Gil-Won Yoon

Noninvasive total hemoglobin measurement

Wavelength selection and prediction algorithm for determining total hemoglobin concentration are investigated. A model based on the difference in optical density induced by the pulsation of the heart beat is developed by taking an approximation of Twerskys theory on the assumption that the variation of blood vessel size is small during arterial pulsing. A device is constructed with a five-wavelength light emitting diode array as the light source. The selected wavelengths are two isobestic points and three in compensation for tissue scattering. Data are collected from 129 outpatients who are randomly grouped as calibration and prediction sets. The ratio of the variations of optical density between systole and diastole at two different wavelengths is used as a variable. We selected several such variables that show high reproducibility among all variables. Multiple linear regression analysis is made in order to predict total hemoglobin concentration. The correlation coefficient is 0.804 and the standard deviation is 0.864 g/dL for the calibration set. The relative percent error and standard deviation of the prediction set are 8.5% and 1.142 g/dL, respectively. We successfully demonstrate the possibility of noninvasive hemoglobin measurement, particularly, using the wavelengths below 1000 nm.

Determination of glucose in whole blood samples by mid-infrared spectroscopy

We have determined the glucose concentration of whole blood from mid-infrared spectra without sample preparation or use of chemical reagents. We selected 1119-1022 cm(-1) as the optimal wavelength range for our measurement by making a first-loading vector analysis based on partial least-squares regression. We examined the influence of hemoglobin on samples by using different calibration and prediction sets. The accuracy of glucose prediction depended on the hemoglobin level in the calibration model; the sample set should represent the entire range of hemoglobin concentration. We obtained an accuracy of 5.9% in glucose prediction, and this value is well within a clinically acceptable range.

Smartphone-Based Mobile Health Monitoring

We developed a health monitoring system based on the smartphone. A compact and low-power-consuming biosignal monitoring unit (BMU) measured electrocardiogram (ECG), photoplethysmogram (PPG), temperature, oxygen saturation, energy expenditure, and location information. The 2.4 GHz Bluetooth(®) (Bluetooth SIG) network in the BMU communicated with a smartphone. Health information was sent to a remote healthcare server through a built-in 3G or Wi-Fi network in the smartphone. The remote server monitored multiple users in real-time. Normally data of vital signs were being transmitted to the server. In an emergency or for a special care case, additional information such as the waveform of the ECG and PPG were displayed at the server. For increased transmission efficiency, data compression and a simple error correction algorithm were implemented. Using a widespread smartphone, an efficient personal health monitoring system was developed and tested successfully for multiple users.

Determination of glucose concentration in a scattering medium based on selected wavelengths by use of an overtone absorption band.

A method and device for measuring glucose concentration in a scattering medium have been developed. A spectral range of 800-1800 nm is considered for wavelength selection because of its deeper penetration into biological tissue and the presence of a glucose absorption band. An algorithm based on selected wavelengths is proposed to minimize interference from other components. The optimal distance between the light source and the detector for diffuse reflectance measurement minimizes the influence of medium scattering. The proposed algorithm and measuring device are tested with a solution containing milk with added glucose. Glucose concentrations between 0 and 2000 mg/dl are determined with a correlation coefficient of 0.977. We also investigate the influence of concentration variations of other substances such as water, hemoglobin, albumin, and cholesterol when they are mixed in a scattering medium.

Multicomponent assay for human serum using mid-infrared transmission spectroscopy based on component-optimized spectral region selected by a first loading vector analysis in partial least-squares regression

Mid-infrared transmission spectroscopy with partial least-squares regression was used to determine the concentrations of blood components such as total protein, albumin, globulin, total cholesterol, HDL (high density lipoprotein) cholesterol, triglycerides, glucose, BUN (blood urea nitrogen), and uric acid in human serum. The optimal spectral region for each component was selected by first loading vector analysis. Positive peaks with positive value were assigned by first loading vector analysis. Because blood components in serum show a correlation among several components, a useful spectral region for predicting a particular component was selected such that its spectral feature was not overlapped by those of other components. Several regions with positive peaks by first loading vector were used to establish calibration models. The proposed method proved to be effective for a multicomponent assay and can also be used even when a single component spectrum in aqueous solution for all components is not known. Total protein, albumin, globulin, total cholesterol, triglycerides, and glucose have a mean percentage error of cross-validation (MPECV) of less than 5%. But HDL cholesterol, BUN, and uric acid have MPECVs between 12 and 18%. In terms of both the percentage error of cross-validation and clinically allowable error, six serum components, excepting HDL-cholesterol, BUN, and uric acid, were determined successfully.

