13 research outputs found
Implementation of machine learning techniques for the prediction of COVID-19 mortality. Preliminary results
No full textUniversitatea de Stat de Medicină şi Farmacie „Nicolae Testemiţanu”, Chişinău, Republica MoldovaIntroducere. Pandemia COVID-19 a reprezentat o provocare majoră pentru sistemele de sănătate la nivel global. Progre sele în domeniul învățării automate (ML) oferă oportunități în gestionarea pacienților de la diagnostic și prognoză până la tratament personalizat și alocarea resurselor critice. În special, utilizarea tehnicilor de bază de învățare automată se dezvoltă rapid în predicția mortalității COVID-19, deoarece un model de predicție a mortalității ar putea fi rapid și eficient în luarea deciziilor clinice pentru pacienții cu risc iminent de deces. Integrarea acestor tehnologii în practica medicală poate transforma modul în care ar putea fi gestionate viitoarele pandemii. Actualmente, nu există un scor universal de predicție a mortalității, abordarea învățării au tomate fiind folosită mai puțin. Scopul lucrării. Elaborarea modelelor predictive pentru identificarea pacienților cu risc major de deces, bazate pe învățarea automată pentru stratificarea pacienților și optimizarea managementul clinic pacienților COVID-19 în UTI. Material și metode. Studiu interdisciplinar retrospectiv analitic de cohortă unicentric. Populația selectată (3200 pacienți) a fost pacienții internați în UTI din cadrul Institutului de Medicină Urgentă în perioada 2019-2022. Au fost elaborate 8 modele predictive, respondenții fiind divizați în lotul de antrenare pentru elab orarea modelelor (80%) și lotul de testare pentru a evalua capacitățile modelului de a prezice rezultatul cercetat (20%), toate modelele fiind aplicate fără hiperparametri în plus. Rezultate. Rata mortalității a fost estimată la un nivel de 30%. Algoritmul “support vector machine” a prezentat metrici optime având acuratețea estimată la nivel de 62.1%. Concluzii. Datele preliminare obținute permit de a conside ra această direcție ca fiind de perspectivă. Însă, este absolut necesar de a îmbunătăți metricele estimate prin aplicarea hiperparametrilor.Background. The COVID-19 pandemic has represented a major challenge for health systems globally. Advances in the field of machine learning (ML) offer opportunities to manage patients from diagnosis and prognosis to personalized treatment and the allocation of critical resources. In particular, the use of basic machine learning techniques is rapidly developing in the prediction of COVID-19 mortality, as a mortality prediction model could be fast and efficient in clinical decision-making for patients at imminent risk of death. Integrating these technologies into medical practice can transform the way future pandemics and other complex diseases are managed. Currently, there is no universal mor tality prediction score, so the machine learning approach is being used less. Objective of the study. Development of predictive models for identifying patients at high risk of dy ing, based on machine learning for patient stratification and optimizing the clinical management of COVID-19 patients in the ICU. Material and methods. Interdisciplinary retrospective analytic single-center cohort study. The selected population (3200 patients) were the patients admitted to the Intensive Care Unit of the Institute of Emergency Med icine during 2019-2022. 8 predictive models were developed, the respondents being divided into the training group to develop the models (80%) and the test group to evaluate the capabilities of the model to predict the researched outcome (20%), all models being applied without addition al hyperparameters. Results. The mortality rate was estimated at 30%. The “support vector machine” algorithm had presented optimal metrics with an estimated accuracy of 62.1%. Conclusion. The obtained preliminary data allow us to consider this direction as perspective. However, it is absolutely necessary to improve the estimated metrics by applying hyperparameter
Prognostic value of D-dimers in patients with COVID-19: narrative synthesis
No full textIntroduction. Contemporary researchers have suggested and demonstrated the hypothesis that the elevated level of D-dimers, which is a valuable marker of coagulation and fibrinolysis activation, can predict the severity of COVID-19, pulmonary complications, and thromboembolic events before they occur.
Material and methods. The bibliographic resources were analyzed and selected from databases such as PubMed, Hinari, SpringerLink, and Google Search using keywords such as “COVID-19,” “SARS-CoV-2,” “coronavirus,” “D-dimers,” “biomarkers,” and “severity prediction,” which were used in various combinations to maximize search efficiency. Therefore, the manuscript contains 51 representative articles for the purpose of this synthesis article.
Results. The D-dimer levels are significantly higher in patients with severe forms of COVID-19 compared to those with non-severe forms, in patients with acute respiratory distress syndrome compared to those without acute respiratory distress syndrome, and in deceased patients compared to those who have survived. D-dimers positively correlate with the degree of severity and the increased risk of progression to severe disease, inversely proportional to the survival rate. They can predict prognosis, determine therapeutic strategies, prevent complications, positively influence the disease’s course, and monitor the prognosis.
Conclusions. D-dimers should be used as a pre-radiographic screening tool as early as possible after admission and as an indicator for risk stratification of venous thromboembolism in hospitalized patients with COVID-19. Based on the increase in D-dimer levels, adjusting therapeutic doses of anticoagulants is more beneficial for patients compared to administering prophylactic doses
Prognostic value of D-dimers in patients with COVID-19: narrative synthesis
No full textIntroduction. Contemporary researchers have suggested and demonstrated the hypothesis that the elevated level of D-dimers, which is a valuable marker of coagulation and fibrinolysis activation, can predict the severity of COVID-19, pulmonary complications, and thromboembolic events before they occur.
Material and methods. The bibliographic resources were analyzed and selected from databases such as PubMed, Hinari, SpringerLink, and Google Search using keywords such as “COVID-19,” “SARS-CoV-2,” “coronavirus,” “D-dimers,” “biomarkers,” and “severity prediction,” which were used in various combinations to maximize search efficiency. Therefore, the manuscript contains 51 representative articles for the purpose of this synthesis article.
Results. The D-dimer levels are significantly higher in patients with severe forms of COVID-19 compared to those with non-severe forms, in patients with acute respiratory distress syndrome compared to those without acute respiratory distress syndrome, and in deceased patients compared to those who have survived. D-dimers positively correlate with the degree of severity and the increased risk of progression to severe disease, inversely proportional to the survival rate. They can predict prognosis, determine therapeutic strategies, prevent complications, positively influence the disease’s course, and monitor the prognosis.
Conclusions. D-dimers should be used as a pre-radiographic screening tool as early as possible after admission and as an indicator for risk stratification of venous thromboembolism in hospitalized patients with COVID-19. Based on the increase in D-dimer levels, adjusting therapeutic doses of anticoagulants is more beneficial for patients compared to administering prophylactic doses
Factors to consider when assessing the severity of COVID-19
No full textBackground: Analysis and evaluation of the multitude of parameters that impact and mirror clinical evolution of COVID-19 infection. Narrative literature review type of study. Bibliographic search of the PubMed database, applying the keywords: ”SARS-CoV-2”, ”COVID-19”, ”risk score”, ”laboratory
parameters”, ”pathophysiology”, ”cytokine storm”, ”imaging evaluation”, “outcomes”, “clinical evolution”, which were combined with each other. There
were selected English-language publications, in extenso, published in recognized journals from March 2020. Priority in the analysis was given to articles
of critical synthesis of literature, randomized studies, those with large samples of patients. One of the clinically important symptoms that reflects severe
or critical clinical evolution is persistent fever during the time. The presence of comorbidities, especially associated with obesity, represents a high risk of
severe evolution. Proinflammatory, prothrombotic and systemic endothelial damage processes are represented by changes in platelet count, lymphocytes,
neutrophil / lymphocyte ratio, C-reactive protein, D-dimers, fibrinogen, procalcitonin, urea, creatinine, ALS, AST, interleukin-6 and serum ferritin. Bacterial and fungal infections negatively influence clinical evolution. Common prediction scores have low value in COVID-19 patients and need adaptation.
Imaging evaluation identifies the type of lung injury and correlates with the severity degree and outcome.
Conclusions: COVID-19 disease caused by SARS-CoV-2 virus includes a multitude of pathophysiological changes that through its mechanism represent
a systemic nosology. The complete analysis of all the factors and parameters that can influence its clinical evolution is a basic component of the decisionmaking steps and treatment approach
Artificial intelligence-based techniques for predicting outcomes in COVID-19 patients
No full textIntroduction. Currently, extensive research has shown that almost all published prediction models are poorly studied
and have significant limitations, leading to their predictive performance often being overestimated. Additionally, there is
still no universally accepted scoring system, primarily due to the need for adaptation to heterogeneous patient samples
(including patient numbers, clinical profiles, and risk factors) and/or ongoing differences in the organization of healthcare
systems across various countries.
Materials and methods. This is a narrative literature review. A bibliographic search was conducted in the PubMed, Hinari,
SpringerLink, National Center for Biotechnology Information, and Medline databases. Articles published between 2000
and 2024 were selected based on keyword combinations such as “artificial intelligence”, “prediction model”, “algorithm”,
“machine learning”, and “COVID-19”. Information on machine learning predictive models was selected and processed to
identify characteristics that can be used to predict diagnosis, severity, length of hospital stay, ICU admission, treatment,
vaccination, and mortality in COVID-19 patients. After processing the data according to the search criteria, 125 full-text
articles were identified. The final bibliography includes 52 relevant sources, which were considered representative of the
literature on this synthesis article topic.
Results. Artificial intelligence techniques are increasingly being used to predict outcomes in COVID-19 patients, particularly
in estimating mortality among individuals infected with SARS-CoV-2, which can rapidly and effectively support clinical
decision-making. According to the analysis of multiple studies, strong predictors of mortality in COVID-19 patients include
advanced age, male gender, comorbidities, reduced levels of calcium, albumin, red blood cells, and oxygen saturation, as
well as lymphopenia, elevated blood urea nitrogen, creatinine, lactate dehydrogenase, D-dimers, neutrophils, interleukin-6,
procalcitonin, bilirubin, ferritin, aspartate aminotransferase, and troponin.
Conclusions. Artificial intelligence techniques provide potential advantages over conventional assessment methods. The
information obtained from machine learning and deep learning algorithms, including easily accessible and interpretable
data, can assist healthcare workers in making accurate decisions for the appropriate and timely care of COVID- 19 patients.
This can improve patient outcomes, reduce the burden on healthcare systems, and ultimately decrease mortality rates
Anxiety, but not pain catastrophizing, represents a risk factor for severe acute postoperative pain: a prospective, observational, cohort study
No full textState University of Medicine and Pharmacy “Nicolae Testemitanu”, Chisinau, Republic of Moldova
National Scientific and Practical Centre of Emergency Medicine, Chisinau, Republic of MoldovaAbstract
Introduction: The prevalence of severe acute postoperative pain (SAPP), i.e. pain intensity > 5/10
measured with Numeric Rating Scale (NRS), is still high, 24-46% in Western European countries and 64%
in Republic of Moldova.
Objective of the study: We tested the hypothesis that anxiety and pain catastrophizing perception
(interpreted as hypervigilance) represent risk factors for SAPP.
Materials and methods: 176 patients scheduled for abdominal surgery under general anaesthesia
were enrolled in this study, after approval by the University’s Research Ethics Committee and after obtaining
patient written informed consent. Preoperatively, all patients filled a Pain Catastrophizing Scale (PCS)
questionnaire and self-assessed the anxiety level on a numeric rating scale that was bounded by 0 (denoting
no anxiety) and 10 (denoting maximal imaginable anxiety). Duration of surgery, intraoperative administration
of fentanyl and pain intensity at 24 hours postoperatively on NRS was also recorded. Statistical analysis
comprised the following tests: odds ratio (OR), relative risk (RR), positive and negative predictive values
(PPV and NPV, respectively), likelihood ratio, receiver operating characteristic (ROC) curves, and Pearson
correlation test.
Results: “Hypervigilant” patients did not show an increased risk for SAPP based on histogram calculations
(OR = 1.51 [95CI = 0.62-3.65], p = 0.39). However, based on ROC curve calculations (OR = 2.34 [1.13-
4.83], p = 0.0029), these patients showed a risk for SAPP. On average, anxiety determined a fivefold
increase of the SAPP risk (OR = 5.1 [95CI = 1.44-18.50], p = 0.011). Intraoperative fentanyl consumption
had a weak but significant correlation with pain intensity at 24 h postoperatively (Pearson r = 0.26; p =
0.0008). Surgery duration did not correlate with pain intensity (Pearson r = -0.10; p = 0.46).
Conclusion: Anxiety, but not pain catastrophizing, represents a risk factor for SAPP. Intraoperative
fentanyl consumption had a weak correlation with postoperative pain intensit
Predictable severity biomarkers in Covid-19
No full textIntroduction: The recorded studies suggest that there is clear evidence-based association between various laboratory biomarkers and COVID-19
disease severity. These marker levels reflect the intensity of the cytokine-mediated hyperinflammatory response, which is strongly associated with a poor
outcome of SARS-CoV-2 infection.
Conclusions: C-reactive protein is not only a systemic inflammatory marker, but also an important regulator of inflammatory processes. The level
of this protein is positively correlated and can be widely used to predict the severity, prognosis and mortality in COVID-19 patients, additionally to
vital signs monitoring, supportive care, oxygen therapy, ventilation and circulatory support. Procalcitonin is an indicator of disease severity, which
can facilitate timely clinical decision-making, and determination of procalcitonin levels during COVID-19 patients’ follow-up, as well as being used
in assessing risk, predicting prognosis, and improving patient survival. The assessment of hematological laboratory parameters upon admission,
which help in differentiating between severe and non-severe cases, high-risk and low-risk cases of mortality, allows raising awareness, monitoring and
timely treatment of patients with COVID-19, as well as their early improvement of clinical condition. Inflammatory biochemical and hemocytometric
measures are feasible, easily interpretable, and widely available biomarkers in most healthcare settings, favorable for being used in treatment and severity
determination, in predicting clinical outcomes, and in the prognosis of patients with COVID-19. However, the assessment of the accuracy of these
biomarkers needs to be determined in further more relevant worldwide studies, showing a more precise design, more accurate performance, and having
larger sample sizes
IDENTIFICATION OF RISK FACTORS FOR POSTOPERATIVE ACUTE SEVERE PAIN IN ABDOMINAL SURGERY.
No full textCulegere de Rezumate Stiinţifice a Congresului SRATI 2012:
AL 38-LEA CONGRES AL SOCIETĂŢII ROMÂNE DE ANESTEZIE ŞI TERAPIE INTENSIVĂ;
AL 6-LEA CONGRES ROMÂNO - FRANCEZ DE ANESTEZIE ŞI TERAPIE INTENSIVĂ;
AL 4-LEA SIMPOZION ROMÂNO - ISRAELIAN DE ACTUALITĂŢI ÎN ANESTEZIE ŞI TERAPIE INTENSIVĂ;
AL 11-LEA CONGRES AL ASISTENŢILOR DE ANESTEZIE ŞI TERAPIE INTENSIVĂ;
AL 10-LEA CONGRES AL SOCIETĂŢII ROMÂNE DE SEPSISIntroducere: În pofida măsurilor luate, prevalenţa durerii postoperatorii acute intense, DPOI (≥5/10 pe SVN) rămâne înaltă (24-
46% – în Europa de Vest şi 64% – în Republica Moldova). Strategiile preventive pentru DPOI trebuie să ia în consideraţie şi factorii
de risc.
Scopul lucrării: Identificarea factorilor de risc pentru DPOI după intervenţii pe abdomen (herniorafie, apendectomie, colecistectomie)
prin screening-ul unor condiţii pre- şi intraoperatorii suspecte.
Introduction: Despite recent acivements, the prevalence of postoperative acute severe pain, PASP (≥5/10, VNS) is high (24-46% –
in West European countries and 64% – in Republic of Moldova). Prevention strategies for PASP should take into account the risk
factors.
