Hon-Yi Shi

Comparison of Artificial Neural Network and Logistic Regression Models for Predicting In-Hospital Mortality after Primary Liver Cancer Surgery

Background Since most published articles comparing the performance of artificial neural network (ANN) models and logistic regression (LR) models for predicting hepatocellular carcinoma (HCC) outcomes used only a single dataset, the essential issue of internal validity (reproducibility) of the models has not been addressed. The study purposes to validate the use of ANN model for predicting in-hospital mortality in HCC surgery patients in Taiwan and to compare the predictive accuracy of ANN with that of LR model. Methodology/Principal Findings Patients who underwent a HCC surgery during the period from 1998 to 2009 were included in the study. This study retrospectively compared 1,000 pairs of LR and ANN models based on initial clinical data for 22,926 HCC surgery patients. For each pair of ANN and LR models, the area under the receiver operating characteristic (AUROC) curves, Hosmer-Lemeshow (H-L) statistics and accuracy rate were calculated and compared using paired T-tests. A global sensitivity analysis was also performed to assess the relative significance of input parameters in the system model and the relative importance of variables. Compared to the LR models, the ANN models had a better accuracy rate in 97.28% of cases, a better H-L statistic in 41.18% of cases, and a better AUROC curve in 84.67% of cases. Surgeon volume was the most influential (sensitive) parameter affecting in-hospital mortality followed by age and lengths of stay. Conclusions/Significance In comparison with the conventional LR model, the ANN model in the study was more accurate in predicting in-hospital mortality and had higher overall performance indices. Further studies of this model may consider the effect of a more detailed database that includes complications and clinical examination findings as well as more detailed outcome data.

The effect of preoperative transarterial chemoembolization of resectable hepatocellular carcinoma on clinical and economic outcomes

Hepatocellular carcinoma (HCC) is one of the most malignant cancers in the world. The effect of preoperative transarterial chemoembolization (TACE) for resectable HCC is still controversial and cost‐associated treatments are unknown.

Responsiveness of the Harris Hip Score and the SF-36: five years after total hip arthroplasty

PurposeComparing the responsiveness over time of the Harris Hip Score (HHS) and the SF-36 in patients who underwent total hip arthroplasty (THA) and assessing variation in the responsiveness of these measures by the number of co-morbidities.MethodsThis prospective study analyzed 335 THA patients treated at two southern Taiwan hospitals from 1997 to 2000. Magnitude of change in HRQoL was compared by generalized estimating equation. Bias-corrected and accelerated bootstrapping was used to measure magnitude of change in HHS and SF-36 subscale scores for five different time intervals spanning a 5-year period.ResultsThe analytical results indicated that the pain and physical function subscales of the HHS are more responsive than those of the SF-36 for short-term (within 1 year post-surgery) measurements but are less responsive for long-term measurements. At various follow-up intervals, the HHS and the SF-36 significantly differed in ES of changes in pain and physical function subscale scores for patients with one co-morbidity and for patients with two or more co-morbidities.ConclusionFor long-term evaluation of THA patients, clinicians and health researchers should weight both measures equally and should also consider co-morbidities.

Effects of Provider Patient Volume and Comorbidity on Clinical and Economic Outcomes for Total Knee Arthroplasty A Population-Based Study

Our study examined how provider patient volume, postoperative infection rate, and perioperative complication affect length of stay, hospitalization charges, and adverse outcomes for patients undergoing total knee arthroplasty (TKA). The study sample included patients who had undergone total knee arthroplasty at all acute care hospitals in Taiwan between 2000 and 2003. Two economic indicators revealed linear associations with surgeons patient volume, hospitals patient volume, and comorbidity score. Patients who developed postoperative infections remained hospitalized an average of 8.49 days longer than did patients with no infection. Postoperative infection was associated with surgeon experience. Our findings indicate that a surgeons patient volume has a more significant effect than a hospitals patient volume on clinical outcomes. However, patient volumes for both surgeon and hospital are equally important in economic outcomes.

