Search

Earlier we developed signal analysis for nystagmus measured from otoneurological patients suffering from vertigo and dizziness. It was based on three rotation directions of the eye: horizontal, vertical and torsional. However, nystagmus frequently appears only in two of the former directions. In order to enable two-dimensional analysis approach on the basis of various pairs from the original three, we designed and implemented an advanced method from our earlier one. These signals can be used to investigate and model the dysfunction of semicircular canals in the inner ear. Nystagmus has to be stimulated for healthy subjects, but in patients it can also be spontaneous. The method developed was tested with the signals of 30 otoneurological patients. [ABSTRACT FROM AUTHOR]

In this study, we examine the applicability of association rules for analysing high-dimensional data concerning age-related hearing impairment (ARHI). The ARHI data of the study contain hundreds of variables concerning phenotype, genotype and environmental factors. The number of association rules produced from the data is too large for manual exploration in the raw and furthermore, the rules are overlapping. Thus, the focus of our study is to develop an approach to cluster association rules into subsets and to summarise and represent the found rule subsets for easier exploration of rules. The results show that it is possible to efficiently extract rules representing interesting environmental factor-gene or gene-gene interactions. Finding suitable parameters for the association rule mining and the possibility to post-process the mined rules is essential. The developed approach facilitates rule exploration by grouping rules with items concerning the same phenomenon to the same subset and byrevealing overlapping rules. [ABSTRACT FROM AUTHOR]

The human postural control system is complex and it combines information from different sources. The most important information comes from vestibular, visual and proprioceptive senses. We studied the effects of removing the visual and proprioceptive information simultaneously. The force feedback from the ground was removed with vibrators attached on the musculus soleus of both calves. By using features of force platform signals, when the vibrators were on, the lengths of the swaying paths were four times the lengths when the vibrators were off. Our results show that it is possible to separate the effects of the visual and proprioceptive senses from that of vestibular sense, which is very useful for investigations of balance problems in otoneurology. This supports our future aim to classify between healthy subjects and different otoneurological patients with signal analysis and pattern recognition methods to be used for force platform signals. [ABSTRACT FROM AUTHOR]

Nystagmus needs to be stimulated for healthy subjects, but in patients it can also be spontaneous. By recording spontaneous nystagmus it is possible to reveal underlying disorders of the semicircular canals of the inner ear. We developed a signal analysis technique for this purpose and tested it with 28 otoneurological patients who had disorders in their semicircular canals. Our test results support the postulate how nystagmus should appear under these conditions and agree with results presented in the otoneurological literature. [ABSTRACT FROM AUTHOR]

A dataset including cases of six otoneurological diseases was analysed using machine learning methods to investigate the classification problem of these diseases and to compare the effectiveness of different methods for this data. Linear discriminant analysis was the best method and next multilayer perceptron neural networks provided that the data was input into a network in the form of principal components. Nearest neighbour searching, k-means clustering and Kohonen neural networks achieved almost as good results as the former, but decision trees slightly worse. Thus, these methods fared well, but Naïve Bayes rule could not be used since some data matrices were singular. Otoneurological cases subject to the six diseases given can be reliably distinguished. [ABSTRACT FROM AUTHOR]

We constructed a signal analysis and recognition system to compute revealing results from vestibulo-ocular reflex eye movements and their stimulation head movements of impulsive type that were generated as passive head movements with a special device designed by us for this purpose. Further, we implemented perceptron neural networks to separate healthy subjects from patients suffering from dysfunction of vestibulo-ocular reflex in either ear. We gained high classification accuracies with this method ready for routine use. [ABSTRACT FROM AUTHOR]

The present aim was to describe the first phase attempts to recognise voluntarily produced changes in electromyographic signals measured from two facial muscles. Thirty subjects voluntarily activated two facial muscles, corrugator supercilii and zygomaticus major. We designed a neural network based recognition system that screened out muscle activations from the electromyographic signals. When several subjects were tested according to the same test protocol, the neural network system was able to correctly recognise more than 95 % of all muscle activations. This is a promising result and we shall next proceed to modify the system for real-time functioning and then design its utilisation for various multimodal human-computer interaction techniques. The subsequent phase in the future will be the interaction backwards: when a computer program first recognised the use of the facial muscles, it will then follow the instructions given by the user. For instance, by using the facial muscles the subject could select or activate objects on the computer screen. This would be one of the opportunities that we develop to help, e.g., disabled persons, who are unable to use their hands. [ABSTRACT FROM AUTHOR]

We studied how the splitting of a multi-class classification problem into multiple binary classification tasks, like One-vs-One (OVO) and One-vs-All (OVA), affects the predictive accuracy of disease classes. Classifiers were tested with an otoneurological data using 10-fold cross-validation 10 times with k-Nearest Neighbour (k-NN) method and Support Vector Machines (SVM). The results showed that the use of multiple binary classifiers improves the classification accuracies of disease classes compared to one multi-class classifier. In general, OVO classifiers worked out better with this data than OVA classifiers. Especially, the OVO with k-NN yielded the highest total classification accuracies.

Nystagmus needs to be stimulated for healthy subjects, but in patients it can also be spontaneous. By recording spontaneous nystagmus it is possible to reveal underlying disorders of the semicircular canals of the inner ear. We developed a signal analysis technique for this purpose and tested it with 28 otoneurological patients who had disorders in their semicircular canals. Our test results support the postulate how nystagmus should appear under these conditions and agree with results presented in the otoneurological literature.

