Chien-Yeh Hsu

Taipei Medical University, T’ai-pei, Taipei, Taiwan

Are you Chien-Yeh Hsu?

Claim your profile

Publications (64)81.77 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: Background We developed an artificial neural network (ANN) model to predict prostate cancer pathological staging in patients prior to when they received radical prostatectomy as this is more effective than logistic regression (LR), or combined use of age, prostate-specific antigen (PSA), body mass index (BMI), digital rectal examination (DRE), trans-rectal ultrasound (TRUS), biopsy Gleason sum, and primary biopsy Gleason grade. Methods Our study evaluated 299 patients undergoing retro-pubic radical prostatectomy or robotic-assisted laparoscopic radical prostatectomy surgical procedures with pelvic lymph node dissection. The results were intended to predict the pathological stage of prostate cancer (T2 or T3) after radical surgery. The predictive ability of ANN was compared with LR and validation of the 2007 Partin Tables was estimated by the areas under the receiving operating characteristic curve (AUCs). Results Of the 299 patients we evaluated, 109 (36.45%) displayed prostate cancer with extra-capsular extension (ECE), and 190 (63.55%) displayed organ-confined disease (OCD). LR analysis showed that only PSA and BMI were statistically significant predictors of prostate cancer with capsule invasion. Overall, ANN outperformed LR significantly (0.795 ± 0.023 versus 0.746 ± 0.025, p = 0.016). Validation using the current Partin Tables for the participants of our study was assessed, and the predictive capacity of AUC for OCD was 0.695. Conclusion ANN was superior to LR at predicting OCD in prostate cancer. Compared with the validation of current Partin Tables for the Taiwanese population, the ANN model resulted in larger AUCs and more accurate prediction of the pathologic stage of prostate cancer.
    Journal of the Chinese Medical Association 10/2014; · 0.75 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: To surveyed the quantities, types, and related information of potential drug-drug interactions (DDIs) and estimate the off-label use percentage of pediatric outpatient prescriptions for newborns and infants from the National Health Insurance Research Database (NHIRD) of Taiwan. Adverse drug reactions (ADR) may cause morbidity and mortality, potential drug-drug interactions (DDI) increase the probability of ADR. Research on ADR and DDI in infants is of particular urgency and importance but the related profiles in these individuals are not well known. All prescriptions written by physicians in 2000 were analyzed to identify potential DDIs among drugs appearing on the same prescription sheet. Of a total of 150.6 million prescription sheets, with 669.5 million prescriptions registered in the NHIRD of Taiwan, six million (3.99%) prescription sheets were for 2.1 million infants with 19.4 million (2.85%) prescriptions. There were 672,020 potential DDIs in this category, accounting for 3.53% per prescription; an estimated one DDI in every three patients. The interactions between aspirin and aluminum/magnesium hydroxide were most common (4.42%). Of the most significant drug-drug interactions, the interaction of digoxin with furosemide ranked first (20.14%), followed by the interactions of cisapride with furosemide and erythromycin (6.02% and 4.85%, respectively). The interactions of acetaminophen and anti-cholinergic agents comprised most types of drug-drug interactions (6.62%). Although the prevalence rates of DDIs are low, life-threatening interactions may develop. Physicians must be reminded of the potential DDIs when prescribing medications for newborns and infants.
    Computer methods and programs in biomedicine 09/2013; · 1.56 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Slowing of average electroencephalography (EEG) frequency in Alzheimer's disease (AD) is well established, but whether EEG changes are able to reflect the severity of AD is uncertain. We attempt to establish quantitative EEG parameters that are suitable for evaluating AD in clinical practice. Ninety-five patients with newly diagnosed AD at different stages from four neurologic institutes were enrolled for the study. Standard scalp resting EEG data were collected for quantitative analysis. Global band power ratio and interhemispheric alpha band coherence were calculated. Patients with advanced AD had a greater slow-to-fast wave power ratio. Among several power ratio parameters, global theta and delta to alpha and beta band power ratio showed the best correlation with stages of AD (p < 0.05 between any two patient groups). Patients with advanced AD had decreased coherence in multiple brain regions. The phenomenon was most prominent in the centroparietal region (p < 0.05 between any two patient groups). Increased global slow-to-fast power ratio and decreased centroparietal interhemispheric alpha band coherence are strongly correlated with disease progress in AD patients. These two quantitative EEG parameters may help evaluate AD patients in daily clinical practice. Global power ratio changes may suggest a shift of dominant frequency, and decreased interhemispheric alpha band coherence may suggest functional disconnection and corpus callosum abnormalities in AD patients.
