals
The integration of AI in healthcare is met with varying levels of enthusiasm across different generations of doctors. A young physician from Vinogradska Hospital shed light on this divide, noting that while younger doctors are generally more open and adaptable to using AI tools in their practice, older generations often exhibit hesitation. This reluctance, she explained, stems from a combination of unfamiliarity with the technology and concerns over its opaque decision-making processes.
The panelists agreed that effective education is key to bridging this gap. They highlighted that training programs need to be comprehensive, focusing not only on how to use AI tools but also on understanding their limitations and capabilities. The aim is to equip healthcare workers with the confidence and knowledge to integrate AI effectively into their workflow, ensuring that it complements rather than complicates their practice.
AI as an ally, not a replacement
The panel’s discussion underscored a crucial message: AI in healthcare is a tool, not a replacement for human expertise. AI can undoubtedly streamline processes and improve efficiency, but it cannot assume the responsibility inherent in healthcare decisions. As Ognjen Bagatin, emphasized, “Responsibility remains with the human, AI can assist, but it’s ultimately up to us to interpret and apply these recommendations appropriately.”
As the AI2MED project and similar initiatives show, a thoughtful balance between technology and human insight can lead to a more effective and empathetic healthcare system for all.
