Diagnostic tests that help doctors choose treatments are often too narrowly focused—for example, checking for the presence of a single DNA mutation. Consequently, doctors must make decisions based on very limited information rather than a comprehensive insight into the individual and often highly complex condition of the patient. As a result, medications often do not work as effectively as we would like. "Beyond a simple yes or no, we need diagnostics to provide answers like: yes, and also this... or no, but let's try this instead," said during the discussion How the power of artificial intelligence is transforming cancer diagnostics and treatment, which was March 27, 2025 organized by Slovenian Technology Forum, emphasized Dr. Rafael Rosengarten, co-founder and CEO of Genialis.
According to him, artificial intelligence provides insight into the biological mechanisms of diseasethat simple DNA tests fail to capture—which is why Genialis relies on RNA data, as RNA reflects what is actually happening in the cell, not just what is encoded within it. Using this data, their machine learning models predict drug efficacy for each individual patient.
Genialis (www.genialis.com), a member of the Slovenian Technology Forum, is an international company transforming precision oncology using RNA and artificial intelligence. The international team of approximately 40 experts from various fields operates in Boston, Houston (USA), and Ljubljana.
In October 2024, the company introduced Genialis™ Supermodel, the first foundation model for cancer biology based on RNA data. The model incorporates hundreds of thousands of patient samples from around the world, and its goal is to support all aspects of clinical development in oncology—for every target, drug, and patient.
Many companies, including Genialis, are strives to ensure that cancer is no longer considered an incurable disease. "I am inspired by the speed of technological innovation, especially in companies and technologies for the early detection and prevention of disease. The good news is that we are becoming increasingly aware of the lifestyle factors that influence the development and progression of disease. At least in the US, we are seeing a decrease in cancer rates among certain demographic groups. Unfortunately, however, it is increasing in other groups, especially among women and young people," Rosengarten emphasized in the interview.
Genialis collaborates with pharmaceutical companies developing new cancer drugs. Their technology is used in the research and clinical development of targeted therapies and helps determine who will benefit most from a particular drug, how it works, or even why a drug might not be successful. They also work with companies developing diagnostic tests that doctors can use when making treatment decisions for individual patients.
But as Rosengarten says, one of the main challenges is certainly the accessibility of high-quality research data from existing cancer patients, from which machine learning models could derive relevant insights. In this area, they are working with organizations around the world to secure patient data from diverse populations – from Asia, South and North America, as well as from Slovenia (they are collaborating with the University Clinic Golnik). "This is the only way to ensure diverse and, consequently, high-quality data. Machine learning models are only as good as the data we train them on."
Another challenge is certainly research funding. Pharmaceuticals and biotechnology are highly capital-intensive industries. For a drug to reach the second phase of clinical trials, a company invests hundreds of millions of dollars, and it takes about two billion dollars for a molecule to become a drug. At the same time, 96% of oncology drugs never reach the final stage of clinical trials – and the patients who need them. Genialis is also currently preparing for its next round of fundraising.
There has been a long-standing global debate about the regulation of artificial intelligence. Rosengarten responds that in the healthcare sector, he is not actually worried about what this regulation will look like, as healthcare is already highly regulated by nature. "In the US, every product for medical purposes, regardless of whether it uses artificial intelligence or not, must go through clearly defined protocols set by the US Food and Drug Administration (FDA). I am much more concerned about other areas where artificial intelligence is being misused, such as direct consumer disinformation or deepfakes."
Rosengarten believes the key remains ensuring that doctors are both able and permitted to use existing technology. "I cannot speak for Slovenia, but I can say that in the US, it is very difficult to get reimbursement for tests that are already available. Those who pay, meaning insurance companies and other decision-makers, will have to grasp the importance of providing every patient with all appropriate tests so that healthcare professionals are aware of the best available treatment options."
