Dong Gyu Lee

Air Pollution Monitoring System based on Geosensor Network

Environment Observation and Forecasting System(EOFS) is a application for monitoring and providing a forecasting about environmental phenomena. We design an air pollution monitoring system which involves a context model and a flexible data acquisition policy. The context model is used for understanding the status of air pollution on the remote place. It can provide an alarm and safety guideline depending on the condition of the context model. It also supports the flexible sampling interval change for effective the tradeoff between sampling rates and battery lifetimes. This interval is changed depending on the pollution conditions derived from the context model. It can save the limited batteries of geosensors, because it reduces the number of data transmission.

Discovering Medical Knowledge using Association Rule Mining in Young Adults with Acute Myocardial Infarction

The knowledge discovery has been widely applied to mine significant knowledge from medical data. Nevertheless, previous studies have produced large numbers of imprecise patterns. To reduce the number of imprecise patterns, we need an approach that can discover interesting patterns that connote causality between antecedent and consequence in a pattern. In this paper, we propose association rule mining method that can discover interesting patterns that include medical knowledge in Korean acute myocardial infarction registry that consists of 1,247 young adults collected by 51 participating hospitals since 2005. Proposed method can remove imprecise patterns and discover target patterns that include associations between blood factors and disease history. The association that blood factors affect to disease history is defined as target pattern. In our experiments, the interestingness of a target pattern is evaluated in terms of statistical measures such as lift, leverage, and conviction. We discover medical knowledge that glucose, smoking, triglyceride total cholesterol, and creatinine are associated with diabetes and hypertension in Korean young adults with acute myocardial infarction.

A Framework of In-Situ Sensor Data Processing System for Context Awareness

We propose a framework of the context awareness system which processes a large amount of sensor data from the application areas. The proposed framework consists of a context acquisition, a knowledge base, a rule manager, and a context information manager, etc. we implement the proposed framework of in-situ sensor data processing system that manages the data transmitted from various sensors and notifies the manager of the alarm message for specific conditions. Our proposed framework is able to be applied to the prevention of a forest fire, the warning system for detecting environmental pollution, etc.

Design of Sensor Data Processing Steps in an Air Pollution Monitoring System

Environmental monitoring is required to understand the effects of various kinds of phenomena such as a flood, a typhoon, or a forest fire. To detect the environmental conditions in remote places, monitoring applications employ the sensor networks to detect conditions, context models to understand phenomena, and computing technology to process the large volumes of data. In this paper, we present an air pollution monitoring system to provide alarm messages about potentially dangerous areas with sensor data analysis. We design the data analysis steps to understand the detected air pollution regions and levels. The analyzed data is used to track the pollution and to give an alarm. This implemented monitoring system is used to mitigate the damages caused by air pollution.

Trigger Learning and ECG Parameter Customization for Remote Cardiac Clinical Care Information System

Coronary heart disease is being identified as the largest single cause of death along the world. The aim of a cardiac clinical information system is to achieve the best possible diagnosis of cardiac arrhythmias by electronic data processing. Cardiac information system that is designed to offer remote monitoring of patient who needed continues follow up is demanding. However, intra- and interpatient electrocardiogram (ECG) morphological descriptors are varying through the time as well as the computational limits pose significant challenges for practical implementations. The former requires that the classification model be adjusted continuously, and the latter requires a reduction in the number and types of ECG features, and thus, the computational burden, necessary to classify different arrhythmias. We propose the use of adaptive learning to automatically train the classifier on up-to-date ECG data, and employ adaptive feature selection to define unique feature subsets pertinent to different types of arrhythmia. Experimental results show that this hybrid technique outperforms conventional approaches and is, therefore, a promising new intelligent diagnostic tool.

Comparison of clinical outcomes between culprit vessel only and multivessel percutaneous coronary intervention for ST-segment elevation myocardial infarction patients with multivessel coronary diseases

