3D dissimilar-siamese-u-net for hyperdense Middle cerebral artery sign segmentation.

The hyperdense middle cerebral artery sign (HMCAS) representing a thromboembolus has been declared as a vital CT finding for intravascular thrombus in the diagnosis of acute ischemia stroke. Early recognition of HMCAS can assist in patient triage and subsequent thrombolysis or thrombectomy treatment. A total of 624 annotated head non-contrast-enhanced CT (NCCT) image scans were retrospectively collected from multiple public hospitals in Hong Kong. In this study, we present a deep Dissimilar-Siamese-U-Net (DSU-Net) that is able to precisely segment the lesions by integrating Siamese and U-Net architectures. The proposed framework consists of twin sub-networks that allow inputs of left and right hemispheres in head NCCT images separately. The proposed Dissimilar block fully explores the feature representation of the differences between the bilateral hemispheres. Ablation studies were carried out to validate the performance of various components of the proposed DSU-Net. Our findings reveal that the proposed DSU-Net provides a novel approach for HMCAS automatic segmentation and it outperforms the baseline U-Net and many state-of-the-art models for clinical practice.

Development of a data-driven COVID-19 prognostication tool to inform triage and step-down care for hospitalised patients in Hong Kong: a population-based cohort study.

BACKGROUND: This is the first study on prognostication in an entire cohort of laboratory-confirmed COVID-19 patients in the city of Hong Kong. Prognostic tool is essential in the contingency response for the next wave of outbreak. This study aims to develop prognostic models to predict COVID-19 patients' clinical outcome on day 1 and day 5 of hospital admission. METHODS: We did a retrospective analysis of a complete cohort of 1037 COVID-19 laboratory-confirmed patients in Hong Kong as of 30 April 2020, who were admitted to 16 public hospitals with their data sourced from an integrated electronic health records system. It covered demographic information, chronic disease(s) history, presenting symptoms as well as the worst clinical condition status, biomarkers' readings and Ct value of PCR tests on Day-1 and Day-5 of admission. The study subjects were randomly split into training and testing datasets in a 8:2 ratio. Extreme Gradient Boosting (XGBoost) model was used to classify the training data into three disease severity groups on Day-1 and Day-5. RESULTS: The 1037 patients had a mean age of 37.8 (SD ± 17.8), 53.8% of them were male. They were grouped under three disease outcome: 4.8% critical/serious, 46.8% stable and 48.4% satisfactory. Under the full models, 30 indicators on Day-1 and Day-5 were used to predict the patients' disease outcome and achieved an accuracy rate of 92.3% and 99.5%. With a trade-off between practical application and predictive accuracy, the full models were reduced into simpler models with seven common specific predictors, including the worst clinical condition status (4-level), age group, and five biomarkers, namely, CRP, LDH, platelet, neutrophil/lymphocyte ratio and albumin/globulin ratio. Day-1 model's accuracy rate, macro-/micro-averaged sensitivity and specificity were 91.3%, 84.9%/91.3% and 96.0%/95.7% respectively, as compared to 94.2%, 95.9%/94.2% and 97.8%/97.1% under Day-5 model. CONCLUSIONS: Both Day-1 and Day-5 models can accurately predict the disease severity. Relevant clinical management could be planned according to the predicted patients' outcome. The model is transformed into a simple online calculator to provide convenient clinical reference tools at the point of care, with an aim to inform clinical decision on triage and step-down care.

Automated Hierarchy Evaluation System of Large Vessel Occlusion in Acute Ischemia Stroke.

