Shan-Yueh Chang

Tuberculosis (TB) remains a major global health challenge, and early, accurate diagnosis is essential for effective disease control. Chest radiography (CXR) is widely used for TB screening because of its accessibility, yet its limited specificity necessitates confirmatory molecular testing such as polymerase chain reaction (PCR) assays. This study aimed to evaluate the diagnostic performance of a deep learning model (DLM) for TB detection using CXR and to compare its predictive accuracy with PCR results, specifically in a low-burden region. A retrospective dataset of CXR images and corresponding PCR findings was obtained from two hospitals. The DLM, based on the CheXzero vision transformer, was trained on a large imaging dataset and evaluated using receiver operating characteristic (ROC) curves and area under the curve (AUC) metrics. Internal and external validation sets assessed sensitivity, specificity, and predictive values, with subgroup analyses according to imaging modality, demographics, and comorbidities. The model achieved an AUC of 0.915 internally and 0.850 externally, maintaining good sensitivity and specificity, though performance declined when limited to PCR-confirmed cases. Accuracy was lower for older adults and those with chronic kidney disease, chronic obstructive pulmonary disease, or heart failure. These findings suggest AI-assisted CXR screening may support TB detection in resource-limited settings, but PCR confirmation remains essential.

Background Anemia is a prevalent global health issue, affecting individuals across all age groups and leading to significant morbidity. Traditional methods for diagnosing anemia rely on laboratory assessments of hemoglobin concentration, which are invasive and costly. This study investigates the potential of a chest X-ray (CXR)-based deep learning model (DLM) to predict anemia, offering a non-invasive and efficient diagnostic alternative. Patients and methods We conducted a retrospective cohort study using data from Tri-Service General Hospital, Taipei, Taiwan, collected between June 2016 and February 2022. A total of 305,793 patients aged 20 years and older who had at least one CXR were included. The dataset was divided into development, tuning, internal validation, and external validation sets. All data originated from Tri-Service General Hospital, except for the external validation set, which was obtained from the Jingjhou branch. A DLM based on a 50-layer SE-ResNeXt architecture was developed to predict anemia from CXR images. The model's performance was compared to predictions made using traditional clinical data to assess its predictive accuracy. Results The DLM for predicting anemia showed strong performance, with an AUC of 0.845 in the internal validation set (sensitivity 68.5%, specificity 84.7%) and 0.852 in the external set (sensitivity 71.5%, specificity 83.2%). Subgroup analysis revealed better diagnostic accuracy with computed radiography, posteroanterior (PA) view, and in males aged 55 to 64, while younger females had lower accuracy. The DLM, especially when combined with patient data, outperformed other clinical models. Conclusion Our study introduces a deep learning tool for detecting anemia from chest images, offering faster decision-making and interventions in clinical settings. With further refinement, this model aims to align more closely with real-time clinical conditions, enhancing its utility. Ultimately, it has the potential to improve patient care through quicker anemia detection and better clinical outcomes.

The development of lung cancer is a complex process that involves many genetic and epigenetic changes. Sex-determining region Y (SRY)-box (SOX) genes encode a family of proteins that are involved in the regulation of embryonic development and cell fate determination. SOX1 is hypermethylated in human cancers. However, the role of SOX1 in the development of lung cancer is unclear. We used quantitative methylation-specific polymerase chain reaction (MSP), quantitative reverse transcription polymerase chain reaction (RT–PCR) analysis, and web tools to confirm the frequent epigenetic silencing of SOX1 in lung cancer. Stable overexpression of SOX1 repressed cell proliferation, anchorage-independent growth, and invasion in vitro as well as cancer growth and metastasis in a xenograft mouse model. Knockdown of SOX1 by the withdrawal of doxycycline partly restored the malignant phenotype of inducible SOX1-expressing NSCLC cells. Next, we discovered the potential downstream pathways of SOX1 using RNA-seq analysis and identified HES1 as a direct target of SOX1 using chromatin immunoprecipitation (ChIP)-PCR. Furthermore, we performed phenotypic rescue experiments to prove that overexpression of HES1-FLAG in SOX1-expressing H1299 cells partly reversed the tumor-suppressive effect. Taken together, these data demonstrated that SOX1 acts as a tumor suppressor by directly inhibiting HES1 during the development of NSCLC.