Ubiquitous Health Monitoring System for Multiple Users Using a ZigBee and WLAN Dual-Network

A ubiquitous health monitoring system for multiple users was developed based on a ZigBee and wireless local area network (WLAN) dual-network. A compact biosignal monitoring unit (BMU) for measuring electrocardiogram (ECG), photoplethysmogram (PPG), and temperature was also developed. A single 8-bit microcontroller operated the BMU including most of digital filtering and wireless communication. The BMU with its case was reduced to 55 x 35 x 15 mm and 33 g. In routine use, vital signs of 6 bytes/sec (heart rate, temperature, pulse transit time) per each user were transmitted through a ZigBee module even though all the real-time data were recorded in a secure digital memory of the BMU. In an emergency or when need arises, a channel of a particular user was switched to another ZigBee module, called the emergency module, that sent all ECG and PPG waveforms in real time. Each emergency ZigBee module handled up to a few users. Data from multiple users were wirelessly received by the ZigBee receiver modules in a controller called ZigBee-WLAN gateway, where the ZigBee modules were connected to a WLAN module. This WLAN module sent all data wirelessly to a monitoring center. Operating the dual modes of ZigBee/WLAN utilized an advantage of ZigBee by handling multiple users with minimum power consumption, and overcame the ZigBee limitation of low data rate. This dual-network system for LAN is economically competitive and reliable.

Data preprocessing and partial least squares regression analysis for reagentless determination of hemoglobin concentrations using conventional and total transmission spectroscopy.

Visible-near infrared spectroscopy was successfully used for the determination of total hemoglobin concentration in whole blood. Absorption spectra of whole blood samples, whose hemoglobin concentrations ranged between 6.6 and 17.2 g/dL, were measured from 500 to 800 nm. Two different types of transmission were measured: conventional transmission spectroscopy which collected primarily collimated radiation transmitted through the sample, and total transmission spectroscopy which used an integrating sphere to collect all scattered light as well. Different preprocessing techniques in conjunction with a partial least squares regression calibration model to predict hemoglobin concentrations were applied to the above two types of transmission. Depending on different preprocessing methods, the standard error of predictions ranged from 0.37 to 2.67 g/dL. Mean centering gave the most accurate prediction in our particular data set. Preprocessing methods designed for compensation of the scattering effect produced the worst results contrary to expectations. For univariate analysis, better prediction was achieved by total transmission measurement than by conventional transmission measurement. No significant difference was observed for multivariate analysis on the other hand. Careful selection of the data preprocessing methods and of the multivariate statistical model is required for reagentless determination of hemoglobin concentration in whole blood.

Digital envelope detector for blood pressure measurement using an oscillometric method

Blood pressure measurement in the finger artery offers some advantages compared with that in the brachial artery. However, volume oscillometric signals obtained from finger artery measurement are often influenced by motion artefact due to respiration, speaking, involuntary or voluntary movement, etc. In this paper, we developed a digital envelope detector to detect the maximum oscillation criterion in blood pressure measurement for the first time. The digital envelope detector is robust to noise signals generated by motion artefact and filters out the carrier frequency efficiently. To verify the feasibility of our method, we measured blood pressure for eight subjects using our developed system. The results were compared with the auscultation method. In the case of using a digital envelope detector, we could reduce the mean difference error and standard deviation by 30-40%. Our proposed digital envelope detector is a useful tool to improve the accuracy of blood pressure measurement in finger artery.

Telemonitoring System of Fall Detection for the Elderly

The population of elderly people increases rapidly as our society moves towards the aged one. Healthcare for the elderly becomes an important issue and falling down is one of the critical problems although not well recognized. In this study, a fall detection system was developed using a 3-axis accelerometer. Analyzing fall patterns, we took into account the degree of impact, posture angle, the repetitions of similar movements and the activities after a potential fall and proposed an algorithm of fall detection. Information of the fall sensor was sent to a remote healthcare server through the wireless networks of Zigbee and WLAN. Our system was designed to monitor multiples users. 12 persons participated in experiment and each one performed 24 different movements. Our proposed algorithm was compared with other reported ones. Our method produced the excellent results having a sensitivity of 96.4 % and a specificity of 100 % whereas other methods had a sensitivity range between 87.5 % and 94.8 % and a specificity range between 63.5 % and 83.3 %.

Multiple diagnosis based on photoplethysmography: hematocrit, SpO2, pulse, and respiration

Photo-plethysmography measures pulsatile blood flow in real-time and non-invasively. One of widely known applications of PPG is the measurement of saturated oxygen in arterial blood(SpO2). In our work, using several wavelengths more than those used in a pulse oximeter, an algorithm and instrument have been developed to measure hematocrit, saturated oxygen, pulse and respiratory rates simultaneously. To predict hematocrit, a dedicated algorithm is developed based on scattering of RBC and a protocol for detecting outlier signals is used to increase accuracy and reliability. Digital filtering techniques are used to extract respiratory rate signals. Utilization of wavelengths under 1000nm and a multi-wavelength LED array chip and digital-oriented electronics enable us to make a compact device. Our preliminary clinical trials show that the achieved percent errors are ±8.2% for hematocrit when tested with 594 persons, R2 for SpO2 fitting is 0.99985 when tested with a Bi-Tek pulse oximeter simulator and the SpO2 error for in vivo test is ±2.5% over the range of 75~100%. The error of pulse rates is less than ±5%. We obtained a positive predictive value of 96% for respiratory rates in qualitative analysis.