Jongshill Lee

Nonlinear Support Vector Machine Visualization for Risk Factor Analysis Using Nomograms and Localized Radial Basis Function Kernels

Nonlinear classifiers, e.g., support vector machines (SVMs) with radial basis function (RBF) kernels, have been used widely for automatic diagnosis of diseases because of their high accuracies. However, it is difficult to visualize the classifiers, and thus difficult to provide intuitive interpretation of results to physicians. We developed a new nonlinear kernel, the localized radial basis function (LRBF) kernel, and new visualization system visualization for risk factor analysis (VRIFA) that applies a nomogram and LRBF kernel to visualize the results of nonlinear SVMs and improve the interpretability of results while maintaining high prediction accuracy. Three representative medical datasets from the University of California, Irvine repository and Statlog dataset-breast cancer, diabetes, and heart disease datasets-were used to evaluate the system. The results showed that the classification performance of the LRBF is comparable with that of the RBF, and the LRBF is easy to visualize via a nomogram. Our study also showed that the LRBF kernel is less sensitive to noise features than the RBF kernel, whereas the LRBF kernel degrades the prediction accuracy more when important features are eliminated. We demonstrated the VRIFA system, which visualizes the results of linear and nonlinear SVMs with LRBF kernels, on the three datasets.

Comparison of cortical thickness in patients with early-stage versus late-stage amnestic mild cognitive impairment.

Disappointing outcomes from clinical trials involving amyloid‐modifying therapies for Alzheimers disease (AD) have prompted more focus on the concept of early‐stage (E) amnestic mild cognitive impairment (E‐aMCI). However, limited evidence suggests that E‐aMCI may represent aMCI at a very early stage of AD. Furthermore, the nature of the progression of E‐aMCI to late‐stage aMCI (L‐aMCI) remains unclear. Therefore, the aim of the present study was to characterize patterns of cortical thinning in both E‐aMCI and L‐aMCI patients.

Personal Identification Based on Vectorcardiogram Derived from Limb Leads Electrocardiogram

We propose a new method for personal identification using the derived vectorcardiogram (dVCG), which is derived from the limb leads electrocardiogram (ECG). The dVCG was calculated from the standard limb leads ECG using the precalculated inverse transform matrix. Twenty-one features were extracted from the dVCG, and some or all of these 21 features were used in support vector machine (SVM) learning and in tests. The classification accuracy was 99.53%, which is similar to the previous dVCG analysis using the standard 12-lead ECG. Our experimental results show that it is possible to identify a person by features extracted from a dVCG derived from limb leads only. Hence, only three electrodes have to be attached to the person to be identified, which can reduce the effort required to connect electrodes and calculate the dVCG.

Videographic Analysis of Glottic View With Increasing Cricoid Pressure Force

STUDY OBJECTIVE Cricoid pressure may negatively affect laryngeal view and compromise airway patency, according to previous studies of direct laryngoscopy, endoscopy, and radiologic imaging. In this study, we assess the effect of cricoid pressure on laryngeal view with a video laryngoscope, the Pentax-AWS. METHODS This cross-sectional survey involved 50 American Society of Anesthesiologists status I and II patients who were scheduled to undergo elective surgery. The force measurement sensor for cricoid pressure and the video recording system using a Pentax-AWS video laryngoscope were newly developed by the authors. After force and video were recorded simultaneously, 11 still images were selected per 5-N (Newton; 1 N = 1 kg·m·s(-2)) increments, from 0 N to 50 N for each patient. The effect of cricoid pressure was assessed by relative percentage compared with the number of pixels on an image at 0 N. RESULTS Compared with zero cricoid pressure, the median percentage of glottic view visible was 89.5% (interquartile range [IQR] 64.2% to 117.1%) at 10 N, 83.2% (IQR 44.2% to 113.7%) at 20 N, 76.4% (IQR 34.1% to 109.1%) at 30 N, 51.0% (IQR 21.8% to 104.2%) at 40 N, and 47.6% (IQR 15.2% to 107.4%) at 50 N. The number of subjects who showed unworsened views was 20 (40%) at 10 N, 17 (34%) at 20 and 30 N, and 13 (26%) at 40 and 50 N. CONCLUSION Cricoid pressure application with increasing force resulted in a worse glottic view, as examined with the Pentax-AWS Video laryngoscope. There is much individual difference in the degree of change, even with the same force. Clinicians should be aware that cricoid pressure affects laryngeal view with the Pentax-AWS and likely other video laryngoscopes.

Digital recording system of sphygmomanometry

ObjectivesValidation of blood pressure measurement devices is always difficult because the gold standard method depends on the observers manual assessment. To improve the assessment algorithm in automatic blood pressure measurement devices, knowing the reference blood pressure values and analyzing various data are always necessary. A digital system to record Korotkoff sound and cuff pressure is suggested. MethodsThe recording system consists of cuff, microphone, amplifier, and all the components of an oscillometric measurement device, which enable the collection of various data from patients. After collecting the data, the browsing software can playback the recorded sound and pressure variation on a computer. ResultsThe system records the Korotkoff sound faithfully. The sound can be played with sound card on a personal computer. Cuff pressure can also be displayed with Korotkoff sound at the same time. With this playback software, the observer can assess systolic blood pressure and diastolic blood pressure . The observer can listen again and discuss with others when the decision is not clear. ConclusionThe digital recording system of sphygmomanometry can be used for validation of blood pressure measurement devices, collecting of oscillometric data for research, and educating the students who learn blood pressure measurement.

