1. Weakly-supervised multilingual medical NER for symptom extraction for low-resource languagesRigon Sallauka, Umut Arioz, Matej Rojc, Izidor Mlakar, 2025, izvirni znanstveni članek Opis: Patient-reported health data, especially patient-reported outcomes measures, are vital for improving clinical care but are often limited by memory bias, cognitive load, and inflexible questionnaires. Patients prefer conversational symptom reporting, highlighting the need for robust methods in symptom extraction and conversational intelligence. This study presents a weakly-supervised pipeline for training and evaluating medical Named Entity Recognition (NER) models across eight languages, with a focus on low-resource settings. A merged English medical corpus, annotated using the Stanza i2b2 model, was translated into German, Greek, Spanish, Italian, Portuguese, Polish, and Slovenian, preserving the entity annotations medical problems, diagnostic tests, and treatments. Data augmentation addressed the class imbalance, and the fine-tuned BERT-based models outperformed baselines consistently. The English model achieved the highest F1 score (80.07%), followed by German (78.70%), Spanish (77.61%), Portuguese (77.21%), Slovenian (75.72%), Italian (75.60%), Polish (75.56%), and Greek (69.10%). Compared to the existing baselines, our models demonstrated notable performance gains, particularly in English, Spanish, and Italian. This research underscores the feasibility and effectiveness of weakly-supervised multilingual approaches for medical entity extraction, contributing to improved information access in clinical narratives—especially in under-resourced languages.
Ključne besede: low-resource languages, machine translation, medical entity extraction, NER, NLP, patient-reported outcomes, weakly-supervised learning
Objavljeno v DKUM: 19.05.2025; Ogledov: 0; Prenosov: 2
 Celotno besedilo (338,94 KB) |
2. Natural Deep Eutectic Solvent-Based Matrix Solid Phase dispersion (MSPD) extraction for determination of bioactive compounds from sandy everlasting (Helichrysum arenarium L.) : a case of stability studyMilena Ivanović, Peter Krajnc, Aleš Mlinarič, Maša Islamčević Razboršek, 2022, izvirni znanstveni članek Opis: In the present study, vortex-assisted matrix solid-phase dispersion (VA-MSPD) extraction
was used to isolate the major bioactive compounds from H. arenarium. To reduce the negative environmental impact of the conventionally used organic solvents, four different choline chloride-based
natural deep eutectic solvents (NADES) were investigated as possible eluents. The most influential
VA-MSPD extraction parameters: stationary phase (adsorbent), adsorbent/sample ratio, vortex time,
and volume of extraction solvent were systematically optimized. Ultrasound-assisted extraction
with 80% MeOH was used as the standard method for the comparison of results. The stability of the
obtained extracts was studied over a period of 0 to 60 days at three different temperatures (−18 ◦C,
4
◦C, and 25 ◦C). All extracts were evaluated both spectrophotometrically (determination of total
phenolic content (TPC) and antioxidant activity by ABTS and FRAP assay) and chromatographically
(HPLC-UV). NADES based on choline chloride and lactic acid (ChCl-LA) was selected as the most
effective extractant, with a determined TPC value of its extract of 38.34 ± 0.09 mg GA/g DW (27%
higher than the methanolic VA-MSPD extract) and high antioxidant activity. The content of individual phenolic compounds (chlorogenic acid, dicaffeoylquinic acid isomers, naringenin isomers,
and chalcones) in the ChCl-LA extract, determined by HPLC-UV, was comparable to that of the
conventionally obtained one. Moreover, the stabilization effect of ChCl-LA was confirmed for the
studied compounds: chlorogenic acid, naringenin-40
-O-glucoside, tomoroside A, naringenin-5-Oglucoside, isosalipurposide, and naringenin. The optimum VA-MSPD conditions for the extraction of
H. arenarium polyphenols were: florisil/sample ratio of 0.5/1, a vortex time of 2 min, and an elution
volume of ChCl-LA of 10 mL.
