Effectiveness of ulinastatin in critical care patients in real world: a retrospective multi-center study.

OBJECTIVES: Ulinastatin has been applied in various diseases associated with inflammation, but its effectiveness lacks real-world evidence. We aimed to evaluate the effectiveness of ulinastatin based on a real-world database in the intensive care unit (ICU) patients. METHODS: This was a retrospective cohort study. ICU patient data from multi-centers in China were collected. Propensity score matching (PSM) was applied. The effectiveness of ulinastatin was evaluated by mortality, length of stay in the ICU and mechanical ventilation duration. Kaplan-Meier method was used to estimate the hazard ratio and plot the survival curve. RESULTS: A total of 2036 patients were analyzed after PSM. Mortality was significantly lower in the ulinastatin group than in the non-ulinastatin group (hazard ratio for death: 0.77; 95% confidence interval: 0.62-0.96; p = 0.018). Ulinastatin significantly reduced mortality at 28 days (10.0% vs. 13.6%), 60 days (13.9% vs. 18.2%) and 90 days (14.7% vs. 18.5%), length of stay in the ICU (median 8.0 d vs. 13.0 d), and mechanical ventilation duration (median 24.0 h vs. 25.0 h; p < 0.05). CONCLUSIONS: Ulinastatin was beneficial for patients in the ICU, mainly by reducing mortality, length of ICU stay, and mechanical ventilation duration. This study provides evidence of the clinical effectiveness of ulinastatin.

Forced randomization: the what, why, and how.

BACKGROUND: When running a randomized controlled trial (RCT), a clinical site may face a situation when an eligible trial participant is to be randomized to the treatment that is not available at the site. In this case, there are two options: not to enroll the participant, or, without disclosing to the site, allocate the participant to a treatment arm with drug available at the site using a built-in feature of the interactive response technology (IRT). In the latter case, one has employed a "forced randomization" (FR). There seems to be an industry-wide consensus that using FR can be acceptable in confirmatory trials provided there are "not too many" instances of forcing. A better understanding of statistical properties of FR is warranted. METHODS: We described four different IRT configurations with or without FR and illustrated them using a simple example. We discussed potential merits of FR and outlined some relevant theoretical risks and risk mitigation strategies. We performed a search using Cortellis Regulatory Intelligence database (IDRAC) ( www.cortellis.com ) to understand the prevalence of FR in clinical trial practice. We also proposed a structured template for development and evaluation of randomization designs featuring FR and showcased an application of this template for a hypothetical multi-center 1:1 RCT under three experimental settings ("base case", "slower recruitment", and "faster recruitment") to explore the effect of four different IRT configurations in combination with three different drug supply/re-supply strategies on some important operating characteristics of the trial. We also supplied the Julia code that can be used to reproduce our simulation results and generate additional results under user-specified experimental scenarios. RESULTS: FR can eliminate refusals to randomize patients, which can cause frustration for patients and study site personnel, improve the study logistics, drug supply management, cost-efficiency, and recruitment time. Nevertheless, FR carries some potential risks that should be reviewed at the study planning stage and, ideally, prospectively addressed through risk mitigation planning. The Cortellis search identified only 9 submissions that have reported the use of FR; typically, the FR option was documented in IRT specifications. Our simulation evidence showed that under the considered realistic experimental settings, the percentage of FR is expected to be low. When FR with backfilling was used in combination with high re-supply strategy, the final treatment imbalance was negligibly small, the proportion of patients not randomized due to the lack of drug supply was close to zero, and the time to complete recruitment was shortened compared to the case when FR was not allowed. The drug overage was primarily determined by the intensity of the re-supply strategy and to a smaller extent by the presence or absence of the FR feature in IRT. CONCLUSION: FR with a carefully chosen drug supply/re-supply strategy can result in quantifiable improvements in the patients' and site personnel experience, trial logistics and efficiency while preventing an undesirable refusal to randomize a patient and a consequential unblinding at the site. FR is a useful design feature of multi-center RCTs provided it is properly planned for and carefully implemented.

[Prediction of hemorrhage rate after tonsil surgery in children based on random forest model].