    Journal of the Formosan Medical Association 08/2013; · 1.70 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: INTRODUCTION: Adverse drug reactions (ADR) increase morbidity and mortality; potential drug-drug interactions (DDI) increase the probability of ADR. Studies have proven that computerized drug-interaction alert systems (DIAS) might reduce medication errors and potential adverse events. However, the relatively high override rates obscure the benefits of alert systems, which result in barriers for availability. It is important to understand the frequency at which physicians override DIAS and the reasons for overriding reminders. METHOD: All the DDI records of outpatient prescriptions from a tertiary university hospital from 2005 and 2006 detections by the DIAS are included in the study. The DIAS is a JAVA language software that was integrated into the computerized physician order entry system. The alert window is displayed when DDIs occur during order entries, and physicians choose the appropriate action according to the DDI alerts. There are seven response choices are obligated in representing overriding and acceptance: (1) necessary order and override; (2) expected DDI and override; (3) expected DDI with modified dosage and override; (4) no DDI and override; (5) too busy to respond and override; (6) unaware of the DDI and accept; and (7) unexpected DDI and accept. The responses were collected for analysis. RESULTS: A total of 11,084 DDI alerts of 1,243,464 outpatient prescriptions were present, 0.89% of all computerized prescriptions. The overall rate for accepting was 8.5%, but most of the alerts were overridden (91.5%). Physicians of family medicine and gynecology-obstetrics were more willing to accept the alerts with acceptance rates of 20.8% and 20.0% respectively (p<0.001). Information regarding the recognition of DDIs indicated that 82.0% of the DDIs were aware by physicians, 15.9% of DDIs were unaware by physicians, and 2.1% of alerts were ignored. The percentage of total alerts declined from 1.12% to 0.79% during 24 months' study period, and total overridden alerts also declined (from 1.04% to 0.73%). CONCLUSION: We explored the physicians' behavior by analyzing responses to the DDI alerts. Although the override rate is still high, the reasons why physicians may override DDI alerts were well analyzed and most DDI were recognized by physicians. Nonetheless, the trend of total overrides is in decline, which indicates a learning curve effect from exposure to DIAS. By analyzing the computerized responses provided by physicians, efforts should be made to improve the efficiency of the DIAS, and pharmacists, as well as patient safety staffs, can catch physicians' appropriate reasons for overriding DDI alerts, improving patient safety.
    Computer methods and programs in biomedicine 04/2013; · 1.56 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSE: We developed a nursing process decision support system (NPDSS) based on three clinical pathways, including benign prostatic hypertrophy, inguinal hernia, and urinary tract stone. NPDSS included six major nursing diagnoses - acute pain, impaired urinary elimination, impaired skin integrity, anxiety, infection risk, and risk of falling. This paper aims to describe the design, development and validation process of the NPDSS. METHODS: We deployed the Delphi method to reach consensus for decision support rules of NPDSS. A team of nine-member expert nurses from a medical center in Taiwan was involved in Delphi method. The Cronbach's α method was used for examining the reliability of the questionnaire used in the Delphi method. The Visual Basic 6.0 as front-end and Microsoft Access 2003 as back-end was used to develop the system. A team of six nursing experts was asked to evaluate the usability of the developed systems. A 5-point Likert scale questionnaire was used for the evaluation. The sensitivity and specificity of NPDSS were validated using 150 nursing chart. RESULTS: The study showed a consistency between the diagnoses of the developed system (NPDSS) and the nursing charts. The sensitivities of the nursing diagnoses including acute pain, impaired urinary elimination, risk of infection, and risk of falling were 96.9%, 98.1%, 94.9%, and 89.9% respectively; and the specificities were 88%, 49.5%, 62%, and 88% respectively. We did not calculate the sensitivity and specificity of impaired skin integrity and anxiety due to non-availability of enough sample size. CONCLUSIONS: NPDSS can help nurses in decision making of nursing diagnoses. Besides, it can help them to generate nursing diagnoses based on patient-specific data, individualized care plans, and implementation within their usual nursing workflow.