Ključne besede: matrix solid phase dispersion extraction, MSPD, natural deep eutectic solvents, NADES, extraction optimization, sandy everlasting, phenolic compounds, stability studies, antioxidant activity
Objavljeno v DKUM: 09.05.2025; Ogledov: 0; Prenosov: 3
Celotno besedilo (2,47 MB)
Gradivo ima več datotek! Več... |
3. Synergistic effect of supercritical and ultrasound-assisted ginger (Zingiber officinale Roscoe) extractsTaja Žitek Makoter, Nika Kučuk, Vesna Postružnik, Maja Leitgeb, Željko Knez, Mateja Primožič, Maša Knez Marevci, 2022, izvirni znanstveni članek Opis: Proper processing of natural material is crucial to obtain an extract with high content of
biologically active components. Dried, grinded ginger roots were extracted by ultrasonic method
and supercritical extraction with CO2
. The aim of the study was to determine if a mixture of the two
types of extracts attained by different methods and solvents exhibits better bioavailability than each
extract itself. Therefore, both extracts were analytically evaluated and then mixed in a ratio of 1:1.
The supercritical extract (SCG extract) and the mixed extract (mixG extract) had high antioxidant
activity (78% and 73%) and total phenols (827 mg/g ext. and 1455 mg/g ext.), which is also consistent
with the levels of gingerol (303 mg/g ext. and 271 g/g ext.) and shogaol (111 mg/g ext. and 100 g/g
ext.) in the extracts. In comparison to both pure extracts higher levels of total phenols were found
in the extract mixG. This could be the reason for the significant inhibition of melanoma cells and
antimicrobial potential (against Staphylococcus aureus, Escherichia coli, and Candida albicans). The
combination of the extracts resulted in a significant increase in the inhibition of selected microbial
and melanoma cells WM-266-4 compared to the control. Cell viability decreased below 60% when
mixG extract was applied. Antimicrobial activity has been confirmed.
Ključne besede: ginger, supercritical fluid extraction, ultrasound-assisted extraction, antimicrobial activity
Objavljeno v DKUM: 09.05.2025; Ogledov: 0; Prenosov: 2
Celotno besedilo (1,59 MB)
Gradivo ima več datotek! Več... |
4. Optimisation of the green process of industrial hemp - preparation and its extract characterisationTaja Žitek Makoter, Petra Kotnik, Teo Makoter, Vesna Postružnik, Željko Knez, Maša Knez Marevci, 2022, izvirni znanstveni članek Opis: Natural medicines and products are becoming increasingly important in the pharmaceutical
and food industries. The most important step in obtaining a natural remedy is the processing of
the natural material. This study offers the separation of the industrial hemp plant into fractions by
mechanical treatment, which has a significant impact on the selectivity of the obtained fractions. This
study also offers a solution to reduce waste by fractionating industrial hemp, focusing on the fraction
with the highest cannabinoid content (49.5% of CBD). The study confirmed the anticancer potential
of the extract, which prevents further division of WM-266-4 melanoma cells at a concentration of
10−3 mg/mL. However, application of the extract (c = 10−3 mg/mL) to normal human epidermal
melanocytes proved to be insignificant, as the metabolic activity of the cells was the same as in the
control cell group.