Objective:Hemorrhage after tonsil surgery in children is a serious and potentially life-threatening complication. The purpose of this study was to establish a risk warning model for hemorrhage after tonsil surgery in children through a national multi-center retrospective study, providing a basis for hierarchical management after tonsil surgery in children. Methods:Stratified sampling was performed on 8 854 children who underwent tonsillectomy under general anesthesia from 15 research centers in different provinces from January 15, 2022 to May 15, 2023. The sample size of this study was 2 724 cases, including 1 096 males and 1 628 females. Children were divided into bleeding and non-bleeding groups according to whether or not they had bleeding after surgery. The random forest algorithm was used to build a risk warning model. By continuously exploring the optimized model, the accuracy of predicting the postoperative bleeding rate of tonsils in children was improved, and the prediction effectiveness of the model was verified by ten-fold cross-validation. Results:Among 2 724 children, 117 had postoperative bleeding after tonsillectomy, with a bleeding rate of 4.30%. The model constructed by the random forest algorithm for the training set was verified in the test set, and the obtained prediction accuracy was 98.72%, the recall rate was 78.95%, and the area under the ROC curve AUC was 0.96. Conclusion:Although the recall rate of the random forest model needs to be improved, the overall accuracy is quite excellent. It can effectively avoid misjudging positive cases as negative cases. It is a useful tool that can be used to predict the postoperative bleeding rate of tonsils and clinical medical decision-making, laying a good foundation for subsequent optimization and improvement.

Dedifferentiated liposarcoma of the extremities: a Korean multi-center study of 107 cases.

BACKGROUND: Dedifferentiated liposarcoma of the extremities (DDL-E) is rare in comparison to that of the retroperitoneum. Its clinical features and surgical principle for resection margins at the dedifferentiated and the well-differentiated components are yet to be elucidated. METHODS: This retrospective multi-center study examined patients diagnosed with DDL-E from August 2004 to May 2023 at 5 sarcoma centers. Clinical features, oncologic outcomes, and prognostic factors were analyzed. RESULTS: A total of 107 patients were reviewed. The 5-year local recurrence free survival (LRFS), metastasis-free survival (MFS) and disease specific survival (DSS) were 84.7%, 78.6%, and 87.8%, respectively. Other primary malignancies and extrapulmonary metastasis were observed in 27 and 4 patients, respectively. The independent risk factor for local recurrence was R1/2 margin at the dedifferentiated component of the tumor. Metastasis was associated with tumor size in univariate analysis. The independent risk factor for DSS was tumor grade. Previous unplanned excision, de novo presentation, tumor depth, absence of the well-differentiated component, infiltrative border, R1/2 margin at the well-differentiated component were not associated with oncologic outcomes. CONCLUSIONS: This is the largest study examining DDL-E to-date. Localized DDL-E has low potential for metastasis and carries an excellent prognosis. Other primary malignancy and extrapulmonary metastasis are more frequent in DDL-E, thus close monitoring of other sites during follow-up is recommended. While wide resection margin is the standard surgical approach for DDL-E, further investigation into moderated wide resection margin at the well-differentiated component is warranted.

The humanistic care ability of nurses in 27 provinces in China: a multi-center cross-sectional study.

Background: Currently, studies found that the humanistic care ability of nurses is at low level in China, resulting in patients' concerns and dissatisfaction regarding the lack of empathy among nurses. We aimed to explore the factors that influence nurses' humanistic care ability, providing a new perspective on improving patient satisfaction and promote high quality medical services. Methods: A multi-center cross-sectional study recruited nurses from tertiary and secondary hospitals in China between July 2022 and August 2022. Data concerning self-developed questions on nurses' socio-demographic data and Caring Ability Inventory (CAI) were collected through the Questionnaire Star Platform, using a multi-stage sampling method. Results: The total score for the level of caring ability among the 15,653 surveyed Chinese nurses was 192.16 ± 24.94. Various factors significantly influence the level of humanistic care ability, including professional title, department, degree of passion for the job, job satisfaction, emphasis on self-care, participation in humanistic care training, support from family for the job, relationships with colleagues, satisfaction with salary, and previous experience working in pilot wards emphasizing humanistic care (p < 0.01). Conclusion: At present, nurses exhibit a comparatively modest proficiency in humanistic care ability. Numerous factors contribute to this situation. Nursing administrators ought to enhance the scope of humanistic care practices, conduct consistent professional training sessions, advocate for the implementation of model wards emphasizing humanistic care, foster a supportive organizational culture conducive to nurses, and underscore the significance of both nurturing nurses and promoting self-care among them.

Investigation of ComBat Harmonization on Radiomic and Deep Features from Multi-Center Abdominal MRI Data.

