Explore the vast range of titles available.

Dual-energy X-ray absorptiometry (DXA) is underutilized to measure bone mineral density (BMD) and evaluate fracture risk. We present an automated tool to identify fractures, predict BMD, and evaluate fracture risk using plain radiographs. The tool performance is evaluated on 5164 and 18175 patients with pelvis/lumbar spine radiographs and Hologic DXA. The model is well calibrated with minimal bias in the hip (slope = 0.982, calibration-in-the-large = −0.003) and the lumbar spine BMD (slope = 0.978, calibration-in-the-large = 0.003). The area under the precision-recall curve and accuracy are 0.89 and 91.7% for hip osteoporosis, 0.89 and 86.2% for spine osteoporosis, 0.83 and 95.0% for high 10-year major fracture risk, and 0.96 and 90.0% for high hip fracture risk. The tool classifies 5206 (84.8%) patients with 95% positive or negative predictive value for osteoporosis, compared to 3008 DXA conducted at the same study period. This automated tool may help identify high-risk patients for osteoporosis. Dual-energy X-ray absorptiometry and the Fracture Risk Assessment Tool are recommended tools for osteoporotic fracture risk evaluation, but are underutilized. Here, the authors present an opportunistic tool to identify fractures, predict bone mineral density and evaluate fracture risk using plain pelvis and lumbar spine radiographs.

Large Language Models (LLMs) are capable of transforming healthcare by demonstrating remarkable capabilities in language understanding and generation. They have matched or surpassed human performance in standardized medical examinations and assisted in diagnostics across specialties like dermatology, radiology, and ophthalmology. LLMs can enhance patient education by providing accurate, readable, and empathetic responses, and they can streamline clinical workflows through efficient information extraction from unstructured data such as clinical notes. Integrating LLM into clinical practice involves user interface design, clinician training, and effective collaboration between Artificial Intelligence (AI) systems and healthcare professionals. Users must possess a solid understanding of generative AI and domain knowledge to assess the generated content critically. Ethical considerations to ensure patient privacy, data security, mitigating biases, and maintaining transparency are critical for responsible deployment. Future directions for LLMs in healthcare include interdisciplinary collaboration, developing new benchmarks that incorporate safety and ethical measures, advancing multimodal LLMs that integrate text and imaging data, creating LLM-based medical agents capable of complex decision-making, addressing underrepresented specialties like rare diseases, and integrating LLMs with robotic systems to enhance precision in procedures. Emphasizing patient safety, ethical integrity, and human-centered implementation is essential for maximizing the benefits of LLMs, while mitigating potential risks, thereby helping to ensure that these AI tools enhance rather than replace human expertise and compassion in healthcare.

Prior research on technology usage had largely overlooked the issue of user resistance or barriers to technology acceptance. Prior research on the Electronic Medical Records had largely focused on technical issues but rarely on managerial issues. Such oversight prevented a better understanding of users’ resistance to new technologies and the antecedents of technology rejection. Incorporating the enablers and the inhibitors of technology usage intention, this study explores physicians’ reactions towards the electronic medical record. The main focus is on the barriers, perceived threat and perceived inequity. 115 physicians from 6 hospitals participated in the questionnaire survey. Structural Equation Modeling was employed to verify the measurement scale and research hypotheses. According to the results, perceived threat shows a direct and negative effect on perceived usefulness and behavioral intentions, as well as an indirect effect on behavioral intentions via perceived usefulness. Perceived inequity reveals a direct and positive effect on perceived threat, and it also shows a direct and negative effect on perceived usefulness. Besides, perceived inequity reveals an indirect effect on behavioral intentions via perceived usefulness with perceived threat as the inhibitor. The research finding presents a better insight into physicians’ rejection and the antecedents of such outcome. For the healthcare industry understanding the factors contributing to physicians’ technology acceptance is important as to ensure a smooth implementation of any new technology. The results of this study can also provide change managers reference to a smooth IT introduction into an organization. In addition, our proposed measurement scale can be applied as a diagnostic tool for them to better understand the status quo within their organizations and users’ reactions to technology acceptance. By doing so, barriers to physicians’ acceptance can be identified earlier and more effectively before leading to technology rejection.

