Projects

Ongoing projects

This project is aimed to develop and evaluate deep learning models for predicting spinal curvature from 3D back surface scans as a radiation-free alternative to traditional X-ray methods. Two primary approaches were investigated: a Convolutional Neural Network using depth maps and a Point Cloud Transformer using 3D point clouds. The models were optimized and compared […]
Read more
Creating an AI platform for the drug supply prediction for 25 million users across 1752 location across Côte d’Ivoire. The platform has also been designed to perform and control the medicine supply chain in Côte d’Ivoire. The goal of this project is to contribute in improving the health system and the overall quality of life […]
Read more

Partners

The Innovation Fund of the Republic of Serbia
Read more
The project contains comprehensive study aimed at improving the prediction of melanoma recurrence using machine learning (ML) techniques. Drawing from a dataset provided by the Serbian national melanoma registry, it aims to employ a variety of statistical and ML methods to refine the prognosis for melanoma patients. The study on analyzing survival rates, recurrence, and […]
Read more
In collaboration with a startup from Japan, we are researching and developing a platform for the design of small molecules. The platform includes state-of-the-art AI models for predicting molecular properties based on small sets of experimental data, as well as generative molecule design to optimize the desired property.
Read more
As part of this pilot project, an analysis of user emails was carried out in cooperation with Telekom company to enhance the quality of customer support and user satisfaction. After anonymization, the emails were processed using the topic modeling method, leveraging state-of-the-art language models for the Serbian language. The developed system enables fast and automatic […]

Partners

Telekom Srbija
Read more
The aim of this pilot project, launched jointly with three large healthcare centers in Serbia – the University Clinical Centre of Serbia, the University Children’s Clinic Tiršova, and Zemun Hospital – and in cooperation with the innovative Japanese pharmaceutical company Takeda, is to reduce the complexity of Fabry disease diagnosis by employing modern AI techniques. […]

Partners

Takeda
Read more
This project primarily focuses on verifying the accuracy of information in texts generated by generative language models within the biomedical field. The developed system aims to provide users with accurate, real-time responses to their queries supported by relevant sources. This project involves fine-tuning and prompting generative language models, lexical, semantic, and hybrid search, and metrics […]

Partners

Bayer AG
Read more
NGI
The objective of this project was to leverage transfer learning capabilities to develop a method enabling the training of a model to recognize Named Entities for which it has not been trained, with no or minimal supporting training examples. This method necessitates the input data transformation for the binary classification of tokens and employs a […]

Partners

Bayer AG
Read more
The project revolves around the development of new Natural Language Understanding (NLU) tools designed to help analyse free form medical texts in Serbian. These will be used to aid in the targeted screening of Fabry disease (FD) and Hunter syndrome (MPSII) in order to improve the diagnosis and treatment of FD and MPSII patients.

Partners

Takeda
Read more
Green program of cooperation between science and economy of The Science Fund of the Republic of Serbia
Read more
The use of electronic noses in industry enables the monitoring and control of a wide range of manufacturing processes in an industry in a minimally invasive way. Its application in the food and beverage industry is particularly noteworthy, where it can be used for quality control of products and monitoring of certain biochemical processes. The […]
Read more
In collaboration with a startup from the USA, we are developing a generative AI-based platform for peptide design targeting specific proteins. In addition to the generative process, the platform includes candidate filtering using the state-of-the-art Alpha Fold 2 model to predict the binding affinities of designed peptides and the target protein.
Read more
With the help of our Chinese colleagues we have been able to create a 2.7 billion parameter NLU model able to translate between Serbian and 52 other languages. This is just a starting point and we hope to improve the model’s performance in the time to come, but you can try it out now if […]

Partners

Peng Cheng Laboratory
Read more
Based on the premise that it takes two to tango, the project aims to build a synergistic approach to human-machine decision making The international consortium behind TANGO will develop the theoretical basis and computational framework for hybrid decision support systems (HDSS) in which humans and machines are aligned in terms of values and goals, know […]
Read more
The increase of textual data volumes necessitates a swift progression in data science, particularly in Natural Language Processing (NLP). The UniDive action is centered on the task of preserving linguistic diversity, with a focus on understanding both the differences between languages and the nuances within them. UniDive reframes technology from a potential threat to an […]
Read more

Past projects

This project aimed to determine the underlying causes of vaccine hesitancy regarding COVID-19 among young individuals in Serbia, thereby contributing to an increase in vaccination rates. The dataset employed in this research comprised tweets related to COVID-19 vaccination. By integrating sentiment analysis using advanced language models and topic modeling techniques, we identified several categories of […]

Partners

USAID | UNDP
Read more

Would you like to work with us?

Send us an inquiry via the Contact page and our team will contact you as soon as possible!

This site is registered on wpml.org as a development site.