Search

Arousals are considered one of the main causes of daytime sleepiness. They impede the proper flow of sleep cycles and cause weariness. Manual scoring of arousals is time-consuming, requires expert knowledge, and has high inter-scorer variability. A major difficulty in detecting arousals automatically is the existing variance across patients. Based on data mining techniques, we present a different approach to the automatic detection of arousals that overcomes the hurdle of differences in signal characteristics across patients. Offline we used a training-set of adult patients to define a set of general rules to detect arousals (termed meta-rules). This was done by analyzing the correlations between occurrences of arousals and the EEG, EMG, pulse and SaO2 signals as follows: (1) each signal was mathematically projected into several spaces (termed projected-signals); (2) from each such projected-signal, the algorithm extracted time points that indicated meaningful changes (termed critical-points); (3) data mining techniques were applied to all the critical-points to discover patterns of repeating behavior; (4) classes of patterns which were highly correlated with manually scored arousals were formalized as meta-rules. Online we used a test-set of adult patients from two other different sleep laboratories. Using the meta-rules, the algorithm extracted individual rules for each patient (termed actual-rules), and used them to automatically detect the patients' arousals. These arousals were significantly correlated (R=0.88, p<0.0001; sensitivity=75.2%, positive predictive value=76.5%) with those detected manually by experts. Since the total number of arousals is a measure of sleep quality, this algorithm constitutes a novel approach to automatically estimate sleep quality.

Data Mining, or knowledge discovery, is the computer-assisted process of digging through and analyzing large quantity of data in order to extract meaningful knowledge. Data mining methods are used in many studies to identify phenomena quicker and better than human experts. One class of these methods was designed for dealing with time series data. However, when several channels of data are collected simultaneously, data mining algorithms encounter numerous difficulties since channels may be measured in different units, may be recorded at different sampling-rates, or may have completely different characteristics. Furthermore, as the size of these data increases, the amount of irrelevant data usually increases as well and the process becomes impractical. Hence, in such cases, the analyst must be capable of focusing on the informational parts while ignoring the noise data. These kinds of difficulties complicate the analysis of multichannel data as compared to the analysis of single-channel data. This paper presents a useful technique for preprocessing multi channel data. Our technique supplies tools for coping with all the above-mentioned difficulties, and prepares the data for further analysis (using common algorithms, especially from the data mining field). The paper is divided as follows. After the introduction (Section I) we describe the state of the art (Section II), follows by the main section-methodology (Section III) which is divided to four steps (3.2-3.5). The results are described in a separate section (Section IV). Then, a discussion and conclusions of the proposed methodology are given in (Sections V and VI). Acknowledgements and the references follow.

Despite many reports indicating the existence of precise firing sequences in cortical activity, serious objections have been raised regarding the statistics used to detect them and the relations of these sequences to behavior. We show that in behaving monkeys, pairs of spikes from different neurons tend to prefer certain time delays when measured in relation to a specific behavior. Single-unit activity was recorded from eight microelectrodes inserted into the motor and premotor cortices of two monkeys while they were performing continuous drawinglike hand movements. Repeated scribbling paths, termed drawing components, were extracted by data-mining techniques. The set of the least predictable relations between drawing components and pairs of neurons was determined and represented by one statistic termed the relations score. The chance probability of the relations score was evaluated by teetering the spike times: 1,000 surrogates were generated by randomly teetering the original time of each spike in a small window. In nine of 13 experimental days the precision was better than 12 ms and, in the best case, spike precision reached 0.5 ms.

We show that times of spikes can be very precise. In the cerebral cortex, where each nerve cell is affected by thousands of others, it is the common belief that the exact time of a spike is random up to an averaged firing rate over tens of milliseconds. In a brain slice, precise time relations of several neurons have been observed. It remained unclear whether this phenomenon can also be observed in brains of behaving animals. Here we show, in behaving monkeys, that time intervals between spikes, measured in correspondence to a specific behavior, may be controlled to within the milliseconds range.