ComBat harmonization has been developed to remove non-biological variations for data in multi-center research applying artificial intelligence (AI). We investigated the effectiveness of ComBat harmonization on radiomic and deep features extracted from large, multi-center abdominal MRI data. A retrospective study was conducted on T2-weighted (T2W) abdominal MRI data retrieved from individual patients with suspected or known chronic liver disease at three study sites. MRI data were acquired using systems from three manufacturers and two field strengths. Radiomic features and deep features were extracted using the PyRadiomics pipeline and a Swin Transformer. ComBat was used to harmonize radiomic and deep features across different manufacturers and field strengths. Student's t-test, ANOVA test, and Cohen's F score were applied to assess the difference in individual features before and after ComBat harmonization. Between two field strengths, 76.7%, 52.9%, and 26.7% of radiomic features, and 89.0%, 56.5%, and 0.1% of deep features from three manufacturers were significantly different. Among the three manufacturers, 90.1% and 75.0% of radiomic features and 89.3% and 84.1% of deep features from two field strengths were significantly different. After ComBat harmonization, there were no significant differences in radiomic and deep features among manufacturers or field strengths based on t-tests or ANOVA tests. Reduced Cohen's F scores were consistently observed after ComBat harmonization. ComBat harmonization effectively harmonizes radiomic and deep features by removing the non-biological variations due to system manufacturers and/or field strengths in large multi-center clinical abdominal MRI datasets.

Comparison of risk factors between lean and non-lean MASLD in individuals with type 2 diabetes: a multicenter study.

OBJECTIVE: A multi-center study in patients with type 2 diabetes mellitus (T2DM) was performed to assess the differences of liver steatosis and fibrosis between lean and non-lean individuals. METHODS: Patients with T2DM from16 centers were recruited and underwent transient elastography examination for diagnosis of liver steatosis and fibrosis. Clinical information including diabetes status, serum lipids profiles, and inflammatory markers were collected. Potential risk factors of liver steatosis and fibrosis in lean (BMI＜23 kg/m2) and non-lean (BMI≥23 kg/m2) groups were analyzed. RESULTS: A total of 1762 patients were included. The prevalence of liver steatosis and fibrosis in lean group were 44.7% and 23.4%. Prevalence of hypertension and cardiovascular disease were higher in lean patients compared to non-lean group. Lean patients with liver steatosis or fibrosis were older, had longer diabetes duration, lower levels of HOMA-IR and serum lipids. Body mass index (BMI), visceral fat area (VFA) and triglyceride were among the most significant correlators of liver steatosis for both non-lean and lean patients. However, lipid profiles were different between two groups. Besides, insulin resistance, BMI and lipid levels were not observed to be associated with fibrosis in lean group. CONCLUSION: Liver steatosis and fibrosis in lean patients with T2DM were less associated with insulin resistance. Risk factors of liver steatosis were different between lean and non-lean patients, indicating the necessity of risk stratification and tailored management strategies.

A cost-effectiveness analysis of intraosseous basivertebral nerve ablation for the treatment of chronic low back pain.

BACKGROUND CONTEXT: Randomized trials have demonstrated the superiority of intraosseous basivertebral nerve ablation (BVNA) compared with sham and standard care in terms of improvements in pain, disability, and health-related quality of life in patients with vertebrogenic chronic low back pain (cLBP). PURPOSE: To assess the cost effectiveness of BVNA in patients with vertebrogenic cLBP compared to standard care alone. STUDY DESIGN/SETTING: A model-based economic analysis. PATIENT SAMPLE: Base case analysis used INTRACEPT, a randomized trial comparing BVNA with standard care in 140 patients with vertebrogenic cLBP, recruited from 23 sites across the United States, with a follow-up, up to 5 years. Scenario analyses compared data from the Surgical Multi-center Assessment of Radiofrequency Ablation for the Treatment of Vertebrogenic Back Pain (SMART) randomized trial against a sham control, and a single-arm study. OUTCOME MEASURES: Costs and quality-adjusted life years (QALYs) were calculated to determine the incremental cost-effectiveness ratio (ICER). METHODS: A cost-effectiveness model was built in Microsoft Excel® to evaluate the costs and health outcomes of patients undergoing BVNA using the Intracept Procedure (Relievant Medsystems) to treat vertebrogenic cLBP from a US payor perspective. Alternative scenario sensitivity analyses and probabilistic sensitivity analyses were conducted to assess the robustness of the model results. QALYs were discounted at 3.0% per year. RESULTS: Base case analysis showed that BVNA relative to standard care alone was a cost-effective strategy for the management of patients with vertebrogenic cLBP, with an ICER of US$11,376 per QALY at a 5-year time horizon from introduction of the procedure. Modeling demonstrated a >99% probability that this was cost effective in the US, based on a willingness-to-pay threshold of US$100,000 to US$150,000. Various sensitivity and scenario analyses produced ICERs that all remained below this threshold. CONCLUSIONS: BVNA with the Intracept Procedure offers patients with vertebrogenic cLBP, clinicians, and healthcare systems a cost-effective treatment compared to standard care alone.