BACKGROUND: The detection of large vessel occlusion (LVO) plays a critical role in the diagnosis and treatment of acute ischemic stroke (AIS). Identifying LVO in the pre-hospital setting or early stage of hospitalization would increase the patients' chance of receiving appropriate reperfusion therapy and thereby improve neurological recovery. METHODS: To enable rapid identification of LVO, we established an automated evaluation system based on all recorded AIS patients in Hong Kong Hospital Authority's hospitals in 2016. The 300 study samples were randomly selected based on a disproportionate sampling plan within the integrated electronic health record system, and then separated into a group of 200 patients for model training, and another group of 100 patients for model performance evaluation. The evaluation system contained three hierarchical models based on patients' demographic data, clinical data and non-contrast CT (NCCT) scans. The first two levels of modeling utilized structured demographic and clinical data, while the third level involved additional NCCT imaging features obtained from deep learning model. All three levels' modeling adopted multiple machine learning techniques, including logistic regression, random forest, support vector machine (SVM), and eXtreme Gradient Boosting (XGboost). The optimal cut-off for the likelihood of LVO was determined by the maximal Youden index based on 10-fold cross-validation. Comparisons of performance on the testing group were made between these techniques. RESULTS: Among the 300 patients, there were 160 women and 140 men aged from 27 to 104 years (mean 76.0 with standard deviation 13.4). LVO was present in 130 (43.3%) patients. Together with clinical and imaging features, the XGBoost model at the third level of evaluation achieved the best model performance on testing group. The Youden index, accuracy, sensitivity, specificity, F1 score, and area under the curve (AUC) were 0.638, 0.800, 0.953, 0.684, 0.804, and 0.847, respectively. CONCLUSION: To the best of our knowledge, this is the first study combining both structured clinical data with non-structured NCCT imaging data for the diagnosis of LVO in the acute setting, with superior performance compared to previously reported approaches. Our system is capable of automatically providing preliminary evaluations at different pre-hospital stages for potential AIS patients.

Burden of large vessel occlusion stroke and the service gap of thrombectomy: A population-based study using a territory-wide public hospital system registry.

BACKGROUND: Ischemic stroke due to large vessel occlusion can be effectively treated with thrombectomy but access to this treatment is limited in many parts of the world. Local incidence of large vessel occlusion is critical in determining the development of thrombectomy service, but reliable data from Asian countries are lacking. AIMS: We performed a population-based study to estimate the burden of large vessel occlusion and the service gap for thrombectomy in Hong Kong. METHODS: All acute ischemic stroke patients admitted in 2016 to the public healthcare system, which provided 90% of the emergency healthcare in the city, was identified from the Hong Kong Hospital Authority's central electronic database. The diagnosis of large vessel occlusion was retrospectively verified by two independent cerebrovascular specialists in a randomly sampled cohort based on clinical and neuroimaging data. The incidence of large vessel occlusion in the population was estimated through weighting the sample results and compared with the thrombectomy data in the same period. RESULTS: There were 6859 acute ischemic stroke patients treated in the public health system in 2016. Amongst the 300 patients randomly sampled according to diagnosis coding, 130 suffered from anterior circulation large vessel occlusion. This translated to 918 patients (95% CI 653-1180) and 13.3% of all ischemic stroke patients. The estimated incidence of anterior circulation large vessel occlusion was 12.5 per 100,000 persons per year (95% CI 11.7-13.4). Large vessel occlusion stroke patients were more commonly female than male (67.4% vs. 31.6%, p = 0.003), and were older than non-large vessel occlusion stroke patients (mean of 80.5 years vs. 71.4 years, p = < 0.001). They also had higher 30-day mortality rate (31.1% vs. 4.6%, p = < 0.001), and longer hospital stay (mean 38.6 vs. 21.1 days, p = 0.003) than non-large vessel occlusion stroke. In the same period, 83 thrombectomies for large vessel occlusion were performed, representing 9.1% of the estimated large vessel occlusion incidence. CONCLUSION: The estimated incidence of anterior circulation large vessel occlusion in the Hong Kong Chinese population is lower than that in the West. There is however a substantial service gap for endovascular thrombectomy with less than 10% of large vessel occlusion patients receiving thrombectomy.

Effects of Patient Empowerment Programme (PEP) on clinical outcomes and health service utilization in type 2 diabetes mellitus in primary care: an observational matched cohort study.

