sv.
- MeSH
- Medical Informatics MeSH
- Neural Networks, Computer MeSH
- Publication type
- Periodical MeSH
- Conspectus
- Lékařské vědy. Lékařství
- NML Fields
- lékařská informatika
elektronický časopis
- Conspectus
- Patologie. Klinická medicína
- NML Fields
- neurovědy
- neurologie
- NML Publication type
- elektronické časopisy
sv.
- MeSH
- Neural Networks, Computer MeSH
- Publication type
- Periodical MeSH
- Conspectus
- Automatizační a řídicí technika
- NML Fields
- neurovědy
- technika
BACKGROUND: The recent big data revolution in Genomics, coupled with the emergence of Deep Learning as a set of powerful machine learning methods, has shifted the standard practices of machine learning for Genomics. Even though Deep Learning methods such as Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) are becoming widespread in Genomics, developing and training such models is outside the ability of most researchers in the field. RESULTS: Here we present ENNGene-Easy Neural Network model building tool for Genomics. This tool simplifies training of custom CNN or hybrid CNN-RNN models on genomic data via an easy-to-use Graphical User Interface. ENNGene allows multiple input branches, including sequence, evolutionary conservation, and secondary structure, and performs all the necessary preprocessing steps, allowing simple input such as genomic coordinates. The network architecture is selected and fully customized by the user, from the number and types of the layers to each layer's precise set-up. ENNGene then deals with all steps of training and evaluation of the model, exporting valuable metrics such as multi-class ROC and precision-recall curve plots or TensorBoard log files. To facilitate interpretation of the predicted results, we deploy Integrated Gradients, providing the user with a graphical representation of an attribution level of each input position. To showcase the usage of ENNGene, we train multiple models on the RBP24 dataset, quickly reaching the state of the art while improving the performance on more than half of the proteins by including the evolutionary conservation score and tuning the network per protein. CONCLUSIONS: As the role of DL in big data analysis in the near future is indisputable, it is important to make it available for a broader range of researchers. We believe that an easy-to-use tool such as ENNGene can allow Genomics researchers without a background in Computational Sciences to harness the power of DL to gain better insights into and extract important information from the large amounts of data available in the field.
- MeSH
- Genomics MeSH
- Neural Networks, Computer * MeSH
- Protein Structure, Secondary MeSH
- Machine Learning * MeSH
- Publication type
- Journal Article MeSH
This article focuses on the development of algorithms for a smart neurorehabilitation system, whose core is made up of artificial neural networks. The authors of the article have proposed a completely unique transfer of ACE-R results to the CHC model. This unique approach allows for the saturation of the CHC model domains according to modified ACE-R factor analysis. The outputs of the proposed algorithm thus enable the automatic creation of a personalized and optimized neurorehabilitation plan for individual patients to train their cognitive functions. A set of tasks in 6 levels of difficulty (level 1 to level 6) was designed for each of the nine CHC model domains. For each patient, the results of the ACE-R screening helped deter-mine the specific CHC domains to be rehabilitated, as well as the initial gaming level for rehabilitation in each domain. The proposed artificial neural network algorithm was adapted to real data from 703 patients. Experimental outputs were compared to the outputs of the initially designed fuzzy expert system, which was trained on the same real data, and all outputs from both systems were statistically evaluated against expert conclusions that were available. It is evident from the conducted experimental study that the smart neurorehabilitation system using artificial neural networks achieved significantly better results than the neurorehabilitation system whose core is a fuzzy expert system. Both algorithms are implemented into a comprehensive neurorehabilitation portal (Eddie), which was supported by a research project from the Technology Agency of the Czech Republic.