Conception vessel acupuncture research regularity in the treatment of diminished ovarian reserve: a multi-center, large-sample prospective cohort study protocol.

Background: Diminished ovarian reserve (DOR) refers to a decrease in the number or quality of oocytes in the ovarian cortex, which is a degenerative disease of the reproductive system, and can further develop into premature ovarian failure. There are few studies on acupuncture and moxibustion for DOR, which are still in the exploratory stage. Methods/design: This study was a real-world case registry study. According to whether the subjects received conception vessel acupuncture or not, they were divided into the basic treatment combined with conception vessel acupuncture group and the basic treatment group. A total of 1221 patients with DOR were enrolled and treated for 12 weeks. The percentage of patients with ≥30% improvement in anti-Müllerian hormone (AMH) was evaluated at the end of week 12. Secondary outcomes included Antral follicle count (AFC), modified Kupperman scale, basal FSH level, LH level, FSH/LH ratio, positive pregnancy, clinical pregnancy, early spontaneous abortion, ongoing pregnancy, and ectopic pregnancy. Discussion: This study provides clinical evidence and theoretical support for the treatment of DOR with conception vessel acupuncture and moxibustion, so as to guide and improve the efficacy of acupuncture and moxibustion. Trial registration: Acupuncture-Moxibustion Clinical Trial Registry ChiCTR2400080471. Registered on 30 January 2024.

A Multi-Center Study on the Implementation and Challenges of Nursing Three-Level Ward Rounds in Tertiary Hospitals in Shanxi Province.

Background: The "Key Points of the Core System of Medical Quality and Safety" (hereinafter referred to as the "Key Points") was promulgated by the National Health Commission of China in 2018, requiring that nursing ward rounds should be carried out with reference to the three-level ward round system; In 2020 and 2022 editions of the "Evaluation Standards for Tertiary Hospitals", which were implemented in China, required that nursing ward rounds should be carried out with reference to the "Key Points". Additionally, the Action Plan for Comprehensively Improving Medical Quality (2023-2025) also mentions the need to improve the quality of three-level ward rounds. However, there are no detailed guidelines regarding implementing "Nursing Three-level Ward Rounds". Purpose: This study aimed to investigate the current situation of nursing three-level ward rounds in tertiary hospitals after the promulgation of the "Key Points of the Core System of Medical Quality and Safety" to provide insights and guidelines regarding relevant standards, so as to better implement of the requirements of "nursing ward rounds" in the "Evaluation Standards for Tertiary Hospitals" and "improving the quality of three-level ward rounds" in the "Action Plan". Methods: A multi-center study was conducted in February 2024, including all tertiary public hospitals in the Shanxi Province, China. A questionnaire survey using the self-designed "Questionnaire on the Implementation of Nursing Three-level Ward Rounds" was carried out. The questionnaire included the basic information of the hospital and the implementation of the three-level (namely I, II, and III) rounds (including "five aspects": ward round personnel, object, content, frequency, and record), which is expressed by quantity and composition ratio. Next is the text analysis method. First, the "five aspects" of the hospital that filled in the questionnaire survey with "nursing three-level ward rounds have been carried out" were assessed. Second, the five aspects of each hospital were assessed for consistency with the "Nursing Three-level Ward Rounds System" (hereinafter referred to as the "System") of their respective hospitals.Third, the consistency of the "System" of the hospital with the "Key Points" was assessed. The results of the analysis of the former are expressed in terms of quantity and composition ratio; the results of the latter two were analyzed using Fisher's exact test method to compare any differences. Results: Notably, 14 of the 67 tertiary public hospitals (20.9%) carried out nursing three-level ward rounds. There were 4-10 situations in the five aspects of I, II, and III ward rounds filled in by the hospitals. The five aspects of the I, II, III ward rounds in 14 hospitals were significantly comparable with the "System", which, in turn, was comparable with the "Key Points" (P < 0.05). Conclusion: Not all tertiary public hospitals in the Shanxi Province have not all carried out nursing three-level ward rounds. Furthermore, the five aspects of the hospitals that carried out nursing three-level ward rounds were not entirely consistent in terms of ward round personnel, object, content, frequency, and record. The filling in of the nursing three-level ward rounds carried out by the hospitals is inconsistent with the respective "System"; the "System" of the hospital is not in line with the "Key Points". Impact on Nursing Work: Nursing administrators should be aware of the newly issued norms and requirements in their workplace, and revise the relevant systems in accordance with the norms and requirements in a timely manner. Additionally, the revision of the system should cover the core requirements of the norms and be practicable. The system should be supervised to ensure that 100% of the implementation is in accordance with the system.