BACKGROUND: To evaluate the effects of a large population-based patient empowerment programme (PEP) on clinical outcomes and health service utilization rates in type 2 diabetes mellitus (T2DM) patients in the primary care setting. RESEARCH DESIGN AND SUBJECTS: A stratified random sample of 1,141 patients with T2DM enrolled to PEP between March and September 2010 were selected from general outpatient clinics (GOPC) across Hong Kong and compared with an equal number of T2DM patients who had not participated in the PEP (non-PEP group) matched by age, sex and HbA1C level group. MEASURES: Clinical outcomes of HbA1c, SBP, DBP and LDL-C levels, and health service utilization rates including numbers of visits to GOPC, specialist outpatient clinics (SOPC), emergency department (ED) and inpatient admissions, were measured at baseline and at 12-month post-recruitment. The effects of PEP on clinical outcomes and health service utilization rates were assessed by the difference-in-difference estimation, using the generalized estimating equation models. RESULTS: Compared with non-PEP group, PEP group achieved additional improvements in clinical outcomes over the 12-month period. A significantly greater percentage of patients in the PEP group attained HbA1C≤7% or LDL-C≤2.6 mmol/L at 12-month follow-up compared with the non-PEP group. PEP group had a mean 0.813 fewer GOPC visits in comparison with the non-PEP group. CONCLUSIONS: PEP was effective in improving the clinical outcomes and reduced the general outpatient clinic utilization rate over a 12-month period. Empowering T2DM patients on self-management of their disease can enhance the quality of diabetes care in primary care. TRIAL REGISTRATION: ClinicalTrials.gov NCT01935349.

Evaluation of the quality of care of a multi-disciplinary risk factor assessment and management programme (RAMP) for diabetic patients.

BACKGROUND: Type 2 Diabetes Mellitus (DM) is a common chronic disease associated with multiple clinical complications. Management guidelines have been established which recommend a risk-stratified approach to managing these patients in primary care. This study aims to evaluate the quality of care (QOC) and effectiveness of a multi-disciplinary risk assessment and management programme (RAMP) for type 2 diabetic patients attending government-funded primary care clinics in Hong Kong. The evaluation will be conducted using a structured and comprehensive evidence-based evaluation framework. METHOD/DESIGN: For evaluation of the quality of care, a longitudinal study will be conducted using the Action Learning and Audit Spiral methodologies to measure whether the pre-set target standards for criteria related to the structure and process of care are achieved. Each participating clinic will be invited to complete a Structure of Care Questionnaire evaluating pre-defined indicators which reflect the setting in which care is delivered, while process of care will be evaluated against the pre-defined indicators in the evaluation framework.Effectiveness of the programme will be evaluated in terms of clinical outcomes, service utilization outcomes, and patient-reported outcomes. A cohort study will be conducted on all eligible diabetic patients who have enrolled into RAMP for more than one year to compare their clinical and public service utilization outcomes of RAMP participants and non-participants. Clinical outcome measures will include HbA1c, blood pressure (both systolic and diastolic), lipids (low-density lipoprotein cholesterol) and future cardiovascular diseases risk prediction; and public health service utilization rate will include general and specialist outpatient, emergency department attendances, and hospital admissions annually within 5 years. For patient-reported outcomes, a total of 550 participants and another 550 non-participants will be followed by telephone to monitor quality of life, patient enablement, global rating of change in health and private health service utilization at baseline, 6, 12, 36 and 60 months. DISCUSSION: The quality of care and effectiveness of the RAMP in enhancing the health for patients with type 2 diabetes will be determined. Possible areas for quality enhancement will be identified and standards of good practice can be established. The information will be useful in guiding service planning and policy decision making.

Prognostication in severe acute respiratory syndrome: a retrospective time-course analysis of 1312 laboratory-confirmed patients in Hong Kong.

