A student of NSU, together with a team of ITMO students, won the hackathon of the University of California

The international Caltech Longevity Hackathon unites participants from various fields of biology, medicine, artificial intelligence, data analysis, and entrepreneurship to develop solutions in health and longevity. The hackathon took place from May 22 to 24 on the campus of the California Institute of Technology. It was participated in remotely byPolina Karitskaya, a 2nd-year master’s degree student of the Faculty of Information Technologies (FIT) at NGU. Together with students of ITMOAnnoy Chesnokova,Lvov EidelmanиDastan AbdyldinThey developed the LongevMarker AI project — a system based on artificial intelligence for selecting biomarkers.

—My friends from ITMO told me about the competition and suggested I join their team as a specialist in machine learning and data analysis. Although we were all connected to bioinformatics one way or another, each participant had their own specialization, so we managed to distribute responsibilities effectively. Anna and Dastan were engaged in searching, analyzing, and structuring biological data, as well as assisting with the biological interpretation of the results.Myself and Lev were responsible for the technical part of the project: we designed the architecture of the machine learning model, trained it, tested various approaches, and analyzed the quality of the predictions. The preparation of the final presentation and interpretation of the results was handled by the entire team together., — says about the team Polina.

At the hackathon, the team developed the LongevMarker AI project—a system based on artificial intelligence for selecting biomarkers that can be used to assess the effectiveness of drugs and other interventions aimed at extending healthy life. One of the key problems in longevity research is that evaluating the real effect of therapy can take years or even decades. Therefore, scientists use so-called surrogate biomarkers—measurable indicators that allow much earlier assessment of whether a drug has the expected effect.However, the search for suitable biomarkers requires the analysis of a huge number of scientific publications and expert knowledge. Therefore, the team set the task of automating this process.

—In the course of the hackathon, we gathered and structured data from scientific literature, formed a database of known connections between drug interventions and biomarkers, developed a machine learning model, and tested modern neural network methods for analyzing the chemical structure of molecules. As a result, a recommendation system was created that obtains information about a compound or drug and suggests the most promising biomarkers for evaluating its effectiveness, ranking them by degree of relevance.— added Polina.

The development by Polina’s team became the winner of the Caltech Longevity Hackathon 2026, receiving a cash prize of 1500 US dollars and a valuable prize — interest in the project from investors and experts in the startup sphere.

The project is currently at the stage of a minimally viable product. At the hackathon, the team managed to demonstrate that artificial intelligence can indeed help researchers select promising biomarkers for longevity studies.

Ahead of the team is a stage of work to expand the data volume, improve the quality of recommendations, and conduct deeper model validation in collaboration with experts in the fields of biomedicine and gerontology. In addition, a remote meeting is planned in the near future with representatives of the venture fund XYZ Ventures from San Francisco, which invests in technological startups at early stages. For us, this is an opportunity to receive expert evaluation of the project and discuss prospects for its further development as a potential product for research organizations and biotechnological companies.

—I was immediately interested in the hackathon format. Since I am engaged in bioinformatics and the application of artificial intelligence methods in biomedical research, participating in such a hackathon seemed to me an excellent opportunity to work on a relevant scientific problem together with a strong interdisciplinary team and receive feedback from industry experts, including representatives of the American company Insilico Medicine. This was my first experience participating in an international hackathon of this level. Among the participants were students and researchers from leading universities in the world, including Harvard, MIT, and other prestigious universities.Therefore, initially I did not set myself the task of necessarily winning — I was interested in working on a real task in a good team and gaining invaluable experience.— summarizes Polina.