Outcome of hepatectomy after systemic therapy for hepatocellular carcinoma: a Japanese multicenter study.

BACKGROUND AND PURPOSE: In recent years, new systemic therapies have been developed for hepatocellular carcinoma (HCC). The aim of this study was to evaluate the prognosis of patients with unresectable HCC treated with R0 hepatectomy after systemic therapy. METHODS: Data from 27 patients who underwent hepatectomy for HCC after systemic therapy at six facilities were analyzed retrospectively. Cancer-specific survival (CSS) and recurrence-free survival (RFS) after hepatectomy were investigated using Kaplan-Meier curves. We examined the prognostic value of the oncological criteria of resectability for HCC reported by the Japanese Expert Consensus 2023. RESULTS: R0 resection was performed in 24 of the 27 patients. Using the Response Evaluation Criteria in Solid Tumors, 0 patient had a complete response, 16 had a partial response, 6 had stable disease, and 2 had progressive disease. Median CSS was not evaluated, but the median RFS was 17.8 months. Patients with resectable and borderline resectable (BR) 1 cancers had a better prognosis than those with BR2 cancers. The group whose oncological criteria were improved by systemic therapy had a lower recurrence rate than the group whose oncological criteria were maintained, but no difference was observed in CSS. CONCLUSIONS: The findings of this study suggest that hepatectomy after systemic therapy may improve the prognosis of HCC patients.

Peripheral neuropathy and associated factors among type 2 diabetic patients attending referral hospitals in the Amhara region, a multi-center cross-sectional study in Ethiopia.

Diabetic peripheral neuropathy is one of the diabetes most common microvascular complications. It is very prevalent in Sub-Saharan Africa due to a combination of causes, including rising diabetes prevalence, limited healthcare resources, and a lack of access to competent medical care. However, just a few studies have been undertaken in the study area. Institution-based cross-sectional study was conducted in the Amhara region referral hospitals, in 2022. By using a systematic random sampling technique, a total of 627 respondents were included. The data was entered into EPI Data version 4.6 and exported to SPSS version 25 for further analysis. A binary logistic regression was used to determine the relationship between the dependent and predictor variables. The association between predictor variables and the dependent variable was determined using multivariate logistic regression [p value < 0.05, 95% confidence interval]. The overall prevalence of diabetic peripheral neuropathy among the study participants was 48.2% (95% CI; 44.2, 52.1). Aged between 40 and 60 years (AOR = 4:27; 95% CI 2.62, 6.94), and 60 years and older (AOR = 4:47; 95% CI 2.40, 8.35), participants who have lived alone (AOR = 2:14; 95% CI 1.21, 3.79), patients with comorbidity (AOR = 1:83; 95% CI 1.22, 2.76), and being physically inactive (AOR = 1:69; 95% CI 1.14, 2.49) were significantly associated with Diabetic peripheral neuropathy. Diabetic peripheral neuropathy was high among diabetic patients. Healthcare providers should prioritize regular screening and early intervention for individuals at higher risk, particularly those aged 40 and above, those living alone, patients with comorbid conditions, and those who are physically inactive. Implementing community-based support programs, encouraging physical activity, and providing comprehensive management plans for diabetes and associated comorbidities can help mitigate the risk and improve the quality of life for these patients.

How to customize common data models for rare diseases: an OMOP-based implementation and lessons learned.