    International Journal of Medical Informatics 03/2013; · 2.72 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Public health informatics has been defined as the systematic application of information and computer science and technology to public health practice, research, and learning [1]. Unfortunately, limited reports exist concerning to the capacity building strategies to improve public health informatics workforce in limited-resources setting. In Indonesia, only three universities, including Universitas Gadjah Mada (UGM), offer master degree program on related public health informatics discipline. UGM started a new dedicated master program on Health Management Information Systems in 2005, under the auspice of the Graduate Program of Public Health at the Faculty of Medicine. This is the first tracer study to the alumni aiming to a) identify the gaps between curriculum and the current jobs and b) describe their perception on public health informatics competencies. We distributed questionnaires to 114 alumni with 36.84 % response rate. Despite low response rate, this study provided valuable resources to set up appropriate competencies, curriculum and capacity building strategies of public health informatics workforce in Indonesia.
    Studies in health technology and informatics 01/2013; 192:1076.
  • Source
    The American journal of tropical medicine and hygiene 12/2012; 87(6):1152. · 2.53 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Sex-related factors play an important role in the pathophysiology of heart failure (HF). However, trends in sex-related differences in hospital management for HF are not clear. We identified patients hospitalized for HF through a nationwide database (National Health Insurance in Taiwan), containing 722,272 subjects from 1999 to 2008. Higher incidences of diabetes mellitus (37 vs. 25 %, p < 0.001), thyroid dysfunction (2 vs. 0 %, p < 0.001), and transient cerebral ischemia (2 vs. 1 %, p < 0.05), as well as a lower incidence of chronic lung disease (14 vs. 22 %, p < 0.001) differentiated female HF patients from male HF patients. During this 10-year period, both percentage of HF hospitalization and age-adjusted HF rates significantly increased for total HF sample (1.92 vs. 2.49 ‰, p < 0.05, and 20.44 vs. 27.38/100,000, p < 0.05) and for female (1.76 vs. 2.86 ‰, p < 0.05, and 20.94 vs. 32.12/100,000, p < 0.05), but such changes did not occur among male patients (2.12 vs. 2.09 ‰, p > 0.05, and 19.93 vs. 22.51/100,000, p > 0.05). The age at the time of hospitalization and the length of the hospital stay increased significantly for all HF patients during the 10-year study period. However, the daily cost of hospitalization increased in males, but not in females. Compared to the survivors, patients who died were older and had a longer hospitalization and higher daily cost both in males and females. Through our analysis of the NHI database, we observed trends in factors related to hospitalization of HF patients in Taiwan that may be attributable to sex-related differences in the pathophysiology and treatment strategies for HF.
    Heart and Vessels 10/2012; · 2.13 Impact Factor
  • Anis Fuad, Chien-Yeh Hsu, Aprisa Chrysantina
    Medical Teacher 08/2012; · 2.05 Impact Factor
  • Journal of Evaluation in Clinical Practice 08/2012; 18(4):925. · 1.58 Impact Factor
  • Anis Fuad, Chien-Yeh Hsu
    Journal of thoracic oncology: official publication of the International Association for the Study of Lung Cancer 06/2012; 7(6):1064. · 4.55 Impact Factor
  • Anis Fuad, Chien-Yeh Hsu
    International Journal of Medical Informatics 05/2012; 81(9):649-50. · 2.72 Impact Factor
  • Yuchuan Chen, Chien-Yeh Hsu, Li Liu, Sherry Yang
    [Show abstract] [Hide abstract]
    ABSTRACT: This paper presents a research of constructing a web-based expert system for nutrition diagnosis by utilizing the expert system techniques in artificial intelligence. The research implements Nutritional Care Process and Model (NCPM) defined by American Dietetic Association (ADA) in 2008 and integrate the nutrition diagnosis knowledge from dietetics professionals to establish the basics of building the rule-based expert system with its knowledge base. The system is built using Microsoft Visual Studio 2008 on .NET Framework 3.5SP1 utilizing the built in rule engine which comes with Windows Workflow Foundation.With the help of this system, it is easier for dietetics professionals to adapt to the newly introduced concept of nutrition diagnosis. At the heart of the web based expert system is a knowledge base, it has a rule engine which contains the nutrition diagnosis rules converted from signs and symptoms for nutrition diagnosis from dietetics professionals and are expressed in XML format which are then stored in a SQL database. A knowledge engineer will be able to use a rule editor to add new rules or to update existing rules within the rule database. Dietetics professionals would be able to enter patient’s basic data, anthropometric data, physical exam findings, biochemical data, and food/nutrition history into the program. After dietetics professionals complete nutrition assessment, the program will make inference to the rule base and make nutrition diagnosis. Dietetics professionals could then make the final diagnosis decision for the patient based on the diagnosis report generated by the web based nutrition diagnosis expert system.For this study, I have selected 100 chronic kidney disease patients under hemodialysis from a university hospital, recorded their albumin, cholesterol, creatinine before dialysis, height, and dry weight and then use these data to perform nutrition diagnosis with both the expert system and a practicing dietitian. After comparing the result, I found that the expert system is faster and more accurate than human dietitian.
