Search

SGS20/214/OHK3/3T /37 ”Active learning of time series models”, the project No.FV −20696 “Personal health monitoring and assistive systems” The aim of this work is to assess the minimal technical requirements for using a simple RGB camera for motion sensing in a standard home environment. We experimentally verify the recording requirements for subsequent motion analysis. Our work contributes to the development of physical telerehabilitation without the need to use special HW and thus enable telerehabilitation for the general public, which is especially important during the COVID-19 lockdown. We have found out that such a system can work surprisingly well even with a low-cost camera in poor recording conditions and slow 3G internet connection.

Multi-robotics systems are currently subject of major interest in the robotics literature. In the leading journals can be found hundreds of articles, published in the last few years, concerning applications and theoretical studies of small groups maintaining in fixed formation (Fua et al., 2007; Kaminka et al., 2008) as well as swarms of thousands robots (Derenick & Spletzet, 2007; Daigle et. al, 2008). The large systems can be moreover represented by work published in (Peasgood et al., 2008) where collision free trajectories to reach individual goals are designed for 100 robots. The method using graph and spanning tree representation is developed for utilization in underground mine environment. In another example (Kloetzer & Belta,2007), a large swarm of robots is controlled using hierarchical abstractions. Inter-robot collision avoidance and environment containment are there guaranteed applying centralized communication architecture. Finally work presented in (Milutinovi & Lima, 2006) applies a Stochastic Hybrid Automation model for modeling and control of multi-agent population composed of a large number of agents. In this method probabilistic description of task allocation as well as distribution of the population over the work space is considered. As an example of common multi-robots application highway traffic coordination can be mentioned. In (Pallotino et al., 2007) is presented decentralized approach using traffic rules for control of tens vehicles. The method enables dynamically adding and removing of the vehicles and is based only on local communication which makes the algorithm scalable. Algorithms designed for smaller groups of robots are usually aimed at maintaining of vehicles in a predefined formation for the purpose of cooperative tasks accomplishing (as can be e.g. box pushing (Vig & Adams, 2006), load carrying (Tanner et al., 2003), snow shoveling (Saska et al., 2008) or aircraft as well as satellites cooperative mapping (Ren & Beard, 2003; Kang & Sparks, 2000). Another interesting application of formation driving is presented in (Fahimi, 2007) where autonomous boats are maintained in formations under sliding mode, which provides faster movement. The hot research topics in formations of autonomous robots, investigated nowadays, include e.g. data fusion: (Kaminka et al., 2008) represents the sensing capabilities using a monitoring multigraph. This approach allows the robots to adjust to sensory failures by switching of control graphs on-line. An application of data fusion can be cooperative localization of mobile formations: (Mourikis & Roumeliotis

The quality of life of the elderly is becoming more and more important. Prevention and early evaluation of falls in elderly people is a very important issue. Achieving the best comfort can be possible by contactless methods. The paper describes machine learning approach to a fall detection based on body posture classification using 8x8 pixel image acquired by Grid-EYE array temperature sensor. State-of-the-art computer vision tool - deep neural network - is used. The best result is achieved if classification into three or five classes is assumed. Even with such low resolution thermal image sensor, the final sensitivity and specificity of the class “laying” which corresponds to posture of a fallen person reaches 0.85 and 0.93, respectively.

Detection of artifacts in sleep electroencephalography (EEG) is one of the important tasks on the preprocessing step. Despite many algorithms of artifact detection developed through years, many of them lose their benefits in sleep EEG application. This study proposes a method of artifact detection based on a classification of quasi-stationary EEG epochs with random forest classifier. The method was tested on data of three sleep stages and pre-sleep wake EEG. Results showed 16% increase in $F_{1\,} $for the wake and 9%, 5% and 16% for different sleep stages in comparison to a baseline. All false detection at every presented sleep stage is investigated.

Background The classification of sleep signals is a subjective and time consuming task. A large number of automatic classifiers have been published in the past decade but a sleep community has no strong confidence to use them in clinical practice and still remains using a standard manual scoring according standardized rules. New method We developed a semi-supervised data-driven approach for objective and efficient evaluation of polysomnographic (PSG) data. The proposed algorithm finds a representative set of signal segments that are subsequently scored by a sleep neurologist. The remaining part of the recording is then automatically classified using these templates. Results The method was evaluated on 36 PSG recordings (18 chronic insomniacs, 18 healthy controls). We show a faster and objective evaluation of PSG data compared to the manual scoring that is over-performing automated classifiers (accuracy increases ∼14%). The classification results are comparable on both datasets. Comparison with existing method(s) The methodology that we propose has not yet been published in the area of sleep PSG data processing. The performance of our method is comparable to various published automated approaches (a typical published classification accuracy is ∼75–95%). The method allows the evaluation of PSG recordings in more general terms and across different recording devices and standards. Conclusions The proposed solution is not based on a single-purpose rules or heuristics and training model is not trained on other patient's sleep recordings. The method is applicable to wide range of similar tasks and various types of physiological signals.