Sonographic septation is associated with prolonged hospitalization and increased mortality in patients diagnosed with empyema. However, it is unknown whether sonographic septation is associated with complicated parapneumonic effusion (CPPE) or the need for invasive procedures among patients with pneumonia. In this retrospective study, we included 180 patients with non-purulent neutrophilic exudative pleural effusion secondary to pulmonary infections such as pneumonia and lung abscess. We performed univariate and multivariate logistic regression analyses, including baseline clinical characteristics, values from blood samples, and sonographic echogenicity, to identify variables correlated with CPPE and the need for invasive procedures. Seventy of the 180 included patients (38.89%) displayed sonographic septation. Multivariate logistic regression analysis identified that sonographic septation (adjusted OR (AOR)=3.38 (95% CI 1.64 to 6.98), p=0.001) and younger age (AOR=2.63 (95% CI 1.24 to 5.58), p=0.012) were independently associated with CPPE. With regard to treatment strategy, sonographic septation (AOR 9.06 (95% CI 3.71 to 22.11), p<0.001) and total serum protein level (AOR=1.80 (95% CI 1.13 to 2.86), p=0.013) were independently associated with the need for subsequent invasive procedures in patients with CPPE using multivariate logistic regression analysis. Sonographic septation is a useful predictor of CPPE and may imply the need for early invasive procedures.

Epigenetic modification is considered a major mechanism of the inactivation of tumor suppressor genes that finally contributes to carcinogenesis. LIM homeobox transcription factor 1α (LMX1A) is one of the LIM-homeobox-containing genes that is a critical regulator of growth and differentiation. Recently, LMX1A was shown to be hypermethylated and functioned as a tumor suppressor in cervical cancer, ovarian cancer, and gastric cancer. However, its role in lung cancer has not yet been clarified. In this study, we used public databases, methylation-specific PCR (MSP), reverse transcription PCR (RT-PCR), and bisulfite genomic sequencing to show that LMX1A was downregulated or silenced due to promoter hypermethylation in lung cancers. Treatment of lung cancer cells with the demethylating agent 5-aza-2’-deoxycytidine restored LMX1A expression. In the lung cancer cell lines H23 and H1299, overexpression of LMX1A did not affect cell proliferation but suppressed colony formation and invasion. These suppressive effects were reversed after inhibition of LMX1A expression in an inducible expression system in H23 cells. The quantitative RT-PCR (qRT-PCR) data showed that LMX1A could modulate epithelial mesenchymal transition (EMT) through E-cadherin (CDH1) and fibronectin (FN1). NanoString gene expression analysis revealed that all aberrantly expressed genes were associated with processes related to cancer progression, including angiogenesis, extracellular matrix (ECM) remodeling, EMT, cancer metastasis, and hypoxia-related gene expression. Taken together, these data demonstrated that LMX1A is inactivated through promoter hypermethylation and functions as a tumor suppressor. Furthermore, LMX1A inhibits non-small cell lung cancer (NSCLC) cell invasion partly through modulation of EMT, angiogenesis, and ECM remodeling.

Cisplatin is still the primary therapeutic choice for advanced lung cancers without driver mutations. The occurrence of cisplatin resistance is a major clinical problem in lung cancer treatment. The natural extracted agent emetine reportedly has anticancer effects. This study aimed to explore the possible role of emetine in cisplatin resistance. We used cell viability, Western blot, and Wnt reporter assays to show that emetine suppresses proliferation, β-catenin expression, and Wnt/β-catenin signaling in non-small cell lung cancer (NSCLC). The synergism of emetine and cisplatin was assessed by constructing isobolograms and calculating combination index (CI) values using the Chou-Talalay method. Emetine effectively synergized with cisplatin to suppress the proliferation of cancer cells. Furthermore, nuclear β-catenin and cancer stem cell-related markers were upregulated in the cisplatin-resistant subpopulation of CL1-0 cells. Emetine enhanced the anticancer efficacy of cisplatin and synergized with cisplatin in the cisplatin-resistant subpopulation of CL1-0 cells. Taken together, these data suggest that emetine could suppress the growth of NSCLC cells through the Wnt/β-catenin pathway and contribute to a synergistic effect in combination with cisplatin.