BACKGROUND AND OBJECTIVE: The temporal importance of prognostic indicators for severe acute respiratory syndrome (SARS) has not been studied. This study identified the various clinical prognostic factors for SARS and described the temporal evolution of these factors in the course of the SARS illness in Hong Kong in 2003. METHODS: A retrospective analysis of the entire Hong Kong cohort of 1312 laboratory-confirmed SARS patients aged 15-74 years was undertaken. Demographic, clinical and laboratory data at presentation and investigative data during the first 10 days of illness from the time of symptom onset were compiled. Two adverse outcomes were examined: hospital mortality and the development of oxygenation failure based on the estimated PaO(2)/FiO(2) ratio of <200 mm Hg. Logistic regression was used to identify the association between these prognostic factors and outcomes. RESULTS: Based on adjusted odds ratios with a P-value of <0.05, older age, male gender, elevated pulse rate and elevated neutrophil count were all predictive of oxygenation failure and death during the 10-day illness. Raised serum albumin and creatinine phosphokinase (CPK) levels were predictive of hospital mortality during this period. The presenting ALT and CPK level and the day 7 and day 10 platelet counts were predictive of oxygenation failure while the day 7 LDH was predictive of death. Contact exposure outside health-care institutions also appeared to carry higher risk of death. CONCLUSION: This large-scale analysis identified important discriminatory parameters related to the patients' demographic profile (age and gender), severity of illness (pulse rate and neutrophil count), and multisystem derangement (platelet count, CPK, ALT and LDH), all of which prognosticated adverse outcomes during the SARS episode. While age, pulse rate and neutrophil count consistently remained significant prognosticators during the first 10 days of illness, the prognostic impact of other derangements was more time-course dependent. Clinicians should be aware of the time-course evolution of these prognosticators.

Radiographic-clinical correlation in severe acute respiratory syndrome: study of 1373 patients in Hong Kong.

PURPOSE: To retrospectively analyze serial chest radiographs in all patients with severe acute respiratory syndrome (SARS) in Hong Kong for temporal changes and differences between patients who died and those who were discharged from the hospital and to compare radiographic and clinical parameters. MATERIALS AND METHODS: This retrospective study had ethics review board endorsement, and the need for informed consent was waived. Selected serial chest radiographs obtained from the time of presentation until discharge or death in 1373 patients with laboratory-confirmed SARS were scored. Scoring was based on the area and location of lung opacification on radiographs obtained at each of five milestones (presentation, beginning of ribavirin therapy, beginning of corticosteroid therapy, time of most severe radiographic appearance of disease, and before discharge or death). Extents of lung opacification at these five milestones were compared between patients who died and those who survived (by using a repeated-measures analysis of variance model), and the temporal trend of the radiographic-clinical parameters was analyzed (by using Cochran-Armitage trend testing, Kendall tau correlation coefficients, and descriptive graphic analysis). RESULTS: The final cohort consisted of 1373 patients (1212 of whom [485 male and 727 female patients; mean age, 38.4 years] survived and 161 of whom [84 male and 77 female patients; mean age, 63.0 years] died). Among survivors, older patients had more extensive radiographic changes than younger ones. However, among patients who died, older patients had less extensive radiographic opacification at the worst stage of disease and just before death than did younger patients. Despite a higher mortality risk for male patients, both sexes in the same outcome group had similar radiographic findings. For both outcome groups, the rate of radiographic progression was similar for the first 11 days but diverged afterwards. The extent of opacification increased by approximately one zone every 4-5 days for the initial 11 days. Radiographic scores correlated with the ratio of PaO2 to the fraction of inspired oxygen, lymphocyte count, lactate dehydrogenase level, and neutrophil count at each milestone and in terms of changes between milestones (P < .01 for all correlation coefficients, except for radiographic score and neutrophil count between the first two milestones). CONCLUSION: The initial extent of radiographic opacification may be useful for prognostic prediction. Radiographic progression correlates well with that of important clinical and laboratory parameters and may be used as an objective prognostic indicator early in SARS.

Non-invasive versus invasive mechanical ventilation for respiratory failure in severe acute respiratory syndrome.