Deep gray matter atrophy in neuromyelitis optica spectrum disorder and multiple sclerosis

We investigated changes in deep gray matter (DGM) volume and its relationship to cognition and clinical factors in a large cohort of patients with neuromyelitis optica spectrum disorder (NMOSD) and compared them with results from multiple sclerosis (MS).

Does the accuracy of blood pressure measurement correlate with hearing loss of the observer

ObjectiveThe auscultatory method is influenced by the hearing level of the observers. If the observer has hearing loss, it is possible to measure blood pressure inaccurately by misreading the Korotkoff sounds at systolic blood pressure (SBP) and diastolic blood pressure (DBP). Because of the potential clinical problems this discrepancy may cause, we used a hearing loss simulator to determine how hearing level affects the accuracy of blood pressure measurements. Materials and methodsTwo data sets (data set A, 32 Korotkoff sound video clips recorded by the British Hypertension Society; data set B, 28 Korotkoff sound data acquired from the Korotkoff sound recording system developed by Hanyang University) were used and all the data were attenuated to simulate a hearing loss of 5, 10, 15, 20, and 25 dB using the hearing loss simulator. Five observers with normal hearing assessed the blood pressures from these data sets and the differences between the values measured from the original recordings (no attenuation) and the attenuated versions were analyzed. ResultsGreater attenuation of the Korotkoff sounds, or greater hearing loss, resulted in larger blood pressure measurement differences when compared with the original data. When measuring blood pressure with hearing loss, the SBP tended to be underestimated and the DBP was overestimated. The mean differences between the original data and the 25 dB hearing loss data for the two data sets combined were 1.55±2.71 and −4.32±4.21 mmHg for SBP and DBP, respectively. ConclusionThis experiment showed that the accuracy of blood pressure measurements using the auscultatory method is affected by observer hearing level. Therefore, to reduce possible error using the auscultatory method, observers’ hearing should be tested.

Higher C‐peptide levels are associated with regional cortical thinning in 1093 cognitively normal subjects

Recent studies have demonstrated an association between increased insulin secretion and cognitive impairment. However, there is no previous study that directly evaluates the association between increased insulin secretion and cortical thickness to our knowledge. Therefore, our aim was to evaluate the effect of hyperinsulinemia, as measured by C‐peptide level, on cortical thickness in a large sample of cognitively normal individuals.

High-fidelity digital recording and playback sphygmomanometry system: device description and proof of concept.

ObjectivesThis study describes the development of a new digital sphygmocorder (DS-II), which allows the digital recording and playback of the Korotkoff sounds, together with cuff pressure waveform, and its performance in a pilot validation study. Materials and methodsA condenser microphone and stethoscope head detect Korotkoff sounds and an electronic chip, dedicated to audio-signal processing, is used to record high-quality sounds. Systolic and diastolic blood pressure (SBP/DBP) are determined from the recorded signals with an automatic beat detection algorithm that displays the cuff pressure at each beat on the monitor. Recordings of Korotkoff sounds, with the cuff pressure waveforms, and the simultaneous on-site assessments of SBP/DBP were performed during 100 measurements in 10 individuals. The observers reassessed the recorded signals to verify their accuracy and differences were calculated. ResultsThe features of the high-fidelity DS-II, the technical specifications and the assessment procedures utilizing the playback software are described. Interobserver absolute differences (mean±SD) in measurements were 0.7±1.1/1.3±1.3 mmHg (SBP/DBP) with a mercury sphygmomanometer and 0.3±0.9/0.8±1.2 mmHg with the DS-II. The absolute DS-II mercury sphygmomanometer differences were 1.3±1.9/1.5±1.3 mmHg (SBP/DBP). ConclusionThe high-fidelity DS-II device presents satisfactory agreement with simultaneous measurements of blood pressure with a mercury sphygmomanometer. The device will be a valuable methodology for validating new blood pressure measurement technologies and devices.

A Linear Transformation Approach for Estimating Pulse Arrival Time

We propose a new mathematical framework for estimating pulse arrival time (PAT). Existing methods of estimating PAT rely on local characteristic points or global parametric models: local characteristic point methods detect points such as foot points, max points, or max slope points, while global parametric methods fit a parametric form to the anacrotic phase of pulse signals. Each approach has its strengths and weaknesses; we take advantage of the favorable properties of both approaches in our method. To be more precise, we transform continuous pulse signals into scalar timing codes through three consecutive transformations, the last of which is a linear transformation. By training the linear transformation method on a subset of data, the proposed method yields results that are robust to noise. We apply this method to real photoplethysmography (PPG) signals and analyze the agreement between our results and those obtained using a conventional approach.