Get started
Work

The cardiovascular professional of tomorrow: redefining basic research in the age of artificial intelligence

Year: 2025
Type: article
Abstract: Cinzia Perrino is a professor of cardiology at Federico II University in Naples, Italy. Her research focuses on cardiac hypertrophy, heart failure, ischaemia–reperfusion injury, and gut–heart interactions. At the European Society of Cardiology (ESC) Spring Summit and Cardiovascular Round Table 2025 in Paris, she described how artificial intelligence (AI) is reshaping the landscape of biomedical science and what this means for cardiologists of the future (Figure 1). Cinzia Perrino, professor of cardiology at Federico II University in Naples, Italy Can you describe the current landscape and how AI might impact basic science in cardiology? Basic science in cardiology is going through a difficult time: we are witnessing a divide between expertise and the career paths of those working in clinical and basic research. One of the underlying reasons is the lack of interdisciplinary education, and this has been recognised as an area for improvement by the European Society of Cardiology (ESC). Although in theory AI has the potential to bridge this gap, it also carries the risk of widening it in practice. AI can increasingly surpass humans in numerous tasks. As it is evolving rapidly, what we are talking about today may no longer hold true in a year or so. AI has captured much attention in the public media fuelling high expectations from patients, doctors, and governments alike. The 2024 Nobel Prize in Chemistry was awarded to a team that used AI to crack the long-standing problem of protein structure which scientists had been studying for 50 years. Although AI solved the problem in two years, the reality is more nuanced when we consider that AI analysed existing data to generate new information. What does AI have to offer the cardiologist of tomorrow? AI in basic research can unite different entities to achieve precision healthcare, but there are some caveats. Firstly, basic research is changing. In 2017, we highlighted in an ESC Statement the differences between a hypothesis-driven research approach—where we focus on a specific pathway and follow it—and an unbiased big data-based approach. When I first joined a basic science lab as a medical student 29 years ago, we focused on single dots on a single plane, and now I am leading a registry study in Italy getting samples from patients with peripartum cardiomyopathy. We’re doing multiomics integration, and AI algorithms. Science is changing, and we need to keep pace. What are some of the notable innovations AI has introduced? If you look at different stages of basic science, we have applications for every single step: AI can be a tool to generate new scientific hypotheses. We have semi-autonomous self-driving laboratories, and we can replace humans with generative AI digital twins. The most striking innovation from Japan is an ‘AI scientist’ that can do everything. It generates hypotheses, applies for grants, does the research, analyses the data, drafts the paper, and even does peer review. It’s been hypothesised that by 2050 an AI scientist will be able to win a Nobel Prize with a groundbreaking discovery. Our current scientific knowledge represents just a small fraction of what’s discoverable, and although human researchers can expand knowledge to a certain extent, AI has the potential to reach into vast, undefined territories that we can’t even map. Will we be replaced by AI? There was a commentary published in a journal about radiologists, but which also applies to basic scientists saying: ‘AI won’t replace basic science, but it will replace basic scientists who don’t use it.’ We do know that integrating human expertise with AI capabilities can significantly improve scientific outcomes which become faster and more precise, and this is the winning strategy. We cannot ignore AI anymore; it has to be part of our life, but only scientists can frame research questions to ensure that AI explores meaningful biological problems. Because most AI systems use existing data, we need to identify knowledge gaps and generate high quality big data with accuracy, standardisation, and diversity to bridge those gaps. We also need to design experiments that AI cannot fully automate and design. What are some of the things AI cannot do? Although AI excels at pattern recognition, only human scientists can uncover the mechanisms, and in this context it’s useful to think of AI as a medical device. Human scientists need to validate results with real world experiments to avoid false discoveries, ensuring predictability and clinical translation. We also need to consider regulatory approval, preserve ethics and scientific integrity. We need to fine-tune AI because it is only as good as the data and the algorithms it’s trained on. We are in charge of the models, and AI cannot do everything. There are also a lot of biases related to the utilisation of AI: as basic scientists our job is to understand what they are and how to control them. How can we get a grip on this technology? The ESC is already moving towards appropriate training to keep professionals on top of things in an AI-driven landscape. We should start incorporating AI literacy into curricula to understand the capabilities and limitations for teaching. We don’t need to be developers; we can be consumers or translators or whatever we choose. Core curricula are starting to come out now and this is leading to a change in medical schools and education. For example, in Italy, we have 15 medical technology university schools, with Federico II University students starting cardiology education and training this year after an early focus on biomedical engineering. How medical doctors become qualified is going to change, and the new generation MDs with combined expertise in biomedical engineering will have tremendous opportunities. What out the main points about AI to keep in mind? AI in basic science is a powerful tool with the potential to unlock scientific discoveries that were previously out of reach, but we need to keep in mind that it does not replace human curiosity, creativity, and analytical judgment. Human-AI synergy, rather than substitution, is the way to go right now. The success of AI does not just depend on developing the technology, but also on whether professionals accept and adopt it, and patients trust it. All authors declare no disclosure of interest for this contribution.
DOI: 10.1093/eurheartj/ehaf620
Author: Judith Ozkan
Language: English
Field: Medicine
Open Access status: bronze
OpenAlex ID: W4414795318
FWCI:
Cites:
Cited by:
Related to: