ChapterPDF Available

Development of modern systems for biomedical data analysis

Authors:

Abstract and Figures

The options of an intelligent analytical system for the analysis of heterogeneous clinical and biomedical data in order to improve the diagnosis of complex diseases are described. The software implementations of data processing and analysis have been developed − construction of classifiers, selection of informative features, processing of heterogeneous medical and biological data for research in the field of clinical medicine. The system applies effective optimization approaches to solving classification problems and modern software, allowing access to real data, located in cloud environments, and to use the developed tools on technical devices of different classes from a personal computer to a powerful cluster.
Content may be subject to copyright.
A preview of the PDF is not available
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
One of the main problems in functional genomics is the prediction of the unknown gene/protein functions. With the rapid increase of high-throughput technologies, the vast amount of biological data describing different aspects of cellular functioning became available and made it possible to use them as the additional information sources for function prediction and to improve their accuracy.In our research, we have described an approach to protein function prediction on the basis of integration of several biological datasets. Initially, each dataset is presented in the form of a graph (or network), where the nodes represent genes or their products and the edges represent physical, functional or chemical relationships between nodes. The integration process makes it possible to estimate the network importance for the prediction of a particular function taking into account the imbalance between the functional annotations, notably the disproportion between positively and negatively annotated proteins. The protein function prediction consists in applying the label propagation algorithm to the integrated biological network in order to annotate the unknown proteins or determine the new function to already known proteins. The comparative analysis of the prediction efficiency with several integration schemes shows the positive effect in terms of several performance measures.
Article
Full-text available
Background: Entrapment is a challenging and crucial factor in the prehospital setting. Few studies have addressed whether entrapment has an influence on on-scene treatment or on the following hospital course. Objectives: Here we aimed to investigate the influence of entrapment on prehospital management and on the hospital course of polytrauma patients. Methods: We performed a retrospective analysis of consecutive patients with an Injury Severity Score ≥16 and aged 16-65 years that were admitted between 2005 and 2013 to a Level I trauma center. Two groups were built: entrapped (E) and nonentrapped patients (nE). These groups were evaluated for multiple prehospital and clinical parameters, including on-scene time, prehospital interventions, and posttraumatic complications. Results: There were 310 patients (n = 194 no entrapment [Group nE], n = 116 with entrapment [Group E]) enrolled. The on-scene time was significantly longer in Group E than Group nE. Moreover, this group received a significantly higher volume of colloidal solution. Regarding the Injury Severity Score and Abbreviated Injury Scale (AIS), there were no significant differences between the groups, except for the AISextremities, which was significantly increased in Group E. The overall hospital stay and the initial theater time were significantly longer in Group E than Group nE. No significant differences were present for the occurrence of systemic inflammatory response syndrome, multiple organ dysfunction syndrome, and acute respiratory distress syndrome, nor for Acute Physiology and Chronic Health Evaluation II and estimated and final mortality. Conclusion: In polytraumatized patients, entrapment has a minor influence on the outcome and treatment in the prehospital and hospital setting when using physician-based air rescue. However, entrapped patients are prone to sustain more severe trauma to the extremities.
Conference Paper
Full-text available
The paper presents the approach to semi-supervised fuzzy clustering, based on the extended optimization function and the algorithm of the active constraints selection. The approach is tested on the artificial and real data sets. Clustering results, obtained by the proposed approach, are more accurate relative to the ground truth due to utilization of the additional information about the class labels in the most uncertain regions.
Article
Full-text available
This paper 1 presents two novel approaches to evolutionary design of the classifier ensemble. The first one presents the task of one-objective optimization of feature set partitioning together with feature weighting for the construction of the individual classifiers. The second approach deals with multi-objective optimization of classifier ensemble design. The proposed approaches have been tested on two data sets from the machine learning repository and one real data set on transient ischemic attack. The experiments show the advantages of the feature weighting in terms of classification accuracy when dealing with multivariate data sets and the possibility in one run of multi-objective genetic algorithm to get the non-dominated ensembles of different sizes and thereby skip the tedious process of iterative search for the best ensemble of fixed size.
