| Popis: |
Проаналізовано можливості уніфікації структури комп’ютерної вимірювальної системи, орієнтованої на вирішення поширених завдань біоінформатики, на основі застосування обчислювальних потужностей сучасних комп’ютерних засобів, методів цифрового оброблення сигналів і алгоритмів машинного навчання. Виділено три групи методів формування біосигналів і наведено їх класифікацію. Охарактеризовано фактори, які ускладнюють опрацювання біосигналів, і показано перспективність застосування методів машинного навчання для визначення біомедичних показників. Information about the psychophysiological state of humans is important not only in medical practice for the diagnosis of possible diseases but is also for affective informatics, biometrics, rehabilitation engineering, human-machine interaction, etc. Currently biosignals measurement instrumentations are highly specialized and designed to process the separate types of biosignals (ECG, EEG), or to perform the specific tasks, for example, medical diagnosis or biometry. Methods aimed obtaining final top-level information are still "manual" since they rely heavily on the expert’s experience. Purpose of current work is to consider the ways of provision the flexibility and functionality of bioinformatic means on the basis of computing platforms, digital signal processing methods and machine-learning algorithms. Genesis of biosignals is analyzed. Classification of biosignals generation methods is proposed: – biosignals, the primary nature of which is electric (sensing using of electrodes, e.g. EEG); – biosignals that reflect non-electrical processes in the body (formed by sensors, e.g MCG); – biosignals, which are a response to external stimuli (e.g. BIA). Factors that complicate the processing of biosignals are described. Different generation ways and parameter variabilities become the appreciable barrier for the structure unification of the computer-measuring systems. Another barrier is related to the dissimilarity of the algorithms of determining biomedical data. There exist the drivers that offer opportunities in providing the flexibility and functionality of the bioinformatics system. Such an approach makes possible to distribute the structural elements of a computer-measuring system into three groups: – individual items (electrodes, sensors, actuators, measuring cascade, stimulus formatter); – specific group (signal conditioning, ADC and DAC); – universal group (digital processing unit; computer with software, including library of machine learning algorithms). At the final stage an interpretation of the results is carried out. |