Goal of the Study: Identification of risk factors for PASP after abdominal surgery (hernioplasty, appendectomy, cholecystectomy) via
screening of some intra- and postoperative suspected conditions
Identificarea unor factori de risc pentru durerea postoperatorie acută intensă
No full textIMSP Institutul de Medicină Urgentă
Centrul Naţional Ştiinţifico-Practic de Medicină Urgentă, Chişinău, MoldovaRezumat
În pofida progresului în managementul durerii postoperatorii, prevalenţa pacienţilor cu durere postoperatorie intensă (DPOI) rămâne înaltă, de 24-46%. Pentru o calitate mai bună a managementului durerii postoperatorii, anumiţi factori de risc specifi ci trebuie luaţi în consideraţie. În acest scop, studiul nostru, efectuat pe 92 de pacienţi, a apreciat calitatea de factor de risc pentru următoarele entităţi: catastrofi smul durerii, depresia, durere preoperatorie, durata intervenţiei, anxietatea, intervenţie pe cicatrice preexistentă, consumul intranestezic de fentanil. În studiul nostru, calitatea de factor de risc a fost confi rmată doar pentru anxietate (OR=5,1; CI95=1,44-18,50, p<0,0011). De asemenea, a fost identifi cat un grad mediu de corelare dintre consumul intra-anestezic total de fentanil şi intensitatea durerii postoperatorii (r=0,34; p<0,013).
Despite advances in postoperative pain management field, the prevalence of patients with severe postoperative pain is still high, 24-46%. For a better result, the postoperative pain management should take into consideration some specific risk factors for severe postoperative pain. Early identification of the factors in patients at risk of postoperative pain will allow a more effective intervention and a better management. For this aim, in our study were included 92, in which were evaluated the risk factor capacity of the following entities: pain catastrofizing, depression, preoperative pain, duration of the surgical operation, anxiety, redux, intra-anesthetic fentanil consumption. In our study, anxiety and high intra-anesthetic fentanyl doses were the most common predictors for severe postoperative pain. Also, was found a moderate degree of correlation between the total intra-anesthetic fentanil consumption and postoperative pain intensity.
Несмотря на прогресс в лечении послеоперационной боли, распространенность сильной острой послеоперационной боли (СЩПБ) у пациентов остается высокой, 24-46%. Для улучшения качества послеоперационного обезболивания, нужно принять во внимание специфические факторы риска. С этой целью, мы про- вели исследование на 92 пациентах, дабы оценить являются ли факторами риска следующие: катастрофизм боли, депрессия, предоперационная боль, продолжительность операции, тревожности, операций проводимые на существующих уже послеоперационных шрамах, интра-операционное потребление фентанила. В результате исследования мы смогли доказать что тревожности является фактором риска (OR = 5,1, CI95 = 1,44 до 18,50, р <0,0011). Также была определена средняя степень корреляции между потреблением фентанила во время анестезий и послеоперационной интенсивностью боли (р = 0,34, р <0,013)
Intraoperative transfusion practices in Europe
No full text© 2016 The Author. Published by Oxford University Press on behalf of the British Journal of Anaesthesia.Background: Transfusion of allogeneic blood influences outcome after surgery. Despite widespread availability of transfusion guidelines, transfusion practices might vary among physicians, departments, hospitals and countries. Our aim was to determine the amount of packed red blood cells (pRBC) and blood products transfused intraoperatively, and to describe factors determining transfusion throughout Europe. Methods: We did a prospective observational cohort study enrolling 5803 patients in 126 European centres that received at least one pRBC unit intraoperatively, during a continuous three month period in 2013. Results: The overall intraoperative transfusion rate was 1.8%; 59% of transfusions were at least partially initiated as a result of a physiological transfusion trigger- mostly because of hypotension (55.4%) and/or tachycardia (30.7%). Haemoglobin (Hb)- based transfusion trigger alone initiated only 8.5% of transfusions. The Hb concentration [mean (sd)] just before transfusion was 8.1 (1.7) g dl-1 and increased to 9.8 (1.8) g dl-1 after transfusion. The mean number of intraoperatively transfused pRBC units was 2.5 (2.7) units (median 2). Conclusions: Although European Society of Anaesthesiology transfusion guidelines are moderately implemented in Europe with respect to Hb threshold for transfusion (7-9 g dl-1), there is still an urgent need for further educational efforts that focus on the number of pRBC units to be transfused at this threshold
