Preclinical Assessment of a Novel Cardiovascular 1 Telemedicine System 2

Background: Introduction of telemonitoring systems to patient care which provide extensive information about the cardiovascular status of the patient is a promising direction to reduce cardiovascular morbidity and mortality. Our team has developed a telemedical system which is based 15 on the photoplethysmographic detection of the digital arterial pulse wave. The system incorporates a cloud-based automated algorithm which analyses the pulse contour to provide 15 scientifically 17 established parameters for versatile characterization of cardiovascular function. The aim of the current study was to assess the variability of the measurements to test the applicability of the tool before clinical use. We assessed the repeatability of the measurements by detecting stable artificial signals, and also test-retest variability by repeatedly examining the pulse contours of healthy individuals under 21 standardized conditions. Results: Most contour parameters (stiffness index, reflection index, left ventricular ejection time index 23 and mean interbeat intervals) are measured with high repeatability (coefficients of variation (CV) < 1% 24 for each parameter), and exhibit acceptable intrapersonal fluctuations (CVs <10%). However, some 25 parameters derived from the second derivative of the pulse wave seem to be more variable (aging 26 index, d/a ratio). This is explained by the typical alterations of the pulse wave under specific 27 circumstances, which cause the flattening or complete disappearance of c and d inflections on the 28 second derivative. 29 Conclusion: Our measurements proved that our telemonitoring system detects and analyses digital 30 pulse contours with high accuracy and highlighted that second derivative parameters should be 31 interpreted cautiously. We recommend the evaluation of these parameters only in those 32 measurements where c and d points are detected reliably. Pulse contour parameters are stable in 33 healthy individuals under standardized conditions, which allows detection of subtle abnormal 34 alterations by the remote surveillance system.

and also test-retest variability by repeatedly examining the pulse contours of healthy individuals under 21 standardized conditions. 22 Results: Most contour parameters (stiffness index, reflection index, left ventricular ejection time index 23 and mean interbeat intervals) are measured with high repeatability (coefficients of variation (CV) < 1% 24 for each parameter), and exhibit acceptable intrapersonal fluctuations (CVs <10%). However, some 25 parameters derived from the second derivative of the pulse wave seem to be more variable (aging 26 index, d/a ratio). This is explained by the typical alterations of the pulse wave under specific 27 circumstances, which cause the flattening or complete disappearance of c and d inflections on the 28 second derivative. 29 Conclusion: Our measurements proved that our telemonitoring system detects and analyses digital 30 pulse contours with high accuracy and highlighted that second derivative parameters should be 31 interpreted cautiously. We recommend the evaluation of these parameters only in those 32 measurements where c and d points are detected reliably. Pulse contour parameters are stable in 33 healthy individuals under standardized conditions, which allows detection of subtle abnormal 34 alterations by the remote surveillance system. 35 5 interpretation and validity of measurements, information about testing pulse wave analysis systems 90 for these errors is scarce in scientific literature [20] [21]. In the present study we used a multidirectional 91 methodological approach to address both aspects and focused on selected parameters which have 92 well-reported medical significance based on scientific literature. 93 In order to determine the variability caused by measurement error of our telemedicine system, we 94 used a simulator that generates artificial pulse signals which could be detected by a 95 photoplethysmograph. We repeatedly recorded and evaluated the signal with our system. 96 Our CV functioning constantly adapts to the changing environment. Changes of our CV status are 97 reflected by the pulse wave morphology. In order to enhance the accuracy of the measurements, we 98 need to standardize the circumstances of the examination (e.g. resting conditions, ambient 99 temperature, body position, time of the day, time from last meal, coffee, smoking and physical activity) 100  device to 1 kHz. In order to condition the PPG signal a digital band pass filter -fourth order Butterworth 126 -with -3dB points at 0.1 Hz and 10 Hz is applied. Then the algorithm identifies the pulse cycles.