Background: Liver steatosis assessment by 2D ultrasound is widely used but remains subjective. We previously developed a deep learning (DL) algorithm for objective steatosis quantification. This study aimed to (1) establish histology-based cutoffs, (2) evaluate their transferability across different imaging views, and (3) validate performance on a new scanner not included in training. Methods: We retrospectively analyzed 588 ultrasound studies from 457 histology-proven cases and prospectively collected paired scans using a new scanner (Philips Affiniti 70). Images from right intercostal, left hepatic lobe, and subcostal views were processed with the DL algorithm, and mean values from 3–5 images per view were correlated with histology. Results: Across three views, the DL algorithm achieved AUROCs of 0.891–0.936 across steatosis grades, consistently outperforming FibroScan’s controlled attenuation parameter (0.840–0.905), especially in moderate-to-severe steatosis (p < 0.001). Cutoffs established from right intercostal images (N = 565) were applied to images from left hepatic lobe (N = 464) and subcostal views (N = 341), yielding accuracies of 0.792–0.850. On Affiniti 70 images, AUROCs remained high (0.838–0.896), supporting scanner generalizability. Conclusions: The DL algorithm provides accurate, view-independent steatosis grading across different ultrasound scanners and outperforms CAP, supporting its real-world use for objective, reproducible quantification.

In this study, we aimed to develop a deep learning model for identifying bacterial keratitis (BK) and fungal keratitis (FK) by using slit-lamp images. We retrospectively collected slit-lamp images of patients with culture-proven microbial keratitis between 1 January 2010 and 31 December 2019 from two medical centers in Taiwan. We constructed a deep learning algorithm consisting of a segmentation model for cropping cornea images and a classification model that applies different convolutional neural networks (CNNs) to differentiate between FK and BK. The CNNs included DenseNet121, DenseNet161, DenseNet169, DenseNet201, EfficientNetB3, InceptionV3, ResNet101, and ResNet50. The model performance was evaluated and presented as the area under the curve (AUC) of the receiver operating characteristic curves. A gradient-weighted class activation mapping technique was used to plot the heat map of the model. By using 1330 images from 580 patients, the deep learning algorithm achieved the highest average accuracy of 80.0%. Using different CNNs, the diagnostic accuracy for BK ranged from 79.6% to 95.9%, and that for FK ranged from 26.3% to 65.8%. The CNN of DenseNet161 showed the best model performance, with an AUC of 0.85 for both BK and FK. The heat maps revealed that the model was able to identify the corneal infiltrations. The model showed a better diagnostic accuracy than the previously reported diagnostic performance of both general ophthalmologists and corneal specialists.

Deep learning models have shown strong potential for automated fracture detection in medical images. However, their robustness under varying image quality remains uncertain, particularly for small and subtle fractures, such as scaphoid fractures. Understanding how different types of image perturbations affect model performance is crucial for ensuring reliable deployment in clinical practice. This study aimed to evaluate the robustness of a deep learning model trained to detect scaphoid fractures in radiographs when exposed to various image perturbations. We sought to identify which perturbations most strongly impact performance and to explore strategies to mitigate performance degradation. Radiographic datasets were systematically modified by applying Gaussian noise, blurring, JPEG compression, contrast-limited adaptive histogram equalization, resizing, and geometric offsets. Model accuracy was evaluated across different perturbation types and levels. Image quality was quantified using peak signal-to-noise ratio and structural similarity index measure to assess correlations between degradation and model performance. Model accuracy declined with increasing perturbation severity, but the extent varied across perturbation types. Gaussian blur caused the most substantial performance drop, whereas contrast-limited adaptive histogram equalization increased the false-negative rate. The model demonstrated higher resilience to color perturbations than to grayscale degradations. A strong linear correlation was found between peak signal-to-noise ratio-structural similarity index measure and accuracy, suggesting that better image quality led to improved detection. Geometric offsets and pixel value rescaling had minimal influence, whereas resolution was the dominant factor affecting performance. The findings indicate that image quality, especially resolution and blurring, substantially influences the robustness of deep learning-based fracture detection models. Ensuring adequate image resolution and quality control can enhance diagnostic reliability. These results provide valuable insights for designing more accurate and resilient medical imaging models under real-world variability.