    Expert Systems with Applications 02/2012; 39:2132-2156. · 1.97 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: It is not clear whether gender is associated with different hospitalization cost and lengths for acute myocardial infarction (AMI). We identified patients hospitalized for primary diagnosis of AMI with (STEMI) or without (NSTEMI) ST elevation from 1999 to 2008 through a national database containing 1,000,000 subjects. As compared to that in 1999~2000, total (0.35‰  versus 0.06‰, P < 0.001) and male (0.59‰  versus 0.07‰, P < 0.001) STEMI hospitalization percentages were decreased in 2007~2008, but female STEMI hospitalization percentages were not different from 1999 to 2008. However, NSTEMI hospitalization percentages were similar over the 10-year period. The hospitalization age for AMI, STEMI, and NSTEMI was increased over the 10-year period by 14, 9, and 7 years in male, and by 18, 18, and 21 years in female. The female and male hospitalization cost and lengths were similar in the period. As compared to nonmedical center, the hospitalization cost for STEMI in medical center was higher in male patients, but not in female patients, and the hospitalization cost for NSTEMI was higher in both male and female gender. We found significant differences between male and female, medical center and non-medical center, or STEMI and NSTEMI on medical care over the 10-year period.
    The Scientific World Journal 01/2012; 2012:184075. · 1.22 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: This study proposed a recognized system for electroencephalogram (EEG) data classification. In addition to the wavelet-based amplitude modulation (AM) features, the fuzzy c-means (FCM) clustering is used for the discriminant of left finger lifting and resting. The features are extracted from discrete wavelet transform (DWT) data with the AM method. The FCM is then applied to recognize extracted features. Compared with band power features, k-means clustering, and linear discriminant analysis (LDA) classifier, the results indicate that the proposed method is satisfactory in applications of brain-computer interface (BCI).
    Clinical EEG and neuroscience: official journal of the EEG and Clinical Neuroscience Society (ENCS) 01/2012; 43(1):32-8. · 3.16 Impact Factor
  • Cheng-Mei Chen, Chien-Yeh Hsu, Chyi-Huey Bai
    [Show abstract] [Hide abstract]
    ABSTRACT: The aim of this study is to establish a decision system to assist liver cancer patients to choose treatments, and explore whether patients' treatment preferences were concordant with their physicians' recommendations. We analyzed the patients' considerations involved in the evaluation of treatment and combined these considerations with Barcelona Clinic Liver Cancer (BCLC) clinical guidelines to design the surveys and adopted Multi-Attribute Utility Theory (MAUT) to establish a decision model for treatment selections. Without intervening the clinical processes, we interviewed 53 liver cancer patients(and their families) using this decision system to choose preferred treatments. The results showed: (1) There was a difference between liver cancer patients' treatment preferences and their physicians' recommendations, (2) Liver cancer patients with advanced stage, still prefer active treatment, (3) The most concerns on treatments were cure rate, survival rate and self-care ability. Instead of only following physicians' recommendations, this system assist liver cancer patients to evaluate treatment using various factors(considerations) including physiological, psychological and social support. For oncologists, we suggest the psychological processes of patients are taken into consideration, and this study provide an example of framework on cancer care interventions and expect to improve the care of liver cancer population in the future.