Responsiveness and minimal important differences after revision total hip arthroplasty

BackgroundThe health-related quality of life (HRQoL) is currently weighted more heavily when evaluating health status, particularly regarding medical treatments and interventions. However, it is rarely used by physicians to compare responsiveness. Additionally, responsiveness estimates derived by the Harris Hip Score (HHS) and the Short Form 36 (SF-36) before and after revision total hip arthroplasty (THA) have not been clinically compared. This study compared responsiveness and minimal important differences (MID) between HHS and SF-36.MethodsAll revision THA patients completed the disease-specific HHS and the generic SF-36 before and 6 months after surgery. Scores using these instruments were interpreted by generalized estimating equation (GEE) before and after revision THA. The bootstrap estimation and modified Jacknife test were used to derive 95% confidence intervals for differences in the responsiveness estimates.ResultsComparisons of effect size (ES), standardized response means (SRM), relative efficiency (RE) (>1) and MID indicated that the responsiveness of HHS was superior to that of SF-36. The ES and SRM for pain and physical functions in the HHS were significantly larger than those of the SF-36 (p < 0.001).ConclusionThe data in this study indicated that clinicians and health researchers should weight disease-specific measures more heavily than generic measures when evaluating treatment outcomes.

Temporal trends and volume-outcome associations after traumatic brain injury: a 12-year study in Taiwan.

OBJECT The purpose of this study was to evaluate temporal trends in traumatic brain injury (TBI); the impact of hospital volume and surgeon volume on length of stay (LOS), hospitalization cost, and in-hospital mortality rate; and to explore predictors of these outcomes in a nationwide population in Taiwan. METHODS This population-based patient cohort study retrospectively analyzed 16,956 patients who had received surgical treatment for TBI between 1998 and 2009. Bootstrap estimation was used to derive 95% confidence intervals for differences in effect sizes. Hierarchical linear regression models were used to predict outcomes. RESULTS Patients treated in very-high-volume hospitals were more responsive than those treated in low-volume hospitals in terms of LOS (-0.11; 95% CI -0.20 to -0.03) and hospitalization cost (-0.28; 95% CI -0.49 to -0.06). Patients treated by high-volume surgeons were also more responsive than those treated by low-volume surgeons in terms of LOS (-0.19; 95% CI -0.37 to -0.01) and hospitalization cost (-0.43; 95% CI -0.81 to -0.05). The mean LOS was 24.3 days and the average LOS for very-high-volume hospitals and surgeons was 61% and 64% shorter, respectively, than that for low-volume hospitals and surgeons. The mean hospitalization cost was US

Trends and outcome predictors after traumatic brain injury surgery: a nationwide population-based study in Taiwan

7,292.10, and the average hospitalization cost for very-high-volume hospitals and surgeons was 19% and 22% lower, respectively, than that for low-volume hospitals and surgeons. Advanced age, male sex, high Charlson Comorbidity Index score, treatment in a low-volume hospital, and treatment by a low-volume surgeon were significantly associated with adverse outcomes (p < 0.001). CONCLUSIONS The data suggest that annual surgical volume is the key factor in surgical outcomes in patients with TBI. The results improve the understanding of medical resource allocation for this surgical procedure, and can help to formulate public health policies for optimizing hospital resource utilization for related diseases.