Background: Metabolic acidosis is a common complication in patients with chronic kidney disease (CKD). Oral sodium bicarbonate is often used to treat metabolic acidosis and prevent CKD progression. However, there is limited information about the effect of sodium bicarbonate on major adverse cardiovascular events (MACE) and mortality in patients with pre-dialysis advanced CKD. Method: 25599 patients with CKD stage V between January 1, 2001 and December 31, 2019 were identified from the Chang Gung Research Database (CGRD), a multi-institutional electronic medical record database in Taiwan. The exposure was defined as receiving sodium bicarbonate or not. Baseline characteristics were balanced using propensity score weighting between two groups. Primary outcomes were dialysis initiation, all-cause mortality, and major adverse cardiovascular events (MACE) (myocardial infarction, heart failure, stroke). The risks of dialysis, MACE, and mortality were compared between two groups using Cox proportional hazards models. In addition, we performed analyzes using Fine and Gray sub-distribution hazard models that considered death as a competing risk. Result: Among 25599 patients with CKD stage V, 5084 patients (19.9%) were sodium bicarbonate users while 20515 (80.1%) were sodium bicarbonate non-users. The groups had similar risk of dialysis initiation (hazard ratio (HR): 0.98, 95% confidence interval (CI): 0.95-1.02, p < 0.379). However, taking sodium bicarbonate was associated with a significantly lower risks of MACE (HR: 0.95, 95% CI 0.92–0.98, p < 0.001) and hospitalizations for acute pulmonary edema (HR: 0.92, 95% CI 0.88–0.96, p < 0.001) compared with non-users. The mortality risks were significantly lower in sodium bicarbonate users compared with sodium bicarbonate non-users (HR: 0.75, 95% CI 0.74–0.77, p < 0.001). Conclusion: This cohort study revealed that in real world practice, use of sodium bicarbonate was associated with similar risk of dialysis compared with non-users among patients with advanced CKD stage V. Nonetheless, use of sodium bicarbonate was associated with significantly lower rate of MACE and mortality. Findings reinforce the benefits of sodium bicarbonate therapy in the expanding CKD population. Further prospective studies are needed to confirm these findings.

Background: Statins are commonly used for cardiovascular disease (CVD) prevention. Observational studies reported the effects on sepsis prevention and mortality improvement. Patients with chronic kidney disease (CKD) are at high risk for CVD and infectious diseases. Limited information is available for statin use in patients with non-dialysis CKD stage V. Method: The retrospective observational study included patients with non-dialysis CKD stage V, with either de novo statin use or none. Patients who were prior statin users and had prior cardiovascular events were excluded. The key outcomes were infection-related hospitalization, major adverse cardiovascular events (MACE) (non-fatal myocardial infarction, hospitalization for heart failure, or non-fatal stroke), and all-cause mortality. The data were retrieved from the Chang Gung Research Database (CGRD) from January 2001 to December 2019. Analyses were conducted with Cox proportional hazard regression models in the propensity score matching (PSM) cohort. Result: A total of 20,352 patients with CKD stage V were included (1,431 patients were defined as de novo statin users). After PSM, 1,318 statin users were compared with 1,318 statin non-users. The infection-related hospitalization (IRH) rate was 79.3 versus 94.3 per 1,000 person-years in statin users and statin non-users, respectively [hazard ratio (HR) 0.83, 95% confidence interval (CI) 0.74–0.93, p = 0.002]. The incidence of MACE was 38.9 versus 55.9 per 1,000 person-years in statin users and non-users, respectively (HR, 0.72; 95% CI 0.62–0.83, p < 0.001). The all-cause mortality did not differ between statin users and non-users, but statin users had lower infection-related mortality than non-users (HR, 0.59; 95% CI 0.38–0.92, p = 0.019). Conclusion: De novo use of statin in patients with non-dialysis CKD stage V reduced the incidence of cardiovascular events, hospitalization, and mortality for infectious disease. The study results reinforced the benefits of statin in a wide range of patients with renal impairment before maintenance dialysis.