Glycaemia prediction plays a vital role in preventing complications related to diabetes mellitus type 1, supporting physicians in their clinical decisions and motivating diabetics to improve their everyday life. Several algorithms, such as mathematical models or neural networks, have been proposed for blood glucose prediction. An approach of combining several glycaemia prediction models is proposed. The main idea of this framework is that the outcome of each prediction model becomes a new feature for a simple regressive model. This approach can be applied to combine any blood glycaemia prediction algorithms. As an example, the proposed method was used to combine an Autoregressive model with exogenous inputs, a Support Vector Regression model and an Extreme Learning Machine for regression model. The multiple-predictor was compared to these three prediction algorithms on the continuous glucose monitoring system and insulin pump readings of one type 1 diabetic patient for one month. The algorithms were evaluated in terms of root-mean-square error and Clarke error-grid analysis for 30, 45 and 60 min prediction horizons.

Energy consumed in buildings represents a challenge in the context of reduction of greenhouse gases emission. For this reason and due to the growing interest in operative costs reduction the energy used by buildings (tertiary and privates) for heating, ventilating, and air conditioning (HVAC) is even more investigated. Due to the nature of the energy consumption profile a predictive optimization method is one of the solution the scientific literature spreads even more. However optimization techniques need a good and reliable prediction of the variables of interest over a time horizon. This work focuses on methods to obtain a robust and reliable predictor based on artificial neural networks. For the optimization purposes the neural model predicts total heating energy consumption (gas), internal air temperature and aggregated thermal discomfort 12 h ahead. Training and testing data are simulated using a simulator based on heat, air and moisture model for building and systems evaluation (HAMBASE), by which a real office building was modeled. Influence of training data sample size and selection of predictor inputs is examined. Several combinations of early stopping condition and network complexity are tested for different training sample sizes. It is observed that the early stopping mechanism is crucial especially but not only for small training data, because it reliably overcomes overfitting problems. Surprisingly, relatively small networks were sufficient or performed best, although examined range of training sample covered up to five heating seasons. The use of a model tuning is thus supported by the results. Further, two strategies of selection of suitable input variables are demonstrated. While the input selection does not degrade the prediction performance, it is able to reduce the dimensionality and thus to save computational, communication, time, and data acquisition demands. The importance of inputs selection in HVAC modeling is thus pointed out and demonstrated.

This paper introduces the semi-automatic process using active learning methods which could improve the current state, where a human specialist has to annotate a multiple hours long polysomnographical record to sleep stages. This work is focused on the utilization of density-weighted methods of active learning, one of them turned out to be well-suited for this type of task. Moreover, we proposed several criteria for the comparison of active learning methods. The method saves more than 80% of expert’s annotation effort.

A complete real-time fall detection system is presented consisting of camera data acquisition, image processing, pattern recognition, fall alarming, and web interface. Classifiers are trained using only three input features extracted from each video frame using image processing. Among linear and quadratic Bayes, Parzen classifier and 3-nearest neighbors classifier, the last one performed best on a testing set from sensitivity, specificity and area under receiver operating characteristic point of view. Such frame classification is further used in a simple rule triggering the fall alarm process. The fall alarm tested in a real time scenario with 40 falls performed by four persons detected all the falls while having one false positive case.

Clustering is a popular analysis technique in a modern science full of unlabeled data, hidden dependencies and relations between elements in datasets. The presented study proposes a new hybrid hierarchical clustering method suitable for large datasets. It is based on the combination of effective simple methods. The proposed method was tested and compared with a widely used agglomerative clustering method. Two groups of datasets were used for testing. The first group contains data delivered from real biomedical data and related to a real problem of indication of sleep stages. The second group consists of artificially generated large data. Time, memory consumption, and mutual information were compared.

A multiple-steps ahead prediction of glucose level from real time continuous glucose monitoring system (RT-CGMS) device is presented. Both linear and nonlinear autoregressive models with exogenous inputs are used for the system identification. Insulin and nutritional income are used as the exogenous inputs. To better represent the dynamical character of those external factors, simple compartment models are used providing a signal of the influence of the insulin and nutrition. Those signals are used as inputs to regressive models. The main problem of adaptation of those compartment models to particular patient is solved using continuous particle swarm optimization algorithm. The proposed approach is demonstrated on data from type I diabetic patients with RT-CGMS and insulin pump. The results provide the first step of creation of future mobile application for decision support of type 1 diabetics.

The study is devoted to data processing methods in automatic sleep polysomnography (PSG) analysis. The idea is in using covariance matrices a carrier of a discriminative information. In the study, we are challenging with a problem of sleep stage classification. We are trying to solve that problem using spatial geometric analysis. For experiments, we took data from seven patients; data were recorded in National Institute of Mental Health. Artifact-free segments were extracted from the data. The covariance matrix was obtained for each segment. The classification was performed using a minimum distance to a class or in k-nearest-neighbor (KNN) method. A distance between objects was calculated using Riemannian Geometry. Classification methods were tested by cross-validation scheme. Using only covariance matrix of multimodal data and without additional information divided by frequency ranges, it is possible to classify sleep stages with high accuracy: the average accuracy for KNN is 0.929, for minimum distance to a class center it is only 0.816. Advantages of the method are working with data from different domains, adjustability to a different number of channels. Support: project No. 17-20480S of GACR, project “National Institute of Mental Health (NIMH-CZ),” Grant No. ED2.1.00/03.0078 and project No. LO1611.