Background The clinical significance of complete revascularization for ST segment elevation myocardial infarction (STEMI) patients during admission is still debatable. Methods A total of 1406 STEMI patients from the Korean Myocardial Infarction Registry with multivessel diseases without cardiogenic shock who underwent primary percutaneous coronary intervention (PPCI) were analyzed. We used propensity score matching (PSM) to control differences of baseline characteristics between culprit only intervention (CP) and multivessel percutaneous coronary interventions (MP), and between double vessel disease (DVD) and triple vessel disease (TVD). The major adverse cardiac event (MACE) was analyzed for one year after discharge. Results TVD patients showed higher incidence of MACE (14.2% vs. 8.6%, P = 0.01), any cause of revascularization (10.6% vs. 5.9%, P = 0.01), and repeated PCI (9.5% vs. 5.7%, P = 0.02), as compared to DVD patients during one year after discharge. MP reduced MACE effectively (7.3% vs. 13.8%, P = 0.03), as compared to CP for one year, but all cause of death (1.6% vs. 3.2%, P = 0.38), MI (0.4% vs. 0.8%, P = 1.00), and any cause of revascularization (5.3% vs. 9.7%, P = 0.09) were comparable in the two treatment groups. Conclusions STEMI patients with TVD showed higher rate of MACE, as compared to DVD. MP performed during PPCI or ad hoc during admission for STEMI patients without cardiogenic shock showed lower rate of MACE in this large scaled database. Therefore, MP could be considered as an effective treatment option for STEMI patients without cardiogenic shock.

Application of Closed Gap-Constrained Sequential Pattern Mining in Web Log Data

Discovery of information in web log data is a very popular research area in the field of data mining. Two of the objectives of favorite applications are to obtain useful information of web users’ behavior and to analyze the structure of web sites. In this paper, we suggest a novel approach to generate web sequential patterns using the gap-constrained method in web log data. The process of mining task in the proposed approach is described as follows. First, pre-process of the raw web log data is introduced by removing irrelevant or redundant items, gathering the same users and transforming the web log data into a set of tuples (sequence identifier, sequence) constrained by visiting time. Second, web access patterns, which are closed sequential patterns with gap constraints, are generated using the Gap-BIDE algorithm in web log data with two parameters, minimum support threshold and gap constraint. In the experiment, a data set is derived from http://www.vtsns.edu.rs/maja/, which is proposed in [1]. The result shows that, with the application of sequential pattern mining in the web log data presented in this paper, we can find information about navigational behavior of web users and the structure of the web page can be designed more legitimately by the order of obtained patterns.

Neural Connectivity of the Pedunculopontine Nucleus in Relation to Walking Ability in Chronic Patients with Intracerebral Hemorrhage

Objectives: As the mesencephalic locomotor center, the pedunculopontine nucleus (PPN) is known to be involved in control of locomotor function. We investigated neural connectivity of the PPN in relation to walking ability in chronic patients with spontaneous intracerebral hemorrhage. Methods: Forty-three consecutive chronic patients with subcortical hemorrhage and 20 healthy control subjects were recruited. A seed region of interest was manually drawn on the PPN and connectivity of the PPN was measured. Results: In the affected hemisphere, connectivity with the ipsilesional cerebellar locomotor center and connectivity with the contralesional pontine locomotor center were decreased in patients who could not walk, compared with patients who could walk and normal controls (p < 0.05). Conclusions: Connectivity between the PPN and ipsi-lesional cerebellum locomotor center and contralesional pontine locomotor center in the affected hemisphere appears to be related to walking ability.

Highlighting the current issues with pride suggestions for improving the performance of real time cardiac health monitoring

Electrocardiogram (ECG) signal utilized by Clinicians to extract very useful information about the functional status of the heart. Of particular interest systems designed for monitoring people outdoor and detecting abnormalities on the real time. However, there are far from achieving the ideal of being able to perform adequately real time remote cardiac health monitoring in practical life. That is due to problematical challenges. In this paper we discuss all these issues, furthermore our intimations and propositions to relief such concerns are stated.

Superiority real-time cardiac arrhythmias detection using trigger learning method

The Electrocardiogram (ECG) signal uses by Clinicians to extract very useful information about the functional status of the heart, accurate and computationally efficient means of classifying cardiac arrhythmias has been the subject of considerable research efforts in recent years. The contradicting considerations on the unique characteristics of patients activities and the inherent requirements of real-time heart monitoring pose challenges for practical implementation. That is due to susceptibility to potentially changing morphology not only between different patients or patient cluster, but also within the same patient. As a result, the model constructed using an old training data no longer needs to be adapt with the new concepts. Consequently, developing one classifier model to satisfy all patients in different situation using static training datasets is unsuccessful. Our proposed methodology automatically trains the classifier model by up-to-date training data, so as to be identifying with the new concepts. The performance of the trigger method is evaluated using various approaches. The results demonstrate the effectiveness of our proposed technique, and they suggest that it can be used to enhance the performance of new intelligent assistance diagnosis systems.