Intrapersonal variability at standard conditions
To define the size of variability caused by physiological fluctuations of CV functioning, which still 164 remains after standardizing the measurement conditions, we performed 10 repeated 2-minute-long 165 measurements on 10 young healthy individuals (M/F: 5/5, Age: 19-35, Mean ± SD: 25.3 ± 4.3) at 166 standard conditions. The course of successive measurements took approximately 30 minutes. We 167 defined 'standard condition' as the set of measurement conditions which we recommend our users to 168 keep when they perform their daily morning measurements during follow-up. Criteria of standard 169 conditions: measurement takes place in a quiet room at room temperature; in the morning hours at 170 least two hours after the last meal and coffee; in a sitting, resting position, with hands held quietly on 171 a

197
Determination of measurement error by the telemedicine system was assessed by detecting stable 198 artificial signals generated by a pulse oximeter simulator. The overall measurement error may be 199 produced by the data analyzing algorithm, the measurement error of a single pulse oximeter as well 200 as by the variability due to using different pulse oximeter devices to detect the pulse signals. Firstly, in 201 order to assess the combined contribution of the algorithm and the error of a single pulse oximeter to 202 the overall measurement error, we detected the normal pulse signals of the simulator with a single, 203 randomly chosen pulse oximeter and repeated it 5 times (Table 1. Normal condition, 1 st column). The 204 results showed that the measurement was stable: the CI was very close to the mean of the 5 205 measurements, and the coefficient of variation was below 1% for each calculated variable. 206 Then we randomly chose 4 other pulse oximeters of the same release, and repeated the 207 measurements as described above. Then we averaged the results of the 25 measurements. These 208 showed that the output data had low variability as evidenced by narrow CI-s and small (lower than 1%) 209 CV-s for each parameter (Table 1. Normal condition, 2 nd column). 210 After proving that our system detects and analyzes normal pulse signals reliably, we repeated the 211 measurements described above with signal presets of the simulator, which simulate abnormal 212 conditions. For this purpose, we used the 'Abnormal 1' and the 'Abnormal 2' presets. The former 213 preset of the simulator generates a signal with high heart rate (95/min). In this setting, the reliability 214 of pulse detection and analysis was similar to the 'Normal' condition except for the calculation of aging 215 index and d/a parameters -as the second derivative of this preset has no detectable c and d points. 216 (Table 1. 'Abnormal 1' condition) 217 The 'Abnormal 2' signal preset mimics a condition, when the signal is of low intensity (a typical source 218 of error in DVP detection). Similar to what we observed with the 'Abnormal 1' signals, the results of 219 these measurements also showed stable detection and analysis for most parameters, except for the 220 aging index and the d/a ratio -for the same reasons as in Abnormal 1. (Table 1. 'Abnormal 2' condition) 221 222 Test-retest variability was assessed to evaluate intrapersonal variability of the pulse wave parameters 223 under standard conditions. For this purpose, resting measurements were repeated 10 times in 10 224 healthy individuals. After calculating the coefficient of variation for each individual, the CV-s of the 10 225 subjects were averaged. The mean CV-s are presented in Table 2 Table 3. The mean of the measurements of the 4 fingers are presented, 243 showing no relevant difference between the fingers. Moreover, the intraclass correlation coefficients 244 were over 99% for mean interbeat interval, mean heart rate, left ventricular ejection time index 245 indicating that the effect of using different fingers for measurement is negligible. The ICCs for stiffness 246 index and c-d point detection ratio were about 90%, and were over 80% for reflection index, b/a, d/a 247 and aging index. These confirm that the effect of using different fingers on variability is much less than 248 that of the interindividual differences for these parameters, too (see Table 3  The repeatability of the measurements of our telemonitoring system was assessed by calculating the 270 variability of the DVP parameters obtained from successive measurements of stable artificial pulse 271 signals, which simulated healthy pulse waves and were generated by a pulse oximeter simulator 272 device. Such variability can be caused by measurement errors of the pulse oximeter instrument and 273 also the automated algorithm analyzing the detected pulse wave. The combined effect of these 2 274 factors on measurement variability was investigated by testing the agreement between the results of 275 5 successive measurements performed by the same randomly chosen pulse oximeter device. The 276 variation was smaller than the predefined 2% criterion of acceptance for each parameter (Table 1. 277 'Normal' condition). Afterwards, we extended the investigation to 4 additional instruments with which 278 we performed the same measurements. We pooled the 5x5 measurements and calculated the overall 279 CV-s, which now reflect the combined variation caused by measurement error of a single pulse 280 oximeter, analysis by the algorithm, and also the 'inter-instrumental' variability of several pulse ( The pulse oximeter simulator also offers abnormal pulse signals. We repeated the measurements with 285 these settings, too. 'Abnormal 1' setting generates a pulse signal of high heart rate and almost totally 286 absent second derivative c-d points, whereas 'Abnormal 2' a signal simulates a weak pulse wave (e.g. 287 similar to that observed in case of vasoconstriction due to cold). Second derivative c-d points are 288 absent in this setting as well. With these settings the calculation of most parameters was still highly 289 repeatable (CV% below 2%). However, detection of c and d points became less reliable. In accordance 290 with that, c-d point detection ratio, the parameter which expresses the percent of those pulse cycles 291 in which c and d points are recognized by the algorithm, fell below 5% for each setting (Table 1. heart rate (CV-s are lower than 10%; Table 2). Consequently, these parameters are suitable for patient 330 follow-up, as deviation of a measurement from the ordinary individual value of the patient is not likely 331 to be caused by normal intrapersonal variability, but rather indicates pathological alterations.