    Intelligent Systems (GCIS), 2012 Third Global Congress on; 01/2012
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: This study investigated whether diffusion tensor imaging (DTI) could identify potential abnormalities in type 2 diabetes mellitus (T2DM) patients without cognitive complaints compared to healthy controls. In addition, the existence of associations between diffusion measures and clinical parameters was examined. Forty T2DM patients and 97 non-diabetic controls completed a clinical and biochemistry examination. Structural MRI scans (DTI, T1, T2, FLAIR) were subsequently acquired with a 1.5 Tesla scanner. In addition to a global DTI analysis, voxel-based analysis was performed on the fractional anisotropy (FA), mean diffusivity (MD), and axial (AD) and transverse (TD) diffusivity maps to investigate regions that exhibit (i) WM differences between patients and controls; and (ii) associations between clinical measurements and these DTI indices. There were no significant differences in age, gender, and WM hyperintensity scores derived by the conventional MRI scans between controls and T2DM patients. For the T2DM patients, however, the MD of the brain parenchyma was significantly increased compared to controls and was positively correlated with disease duration. The voxel based analyses revealed (i) a significantly decreased FA in the bilateral frontal WM compared to controls which was mainly caused by an increased TD and not a decreased AD within these regions; (ii) a significant association between disease duration and microstructural properties in several brain regions including bilateral cerebellum, temporal lobe WM, right caudate, bilateral cingulate gyrus, pons, and parahippocampal gyrus. Our findings indicate that microstructural WM abnormalities and associations with clinical measurements can be detected with DTI in T2DM patients.
    NeuroImage 09/2011; 59(2):1098-105. · 6.13 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Maintaining a large diagnostic knowledge base (KB) is a demanding task for any person or organization. Future clinical decision support system (CDSS) may rely on multiple, smaller and more focused KBs developed and maintained at different locations that work together seamlessly. A cross-domain inference tool has great clinical import and utility. We developed a modified multi-membership Bayes formulation to facilitate the cross-domain probabilistic inferencing among KBs with overlapping diseases. Two KBs developed for evaluation were non-infectious generalized blistering diseases (GBD) and autoimmune diseases (AID). After the KBs were finalized, they were evaluated separately for validity. Ten cases from medical journal case reports were used to evaluate this "cross-domain" inference across the two KBs. The resultant non-error rate (NER) was 90%, and the average of probabilities assigned to the correct diagnosis (AVP) was 64.8% for cross-domain consultations. A novel formulation is now available to deal with problems occurring in a clinical diagnostic decision support system with multi-domain KBs. The utilization of this formulation will help in the development of more integrated KBs with greater focused knowledge domains.
    Computer methods and programs in biomedicine 09/2011; 104(2):286-91. · 1.56 Impact Factor
  • International journal of cardiology 09/2011; 153(1):89-94. · 6.18 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: According to the Emergency Medical Services Act of Taiwan, rescue personnel are responsible on-site services for emergency patient. They should transport the patient from emergency site to the appropriate medical care institution in the vicinity as directed by Emergency Dispatch Center. To date, unfortunately, the criteria to decide appropriate medical care institution for emergency patient are not very clear. Generally, Emergency Medical Technicians (EMTs) lack the knowledge like variables to refer and prioritize variables while deciding an optimal hospital for emergency patient. The aim of this research is to study what variables can be used as reference by EMT to make decision and to create a computational model to direct the emergency patients to most appropriate hospital. The main purpose of this study is to assist with emergency
    2nd IEEE International Conference on Emergency Management and Management Sciences (ICEMMS), 2011; 08/2011

Publication Stats

202 Citations
81.77 Total Impact Points


  • 2002–2014
    • Taipei Medical University
      • Graduate Institute of Medical Informatics
      T’ai-pei, Taipei, Taiwan
  • 2013
    • Chang Gung University of Science and Technology
      Chang-hua Pei-pu, Taiwan, Taiwan
    • Oriental Institute of Technology, Taiwan
      T’ai-pei, Taipei, Taiwan
  • 2012–2013
    • Gadjah Mada University
      • Department of Public Health
      Yogyakarta, Daerah Istimewa Yogyakarta, Indonesia
  • 2011–2012
    • Wan Fang Hospital
      T’ai-pei, Taipei, Taiwan
  • 2006
    • National Yang Ming University
      • Institute of Public Health
      Taipei, Taipei, Taiwan