- MeSH
- Algorithms MeSH
- Expert Systems * MeSH
- Fuzzy Logic MeSH
- Humans MeSH
- Neural Networks, Computer MeSH
- Neurological Rehabilitation * MeSH
- Check Tag
- Humans MeSH
- Publication type
- Journal Article MeSH
- Research Support, Non-U.S. Gov't MeSH
Podám trombocytárního koncentrátu, jako prevence vzniku krvácivých stavu u pacientů s iatrogenni trombocytopenii je pro lékaře zpravidla spojeno s velmi obtížným rozíiodovánim. Podá-li trombocytárni koncentrát, výrazně sníti riziko závažných krvácivých stavů, avšak aloimunizuje nemocného a sníží tak účinnost každého dalšího náplavu trombocytů. Nepodá-li trombocytární koncentrát, k aloimunizaci sice nedojde, ale zvyšuje se riziko závažných krvácivých projevů, které mohou skončit i letálně. Z uvedeného vyplývá, že rozhodnutí o podání či nepodáni trombocytárniho koncentrálu je vždy velmi závažným krokem a chybná volba může mít pro další pacientův osud závažné důsledky. V tomto příspěvku je popsán způsob výběru markerů, ťj. ze statistického hlediska důležitých parametrů, které měly vliv na krvácení pacienta v dosud známých přípádech pomocí metody GUHA Tyto parametry budou použity pro trénováni vícevrstvé neuronové sítě, která bude sloužit pro predikci krvácivých stavů a bude základem systému pro podporu rozhodování.
Thrombocytopenia and bleeding are dangerous complications in the treatment of hematologic malignancies. Therapy and prophylaxis of bleeding is based only on administration of platelet transfusion. The main adverse effect of this therapy is refractoriness and lowered effect of the subsequent transfusion. Physician's decision whether to administer platelet transfusion or not is based on two facts: 1. Estimation of bleeding risks (80 % of decision). 2. Estimation of refractoriness development risk (20 % of decision). We are solving estimation of bleeding risk (item 1) in oi/r decision support system. We have been searching for significant factors influencing bleeding in the beginning. We have completed database of 22 patients, and 1807 hospitalising days. By means of method GUHA, which generates hypothesis type: „assuming factors A, B, C... Is (Is not) present then bleeding occurs (does not occur)".
- MeSH
- Biomarkers MeSH
- Immunization methods MeSH
- Hemorrhage prevention & control MeSH
- Neural Networks, Computer MeSH
- Thrombocytopenia drug therapy MeSH
- Publication type
- Review MeSH
Optimization of neural network topology, weights and neuron transfer functions for given data set and problem is not an easy task. In this article, we focus primarily on building optimal feed-forward neural network classifier for i.i.d. data sets. We apply meta-learning principles to the neural network structure and function optimization. We show that diversity promotion, ensembling, self-organization and induction are beneficial for the problem. We combine several different neuron types trained by various optimization algorithms to build a supervised feed-forward neural network called Group of Adaptive Models Evolution (GAME). The approach was tested on a large number of benchmark data sets. The experiments show that the combination of different optimization algorithms in the network is the best choice when the performance is averaged over several real-world problems.
The purpose of the study is to determine radon-prone areas in the Czech Republic based on the measurements of indoor radon concentration and independent predictors (rock type and permeability of the bedrock, gamma dose rate, GPS coordinates and the average age of family houses). The relationship between the mean observed indoor radon concentrations in monitored areas (∼22% municipalities) and the independent predictors was modelled using a bagged neural network. Levels of mean indoor radon concentration in the unmonitored areas were predicted using the bagged neural network model fitted for the monitored areas. The propensity to increased indoor radon was determined by estimated probability of exceeding the action level of 300Bq/m(3).
- MeSH
- Radiation Monitoring * MeSH
- Neural Networks, Computer * MeSH
- Air Pollutants, Radioactive analysis MeSH
- Air Pollution, Radioactive statistics & numerical data MeSH
- Radon analysis MeSH
- Models, Theoretical MeSH
- Air Pollution, Indoor statistics & numerical data MeSH
- Publication type
- Journal Article MeSH
- Geographicals
- Czech Republic MeSH