Is wider surgical margin justified for better clinical outcomes in patients with resectable hepatocellular carcinoma

OBJECT The authors sought to analyze trends in hospital resource utilization and mortality rates in a population of patients who had received traumatic brain injury (TBI) surgery. METHODS This nationwide population-based cohort study retrospectively analyzed 18,286 patients who had received surgical treatment for TBI between 1998 and 2010. The multiple linear regression model and Cox proportional hazards model were used for multivariate assessment of outcome predictors. RESULTS The prevalence rate of surgical treatment for patients with TBI gradually but significantly (p < 0.001) increased by 47.6% from 5.0 per 100,000 persons in 1998 to 7.4 per 100,000 persons in 2010. Age, sex, Deyo-Charlson comorbidity index score, hospital volume, and surgeon volume were significantly associated with TBI surgery outcomes (p < 0.05). Over the 12-year period analyzed, the estimated mean hospital treatment cost increased 19.06%, whereas the in-hospital mortality rate decreased 10.9%. The estimated mean time of overall survival after TBI surgery (± SD) was 83.0 ± 4.2 months, and the overall in-hospital and 1-, 3-, and 5-year survival rates were 74.5%, 67.3%, 61.1%, and 57.8%, respectively. CONCLUSIONS These data reveal an increased prevalence of TBI, especially in older patients, and an increased hospital treatment cost but a decreased in-hospital mortality rate. Health care providers and patients should recognize that attributes of the patient and of the hospital may affect hospital resource utilization and the mortality rate. These results are relevant not only to other countries with similar population sizes but also to countries with larger populations.

Long-Term Prediction of Emergency Department Revenue and Visitor Volume Using Autoregressive Integrated Moving Average Model

BACKGROUND/PURPOSE Surgical resection for hepatocellular carcinoma (HCC) is regarded as a curable treatment; however, the postoperative recurrence still poses a challenge to surgeons. The effect of surgical margin on long-term outcome is still controversial, although it has been considered as the treatment-related risk factor for recurrence. A precise assessment of the effects of surgical margin on clinical outcome is required to clarify the issue. METHODS A retrospective study was conducted on 407 patients with microscopically complete resection of HCC; they were divided into three groups with surgical margin negative by 1-5 mm (Group A, n = 156), 6-10 mm (Group B, n = 109), and wider than 10 mm (Group C, n = 142). The groups were compared for clinicopathologic characteristics, perioperative features, postoperative recurrence, and long-term outcome. RESULTS The median follow-up time for all patients was 72.97 months. Recurrence rates were similar among these groups. There was no significant difference in the overall 1-, 3-, and 5-year actual survival rates for the groups on the log-rank test (p = 0.073). After controlling the independent risks for disease-free survival, there was also no significant difference in the 1-, 3-, and 5-year disease-free survival rates for the groups (p = 0.354). The patients with wider surgical margin had worse perioperative outcomes; more patients in this group needed blood transfusion (p < 0.001) and more patients suffered from postoperative complications (p = 0.020). They also had higher in-hospital mortality rate than that of other groups (1.41% vs. 0.64%). CONCLUSION No superiority was seen in patients with wider surgical margin in either perioperative features or long-term outcome.

Health-related quality of life after total hip replacement: a Taiwan study.

This study analyzed meteorological, clinical and economic factors in terms of their effects on monthly ED revenue and visitor volume. Monthly data from January 1, 2005 to September 30, 2009 were analyzed. Spearman correlation and cross-correlation analyses were performed to identify the correlation between each independent variable, ED revenue, and visitor volume. Autoregressive integrated moving average (ARIMA) model was used to quantify the relationship between each independent variable, ED revenue, and visitor volume. The accuracies were evaluated by comparing model forecasts to actual values with mean absolute percentage of error. Sensitivity of prediction errors to model training time was also evaluated. The ARIMA models indicated that mean maximum temperature, relative humidity, rainfall, non-trauma, and trauma visits may correlate positively with ED revenue, but mean minimum temperature may correlate negatively with ED revenue. Moreover, mean minimum temperature and stock market index fluctuation may correlate positively with trauma visitor volume. Mean maximum temperature, relative humidity and stock market index fluctuation may correlate positively with non-trauma visitor volume. Mean maximum temperature and relative humidity may correlate positively with pediatric visitor volume, but mean minimum temperature may correlate negatively with pediatric visitor volume. The model also performed well in forecasting revenue and visitor volume.