BACKGROUND: Severe acute respiratory syndrome is frequently complicated by respiratory failure requiring ventilatory support. We aimed to compare the efficacy of non-invasive ventilation against invasive mechanical ventilation treating respiratory failure in this disease. METHODS: Retrospective analysis was conducted on all respiratory failure patients identified from the Hong Kong Hospital Authority Severe Acute Respiratory Syndrome Database. Intubation rate, mortality and secondary outcome of a hospital utilizing non-invasive ventilation under standard infection control conditions (NIV Hospital) were compared against 13 hospitals using solely invasive ventilation (IMV Hospitals). Multiple logistic regression analyses with adjustments for confounding variables were performed to test for association between outcomes and hospital groups. RESULTS: Both hospital groups had comparable demographics and clinical profiles, but NIV Hospital (42 patients) had higher lactate dehydrogenase ratio and worse radiographic score on admission and ribavirin-corticosteroid commencement. Compared to IMV Hospitals (451 patients), NIV Hospital had lower adjusted odds ratios for intubation (0.36, 95% CI 0.164 - 0.791, P = 0.011) and death (0.235, 95% CI 0.077 - 0.716, P = 0.011), and improved earlier after pulsed steroid rescue. There were no instances of transmission of severe acute respiratory syndrome among health care workers due to the use of non-invasive ventilation. CONCLUSION: Compared to invasive mechanical ventilation, non-invasive ventilation as initial ventilatory support for acute respiratory failure in the presence of severe acute respiratory syndrome appeared to be associated with reduced intubation need and mortality.

Under-reporting of inpatient services utilisation in household surveys -- a population-based study in Hong Kong.

BACKGROUND: Recognising that household interviews may produce biased estimates of health services utilisation, we examined for under- and over-reporting of hospitalisation episodes in three recent, consecutive population-based household surveys in Hong Kong. METHODS: Territory-wide inpatient service utilisation volumes as estimated from the 1999, 2001 and 2002 Thematic Household Surveys (THS) were benchmarked against corresponding statistics derived from routine administrative databases. Between-year differences on net under-reporting were quantified by Cohen's d effect size. To assess the potential for systematic biases in under-reporting, age- and sex-specific net under-reporting rates within each survey year were computed and the F-test was performed to evaluate differences between demographic subgroups. We modelled the effects of age and sex on the likelihood of ever hospitalisation through logistic regression to compare the odds ratios respectively derived from survey and administrative data. RESULTS: The extent of net under-reporting was moderately large in all three years amounting to about one-third of all inpatient episodes. However, there did not appear to be significant systematic biases in the degree of under-reporting by age or sex on stratified analyses and logistic regression modelling. CONCLUSION: Under-reporting was substantial in Hong Kong's THS. Recall bias was likely most responsible for such reporting inaccuracies. A proper full-design record-check study should be carried out to confirm the present findings.

Chest radiograph scores as potential prognostic indicators in severe acute respiratory syndrome (SARS).

OBJECTIVE: We analyzed serial chest radiographic scores for lung opacification in patients with severe acute respiratory syndrome (SARS) for temporal changes and differences between fatal and discharged cases. We sought to establish the earliest radiographic scores sensitive as potential prognostic indicators of fatal outcomes. MATERIALS AND METHODS: Chest radiographs that had been obtained from presentation until the death or discharge of 313 patients with SARS were scored on the basis of the percentage area and location of lung opacification. Profile analysis and univariable logistic regression were performed on these radiographic scores. RESULTS: Despite the increased mortality risks of advanced age and male sex, no significant difference was seen in the percentage area of opacification (AO%) between the sexes in either the group of patients with fatal outcomes or the group of patients who were discharged. No difference existed between age groups (< 65 years vs >/= 65 years), except for the radiograph showing the peak lung opacification in the deceased group in which the lungs of older patients had less opacification than those of younger patients. The radiographic scores obtained by day 7 were the earliest ones with good performance in prognostic prediction. The model showed good discriminatory performance, indicated by high C-indexes for receiver operator characteristic curves (0.86 for AO% and 0.90 for the number of opacified zones). The predicted proportion of patients with fatal outcomes showed high agreement with percentage of patients who died (goodness-of-fit statistic p = 0.18 for AO%, 0.73 for the number of opacified zones). By day 7, crude odds ratio of death was 1.73 per 5% of AO% (p < 0.0001) or 2.93 per lung zone opacified (p < 0.0001). CONCLUSION: Chest radiographic scores (percentage of lung or the number of zones opacified) by day 7 could be used as fatal prognostic indicators.