BACKGROUND: Given the geographical sparsity of Rare Diseases (RDs), assembling a cohort is often a challenging task. Common data models (CDM) can harmonize disparate sources of data that can be the basis of decision support systems and artificial intelligence-based studies, leading to new insights in the field. This work is sought to support the design of large-scale multi-center studies for rare diseases. METHODS: In an interdisciplinary group, we derived a list of elements of RDs in three medical domains (endocrinology, gastroenterology, and pneumonology) according to specialist knowledge and clinical guidelines in an iterative process. We then defined a RDs data structure that matched all our data elements and built Extract, Transform, Load (ETL) processes to transfer the structure to a joint CDM. To ensure interoperability of our developed CDM and its subsequent usage for further RDs domains, we ultimately mapped it to Observational Medical Outcomes Partnership (OMOP) CDM. We then included a fourth domain, hematology, as a proof-of-concept and mapped an acute myeloid leukemia (AML) dataset to the developed CDM. RESULTS: We have developed an OMOP-based rare diseases common data model (RD-CDM) using data elements from the three domains (endocrinology, gastroenterology, and pneumonology) and tested the CDM using data from the hematology domain. The total study cohort included 61,697 patients. After aligning our modules with those of Medical Informatics Initiative (MII) Core Dataset (CDS) modules, we leveraged its ETL process. This facilitated the seamless transfer of demographic information, diagnoses, procedures, laboratory results, and medication modules from our RD-CDM to the OMOP. For the phenotypes and genotypes, we developed a second ETL process. We finally derived lessons learned for customizing our RD-CDM for different RDs. DISCUSSION: This work can serve as a blueprint for other domains as its modularized structure could be extended towards novel data types. An interdisciplinary group of stakeholders that are actively supporting the project's progress is necessary to reach a comprehensive CDM. CONCLUSION: The customized data structure related to our RD-CDM can be used to perform multi-center studies to test data-driven hypotheses on a larger scale and take advantage of the analytical tools offered by the OHDSI community.

Preliminary Evaluation of Fine-Tuning the OpenDeLD Deidentification Pipeline Across Multi-Center Corpora.

Automatic deidentification of Electronic Health Records (EHR) is a crucial step in secondary usage for biomedical research. This study introduces evaluation of an intricate hybrid deidentification strategy to enhance patient privacy in secondary usage of EHR. Specifically, this study focuses on assessing automatic deidentification using OpenDeID pipeline across diverse corpora for safeguarding sensitive information within EHR datasets by incorporating diverse corpora. Three distinct corpora were utilized: the OpenDeID v2 corpus containing pathology reports from Australian hospitals, the 2014 i2b2/UTHealth deidentification corpus with clinical narratives from the USA, and the 2016 CEGS N-GRID identification corpus comprising psychiatric notes. The OpenDeID pipeline employs a hybrid approach based on deep learning and contextual rules. Pre-processing steps involved harmonizing and addressing encoding and format issues. Precision, Recall, F-measure metrics were used to assess the performance. The evaluation metrics demonstrated the superior performance of the Discharge Summary BioBERT model. Trained on three corpora with a total of 4,038 reports, the best performing model exhibited robust deidentification capabilities when applied to EHR. It achieved impressive micro-averaged F1-scores of 0.9248 and 0.9692 for strict and relaxed settings, respectively. These results offer valuable insights into the model's efficacy and its potential role in safeguarding patient privacy in secondary usage of EHR.

Large-scale pretrained frame generative model enables real-time low-dose DSA imaging: An AI system development and multi-center validation study.

BACKGROUND: Digital subtraction angiography (DSA) devices are commonly used in numerous interventional procedures across various parts of the body, necessitating multiple scans per procedure, which results in significant radiation exposure for both doctors and patients. Inspired by generative artificial intelligence techniques, this study proposes GenDSA, a large-scale pretrained multi-frame generative model-based real-time and low-dose DSA imaging system. METHODS: GenDSA was developed to generate 1-, 2-, and 3-frame sequences following each real frame. A large-scale dataset comprising ∼3 million DSA images from 27,117 patients across 10 hospitals was constructed to pretrain, fine-tune, and validate GenDSA. Two other datasets from 25 hospitals were used for evaluation. Objective evaluations included SSIM and PSNR. Five interventional radiologists independently assessed the quality of the generated frames using the Likert scale and visual Turing test. Scoring consistency among the radiologists was measured using the Kendall coefficient of concordance (W). The Fleiss' kappa values were used for inter-rater agreement analysis for visual Turing tests. FINDINGS: Using only one-third of the clinical radiation dose, videos generated by GenDSA were perfectly consistent with real videos. Objective evaluations demonstrated that GenDSA's performance (PSNR = 36.83, SSIM = 0.911, generation time = 0.07 s/frame) surpassed state-of-the-art algorithms. Subjective ratings and statistical results from five doctors indicated no significant difference between real and generated videos. Furthermore, the generated videos were comparable to real videos in overall quality (4.905 vs. 4.935) and lesion assessment (4.825 vs. 4.860). CONCLUSIONS: With clear clinical and translational values, the developed GenDSA can significantly reduce radiation damage to both doctors and patients during DSA-guided procedures. FUNDING: This study was supported by the National Key R&D Program and the National Natural Science Foundation of China.