variable (aging index, d/a). This concurs with the relatively high variations in c-d point detection ratio 334 of consecutive measurements. This also confirms that aging index and d/a should only be involved in 335
clinical evaluation, when c and d points are reliably detected by the algorithm, otherwise their 336 applicability is questionable (see above). 337 In our study, we also provided preliminary data on the interpersonal variability of the studied contour 338 parameters (Figure 4.) In our study we also tested how different anatomical disposition of the fingers affects the results of 349 pulse contour analysis. It is not a question that we recommend our users to use the same finger for 350 each measurement. However, it may occur that for some reason they use another finger sometimes. 351 Therefore, we need to be aware whether this error causes significant alterations of the output results. 352 We could observe that in healthy individuals there was no clinically relevant difference in pulse contour 353 parameters when measured parallel on index and ring fingers of the 2 hands. The calculated ICCs 354 showed that the effect of using different fingers on variability of the outcomes is much less than the 355 effect of interpersonal differences. Therefore, changing to different fingers does not constitute 356 relevant measurement error. However, we need to keep in mind that pathological alterations and 357 diseases of the supplying arterial tree may have an impact on blood flow of the digital arteries. For this 358 reason, at the first patient visit it is recommended to record pulse signals on several fingers on both 359 sides and analyze whether there are differences in the output parameters. 360 361

362
In this study we completed the preclinical assessment of a novel pulse wave analysis based 363 telemonitoring system. We used a multidirectional approach to explore and characterize the possible 364 measurement errors in depth. We showed that our system is capable of measuring most common    Means (and confidence intervals -CI) and coefficients of variation (CV) of pulse contour variables measured by the SCN4ALL telemedicine system. In order to evaluate repeatability of the measurements by the system, we detected and analyzed artificial pulse signals generated by a pulse oximeter simulator device. 3 different signal settings of the simulator were selected (Normal, Abnormal 1, and Abnormal 2). For each setting, measurements were repeated 5 times with a single randomly chosen pulse oximeter (n=5 columns), then these measurements were supplemented with the repeated measurements on 4 other pulse oximeters of the same release (n=25 columns, showing the results of 5x5 measurements).