Young Moon Chae

Data mining approach to policy analysis in a health insurance domain

This study examined the characteristics of the knowledge discovery and data mining algorithms to demonstrate how they can be used to predict health outcomes and provide policy information for hypertension management using the Korea Medical Insurance Corporation database. Specifically, this study validated the predictive power of data mining algorithms by comparing the performance of logistic regression and two decision tree algorithms, CHIAD (Chi-squared Automatic Interaction Detection) and C5.0 (a variant of C4.5) using the test set of 4588 beneficiaries and the training set of 13,689 beneficiaries. Contrary to the previous study, the CHIAD algorithm performed better than the logistic regression in predicting hypertension, and C5.0 had the lowest predictive power. In addition, the CHIAD algorithm and the association rule also provided the segment-specific information for the risk factors and target group that may be used in a policy analysis for hypertension management.

Patient satisfaction with telemedicine in home health services for the elderly

In a pilot study of telemedicine in home health services (HHS) for elderly patients, we implemented and evaluated a telemedicine system with a 33-kbs narrow-band approach to determine its effectiveness in providing quality services. Fifty patients were selected for the study. We found that telemedicine was effective in terms of reducing the number of clinic visits and achieving patient satisfaction. The average number of clinic visits per month was significantly decreased from 0.64 to 0.42 (p < 0.05) after the use of telemedicine. 72% of patients were satisfied with telemedicine, but only patient location showed a significant difference for patient satisfaction (p < 0.05). Patients in their homes (82%) were more satisfied than patients in nursing homes (50%). Of four types of services provided, medical consultation (100%) was the most highly satisfactory service with telemedicine, followed by physical therapy (83.3%). Although the satisfaction scores did not indicate a significant difference in the system characteristics, the quality of verbal communication appeared to be a more important factor in influencing patient satisfaction than set-up time or quality of image. A computer-based patient record was also developed to view a patient summary and to document encounters at the patients home. Since the system is a low-cost approach that is easy to interface with a notebook computer, it can support various other HHSs.

Evaluation of a mobile phone-based diet game for weight control.

We developed an interactive mobile-phone based application, SmartDiet, that analyzes daily nutrition intake and patterns of daily exercise. It provides a personalized diet profile and promotes knowledge about nutrition using a diet game. We evaluated the effectiveness of the SmartDiet application in terms of acquiring dietary information, weight control and user satisfaction. A case-control study was conducted over a six-week period, with 19 people in the intervention group and 17 people in the control group. During the study, a total of 235 successful data transmissions were performed from the mobile phones and there was a mean of 12.4 transmissions per participant. The three body composition measures (fat mass, weight and body mass index) decreased significantly after the intervention in the intervention group, but there were no significant changes in the control group. In a questionnaire survey at the end of the study, the majority of the participants responded that the system was useful for obtaining information and managing the diet process. The SmartDiet mobile weight management application appears to contribute to weight loss in obese adults.

Analysis of Healthcare Quality Indicator using Data Mining and Decision Support System

Abstract This study presents an analysis of healthcare quality indicators using data mining for developing quality improvement strategies. Specifically, important factors influencing the inpatient mortality were identified using a decision tree method for data mining based on 8405 patients who were discharged from the study hospital during the period of December 1, 2000 and January 31, 2001. Important factors for the inpatient mortality were length of stay, disease classes, discharge departments, and age groups. The optimum range of target group in inpatient healthcare quality indicators were identified from the gains chart. In addition, a decision support system (DSS) was developed to analyze and monitor trends of quality indicators using Visual Basic 6.0. Guidelines and tutorial for quality improvement activities were also included in the system. In the future, other quality indicators should be analyzed to effectively support a hospital-wide continuous quality improvement (CQI) activity and the DSS should be well integrated with the hospital order communication system (OCS) to support concurrent review.

The Adoption of Electronic Medical Records and Decision Support Systems in Korea

Objectives To examine the current status of hospital information systems (HIS), analyze the effects of Electronic Medical Records (EMR) and Clinical Decision Support Systems (CDSS) have upon hospital performance, and examine how management issues change over time according to various growth stages. Methods Data taken from the 2010 survey on the HIS status and management issues for 44 tertiary hospitals and 2009 survey on hospital performance appraisal were used. A chi-square test was used to analyze the association between the EMR and CDSS characteristics. A t-test was used to analyze the effects of EMR and CDSS on hospital performance. Results Hospital size and top management support were significantly associated with the adoption of EMR. Unlike the EMR results, however, only the standardization characteristic was significantly associated with CDSS adoption. Both EMR and CDSS were associated with the improvement of hospital performance. The EMR adoption rates and outsourcing consistently increased as the growth stage increased. The CDSS, Knowledge Management System, standardization, and user training adoption rates for Stage 3 hospitals were higher than those found for Stage 2 hospitals. Conclusions Both EMR and CDSS influenced the improvement of hospital performance. As hospitals advanced to Stage 3, i.e. have more experience with information systems, they adopted EMRs and realized the importance of each management issue.