Article
Full-text available
Background Uncertainty persists regarding the optimal ventilatory strategy in trauma patients developing acute respiratory distress syndrome (ARDS). This work aims to assess the effects of two mechanical ventilation strategies with high positive end-expiratory pressure (PEEP) in experimental ARDS following blunt chest trauma. Methods Twenty-six juvenile pigs were anesthetized, tracheotomized and mechanically ventilated. A contusion was applied to the right chest using a bolt-shot device. Ninety minutes after contusion, animals were randomized to two different ventilation modes, applied for 24 h: Twelve pigs received conventional pressure-controlled ventilation with moderately low tidal volumes (VT, 8 ml/kg) and empirically chosen high external PEEP (16cmH2O) and are referred to as the HP-CMV-group. The other group (n = 14) underwent high-frequency inverse-ratio pressure-controlled ventilation (HFPPV) involving respiratory rate of 65breaths · min−1, inspiratory-to-expiratory-ratio 2:1, development of intrinsic PEEP and recruitment maneuvers, compatible with the rationale of the Open Lung Concept. Hemodynamics, gas exchange and respiratory mechanics were monitored during 24 h. Computed tomography and histology were analyzed in subgroups. Results Comparing changes which occurred from randomization (90 min after chest trauma) over the 24-h treatment period, groups differed statistically significantly (all P values for group effect <0.001, General Linear Model analysis) for the following parameters (values are mean ± SD for randomization vs. 24-h): PaO2 (100 % O2) (HFPPV 186 ± 82 vs. 450 ± 59 mmHg; HP-CMV 249 ± 73 vs. 243 ± 81 mmHg), venous admixture (HFPPV 34 ± 9.8 vs. 11.2 ± 3.7 %; HP-CMV 33.9 ± 10.5 vs. 21.8 ± 7.2 %), PaCO2 (HFPPV 46.9 ± 6.8 vs. 33.1 ± 2.4 mmHg; HP-CMV 46.3 ± 11.9 vs. 59.7 ± 18.3 mmHg) and normally aerated lung mass (HFPPV 42.8 ± 11.8 vs. 74.6 ± 10.0 %; HP-CMV 40.7 ± 8.6 vs. 53.4 ± 11.6 %). Improvements occurring after recruitment in the HFPPV-group persisted throughout the study. Peak airway pressure and VT did not differ significantly. HFPPV animals had lower atelectasis and inflammation scores in gravity-dependent lung areas. Conclusions In this model of ARDS following unilateral blunt chest trauma, HFPPV ventilation improved respiratory function and fulfilled relevant ventilation endpoints for trauma patients, i.e. restoration of oxygenation and lung aeration while avoiding hypercapnia and respiratory acidosis.
Article
Background Penetrating thoracic injuries (PTIs) is a medicosurgical challenge for civilian and military trauma teams. In civilian European practice, PTIs are most likely due to stab wounds and mostly require a simple chest tube drainage. On the battlefield, combat casualties suffer severe injuries, caused by high-lethality wounding agents. The aim of this study was to analyse and compare the demographics, injury patterns, surgical management and clinical outcomes of civilian and military patients with PTIs. Methods All patients with PTIs admitted to a Level I Trauma Centre in France or to Role-2 facilities in war theatres between 1 January 2004 and 31 May 2016 were included. Combat casualties’ data were analysed from Role-2 medical charts. The hospital manages military casualties evacuated from war theatres who had already received primary surgical care, but also civilian patients issued from the Paris area. During the study period, French soldiers were deployed in Afghanistan, in West Africa and in the Sahelo-Saharan band since 2013. Results 52 civilian and 17 military patients were included. Main mechanisms of injury were stab wounds for civilian patients, and gunshot wounds and explosive fragments for military casualties. Military patients suffered more severe injuries and needed more thoracotomies. In total, 29 (33%) patients were unstable or in cardiac arrest on admission. Thoracic surgery was performed in 38 (55%) patients (25 thoracotomies and 13 thoracoscopies). Intrahospital mortality was 18.8%. Conclusion War PTIs are associated with extrathoracic injuries and higher mortality than PTIs in the French civilian area. In order to reduce the mortality of PTIs in combat, our study highlights the need to improve tactical en route care with transfusion capabilities and the deployment of forward surgical units closer to the combatants. In the civilian area, our results indicated that video-assisted thoracoscopic surgery is a reliable diagnostic and therapeutic technique for haemodynamically stable patients.
