Laboratory Staff

Adilova Fatima Tuychievna

Tibbiy-biologik informatika ilmiy laboratoriyasi mudiri More details →

Davronov Rifqat Rahimovich

Katta ilmiy xodim More details →

Ikramov Alisher Akramovich

Katta ilmiy xodim More details →

Kushmuratov Samariddin Ibodulla o’g’li

Kichik ilmiy xodim More details →

Misirov Farxod Abdulla o’g’li

Kichik ilmiy xodim More details →

Safarov Ro’zmat Abdiqayum o’g’li

Kichik ilmiy xodim More details →

Scientific Activity

Currently, the laboratory is conducting research in the following directions:

Generative artificial intelligence and quantum technologies. Generative artificial intelligence (AI), "deep technologies," and APIs have been prominent technological trends in recent years. In June 2024, the UN declared 2025 as the International Year of Quantum Science and Technology. Quantum technologies are transforming industrial sectors using the principles of quantum mechanics. Quantum computers can model molecular interactions at unprecedented speeds, which accelerates the development of new drugs, implements personalized medicine, and takes disease diagnostics to a qualitatively new level. Research conducted in this direction focuses on Kolmogorov-Arnold neural networks, cardiovascular disease prediction on multi-ethnic data benchmarks, quantum algorithms, quantum machine learning (QML), and quantum-enhanced image processing. The results not only enrich theoretical AI understanding but also serve as an important tool in the development of modern medicine and technology.

Cluster analysis and intelligent data modeling. Cluster analysis is one of the fundamental tasks in the field of intelligent modeling and data analysis. In this direction, methods of analyzing graph density variation, detecting clusters with complex nonlinear shapes, identifying dense "cores," and reconstructing the complete structure of clusters are studied. The main goal of the research is to develop a new cluster validation index (Core Separation Index, CSI), create its theoretical foundations, and prove its effectiveness through experimental comparison with other well-known indices. These methods reduce the need for manual hyperparameter selection and increase the level of automation in data analysis.

Efficient fine-tuning of large language models. This direction studies Parameter-Efficient Fine-Tuning (PEFT) methods to address the computational and memory cost problems arising from the growing size of large language models. Research is focused on developing and testing a new hybrid method, OLoRA+, which combines the structured initialization of OLoRA with the accelerated optimization of LoRA+. The main goal is to adapt pre-trained models by selectively fine-tuning a small number of additional parameters while keeping most of the original model frozen, and to reveal new dynamics associated with the selection of learning rate ratios. The results enable efficient and economically cost-effective tuning of AI models.

Vehicle detection and road traffic safety. This direction is dedicated to creating Vehicle Make and Model Recognition (VMMR) systems to ensure road traffic safety. Research involves developing a practical system that detects vehicle models and body colors in real CCTV streams, taking into account local conditions (road surface condition, camera placement, lighting practices). Models adapted to the transportation environment conditions of Uzbekistan, which are not adequately represented in open datasets, are being created. These results enable improvement of traffic management, automated monitoring, and security systems.

Statistical analysis in clinical research. This direction studies problems of identifying statistically significant clinical-laboratory and instrumental indicators associated with mortality in patients suffering from diseases such as arterial hypertension and diabetes mellitus, with the aim of applying scientific research results to clinical practice. Research is based on retrospective data and covers long-term follow-up periods (up to 8 years). The main goal is to create reliable statistical models for prediction across various healthcare problems and to assist medical personnel in making clinical decisions.

International Cooperation

The laboratory actively collaborates scientifically with many institutes and universities, including:

• Moscow and Novosibirsk State Universities;

• Kazan (Volga Region) Federal University;

• The Universities of Stanford, North Carolina, North Dakota, Cincinnati (USA);

• Helmholtz Zentrum München (Germany).

Seminars