An open automation system for predatory journal detection.

The growing number of online open-access journals promotes academic exchanges, but the prevalence of predatory journals is undermining the scholarly reporting process. Data collection, feature extraction, and model prediction are common steps in tools designed to distinguish between legitimate and predatory academic journals and publisher websites. The authors include them in their proposed academic journal predatory checking (AJPC) system based on machine learning methods. The AJPC data collection process extracts 833 blacklists and 1213 whitelists information from websites to be used for identifying words and phrases that might indicate the presence of predatory journals. Feature extraction is used to identify words and terms that help detect predatory websites, and the system's prediction stage uses eight classification algorithms to distinguish between potentially predatory and legitimate journals. We found that enhancing the classification efficiency of the bag of words model and TF-IDF algorithm with diff scores (a measure of differences in specific word frequencies between journals) can assist in identifying predatory journal feature words. Results from performance tests suggest that our system works as well as or better than those currently being used to identify suspect publishers and publications. The open system only provides reference results rather than absolute opinions and accepts user inquiries and feedback to update the system and optimize performance.

Using Deep Learning with Thermal Imaging for Human Detection in Heavy Smoke Scenarios.

In this study, we propose using a thermal imaging camera (TIC) with a deep learning model as an intelligent human detection approach during emergency evacuations in a low-visibility smoky fire scenarios. We use low-wavelength infrared (LWIR) images taken by a TIC qualified with the National Fire Protection Association (NFPA) 1801 standards as input to the YOLOv4 model for real-time object detection. The model trained with a single Nvidia GeForce 2070 can achieve >95% precision for the location of people in a low-visibility smoky scenario with 30.1 frames per second (FPS). This real-time result can be reported to control centers as useful information to help provide timely rescue and provide protection to firefighters before entering dangerous smoky fire situations.

Efficiently Classifying Lung Sounds through Depthwise Separable CNN Models with Fused STFT and MFCC Features.

Lung sounds remain vital in clinical diagnosis as they reveal associations with pulmonary pathologies. With COVID-19 spreading across the world, it has become more pressing for medical professionals to better leverage artificial intelligence for faster and more accurate lung auscultation. This research aims to propose a feature engineering process that extracts the dedicated features for the depthwise separable convolution neural network (DS-CNN) to classify lung sounds accurately and efficiently. We extracted a total of three features for the shrunk DS-CNN model: the short-time Fourier-transformed (STFT) feature, the Mel-frequency cepstrum coefficient (MFCC) feature, and the fused features of these two. We observed that while DS-CNN models trained on either the STFT or the MFCC feature achieved an accuracy of 82.27% and 73.02%, respectively, fusing both features led to a higher accuracy of 85.74%. In addition, our method achieved 16 times higher inference speed on an edge device and only 0.45% less accuracy than RespireNet. This finding indicates that the fusion of the STFT and MFCC features and DS-CNN would be a model design for lightweight edge devices to achieve accurate AI-aided detection of lung diseases.

Multicenter evaluation of two chemiluminescence and three lateral flow immunoassays for the diagnosis of COVID-19 and assessment of antibody dynamic responses to SARS-CoV-2 in Taiwan.

This multicenter, retrospective study included 346 serum samples from 74 patients with coronavirus disease 2019 (COVID-19) and 194 serum samples from non-COVID-19 patients to evaluate the performance of five anti-severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) antibody tests, i.e. two chemiluminescence immunoassays (CLIAs): Roche Elecsys® Anti-SARS-CoV-2 Test (Roche Test) and Abbott SARS-CoV-2 IgG (Abbott Test), and three lateral flow immunoassays (LFIAs): Wondfo SARS-CoV-2 Antibody Test (Wondfo Test), ASK COVID-19 IgG/IgM Rapid Test (ASK Test), and Dynamiker 2019-nCoV IgG/IgM Rapid Test (Dynamiker Test). We found high diagnostic sensitivities (%, 95% confidence interval [CI]) for the Roche Test (97.4%, 93.4-99.0%), Abbott Test (94.0%, 89.1-96.8%), Wondfo Test (91.4%, 85.8-94.9%), ASK Test (97.4%, 93.4-99.0%), and Dynamiker Test (90.1%, 84.3-94.0%) after >21 days of symptom onset. Meanwhile, the diagnostic specificity was 99.0% (95% CI, 96.3-99.7%) for the Roche Test, 97.9% (95% CI, 94.8-99.2%) for the Abbott Test, and 100.0% (95% CI, 98.1-100.0%) for the three LFIAs. Cross-reactivity was observed in sera containing anti-cytomegalovirus (CMV) IgG/IgM antibodies and autoantibodies. No difference was observed in the time to seroconversion detection of the five serological tests. Specimens from patients with COVID-19 pneumonia demonstrated a shorter seroconversion time and higher chemiluminescent signal than those without pneumonia. Our data suggested that understanding the dynamic antibody response after COVID-19 infection and performance characteristics of different serological test are crucial for the appropriate interpretation of serological test result for the diagnosis and risk assessment of patient with COVID-19 infection.

