Blue Sky Research is a contest of research projects by young scholars who use artificial intelligence to solve problems at the frontiers of science. It is organized by the Foundation for Support of Innovations and Youth Initiatives of St. Petersburg (The Center for Strategic Research “North-West” and the association Artificial Intelligence in Industry are the contest’s partners).

The term “blue sky research” is used to refer to scientific inquiry in domains where real-world applications are not immediately apparent. That is precisely why it is deemed high risk research. As highlighted by Sergey Salkutsan, the head of the Foundation for Support of Innovations and Youth Initiatives, the contests’ goal is to support such projects and eminent young scholars, as well as to facilitate the development of new technological products and the formation of interuniversity teams.

“We anticipate that the primary breakthroughs might transpire at the junction of certain avenues of research, so it was important for us to bring together professionals in chemistry, biology, and medicine. And in order to accelerate the advancement of their scientific interests, we suggested they utilize artificial intelligence and join up with researchers from organizations working in different fields,” says Mr. Salkutsan.

At the core of the contest lies a model for scientific cooperation, in which scholars are separated into two roles: a “scientific client” and a “scientific contractor”. The former deal with proposing project ideas, putting forward hypotheses, managing the teams, and assessing the results. The latter are responsible for actually solving their problem. 

Teams of researchers had to develop their projects from a concept to a tangible prototype in nine months. They got to showcase the results of their work to representatives from the scientific community, the authorities, and business in the finals. A special program was created by the organizers in order to help young scholars pass through all the stages of the contest. It included the school of Principal Investigators (PI), which counts among its experts Alexander Bukhanovsky, the head of ITMO’s School of Translational Information Technologies, who explained how to use AI to tackle scientific problems. Another notable inclusion was the Blue Sky PR school of science communication and personal branding, where Master’s students in Science Communication and experts from ITMO’s Center for Science Communication  counseled the participants on how to develop a promotion strategy for a scientific brand in cooperation with science communicators so that as many people as possible are made aware of their research.

“As a part of our school Blue Sky PR: New Names in Science, we, along with researchers, analyzed their representation in the media space and academia and discussed promotion strategies. We consider it crucial to tell exciting stories about these research teams and inspire them to communicate proactively. It is great that thanks to the Foundation the finalists have the chance to implement their own communication strategies and projects, as well as increase their media presence,” shared Daria Denisova, the head of Center for Science Communication.

The projects that rose to the top

Nikita Serov and Daniil Kladko, engineers of the Solution Chemistry of Advanced Materials and Technologies (SCAMT) Institute, developed a digital platform to predict the physical and chemical properties of magnetic nanomaterials. Such materials are most prominently utilized in MRI scanners and magnetic hyperthermia – a type of therapy for cancer. But creating a nanomaterial with particular properties can only be achieved via experimentation, which takes up considerable time and resources. The solution by SCAMT scholars will allow materials science chemists to enter the required particle properties and experimental conditions into a specialized template, which will then display the calculated values.

“To begin with, our goal was to create a platform that is friendly to its various users, so that people who study wet chemistry but are not too familiar with artificial intelligence could take advantage of it. We also aim to add in more data and optimize the descriptors,” said Daniil Kladko. “Later, we will be able to offer our product to pharmaceutical companies as well. They could utilize it to develop nanopatterned drugs based on magnetic nanoparticles.”

Another team which included Ekaterina Skorb, the director of the Infochemistry Scientific Center, Evgeny Smirnov, a leading research associate at the Center, Maria Masalovich, a research associate at the Laboratory for Intelligent Technologies in Infochemistry, and Timur Aliyev, an engineer at the laboratory, put forward an application that qualitatively and quantitatively determines the presence of certain antibiotics in powdered milk based on the current-voltage characteristics of the samples. The thing about milk with a high concentration of antibiotics is that it can not be used to make dairy products. Hence, the manufacturers and the suppliers are keen on quickly and accurately assessing the properties of their raw materials. Unlike the already established methods, this new technology can be deployed anywhere and will find out if the mix has antibiotics and which of the 16 compounds it is, all in just 15 minutes. The researchers did not just win the contest, but were also recognized as the best research team according to the Blue Sky Research’s panel of experts.  

“This platform of electrochemicall sensors will come in handy in different fields when determining the qualitative and quantitative properties of antibiotics. As of now we delved into agriculture and the food industry, but the base technology can find its application in medicine, oil refining, and so on,” explained Ekaterina Skorb. “Besides, starting out we were capable of differentiating between just five antibiotics, now we can work with 16, and in the future we will expand up to 78 compounds with plans to include hormone identification.”

SCAMT engineers Yuliya Razlivina and Olga Kapustina, in collaboration with Nikita Serov,  presented SeQuant – a neural network that creates chemical digital models of biopolymers. It functions like this: the network acquires data on reagents and reaction products (generated descriptors, weighted graphs, Coulomb matrices, etc.), and outputs a protein sequence or sequences, consisting of several chains if it is an enzyme.

“At this time, it is hard to produce a complex molecule in a single stage. Generally, it takes multiple tries, which leads to a low or non-existent yield. In turn, if we could pick the right protein for a particular reaction, we could conduct a complex multi-stage reaction in just one stage. The main objective of our project is to teach a computer that a protein is not just zeroes and ones but a set of chemical information. Now we are working on improving our model, and later we will adjust it to fulfill specific industry goals, for instance in the future it could be used to optimize food production or investigate properties of unknown biopolymers,” explains Olga Kapustina.

Yuliya Razlivina and Olga Kapustina also worked on another project – a pharmacologist neural network. Using the information on drugs approved by the United States Food and Drug Administration (FDA) and transcriptomic data on diseases, this project will help medicinal chemists to create new pharmaceuticals. It is anticipated that artificial intelligence technologies will speed up the development of new treatments, reduce resource requirements, and lower the chance of a candidate drug failing final rounds of testing.

“Over the course of this year, we compiled a large database of biological and chemical properties of different diseases. As a result, we developed two prototype models that can generate molecules of medical compounds, and one that can work with biological data. That is our project’s main feature – both datasets enable the development of a treatment for almost any disease, regardless of its nature. Next we have to combine the different models, test them on well-known diseases, and then move on to rarer ones. Afterwards, we will determine the physical, chemical, and pharmacological properties of the generated molecules in order to eventually synthesize them and begin running tests on animals and humans,” said Olga Kapustina.

The experts of the Blue Sky Research contest also highlighted the projects by several other teams of finalists. Among them were:

• Density-functional theory (DFT) calculations using neuromorphic systems. The project was presented by Mikhail Medvedev, head of the Theoretical Chemistry Group (Lab. №24) in the N. D. Zelinsky Institute of Organic Chemistry of the Russian Academy of Sciences and ITMO’s visiting associate professor. His team also received the award as the best research team according to the Foundation for Support of Innovations and Youth Initiatives of St. Petersburg.
• The development of a smart data collection system to forecast myocardial infarction using AI methods.
• A project for improving the resolution of microscopic images utilizing deconvolution and point spread function, where the point is determined via machine learning.
• A data capture system that uses a “swarm” of infrared carbon dioxide sensors to monitor air quality.
• A project about analyzing images of tumors in the central nervous system and running differential diagnostics on them using AI.
• A smart mini-spectrometer to determine food product properties.

All winners received diplomas, statuettes of the “scientific clients” and “contractors,” along with New Year wishes written by an AI. Additionally, the prize pool of 15 million rubles was split between all teams that successfully passed all stages of the contest (team formation, prototype creation, and its presentation).