Blood is carried from the heart to all parts of your body in vessels called arteries. Blood pressure is the force of the blood pushing against the walls of the arteries. Each time the heart beats (about 60 to 70 times a minute at rest), it pumps out blood into the arteries with different value of systolic pressure SP (highest blood pressure when the heart beats) and different value of diastolic pressure DP (lowest blood pressure when the heart relaxes) [3]. Values of SP and DP change during the whole day with dependence on person’s physical and psychical activity. Accuracy of measurement with the modern automatic blood pressure (BP) monitors using oscillometric method is highly depended on condition of cardiovascular system of the monitored person [1]. Especially, with people who suffer from cardiovascular diseases (e.g. arteriosclerosis) the resulting accuracy is much lower when compared to auscultation method. A reasonable solution for improvement of quality of oscillometric method could be an intelligent universal measuring system for evaluation of BP taking into account condition of patient cardiovascular system (CS) of monitored person i.e. the hemodynamics parameters of CS (e.g. heart rate, stroke volume, total peripheral resistance, systemic arterial compliance). Such a system has to be based on an appropriate model of the considered diseases. To create the models, it is very important to establish a database of oscillometric pulsations waveforms (OPW) complemented by the values of “auscultation” blood pressure and information about patients (age, sex, etc.) as well as their diagnosis. This requires a special HW device for measurement of the OPW – we have developed such a device and it has been validated in Czech Certified Metrological Centre, its accuracy is ± 0.5 mmHg in the measuring range 0 to 300 mmHg. We have introduced the concept of oscillometric pulsations waveform (OPW) database that allows testing of oscillometric algorithms for healthy people and mainly for people whose cardiovascular system is not in standard state (arteriosclerosis etc.). The concept is based on oscillometric data retrieving during cuff deflation and on reference BP measurements by auscultation as in [2]. Together with the data, oscillometric pulsations and cuff pressure are saved into the database. For records of OPW we have developed a special HW device that consists of an arm cuff, a pressure sensor, two regulation valves, batteries and electronic circuits. The device can be controlled from PC by a special SW. The connection with the PC is via USB port. The microcontroller controls the pneumatic and the electronic circuits. Cuff pressure is converted into analog voltage by pressure sensor (Piezoresistive Bridge). The analog voltage is amplified by an amplifier TLV2422 and the amplified cuff pressure signal is then separated into 2 channels by a hi-pass filter. Channel 1 is cuff pressure signal (0-300 mm Hg) and channel 2 represents amplified and filtered cuff pulsations (OPW). The 2 signals are digitized by a 12-bit A/D converter in microcontroller ADuC814 with sampling frequency of 200 Hz. The deflation of the cuff is controlled by the regulations valves. The microcontroller communicates with the notebook computer via FTDI chip. Our OPW monitor is connected through the T-pieces and tubes to the cuff, mercury sphygmomanometer and automatic oscillometric blood pressure monitor. Auscultation values are measured by educated staff. Cuff inflation is controlled by microcontroller of the monitor. Then we can directly compare oscillometric and reference (auscultation) method. Moreover, we exactly know the OPW. Nowadays, we have already collected 950 OPW records of 250 people. There are mainly people older than 60 years in the group and we have tried several algorithms (designed in MATLAB ver. 7.00) for evaluation of systolic and diastolic BP [3] and others hemodynamics parameters of the cardiovascular system (mean arterial pressure, heart rate etc.). We have compared values of BP measured by mercury sphygmomanometer and the commercial oscillometric monitor and we have got less than 80% of the measurements results in range of ± 5 mmHg for systolic and diastolic pressure. In more than 20% for both pressures the differences between oscillometric and reference method were greater than ± 5 mmHg. Difference more than -5 mmHg as well as + 5 mmHg was distributed approximately similar. This is a strong motivation for the creation of our database. We have begun apply on measured values some methods of artificial intelligence (AI), especially data mining with system WEKA [5]. We used correlation and searched some association rules. We validated results of standard statistical analysis but we did not find any other strong rules in the data. Nowadays we plan apply these statistical and AI methods directly on measured oscillometric pulsations. This pilot project could be very useful for development of new blood pressure measurement (BPM) methods and also for determination of correct BPM for each group of cardiovascular condition of patients what can be considerable improvement in medical care and patient satisfaction.

FEL ČVUT Praha

Lit.: 4