A Study on User Satisfaction regarding the Clinical Decision Support System (CDSS) for Medication

Objectives Many medication errors can occur when ordering and dispensing medicine in hospitals. The clinical decision support system (CDSS) is widely used in an effort to reduce medication errors. This study focused on the evaluation of user satisfaction with the CDSS for medication at a university hospital. Specifically, this study aimed to identify the factors influencing user satisfaction and to examine user requirements in order to further improve user satisfaction and drug safety. Methods The study was based on survey data from 218 users (103 doctors, 103 nurses, and 15 pharmacists) at a university hospital that uses the CDSS. In order to identify the factors influencing user satisfaction with the CDSS, a multiple linear regression was performed. In order to compare the satisfaction level among the professional groups, an analysis of variance (ANOVA) was performed. Results The reliability of information, decision supporting capability, and departmental support were significant factors in influencing user satisfaction. In addition, nurses were the most satisfied group, followed by pharmacists and doctors according to the ANOVA. Areas for further improvement in enhancing drug safety were real time information searching and decision supporting capabilities to prevent adverse drug events (ADE) in a timely manner. Conclusions We found that the CDSS users were generally satisfied with the system and that it complements the nationwide drug utilization review (DUR) system in reducing ADE. Further CDSS evaluation in other hospitals is needed to improve user satisfaction and drug safety.

Performance Evaluation of Public Hospital Information Systems by the Information System Success Model

Objectives This study was to evaluate the performance of the newly developed information system (IS) implemented on July 1, 2014 at three public hospitals in Korea. Methods User satisfaction scores of twelve key performance indicators of six IS success factors based on the DeLone and McLean IS Success Model were utilized to evaluate IS performance before and after the newly developed system was introduced. Results All scores increased after system introduction except for the completeness of medical records and impact on the clinical environment. The relationships among six IS factors were also analyzed to identify the important factors influencing three IS success factors (Intention to Use, User Satisfaction, and Net Benefits). All relationships were significant except for the relationships among Service Quality, Intention to Use, and Net Benefits. Conclusions The results suggest that hospitals should not only focus on systems and information quality; rather, they should also continuously improve service quality to improve user satisfaction and eventually reach full the potential of IS performance.

Comparison of alternative knowledge model for the diagnosis of asthma

Abstract This paper compared three knowledge models (namely, neural network, case-based reasoning, and discriminant analysis) for the diagnosis of asthma. The data were collected from 294 patients with asthmatic symptoms who visited the Bronchial Asthma Clinics, Internal Medicine Department of Yonsei University Severance Hospital from June 1992 to May 1995. Diagnostic capabilities for the three knowledge models varied. The neural network had the best overall prediction rate (92%) and the best prediction rate for asthma (96%); the discriminant analysis had the best prediction rate for non-asthma (80%); and the CBR had the lowest prediction rates in all categories.

A Comparison of Logistic Regression Analysis and an Artificial Neural Network Using the BI-RADS Lexicon for Ultrasonography in Conjunction with Introbserver Variability

To determine which Breast Imaging Reporting and Data System (BI-RADS) descriptors for ultrasound are predictors for breast cancer using logistic regression (LR) analysis in conjunction with interobserver variability between breast radiologists, and to compare the performance of artificial neural network (ANN) and LR models in differentiation of benign and malignant breast masses. Five breast radiologists retrospectively reviewed 140 breast masses and described each lesion using BI-RADS lexicon and categorized final assessments. Interobserver agreements between the observers were measured by kappa statistics. The radiologists’ responses for BI-RADS were pooled. The data were divided randomly into train (n = 70) and test sets (n = 70). Using train set, optimal independent variables were determined by using LR analysis with forward stepwise selection. The LR and ANN models were constructed with the optimal independent variables and the biopsy results as dependent variable. Performances of the models and radiologists were evaluated on the test set using receiver-operating characteristic (ROC) analysis. Among BI-RADS descriptors, margin and boundary were determined as the predictors according to stepwise LR showing moderate interobserver agreement. Area under the ROC curves (AUC) for both of LR and ANN were 0.87 (95% CI, 0.77–0.94). AUCs for the five radiologists ranged 0.79–0.91. There was no significant difference in AUC values among the LR, ANN, and radiologists (p > 0.05). Margin and boundary were found as statistically significant predictors with good interobserver agreement. Use of the LR and ANN showed similar performance to that of the radiologists for differentiation of benign and malignant breast masses.

Development of medical decision support system for leukemia management

Abstract A prototype Medical Decision Support System (MDSS) for leukemia patients was developed with emphasis on total management approach from patient registration to diagnosis and treatment. Thus, the MDSS consists of four modules: registry, knowledge model, simulator, and Computer-Assisted Instruction (CAI). Integration of each module improves overall patient management capability and knowledge acquisition capability of the system. Four different knowledge models were developed to predict diagnosis: rule-based reasoning, case-based reasoning, neural network, and discriminant analysis. Among the four, rule-based reasoning produced the most accurate prediction in diagnosis. In the future, the method of leukemia registry can further be extended to the hospital-based cancer registry for other types of cancer. In order to be more effective, the registry should also be integrated with the hospital information system for an easier data entry.