Description

Analysis of blood oxygen saturation and heart frequency for sleep apnoea-hypopnoea syndrome (SAHS) diagnosis. Polysomnography is the main source of information for sleep study. Automatic analysis of cardio-respiratory night signs is very useful for disease diagnosis, such as apnoea-hypopnoea syndrome diagnose (SAHS) or chronic obstructive pulmonary disease (COPD). The University of Valladolid’s group of biomedical informatics (GIB in its Spanish acronym) has developed some diagnosis methods through spectral, non-lineal analysis and time-frequency of polysomnographic registers. Development of predictive models based on machine learning. The main goal of these work lines is to develop and validate an e-Health tool within a new protocol of patient screening to significantly reduce the number of PSGs (night polysomnography). The screening test is based on the automatic analysis of blood oxygen saturation (SpO2) registered at the patient’s home. SpO2 can be easily obtained with a finger pulse oximeter. There are portable pulse oximeters which make this home test easier. The methodology is divided into 2 complementary phases: (i) screening of true negatives and (ii) screening of true positives. An expert system based on machine learning will be optimized in each phase in order to detect (i) subjects not suffering from SAHS without there being false negatives and (ii) patients suffering from SAHS without there being false positives. In order to maximize the diagnostic potential of the SpO2 signal, there will be 3 phases of automatic processing: (i) Feature extraction in order to obtain as much useful information as possible, (ii) feature selection in order to determine the optimal subset including relevant and non-redundant information, and (iii) pattern identification through machine learning algorithms. In order to provide this service to Special Care consultations, it is suggested the development of a comprehensive screening tool which uses cloud computing to analyse every SpO2 register. The results will be provided to the specialist in different formats: phone device and web. The user application will automatically manage the register transfer and the result display.

Other information