The human postural control system is complex and it combines information from different sources. The most important information comes from vestibular, visual and proprioceptive senses. We studied the effects of removing the visual and proprioceptive information simultaneously. The force feedback from the ground was removed with vibrators attached on the musculus soleus of both calves. By using features of force platform signals, when the vibrators were on, the lengths of the swaying paths were four times the lengths when the vibrators were off. Our results show that it is possible to separate the effects of the visual and proprioceptive senses from that of vestibular sense, which is very useful for investigations of balance problems in otoneurology. This supports our future aim to classify between healthy subjects and different otoneurological patients with signal analysis and pattern recognition methods to be used for force platform signals.

A dataset including cases of six otoneurological diseases was analysed using machine learning methods to investigate the classification problem of these diseases and to compare the effectiveness of different methods for this data. Linear discriminant analysis was the best method and next multilayer perceptron neural networks provided that the data was input into a network in the form of principal components. Nearest neighbour searching, k-means clustering and Kohonen neural networks achieved almost as good results as the former, but decision trees slightly worse. Thus, these methods fared well, but Naïve Bayes rule could not be used since some data matrices were singular. Otoneurological cases subject to the six diseases given can be reliably distinguished.

We constructed a signal analysis and recognition system to compute revealing results from vestibulo-ocular reflex eye movements and their stimulation head movements of impulsive type that were generated as passive head movements with a special device designed by us for this purpose. Further, we implemented perceptron neural networks to separate healthy subjects from patients suffering from dysfunction of vestibulo-ocular reflex in either ear. We gained high classification accuracies with this method ready for routine use.

Word-finding difficulty (anomia) is the most common linguistic deficit in dementia. It is often measured by picture naming tasks as naming a picture taps all the major processes in word production, i.e., activation of a concept, retrieval of lexical-semantic information on that concept, retrieval of the corresponding word form and articulation. Naming and naming errors have extensively been simulated by neural network models of lexicalization (see e.g. [1,2]). A common feature of these models is that they are static, i.e. non-learning. However, naming is a dynamic process that changes as a function of normal learning or re-learning after neural damage. These important patterns cannot be caught by the static models of lexicalization. Therefore we have developed a learning model of lexicalization based on multi-layer-perceptron (MLP) neural networks. We tested the model by fitting it to the naming data of 22 Finnish-speaking dementia patients and 19 neurologically intact control subjects. The tests showed an excellent fit between the model's and the subjects naming response distributions. Thus our model seems be suitable to simulate naming disorders of dementia patients.

Proximity functions evaluate distances or similarities between objects. Unlike the Euclidean distance, heterogeneous proximity functions process variables differently according to their scale. The correct evaluation of nominal variables, whose values are unordered, is especially important. We compared five heterogeneous functions with the Euclidean distance to study whether functions sensitive to scale are better than a function assuming the same scale. In addition, we were interested of the relative performance of the five heterogeneous functions. The performance of the functions was measured with a nearest neighbor classifier that was applied to 12 medical data sets characterized with different scales. Unexpectedly, the performance of heterogeneous functions did not differ significantly from that of the Euclidean distance. As expected, significant differences between the Heterogeneous Value Difference Metric (HVDM) and the four value-matching-based heterogeneous functions favored HVDM. Additional research is needed to explain why heterogeneous functions did not outperform the Euclidean distance.

The present aim was to develop signal analysis methods to recognize muscle action potentials recorded with electromyography (EMG) from two facial muscle sites, corrugator supercilii (the muscle activated in frowning) and zygomaticus major (the muscle activated in smiling). Fourteen subjects produced volitional activations on both muscle sites, first, on a single muscle site activation basis and finally combinatory activations of both muscle sites. Wavelets and neural networks were used to analyse these voluntarily produced bursts of electric signals. Our results showed well over 90% recognition rates for all signal types. It is possible to utilize these communicative signals, for example, for multimodal human-computer interaction.

In the next study we consider two distance metrics that were presented in the machine leaming literature for mixed-type variables. We show that they are not really metrics, but pseudometrics. The problem arose from missing values. The metrics can be redefined to satisfy the metricity definition. Distance computation can then be performed reliably without a possibility that a distance between exactly similar patient cases would not be zero. We experimented with both procedures and so-called ignore mode using two medical data sets and observed that the redefined procedure was as good as the original approach as measured with the classification ability of one-nearest neighbour searching. It is noticeable that the described problem caused by missing values can be with any distance measure if its missing values are treated in the pessimistic manner as in the original version of these distance measures.

People relying much on vision in the control of posture are known to have an elevated risk of falling. Dependence on visual control is an important parameter in the diagnosis of balance disorders. We have previously shown that virtual reality methods can be used to produce visual stimuli that affect balance, but suitable stimuli need to be found. In this study the effect of six different virtual reality stimuli on the balance of 22 healthy test subjects was evaluated using force platform posturography. According to the tests two of the stimuli have a significant effect on balance.

We studied the nearest neighbour classification of patient cases with benign positional vertigo, Meniere's disease, sudden deafness, traumatic vertigo, vestibular neuritis, and vestibular schwannoma. The classification results were compared to the inference results obtained by an otoneurological expert system ONE whose inference mechanism somewhat resembles the classical nearest neighbour method. With respect to the predictive accuracy, the classification results of these two systems agreed. The best predictive accuracy for the expert system ONE was 79.7% and for the nearest neighbour method 80.5%. However, differences in the true positive rates for sudden deafness, traumatic vertigo, vestibular neuritis, and vestibular schwannoma were found. The nearest neighbour classification results will be used in the refinement of ONE's knowledge base.

The present study investigates the prediction of how successfully hearing returns to normal after an acute acoustic trauma caused by a firearm shot, an explosion or other impulse noise. Study material consisted of 113 conscripts who had suffered an acute acoustic trauma during their military service. A logistic regression model was formed to predict hearing recovery. From several possible variables at least five were found to be important among our dataset, which is still rather limited. However, more data will be acquired later in order to elaborate the research more widely by using other methods also.