Ocular Disease Detection with Deep Learning (Fine-Grained Image Categorization) Applied to Ocular B-Scan Ultrasound Images.

INTRODUCTION: The aim of this work is to develop a deep learning (DL) system for rapidly and accurately screening for intraocular tumor (IOT), retinal detachment (RD), vitreous hemorrhage (VH), and posterior scleral staphyloma (PSS) using ocular B-scan ultrasound images. METHODS: Ultrasound images from five clinically confirmed categories, including vitreous hemorrhage, retinal detachment, intraocular tumor, posterior scleral staphyloma, and normal eyes, were used to develop and evaluate a fine-grained classification system (the Dual-Path Lesion Attention Network, DPLA-Net). Images were derived from five centers scanned by different sonographers and divided into training, validation, and test sets in a ratio of 7:1:2. Two senior ophthalmologists and four junior ophthalmologists were recruited to evaluate the system's performance. RESULTS: This multi-center cross-sectional study was conducted in six hospitals in China. A total of 6054 ultrasound images were collected; 4758 images were used for the training and validation of the system, and 1296 images were used as a testing set. DPLA-Net achieved a mean accuracy of 0.943 in the testing set, and the area under the curve was 0.988 for IOT, 0.997 for RD, 0.994 for PSS, 0.988 for VH, and 0.993 for normal. With the help of DPLA-Net, the accuracy of the four junior ophthalmologists improved from 0.696 (95% confidence interval [CI] 0.684-0.707) to 0.919 (95% CI 0.912-0.926, p < 0.001), and the time used for classifying each image reduced from 16.84 ± 2.34 s to 10.09 ± 1.79 s. CONCLUSIONS: The proposed DPLA-Net showed high accuracy for screening and classifying multiple ophthalmic diseases using B-scan ultrasound images across mutiple centers. Moreover, the system can promote the efficiency of classification by ophthalmologists.

Non-linear relationship between red blood cell distribution width and gastrointestinal bleeding risk in stroke patients: results from multi-center ICUs.

Background: The red blood cell distribution width (RDW) is closely linked to the prognosis of multiple diseases. However, the connection between RDW and gastrointestinal bleeding (GIB) in stroke patients is not well understood. This study aimed to clarify this association. Methods: This retrospective study involved 11,107 hospitalized patients from 208 hospitals in the United States, admitted between January 1, 2014, and December 31, 2015. We examined clinical data from 7,512 stroke patients in the intensive care unit (ICU). Multivariate logistic regression assessed the link between RDW and in-hospital GIB in stroke patients. Generalized additive model (GAM) and smooth curve fitting (penalty spline method) were utilized to explore the non-linear relationship between RDW and GIB in stroke patients. The inflection point was calculated using a recursive algorithm, and interactions between different variables were assessed through subgroup analyses. Results: Among the 11,107 screened stroke patients, 7,512 were included in the primary analysis, with 190 identified as having GIB. The participants had a mean age of (61.67 ± 12.42) years, and a median RDW of 13.9%. Multiple logistic analysis revealed RDW as a risk factor for in-hospital GIB in stroke patients (OR = 1.28, 95% CI 1.21, 1.36, p < 0.05). The relationship between RDW and in-hospital GIB in stroke patients was found to be non-linear. Additionally, the inflection point of RDW was 14.0%. When RDW was ≥14.0%, there was a positive association with the risk of GIB (OR: 1.24, 95% CI: 1.16, 1.33, p < 0.0001). Conversely, when RDW was <14.0%, this association was not significant (OR: 1.02, 95% CI: 0.97-1.07, p = 0.4040). Conclusion: This study showed a substantial non-linear link between RDW and the risk of GIB in stroke patients. Maintaining the patient's RDW value below 14.0% could lower the risk of in-hospital GIB.