Lopinavir/ritonavir did not shorten the duration of SARS CoV-2 shedding in patients with mild pneumonia in Taiwan.

An increase of Ct values was 0.9 per day in 2 cases of COVID-19 treated with lopinavir/ritonavir (LPV/r), an increase was 1.0 per day in 3 cases without LPV/r through illness day 1-10, indicating that LPV/r did not shorten the duration of SARS CoV-2 shedding.

Demographic differences in people living with HIV according to recruitment sources: comparison between health-care systems and social media networks.

With the improvement of internet technology in health applications, the utilization of internet and social media as new survey methodologies and recruitment source for research participants have been encouraged, yet evidence of the feasibility in people living with HIV (PLHIV) study is still lacking. We conducted a cross-sectional survey to determine whether there are differences among PLHIV recruited from social media networks and health-care systems using an HIV stigma and discrimination questionnaire. The result revealed that PLHIV recruited from social media networks were younger, more sexually active, and had higher educational status and awareness of the country's HIV rights protection laws than those recruited from hospitals. By contrast, participants recruited from hospitals were more diverse regarding key population compositions, had lived with HIV for a longer duration, had a higher prevalence of concomitant physical disabilities than those recruited from social media networks, and fit Taiwan PLHIV characteristics described by 2016 census from Taiwan Centres for Disease Control. We conclude that sampling bias exists when utilizing social media networks for PLHIV studies.

Human infection with avian influenza A H6N1 virus: an epidemiological analysis.

BACKGROUND: Avian influenza A H6N1 virus is one of the most common viruses isolated from wild and domestic avian species, but human infection with this virus has not been previously reported. We report the clinical presentation, contact, and environmental investigations of a patient infected with this virus, and assess the origin and genetic characteristics of the isolated virus. METHODS: A 20-year-old woman with an influenza-like illness presented to a hospital with shortness of breath in May, 2013. An unsubtyped influenza A virus was isolated from her throat-swab specimen and was transferred to the Taiwan Centres for Disease Control (CDC) for identification. The medical records were reviewed to assess the clinical presentation. We did a contact and environmental investigation and collected clinical specimens from the case and symptomatic contacts to test for influenza virus. The genomic sequences of the isolated virus were determined and characterised. FINDINGS: The unsubtyped influenza A virus was identified as the H6N1 subtype, based on sequences of the genes encoding haemagglutinin and neuraminidase. The source of infection was not established. Sequence analyses showed that this human isolate was highly homologous to chicken H6N1 viruses in Taiwan and had been generated through interclade reassortment. Notably, the virus had a G228S substitution in the haemagglutinin protein that might increase its affinity for the human α2-6 linked sialic acid receptor. INTERPRETATION: This is the first report of human infection with a wild avian influenza A H6N1 virus. A unique clade of H6N1 viruses with a G228S substitution of haemagglutinin have circulated persistently in poultry in Taiwan. These viruses continue to evolve and accumulate changes, increasing the potential risk of human-to-human transmission. Our report highlights the continuous need for preparedness for a pandemic of unpredictable and complex avian influenza. FUNDING: Taiwan Centres for Disease Control.

Syphilis and neurosyphilis in human immunodeficiency virus-infected patients: a retrospective study at a teaching hospital in Taiwan.

BACKGROUND AND PURPOSE: Some studies have reported that the risk factors for neurosyphilis in patients with human immunodeficiency virus (HIV) and syphilis co-infection, include CD4 cell counts ≤350cells/µL and rapid plasma reagin (RPR) titer ≥1:32. However, neurosyphils can develop even in patients with CD4 cell counts >350cells/µL or RPR titer <1:32. In this study, we evaluated the outcome of syphilis to treatment in HIV-infected patients, and analysed the predictors of neurosyphilis in this population. METHODS: We retrospectively reviewed medical records of HIV-infected patients with syphilis who visited the China Medical University Hospital between January 2000 and December 2009. Neurosyphilis was defined by white blood cell (WBC) counts >20cells/µL in the cerebrospinal fluid (CSF) sample or elevated Venereal Disease Research Laboratory (VDRL) titers of the CSF samples. Treatment failure was defined as less than 4-fold decrease in the serum RPR titer at or beyond 12 months post-treatment in case of early syphilis, and, at or beyond 24 months in case of late syphilis. RESULTS: One hundred and twenty-one HIV-infected patients (average age, 32 years) with syphilis were included in this study. Of 63 patients who had follow-up of serologic responses, 30 (47.6%) failed to respond to treatment. CD4 cell counts ≤200cells/µL was the indicator for treatment failure (P=.029). Lumbar puncture was performed in 65 patients, and 14 patients were diagnosed with neurosyphilis. At the time of lumbar puncture, 31 and 19 of the 65 patients showed CD4 cell counts of >350cells/µL and RPR of <1:32, respectively. An HIV viral load (VL) ≥10000copies/mL was found to be associated with the development of neurosyphilis (P=.016). CONCLUSION: In HIV-infected patients with syphilis, RPR titer should be evaluated more frequently when CD4 count ≤200cell/µL is associated with treatment failure. Lumbar puncture for the diagnosis of neurosyphilis should be considered in patients with HIV and syphilis co-infection, even in patients with CD4 cell counts >350cells/µL, and particularly when the HIV VL ≥10000copies/mL.