Ključne besede: supercritical fluid extraction, ultrasonic extraction, melanoma cells, extraction, WM-266-4
Objavljeno v DKUM: 09.05.2025; Ogledov: 0; Prenosov: 2
Celotno besedilo (1,96 MB)
Gradivo ima več datotek! Več... |
5. Discrete time model for process meta language with fictitious-clockBoštjan Vlaovič, Aleksander Vreže, 2022, izvirni znanstveni članek Opis: Industries like telecommunications, medical, automotive, military, avionics, and aerospace
use complex real-time systems. Specification and Description Language (SDL) is one of the leading
domain specific languages that is formally defined by international standards and well established in
describing such systems. To check system properties abstracted model of the system is prepared in
selected modeling language. We use Spin (Simple Promela Interpreter) model checker that is one of
the leading tools for verification of complex concurrent and reactive systems. This paper focuses on
modeling the SDL timer construct. It is one of the SDL constructs that is not easily modeled with
Promela, but is present in many SDL systems. After an overview of the related work we propose
a new Discrete Time Model for Promela (DTMP) that is seamlessly integrated in our framework
for modeling SDL systems and can be used with the mainstream version of the Spin tool. To the
best of our knowledge, this is not possible with the existing solutions. We describe how DTMP can
be used to model SDL systems that use timers. Experimental results demonstrate its applicability
to non-SDL systems with Fischer’s mutual exclusion protocol and the Parallel Acknowledgment
with Retransmission that were used in prior studies. We compare state-space requirements with
one of the existing solutions DT Promela and DT Spin. With that, virtues and shortcomings of this
high-level solution are exposed. We have shown that DTMP is effective when an extensive range of
timer expiration values are used, which is usually the case in real-life SDL systems.
Ključne besede: formal specifications, formal languages, discrete time, model checking, automated extraction, SDL, Promela, SpinRCP, Sdl2pml
Objavljeno v DKUM: 27.03.2025; Ogledov: 0; Prenosov: 2
Celotno besedilo (660,17 KB)
Gradivo ima več datotek! Več... |
6. Supercritical fluid and conventional extractions of high value-added compounds from pomegranate peels waste: production, quantification and antimicrobial activity of bioactive constituentsKaja Kupnik, Maja Leitgeb, Mateja Primožič, Vesna Postružnik, Petra Kotnik, Nika Kučuk, Željko Knez, Maša Knez Marevci, 2022, izvirni znanstveni članek Opis: This study is focused on different extractions (Cold Maceration (CM), Ultrasonic Extraction
(UE), Soxhlet Extraction (SE) and Supercritical Fluid Extraction (SFE)) of bioactive compounds from
pomegranate (Punica Granatum L.) fruit peels using methanol, ethanol, and acetone as solvents in
conventional extractions and changing operating pressure (10, 15, 20, 25 MPa) in SFE, respectively.
The extraction yields, total phenols (TP) and proanthocyanidins (PAC) contents, and antioxidant
activity of different extracts are revealed. TP and PAC recovered by extracts ranged from 24.22 to
42.92 mg gallic acid equivalents (GAE)/g and 2.01 to 5.82 mg PAC/g, respectively. The antioxidant
activity of extracts ranged from 84.70% to 94.35%. The phenolic compound identification and
quantification in selective extracts was done using the LC-MS/MS method. The contents of different
flavonoids and phenolic acids have been determined. SFE extract, obtained at 20 MPa, contained
the highest content (11,561.84 µg/g) of analyzed total polyphenols, with predominant ellagic acid
(7492.53 µg/g). For the first time, Microbial Growth Inhibition Rates (MGIRs) were determined at
five different concentrations of pomegranate SFE extract against seven microorganisms. Minimal
Inhibitory Concentration (MIC90) was determined as 2.7 mg/mL of SFE pomegranate peel extract in
the case of five different Gram-negative and Gram-positive bacteria.
Ključne besede: antimicrobial activity, antioxidants, bioactive compounds, extraction, LC-MS/MS, phenolics, phytochemistry, pomegranate, Punica Granatum L., secondary metabolites
Objavljeno v DKUM: 24.03.2025; Ogledov: 0; Prenosov: 6
Celotno besedilo (2,02 MB)
Gradivo ima več datotek! Več... |
7. Improved relation extraction through key phrase identification using community detection on dependency treesShuang Liu, Xunqin Chen, Jiana Meng, Niko Lukač, 2025, izvirni znanstveni članek Opis: A method for extracting relations from sentences by utilizing their dependency trees to identify key phrases is presented in this paper. Dependency trees are commonly used in natural language processing to represent the grammatical structure of a sentence, and this approach builds upon this representation to extract meaningful relations between phrases. Identifying key phrases is crucial in relation extraction as they often indicate the entities and actions involved in a relation. The method uses community detection algorithms on the dependency tree to identify groups of related words that form key phrases, such as subject-verb-object structures. The experiments on the Semeval-2010 task8 dataset and the TACRED dataset demonstrate that the proposed method outperforms existing baseline methods.