Article
In this work, to construct classifiers for two linearly inseparable sets, the problem of minimizing the margin of incorrect classification is formulated, approaches to achieving approximate solution, and calculation estimates of the optimal value for this problem, are considered. Results of computational experiments that compare proposed approaches with SVM are presented. The problem of identifying informative features for large-dimensional diagnostic applications is analyzed and algorithms for its solution are developed.
Article
Background: Veno-venous extra-corporeal membrane oxygenation (vv-ECMO) is an established salvage therapy for severe respiratory failure, and may provide an alternative form of treatment for trauma-induced acute respiratory distress syndrome (ARDS) when conventional treatments have failed. The need for systemic anticoagulation is a relative contraindication in patients with bleeding risks, especially in multi traumatic injury. Methods: We describe a case series of four trauma patients with ARDS who were managed with ECMO admitted to the Neuro Critical Care Unit (NCCU) at Addenbrooke's Hospital, Cambridge (UK), from January 2000 to January 2016. We performed a systematic review of the available literature in order to investigate the safety and efficacy of vv-ECMO in post-traumatic ARDS focusing on the use of different anticoagulation strategies and risk of bleeding in polytrauma patients. Results: Thirty-one patients were included. A heparin bolus was given in 16 cases. Eleven patients developed complications during treatment with ECMO with 3 cases of major bleeding. In all documented cases of bleeding a bolus and infusion of heparin was administered, aiming for an ACT target > 150 s. Two patients treated with heparin-free ECMO developed thromboembolic complications. Four patients died and death was never directly or indirectly related to use of ECMO. Conclusions: Vv-ECMO can be lifesaving in respiratory failure. Our experience and our literature review suggests that vv-ECMO should be considered as rescue treatment for the management of severe hypoxemic respiratory failure secondary to ARDS in trauma patients.In patients with a high risk of bleeding, the use of ECMO with no initial anticoagulation could be considered a valid option. In patients with a moderate risk of bleeding, use of a heparin infusion keeping an ACT target < 150s can be appropriate. Level of evidence: Level III - Systematic Review.
Article
Occupational inhalation of indium compounds can cause the so-called “indium lung disease”. Most affected individuals show pulmonary alveolar proteinosis (PAP) and fibrotic interstitial lung disease. In animal experiments, inhalation of indium tin oxide or indium oxide has been shown to cause lung damage. However, the mechanisms by which indium compounds lead to indium lung disease remain unknown. In this study, we constructed a mouse model of indium lung disease and analyzed gene expression in response to indium exposure. Indium oxide (In2O3, 10 mg/kg, primary particle size <100 nm) was administered intratracheally to C57BL/6 mice (male, 8 weeks of age) twice a week for 8 weeks. Four weeks after the final instillation, histopathological analysis exhibited periodic acid-Schiff positive material in the alveoli, characteristic of PAP. Comprehensive gene expression analysis by RNA-Seq, however, revealed expression of fibrosis-related genes, such as surfactant associated protein D, surfactant associated protein A1, mucin 1, and collagen type I and III, was significantly increased, indicating that fibrotic gene expression progresses in early phase of indium lung. These data supported the latest hypothesis that PAP occurs as an acute phase response and is replaced by fibrosis after long-term latency.