Different presentations and outcomes between HIV-infected and HIV-uninfected patients with Cryptococcal meningitis.

BACKGROUND AND PURPOSE: Cryptococcus species are the most common causative agents of fungal meningitis. Different populations may show different clinical manifestations and outcomes. In this retrospective study, we investigated these differences in patients with and without HIV infection. METHODS: From 1995 to 2009, we collected data from HIV-infected or HIV-uninfected patients aged 18 years or over who had cryptococcal meningitis (CM) in a medical center in Taiwan. We reviewed and analyzed their demographic data, clinical manifestations, therapeutic strategies and outcomes. RESULTS: Among the 72 patients with CM, 19 HIV-infected patients were predominantly younger males, and all of them had AIDS status when CM was diagnosed. In contrast, the 53 HIV-uninfected patients were mostly older males with underlying diseases. The time from initial symptoms to diagnosis was shorter in HIV-infected patients (median 10 vs. 18 days, p = 0.048). The HIV-infected patients presented with less pleocytosis (p = 0.003) and lower protein levels in the cerebrospinal fluid (CSF), but a higher proportion had positive results for cryptococci in the CSF (90% vs. 60%, p = 0.02) and blood (53% vs. 21%, p = 0.009) cultures. Surgical drains and repeated lumbar punctures for the management of increased intracranial pressure were performed in 47% of the HIV-infected patients and 38% of the HIV-uninfected patients. A lower mortality rate was observed in the HIV-infected patients (p = 0.038). On multivariate analysis, initial CD4 count ≤20/mm(3) was an indicator of death or relapse in HIV-infected patients. In the HIV-uninfected group, the initial high cryptococcal antigen titer in the CSF (≥1:512) and hydrocephalus were related to unsatisfactory outcomes. CONCLUSION: In addition to well-known differences, we found a lower mortality in HIV-infected patients than in HIV-uninfected patients. Cryptococci and inflammation in the central nervous system may play important roles in the pathogenesis of CM. Low intensity of inflammation and effective surgical CSF drains for increased intracranial pressure and cryptococci removal may contribute to lower mortality in HIV-infected patients.

Eight years experience in treatment of prosthetic joint infections at a teaching hospital in Central Taiwan.

BACKGROUND/PURPOSE: Prosthetic joint infection (PJI) has become an important issue in the management of patients who receive prostheses. We compared the clinical outcomes of PJIs caused by Gram-negative bacteria (GN PJIs) and Gram-positive bacteria (GP PJIs). METHODS: Patients with culture-proven PJIs admitted to China Medical University Hospital between March 2001 and March 2009 were included in this retrospective study. RESULTS: Fifty-nine patients were diagnosed with PJI during the study period. Nineteen patients had GN PJIs (mean age: 68 years) and 40 had GP PJIs (mean age: 61 years). The most common comorbid condition was diabetes mellitus (23.7%) and the most common presentation was joint pain (79.7%). Staphylococcus aureus was the most common pathogen, whereas Klebsiella pneumoniae was the most common Gram-negative pathogen. The GN PJI group included more cases of hematogenous infection (36.8% vs. 20%; p < 0.001), showed a shorter interval between onset of infection symptoms and surgical intervention (median: 8 days vs. 21 days; p = 0.04), and required longer medical treatment (median: 259 days vs. 161 days; p = 0.04). In comparison with patients whose prostheses were eventually removed, patients whose prostheses were not removed had a shorter interval between onset of infection symptoms and surgical intervention (median: 6 days vs. 90 days; p = 0.004 and median: 6 days vs. 44 days; p = 0.04) in the GP PJI and GN PJI groups, respectively. CONCLUSION: GN PJI was less common than GP PJI, but GN PJI was more complicated and required longer treatment. Prospective randomized clinical studies are needed to investigate whether prosthesis implantation should be reserved if the patient undergoes early surgical intervention for PJI.

Analysis of glyphosate, glufosinate and aminomethylphosphonic acid by capillary electrophoresis with indirect fluorescence detection.

A capillary electrophoresis (CE)-indirect fluorescence detection method is described for the simultaneous determination of glufosinate, glyphosate and aminomethylphosphonic acid. The three analytes were separated by CE in 5 min with a 1 mM fluorescein solution at pH 9.5. Fluorescein also functioned as a background fluorophore for the indirect detection of these nonfluorescent species. Linearity of more than two orders of magnitudes was generally obtained. The concentration limits of detection were in the microM range. Precisions of migration times and peak areas were less than 1.7% and 7.4%, respectively. Quantitation of glyphosate and glufosinate in commercial herbicides is demonstrated. In addition, the applicability of the method for the analysis of ground water was examined.