Žiga Kolar

Mr. Žiga Kolar

Department of Intelligent Systems, Jožef Stefan Institute, Ljubljana, Slovenia
orcid

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
29 July 2025 in Healthcare
Verified
Journal Article
Open access
PDF file
MATJAZ GAMSMATJAZ GAMS
MATJAZ GAMS
THTadej Horvat
Tadej Horvat
ŽKŽiga Kolar
Žiga Kolar
PKPrimož Kocuvan
Primož Kocuvan
KMKostadin Mishev
Kostadin Mishev
MSMonika Simjanoska
Monika Simjanoska
Background/Objectives: The demand for accessible and reliable digital health services has increased significantly in recent years, particularly in regions facing physician shortages. HomeDOCtor, a conversational AI platform developed in Slovenia, addresses this need with a nationally adapted architecture that combines retrieval-augmented generation (RAG) and a Redis-based vector database of curated medical guidelines. The objective of this study was to assess the performance and impact of HomeDOCtor in providing AI-powered healthcare assistance. Methods: HomeDOCtor is designed for human-centered communication and clinical relevance, supporting multilingual and multimedia citizen inputs while being available 24/7. It was tested using a set of 100 international clinical vignettes and 150 internal medicine exam questions from the University of Ljubljana to validate its clinical performance. Results: During its six-month nationwide deployment, HomeDOCtor received overwhelmingly positive user feedback with minimal criticism, and exceeded initial expectations, especially in light of widespread media narratives warning about the risks of AI. HomeDOCtor autonomously delivered localized, evidence-based guidance, including self-care instructions and referral suggestions, with average response times under three seconds. On international benchmarks, the system achieved ≥95% Top-1 diagnostic accuracy, comparable to leading medical AI platforms, and significantly outperformed stand-alone ChatGPT-4o in the national context (90.7% vs. 80.7%, p = 0.0135). Conclusions: Practically, HomeDOCtor eases the burden on healthcare professionals by providing citizens with 24/7 autonomous, personalized triage and self-care guidance for less complex medical issues, ensuring that these cases are self-managed efficiently. The system also identifies more serious cases that might otherwise be neglected, directing them to professionals for appropriate care. Theoretically, HomeDOCtor demonstrates that domain-specific, nationally adapted large language models can outperform general-purpose models. Methodologically, it offers a framework for integrating GDPR-compliant AI solutions in healthcare. These findings emphasize the value of localization in conversational AI and telemedicine solutions across diverse national contexts.

21 February 2022 in Healthcare
Verified
Journal Article
Open access
PDF file
MATJAZ GAMSMATJAZ GAMS
MATJAZ GAMS
ŽKŽiga Kolar
Žiga Kolar
ZVZdenko Vuk
Zdenko Vuk
CSChristina Samuelsson
Christina Samuelsson
BJBernhard Jäger
Bernhard Jäger
EDErik Dovgan
Erik Dovgan
The EU PlatformUptake project’s main goal is to investigate the usage of EU open and partly-open platforms in active and healthy aging (AHA) and ambient-assisted living (AAL) domains, from a software viewpoint. The aim of the project was to provide tools for a deeper interpretation and examination of the platforms, gather user feedback, and use it to improve the state-of-the-art approach in the AHA and AAL domains, and define instructions to enhance the platforms within the recommended order. The emphasis is on the software viewpoint for decision makers. In this paper, we present (i) the PlatformUptake methodology for AHA open platform assessments and its main objectives; (ii) clustering of the analyzed platforms; and (iii) the taxonomies generated from the text descriptions of the chosen platforms. With the use of the clustering tools, we present which platforms could be grouped together due to their similarities. Different numbers of clusters were obtained with two clustering approaches, resulting in the most informative two and four cluster groups. The platforms could be rather neatly presented in this way and, thus, potentially guide future platform structuring. Moreover, taxonomies, i.e., decision trees of platforms, were generated to easily determine each specific platform or to find platforms with the desired properties. Altogether, the computer comprehension of the platforms may be important additions to the human way of dealing with the AHA platforms, influencing future design, publications, related work, and research.

1 October 2026 in Information
Verified
Journal Article
Open access
PDF file
MATJAZ GAMSMATJAZ GAMS
MATJAZ GAMS
MKMatjaž Kumin
Matjaž Kumin
AVAndrej Vozlic
Andrej Vozlic
ALAndrej A. Lukšič
Andrej A. Lukšič
ŽKŽiga Kolar
Žiga Kolar
Democratic theory encompasses numerous democracy concepts, yet their conceptual relationships are rarely examined systematically at scale. This study develops a reproducible computational framework for grouping 100 democracy concepts from standardized textual descriptions created with large language model (LLM) assistance and verified against source materials. Two complementary representations were constructed: lexical word n-gram term frequency–inverse document frequency (TF–IDF) vectors and contextual sentence embeddings. Repeated K-means clustering across multiple random seeds and candidate cluster numbers was summarized through representation-specific co-association matrices and integrated into a consensus similarity matrix. A fifteen-cluster solution, with cluster sizes of 2–13 concepts and a consensus silhouette of 0.3625, provided the most interpretable balance of stability, granularity, and cluster structure. Lexical and contextual partitions nevertheless overlapped only moderately (mean Adjusted Rand Index (ARI) = 0.22; mean Adjusted Mutual Information (AMI) = 0.34), even though both were derived from the same standardized descriptions. This persistent divergence suggests that short formal textual descriptions capture only part of the semantic structure of democracy concepts. Overall, consensus clustering recovered interpretable macro-families while revealing that complex political concepts cannot be adequately represented by textual similarity alone.