This study aims to investigate the association between pulmonary embolism (PE) and the risk of psychiatric disorders. A total of 21,916 patients aged ≥20 years with PE between January 1, 2000, and December 31, 2015, were selected from the National Health Insurance Research Database of Taiwan, along with 65,748 (1:3) controls matched for sex and age. Cox regression model revealed the crude HR was 1.539 (95% CI 1.481 to 1.599; p<0.001), and after adjusting all the covariates, the adjusted HR was 1.704 (95% CI 1.435 to 1.991, p<0.001), for the risk of psychiatric disorders in the PE cohort. PE was associated with the overall psychiatric disorders, dementia, anxiety, depression, and sleep disorders, after the exclusion of the psychiatric diagnoses in the first year. PE was associated with the overall psychiatric disorders, dementia, anxiety, and depression, after the exclusion of the psychiatric diagnoses in the first 5 years. The patients with PE were associated with psychiatric disorders. This finding could serve as a reminder to the physicians to be more watchful and aware in the long-term follow-up of patients with PE for their care and potential mental health problems.

This study has investigated the risk of major adverse cardiovascular events (MACEs), including acute myocardial infarction, coronary artery disease, peripheral artery disease, and acute stroke, among children and adolescents (age younger than 20 years) with obstructive sleep apnea (OSA). In this study, the population-based National Health Insurance Research Database of Taiwan was used to identify patients in whom OSA had been first diagnosed between 2000 and 2015. Children and adolescents with OSA (n = 6,535) were included with 1:3 ratio by age, sex, and index year of control participants without OSA (n = 19,605). The Cox proportional regression model was used to evaluate the risk of MACEs in this cohort study. After a 15-year follow-up, the incidence rate of MACEs was higher in the OSA cohort when compared with the non-OSA control cohort (15.97 and 8.20 per 100,000 person-years, respectively). After adjusting for covariates, the risk of MACEs among children and adolescents with OSA was still significantly higher (hazard ratio = 2.050; 95% confidence interval = 1.312-3.107; P = .010). No MACEs were found in the children and adolescents with OSA who received continuous airway positive pressure treatment or pharyngeal surgery. This study found a significantly higher risk of MACEs in children and adolescents with OSA. These findings strongly suggest that clinicians should provide careful follow-up and medical treatment for children and adolescents with OSA.

Objectives Injury is an important issue in public health. Spinal curvature disorders are deformities characterised by excessive curves of the spine. The prevalence of spinal curvature disorders is not low, but its relationship with injury has not been studied. The aim of this study is to investigate whether spinal curvature disorders increase the risk of injury. Design Population-based retrospective cohort study. Setting Using data from the Taiwan National Health Insurance Research Database from 2000 to 2010. Participants and exposure Patients with spinal curvature disorders were selected using codes from the International Classification of Diseases, Ninth Revision, Clinical Modification. A cohort without spinal curvature was randomly frequency-matched to the spinal curvature disorders cohort at a ratio of 2:1 according to age, sex and index year. Primary outcome measures The risk of injury was analysed using Cox’s proportional hazards regression models adjusting for age, sex, comorbidities, urbanisation level and socioeconomic status. Results A total of 20 566 patients with spinal curvature disorders and 41 132 controls were enrolled in this study. The risk of injury was 2.209 times higher (95% CI 2.118 to 2.303) in patients with spinal curvature disorders than in the control group. The spinal curvature disorders cohort exhibited higher risk of developing injury compared with the control group, regardless of age, sex, comorbidities, urbanisation level and subgroup of spinal curvature disorders. Based on the subgroup analysis, the spinal curvature disorders cohort had higher risks of unintentional injury and injury diagnoses such as fracture, dislocation, open wound, superficial injury/contusion, crushing and injury to nerves and spinal cord compared with the control cohort. Conclusions Patients with spinal curvature disorders have a significantly higher risk of developing injury than patients without spinal curvature disorders. Aggressive detection and management of spinal curvature disorders may be beneficial for injury prevention.