Synthetic CT generation for pelvic cases based on deep learning in multi-center datasets.

BACKGROUND AND PURPOSE: To investigate the feasibility of synthesizing computed tomography (CT) images from magnetic resonance (MR) images in multi-center datasets using generative adversarial networks (GANs) for rectal cancer MR-only radiotherapy. MATERIALS AND METHODS: Conventional T2-weighted MR and CT images were acquired from 90 rectal cancer patients at Peking University People's Hospital and 19 patients in public datasets. This study proposed a new model combining contrastive learning loss and consistency regularization loss to enhance the generalization of model for multi-center pelvic MRI-to-CT synthesis. The CT-to-sCT image similarity was evaluated by computing the mean absolute error (MAE), peak signal-to-noise ratio (SNRpeak), structural similarity index (SSIM) and Generalization Performance (GP). The dosimetric accuracy of synthetic CT was verified against CT-based dose distributions for the photon plan. Relative dose differences in the planning target volume and organs at risk were computed. RESULTS: Our model presented excellent generalization with a GP of 0.911 on unseen datasets and outperformed the plain CycleGAN, where MAE decreased from 47.129 to 42.344, SNRpeak improved from 25.167 to 26.979, SSIM increased from 0.978 to 0.992. The dosimetric analysis demonstrated that most of the relative differences in dose and volume histogram (DVH) indicators between synthetic CT and real CT were less than 1%. CONCLUSION: The proposed model can generate accurate synthetic CT in multi-center datasets from T2w-MR images. Most dosimetric differences were within clinically acceptable criteria for photon radiotherapy, demonstrating the feasibility of an MRI-only workflow for patients with rectal cancer.

Automated interpretation of retinal vein occlusion based on fundus fluorescein angiography images using deep learning: A retrospective, multi-center study.

Purpose: Fundus fluorescein angiography (FFA) is the gold standard for retinal vein occlusion (RVO) diagnosis. This study aims to develop a deep learning-based system to diagnose and classify RVO using FFA images, addressing the challenges of time-consuming and variable interpretations by ophthalmologists. Methods: 4028 FFA images of 467 eyes from 463 patients were collected and annotated. Three convolutional neural networks (CNN) models (ResNet50, VGG19, InceptionV3) were trained to generate the label of image quality, eye, location, phase, lesions, diagnosis, and macular involvement. The performance of the models was evaluated by accuracy, precision, recall, F-1 score, the area under the curve, confusion matrix, human-machine comparison, and Clinical validation on three external data sets. Results: The InceptionV3 model outperformed ResNet50 and VGG19 in labeling and interpreting FFA images for RVO diagnosis, achieving 77.63%-96.45% accuracy for basic information labels and 81.72%-96.45% for RVO-relevant labels. The comparison between the best CNN and ophthalmologists showed up to 19% accuracy improvement with the inceptionV3. Conclusion: This study developed a deep learning model capable of automatically multi-label and multi-classification of FFA images for RVO diagnosis. The proposed system is anticipated to serve as a new tool for diagnosing RVO in places short of medical resources.

Survival Analysis Without Sharing of Individual Patient Data by Using a Gaussian Copula.

Cox regression and Kaplan-Meier estimations are often needed in clinical research and this requires access to individual patient data (IPD). However, IPD cannot always be shared because of privacy or proprietary restrictions, which complicates the making of such estimations. We propose a method that generates pseudodata replacing the IPD by only sharing non-disclosive aggregates such as IPD marginal moments and a correlation matrix. Such aggregates are collected by a central computer and input as parameters to a Gaussian copula (GC) that generates the pseudodata. Survival inferences are computed on the pseudodata as if it were the IPD. Using practical examples we demonstrate the utility of the method, via the amount of IPD inferential content recoverable by the GC. We compare GC to a summary-based meta-analysis and an IPD bootstrap distributed across several centers. Other pseudodata approaches are also considered. In the empirical results, GC approximates the utility of the IPD bootstrap although it might yield more conservative inferences and it might have limitations in subgroup analyses. Overall, GC avoids many legal problems related to IPD privacy or property while enabling approximation of common IPD survival analyses otherwise difficult to conduct. Sharing more IPD aggregates than is currently practiced could facilitate "second purpose"-research and relax concerns regarding IPD access.