Ključne besede: community detection algorithms, dependency trees, relation extraction
Objavljeno v DKUM: 17.01.2025; Ogledov: 0; Prenosov: 5
Celotno besedilo (3,12 MB) |
8. Optimization of the supercritical extraction and decarboxylation process of industrial hempTaja Žitek Makoter, Petra Kotnik, Teo Makoter, Željko Knez, Maša Knez Marevci, 2025, izvirni znanstveni članek Opis: Cannabinoids are the main active ingredients of Cannabis sativa L., obtained by a suitable extraction method and decarboxylation, in which the acid forms of the cannabinoids are converted into active forms. Supercritical CO2 extraction method and decarboxylation process were optimized for 7 cannabinoids (CBD, CBDA, THC, THCA, CBGA, CBN and CBC). The optimal extraction conditions for all cannabinoids were determined at a temperature of 60 °C and a pressure of 300 bar to 550 bar. At higher temperatures and reaction times, a significant depletion of neutral cannabinoids was observed. The optimal conditions of the decarboxylation process varied depending on the component. For the CBD component, which could be measured in the highest concentrations (560 mg/g), the optimal conditions were 140 °C and 10 min. The study can be an important model for predicting the behavior of cannabinoids under certain parameters.
Ključne besede: supercritical fluids, extraction, cannabinoids, decarboxilation, Cannabis sativa I
Objavljeno v DKUM: 08.01.2025; Ogledov: 0; Prenosov: 7
Celotno besedilo (10,15 MB) |
9. |
10. Identification and content of astaxanthin and its esters from microalgae Haematococcus pluvialis by HPLC-DAD and LC-QTOF-MS after extraction with various solventsBiljana Todorović, Jaša Veno Grujić, Andreja Urbanek Krajnc, Roman Kranvogl, Jana Ambrožič-Dolinšek, 2021, izvirni znanstveni članek Opis: Haematococcus pluvialis, a unicellular green microalga that produces a secondary metabolite under stress conditions, bears one of the most potent antioxidants, namely xanthophyll astaxanthin. The aim of our study was to determine the content of astaxanthin and its esterified forms using three different solvents—methyl tert-butyl ether (MTBE), hexane isopropanol (HEX -IPA) and acetone (ACE)—and to identify them by using high performance liquid chromatography coupled with diode array detection and the quadrupole time-of-flight mass spectrometry (HPLC-DAD and LC-QTOF-MS) technique. We identified eleven astaxanthin monoesters, which accounted for 78.8% of the total astaxanthin pool, six astaxanthin diesters (20.5% of total), while free astaxanthin represented the smallest fraction (0.7%). Astaxanthin monoesters (C16:2, C16:1, C16:0), which were the major bioactive compounds in the H. pluvialis samples studied, ranged from 10.2 to 11.8 mg g−1 DW. Astaxanthin diesters (C18:4/C18:3, C18:1/C18:3) were detected in the range between 2.3 and 2.6 mg g−1 DW. All three solvents were found to be effective for extraction, but MTBE and hexane-isopropanol extracted the greatest amount of free bioactive astaxanthin. Furthermore, MTBE extracted more low-chain astaxanthin monoesters (C16), and hexane-isopropanol extracted more long-chain monoesters (C18 and above) and more diesters. We can conclude that MTBE is the solvent of choice for the extraction of monoesters and hexane-isopropanol for diesters.
Ključne besede: astaxanthin diesters, astaxanthin monoesters, carotenoids, Hematococcus pluvialis, astaxanthin diesters, solvent extraction
Objavljeno v DKUM: 22.11.2024; Ogledov: 0; Prenosov: 19
Celotno besedilo (1,83 MB)
Gradivo ima več datotek! Več... |
