HyunSoon Shin

Effective PPG sensor placement for reflected red and green light, and infrared wristband-type photoplethysmography

Using a wristband-type Photoplethymography (PPG) sensor, useful biomedical information such as heart rate and oxygen saturation can be acquired. Most of commercially-used wrist-type PPG sensors use green light reflections for its greater absorptivity of hemoglobin compared to other lights; this is important because wrists have comparably low concentration of blood flow. For reliable biomedical signal processing, we propose measurement sites for reflected red, green, infrared light PPG sensors on wrist. Amplitude, detection rate, and accuracy of heart rate are compared to determine the signal quality on measurement sites. Traditionally, wrist-type PPG sensors are implemented in measurement site 2, 3 or between 2 and 3 (between the distal Radius and the head of Ulna). Experiments show that all three reflected light PPG sensors generate good quality of PPG signals on measurement sites 4 and 11 (around the distal of Radius of left hand) in test subjects.

Driving characteristics analysis of young and middle-aged drivers

From the statistics about rental car accidents, traffic accident death by drunken driving, speeding, and centerline invasion occurs frequently to the twenties drivers compared with the other age groups. Specially, speeding is a dangerous driving behavior. In this paper, we analyzed driving characteristics of young and middle-aged drivers using FOT (Field of Test) data which was collected on the real urban, local road and highway. We performed independent samples t-test to know the difference of driving time, average speed and maximum speed dependent on ages and gender. We found that young male drivers drove more fast than middle-aged drivers on the local roads and highways. Also, we performed ANOVA analysis to know the speeding characteristics of drivers. The speeding time of young driver is longer than that of middle-aged driver on the local road and highway. The results of this study can be utilized in the device to limit the speed through monitoring the running speed of the driver. It can be derived indirectly safe driving services to reflect the speed factor of driver in the driving insurance fee calculation method.

Construction of a database of emotional speech using emotion sounds from movies and dramas

In this study, an emotional speech database called Hanbat Emotional Database (HEMO) was constructed using movie and drama scenes in which emotion is abundantly expressed by professional actors. HEMO consists of 454 speech samples classified into seven emotion categories such as anger, happiness, sadness, disgust, surprise, fear, and neutral. In order to evaluate the performance of HEMO, consistent experiments were conducted based on HMM (Hidden Markov Model) and GMM (Gaussian Mixture Model) for both HEMO and the Berlin Emotional Speech Database (EMO). HEMO showed better results than EMO with a positive recognition rate of 78.89%.

Development of HMM-based snoring recognition system for web services

This study presents a Hidden Markov Models(HMM)-based snoring recognition system over the web services environment that consists of the snoring model generation, the recognition system, and the remote device. In design phase, a set of HMM model (snoring and non-snoring) is created from the MFCC feature vectors extracted from the sound corpus consisting of snoring sounds and non-snoring sounds. The recognition system is organized to provide the web services that can be called by the remote device in different platforms with any language. In the test, this system shows that snoring and non-snoring sounds were recognized as 93% and 95.2% for speaker-independent case and 98.3% and 99% for the speaker-dependent case, respectively.

Analysis of irregular breathing using respiration-induced intensity variations (RIIV) from photoplethysmography signals for sleep apnea

In this paper, a technique for self-diagnosis devices of the sleep apnea symptoms (central and obstructed apnea) based on photoplethysmography (PPG) signal is presented. An umbrella term, Sleep apnea is a sleep-disordered breathing for several chronic conditions in which partial or complete cessation of breathing occurs many times throughout the night. A respiratory rate in this paper is extracted from photoplethysmography signals which is modulated by synchronized original breathing rhythm. This paper presents a new method for sleep apnea, central and obstructed experimentally, that recognize the irregular breathing using a features of respiratory rate based on frequency domain.

An Accident Prediction in Military Barracks Using Data Mining

Recently, several accidents have occurred in South Korean military barracks that caused a social concern. In this paper, we argue that these accidents can be prevented. Specifically, we describe an ongoing project that applies well-known data mining techniques to predict accidents in military barracks in South Korea. For this, we first collect various soldiers’ data, such as social media, personal history and medical data, and then, use ranking, clustering, classification and text mining techniques to analyze this data.

Application of Data Mining for Crime Analysis

Data mining can be considered as a powerful tool that enables low enforcement and emergency agencies to discovery meaningful patterns in a large amount of danger situation. In this paper, we propose a data mining framework for predicting crimes. The proposed framework consists of the following modules: test data generation, classification, clustering and data ranking. We performed the experiments using various datasets in order to determine which one is potentially best for performing crime pattern prediction task.