Due to the openness and easy accessibility of online social media (OSM), anyone can easily contribute a simple paragraph of text to express their opinion on an article that they have seen. Without access control mechanisms, it has been reported that there are many suspicious messages and accounts spreading across multiple platforms. Accordingly, identifying and labeling fake news is a demanding problem due to the massive amount of heterogeneous content. In essence, the functions of machine learning (ML) and natural language processing (NLP) are to enhance, speed up, and automate the analytical process. Therefore, this unstructured text can be transformed into meaningful data and insights. In this paper, the combination of ML and NLP are implemented to classify fake news based on an open, large and labeled corpus on Twitter. In this case, we compare several state-of-the-art ML and neural network models based on content-only features. To enhance classification performance, before the training process, the term frequency-inverse document frequency (TF-IDF) features were applied in ML training, while word embedding was utilized in neural network training. By implementing ML and NLP methods, all the traditional models have greater than 85% accuracy. All the neural network models have greater than 90% accuracy. From the experiments, we found that the neural network models outperform the traditional ML models by, on average, approximately 6% precision, with all neural network models reaching up to 90% accuracy.

The sequential organ failure assessment (SOFA) and quick sequential organ failure assessment (qSOFA) scores are new tools which are used to assess sepsis based on the Third International Consensus Definitions for Sepsis and Septic Shock Task Force. This study aimed to evaluate the feasibility of using the SOFA and qSOFA to predict post-ureteroscopic lithotripsy (URSL) sepsis. Patients who underwent URSL due to ureteral stone obstruction were retrospectively reviewed using SOFA and qSOFA scores. Patient characteristics including age, gender, comorbidities, American Society of Anesthesiologists Classification, stone burden, stone location, hydronephrosis status, infectious status, preoperative SOFA and qSOFA score were collected. Preoperative factors were analyzed to determine if they were correlated with postoperative sepsis. A total of 830 patients were included in this study, of whom 32 (3.9%) had postoperative sepsis. Multivariate analysis revealed that older age, proximal ureteral stones, severe hydronephrosis, and high preoperative qSOFA or SOFA score were significantly associated with postoperative sepsis. The areas under the curves of a qSOFA score ≥ 1 and SOFA score ≥ 2 for predicting postoperative sepsis were 0.754 and 0.823, respectively. Preoperative qSOFA and SOFA scores are convenient and effective for predicting post-URSL sepsis. Further preventive strategies should be performed in these high-risk patients.

In social networks, a single user may create multiple accounts to spread his/her opinions and to influence others, by actively commenting on different news pages. It would be beneficial to both social networks and their communities to demote such abnormal activities, and the first step is to detect those accounts. However, the detection of multiple accounts is challenging, because they may have very realistic names and reasonable activity patterns. In this paper, we investigate three different approaches, and propose using graph embedding together with semi-supervised learning to predict whether multiple accounts are created by the same user. We carry out extensive experimental analyses to understand how changes in the input data and algorithmic parameters of optimization affect the prediction performance. We also note that local information have higher importance than the global ones for such prediction, and point out the threshold leading to the best results. We test the proposed approach with 6700 Facebook pages from the Middle East, and achieve the averaged accuracy at 0.996 and AUC (area under curve)at 0.952 for users with the same name; with the U.S. 2016 election dataset, we obtain the best AUC at 0.877 for users with different names.

A core issue in online social networks is identifying content that will result in greater traffic and interaction. In this paper we use a large Facebook dataset and the concepts of bandwagon effect and information cascade from the field of communications to predict a post's life cycle without analyzing its content. Results from two models indicate evidence of (a) both factors being found on most discussion pages and (b) both factors having predictive power regarding the final number of participants in a post. To test for the bandwagon effect, we designed a system to predict article lifecycles, and found that for most Facebook pages, such predictions can be made within 30 minutes. We offer our data and analysis in support of a better understanding of the linkage between online social media research and journalism/information science theory, one that facilitates accurate predictions regarding posts that attract strong interest.

Rationale: Cavernous hemangioma (CH) is not commonly found within the abdomen or the retroperitoneum. We report the first case of CH originating from the gonadal vein. Diagnosis: A retroperitoneal tumor was found incidentally in a 57-year-old female patient. The differential diagnoses from the initial imaging studies included gastrointestinal stromal tumor, carcinoid tumor, neurogenic tumor, metastasis, lymphadenopathy, or another rare tumor. Interventions: A surgical en-bloc excision was performed via a subcostal incision and intravenous CH arising from a gonadal vein was diagnosis by the urological pathologist. Outcomes: After the surgery, no complications were noted. A computed tomography scan was performed after 3 months follow-up and no tumor recurrence was found. Lessons: This case reminds us that CH should be listed as one of the differential diagnoses for a retroperitoneal tumor. A definite diagnosis of CH relies on surgical resection. The prognosis is well if adequate resection is performed.

Online social media (OSM) platforms have become hotbeds for various cybercrimes. Attackers often employ “shotgun” approaches, slightly altering their malicious content to evade detection and distributing it across multiple OSM channels to target more victims. This tactic is used to launch fake news, political propaganda, and social engineering attacks. To address these issues, we propose a novel framework that uses artificial intelligence (AI) to effectively trace the source of disinformation and detect content with abnormal similarities. Our approach not only analyzes the content itself but also focuses on the external uniform resource locators (URLs) associated with the content and the specific pages where the malicious content appears. The framework is designed to filter out suspicious content, external URLs, and channels based on a few verified instances of disinformation, using an adaptive approach that can quickly respond to evolving disinformation tactics. The framework was evaluated using a real dataset and revealed that certain channels on OSM platforms function as critical bridges between normal and disinformation clusters. Our study offers valuable insights for OSM platforms and administrators to proactively detect clusters of disinformation messages, external URLs, and channels with only a small number of manually verified samples. Additionally, it illuminates the role and dynamics of how disinformation evolves and spreads, which can aid in developing more effective strategies to combat disinformation on OSM platforms.