Open Access

Automated question generation holds great promise in many fields, such as education, to reduce the workload and automate an otherwise tedious task. However, major challenges remain regarding the quality of generated questions. To identify and address these challenges generated questions are evaluated either automatically or manually. While several automated metrics, mostly based on the comparison with a gold standard, exist, their usefulness is limited and human evaluation is often used for more accurate assessment. Our research generates questions using several models and methods, including fine-tuning, zero-shot and few-shot. We compare model performance by classifying the generated questions using a multi-label approach. This approach evaluates by sorting generated questions into zero or more binary problem classes and attempting to identify different problems with the generated questions. Our results show that different models tend to generate questions that fit into different problem classes. Additionally, the problem classification evaluation is capable of recognizing these differences and weighing the classes for the models accordingly, creating model-specific distribution characteristics.

Background: Falls are a common problem experienced by people living with HIV yet predictive models specific to this population remain underdeveloped. We aimed to identify, assess and stratify the predictive strength of various physiological, behavioral, and HIV-specific factors associated with falls among people living with HIV and inform a predictive model for fall prevention. Methods: Systematic review and meta-analysis were conducted to explore predictors of falls in people living with HIV. Data was sourced, screened, extracted, and analyzed by two independent reviewers from eight databases up to January 2nd, 2024, following the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) protocol. Evidence quality and bias were assessed using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) and the Mixed Method Appraisal Tool (MMAT), respectively. Pooled odds ratios (OR) with 95% confidence intervals (CI) were computed using random-effects models to establish associations between predictors and falls risk. We applied established criteria (Bradford Hill’s criteria, Rothman’s and Nweke’s viewpoints) to stratify risk factors and create a weighted predictive algorithm. Results: This review included 12 studies on falls/balance dysfunction in 117,638 participants (54,513 people living with HIV), with varying ages (45–50 years), sample sizes (32 − 26,373), study durations (6 months to 15 years), disease stages (CD4 + counts 347.2 cells/mm³ to ≥ 500 cells/µL) and fall definitions (self-reported histories to real-time reporting). Some predictors of falls in people living with HIV including depression, cannabis use, cognitive impairment/neurocognitive adverse effects (NCAE), hypertension, and stavudine—showed perfect risk responsiveness (Ri = 1), indicating their strong association with falls. Notably, cannabis use demonstrated the highest risk weight (Rw = 3.0, p < 0.05, 95%CI:1.51–5.82), followed by NCAE (Rw = 2.3, p < 0.05, 95%CI:1.66–3.21) and frailty with a broad confidence interval (Rw = 2.2, p < 0.05, 95%CI:0.73–14.40). Other significant predictors included hypertension (Rw = 1.8, p < 0.05, 95%CI:1.33–2.33), depression (Rw = 1.6, p < 0.05, 95%CI:1.22–2.18), stavudine use (Rw = 1.5, p < 0.05, 95%CI: 0.95–2.25), neuropathy (Rw = 1.3, p < 0.05, 95%CI:1.26–2.11), and polypharmacy (Rw = 1.2, p < 0.05, 95%CI:1.16–1.96). The fall risk threshold score was 12.8, representing the 76th percentile of the specific and sufficient risk weight. Conclusion: Our meta-analysis identifies predictors of falls in people living with HIV, emphasizing physiological, behavioral, and HIV-specific factors. Integrating these into clinical practice could mitigate falls-related sequelae. We propose a novel approach to falls risk prediction using a novel clinical index, resulting in a HIV-specific falls risk assessment tool.

Background: Cachexia accounts for about 20% of all cancer‐related deaths and indicates poor prognosis. The impact of Fusobacterium nucleatum (Fn), a microbial risk factor for colorectal cancer (CRC), on the development of cachexia in CRC has not been established. Methods: We evaluated the association between Fn abundance in pre‐surgical stool samples and onset of cachexia at 6 months post‐surgery in n = 87 patients with stages I–III CRC in the ColoCare Study. Results: High fecal Fn abundance compared to negative/low fecal Fn abundance was associated with 4‐fold increased risk of cachexia onset at 6 months post‐surgery (OR = 4.82, 95% CI = 1.15, 20.10, p = 0.03). Conclusion: Our findings suggest that high fecal Fn abundance was associated with an increased risk of cachexia at 6 months post‐surgery in CRC patients. This is the first study to link Fn abundance with cachexia in CRC patients, offering novel insights into biological mechanisms and potential management of cancer cachexia. Due to the small sample size, our results should be interpreted with caution. Future studies with larger sample sizes are needed to validate these findings.

Diese Bachelorarbeit untersucht anhand einer empirischen Studie das Potenzial virtueller Influencer (VIs) in der Unternehmenskommunikation, insbesondere im Hinblick auf die Glaubwürdigkeit. Ziel ist es, zu analysieren, wie menschenähnliche (z.B. Lil Miquela) und comicähnliche VIs (z.B. Noonoouri) auf Social Media wahrgenommen werden und welche Rolle ihre visuelle Darstellung für die Glaubwürdigkeitskommunikation spielt. Basierend auf theoretischen Grundlagen zur Glaubwürdigkeit und einer empirischen Online-Befragung wird ein Bewertungsmodell entwickelt, das die Faktoren Vertrauenswürdigkeit, Authentizität, Ähnlichkeit, Attraktivität, Dynamik und vermittelte Realität einbezieht. Die Arbeit zeigt, dass VIs sowohl Chancen als auch Herausforderungen für die Glaubwürdigkeit bieten und liefert wertvolle Implikationen für den strategischen Einsatz von VIs in der Unternehmenskommunikation.

The development of methods for the meta‐analysis of diagnostic test accuracy (DTA) studies is still an active area of research. While methods for the standard case where each study reports a single pair of sensitivity and specificity are nearly routinely applied nowadays, methods to meta‐analyze receiver operating characteristic (ROC) curves are not widely used. This situation is more complex, as each primary DTA study may report on several pairs of sensitivity and specificity, each corresponding to a different threshold. In a case study published earlier, we applied a number of methods for meta‐analyzing DTA studies with multiple thresholds to a real‐world data example (Zapf et al., Biometrical Journal. 2021; 63(4): 699–711). To date, no simulation study exists that systematically compares different approaches with respect to their performance in various scenarios when the truth is known. In this article, we aim to fill this gap and present the results of a simulation study that compares three frequentist approaches for the meta‐analysis of ROC curves. We performed a systematic simulation study, motivated by an example from medical research. In the simulations, all three approaches worked partially well. The approach by Hoyer and colleagues was slightly superior in most scenarios and is recommended in practice.