Background: The aim of this study was to evaluate the efficacy of a deep learning system in pterygium grading and recurrence prediction. Methods: This was a single center, retrospective study. Slit-lamp photographs, from patients with or without pterygium, were collected to develop an algorithm. Demographic data, including age, gender, laterality, grading, and pterygium area, recurrence, and surgical methods were recorded. Complex ocular surface diseases and pseudopterygium were excluded. Performance of the algorithm was evaluated by sensitivity, specificity, F1 score, accuracy, and area under the receiver operating characteristic curve. Confusion matrices and heatmaps were created to help explain the results. Results: A total of 237 eyes were enrolled, of which 176 eyes had pterygium and 61 were non-pterygium eyes. The training set and testing set were comprised of 189 and 48 photographs, respectively. In pterygium grading, sensitivity, specificity, F1 score, and accuracy were 80% to 91.67%, 91.67% to 100%, 81.82% to 94.34%, and 86.67% to 91.67%, respectively. In the prediction model, our results showed sensitivity, specificity, positive predictive value, and negative predictive values were 66.67%, 81.82%, 33.33%, and 94.74%, respectively. Conclusions: Deep learning systems can be useful in pterygium grading based on slit lamp photographs. When clinical parameters involved in the prediction of pterygium recurrence were included, the algorithm showed higher specificity and negative predictive value in prediction.

Background Allograft kidney transplantation has become a treatment of choice for patients with end-stage renal disease (ESRD), and post-transplant diabetes mellitus (PTDM) has been associated with impaired patient and graft survival. Taiwan has the highest incidence and prevalence rates of ESRD with many recipients and candidates of kidney transplantation. However, information about the epidemiologic features of PTDM in Taiwan is incomplete. Therefore, we aimed to investigate the prevalence and incidence of PTDM with subsequent patient and graft outcomes. Methods Using the Taiwan National Health Insurance Research Database (NHIRD), 3663 kidney recipients between 1997 and 2011 were enrolled. We calculated the cumulative incidences of diabetes mellitus (DM) after transplantation. Cox proportional hazards model with competing risk analysis was used to calculate the hazard ratio (HR) and 95% confidence intervals (CI) between three targeted groups (DM, PTDM, non-DM). The outcomes of primary interest were the occurrence of graft failure excluding death with functioning graft, all-cause mortality, death with functioning graft and major adverse cardiovascular events (MACE) including myocardial infarction (MI), cerebrovascular accident (CVA) and congestive heart failure (CHF). Subgroup analysis for graft failure excluding death with functioning graft, MACE and all-cause mortality was performed, and interaction between PTDM and recipient age was examined. Results Of 3663 kidney transplant recipients, 531 (14%) had pre-existing DM and 631 (17%) developed PTDM. Compared with non-DM group, the PTDM and DM groups exhibited higher risk of graft failure excluding death with functioning graft (PTDM: HR 1.65, 95% CI 1.47–1.85; DM: HR 1.33, 95% CI 1.18–1.50), MACE (PTDM: HR 1.51, 95% CI 1.31–1.74; DM: HR 1.64, 95% CI 1.41–1.9), all-cause mortality (PTDM: HR 1.79, 95% CI 1.59–2.01; DM: HR 2.03, 95% CI 1.81–2.18), and death with functioning graft (PTDM: HR 1.94, 95% CI 1.71–2.20; DM: HR 1.94, 95% CI 1.71–2.21). Both PTDM and DM groups had increased cardiovascular disease-related mortality (PTDM: HR 2.14, 95% CI 1.43–3.20, p < 0.001; DM: HR 1.89, 95% CI 1.25–2.86, p = 0.002), cancer-related mortality (PTDM: HR 1.56, 95% CI 1.18–2.07, p = 0.002; DM: HR 1.89, 95% CI 1.25–2.86, p = 0.027), and infection-related mortality (PTDM: HR 1.47, 95% CI 1.14–1.90, p = 0.003; DM: HR 2.25, 95% CI 1.77–2.84, p < 0.001) compared with non-DM group. The subgroup analyses showed that the add-on risks of MACE and mortality from PTDM were mainly observed in patients who were younger and those without associated comorbidities including atrial fibrillation, cirrhosis, CHF, and MI. Age significantly modified the association between PTDM and MACE (p interaction  < 0.01) with higher risk in recipients with PTDM aged younger than 55 years (adjusted HR 1.64, 95% CI 1.40–1.92, p < 0.001). A trend (p interaction  = 0.06) of age-modifying effect on the association between PTDM and all-cause mortality was also noted with higher risk in recipients with PTDM aged younger than 55 years. Conclusions In the present population-based study, the incidence of PTDM peaked within the first year after kidney transplantation. PTDM negatively impacted graft and patient outcomes. The magnitude of cardiovascular and survival disadvantages from PTDM were more pronounced in recipients aged less than 55 years. Further trials to improve prediction of PTDM and to prevent PTDM are warranted.