As a lightweight and non-invasive imaging technique, lung ultrasound (LUS) has gained importance for assessing lung pathologies. The use of Artificial intelligence (AI) in medical decision support systems is promising due to the time- and expertise-intensive interpretation, however, due to the poor quality of existing data used for training AI models, their usability for real-world applications remains unclear. Methods: In a prospective study, we analyze data from 63 COVID-19 suspects (33 positive) collected at Maastricht University Medical Centre. Ultrasound recordings at six body locations were acquired following the BLUE protocol and manually labeled for severity of lung involvement. Anamnesis and complete blood count (CBC) analyses were conducted. Several AI models were applied and trained for detection and severity of pulmonary infection. Results: The severity of the lung infection, as assigned by human annotators based on the LUS videos, is not significantly different between COVID-19 positive and negative patients ($p = 0.89$). Nevertheless, the predictions of image-based AI models identify a COVID-19 infection with 65% accuracy when applied zero-shot (i.e., trained on other datasets), and up to 79% with targeted training, whereas the accuracy based on human annotations is at most 65%. Multi-modal models combining images and CBC improve significantly over image-only models. Conclusion: Although our analysis generally supports the value of AI in LUS assessment, the evaluated models fall short of the performance expected from previous work. We find this is due to 1) the heterogeneity of LUS datasets, limiting the generalization ability to new data, 2) the frame-based processing of AI models ignoring video-level information, and 3) lack of work on multi-modal models that can extract the most relevant information from video-, image- and variable-based inputs. To aid future research, we publish the dataset at: https://github.com/NinaWie/COVID-BLUES.

Background/Objectives: Hepatitis E virus (HEV) is one of the leading causes of acute hepatitis, with immunosuppressed individuals, such as oncology patients, being particularly vulnerable to chronic infections that may progress to liver disease or fatal outcomes. Assay variability complicates HEV prevalence assessment in at-risk groups. This study aimed to compare the reliability and concordance of three HEV antibody assays—Wantai, Euroimmun, and Elecsys^®—in immunosuppressed oncology patients. Methods: In this prospective pilot study, serum samples were obtained from oncology patients between September 2020 and October 2021. Samples were collected both at baseline (treatment-naive) and during ongoing treatment. A healthy control group was retrospectively included for comparative analysis. Anti-HEV IgM and IgG antibodies were tested in all samples using enzyme-linked immunosorbent assays (Wantai, Euroimmun) and an electrochemiluminescence immunoassay (Elecsys^®). Demographic and clinical data, along with information on HEV risk factors, were extracted from medical records and patient questionnaires. Results: HEV IgM prevalence ranged from 0% (Wantai) to 6% (Elecsys^®), while IgG prevalence was 12% (Euroimmun), 38% (Wantai), and 53% (Elecsys^®). Concordance was poor, with Cohen’s Kappa values indicating slight to moderate agreement (κ = 0.000–0.553). Patients with hematological malignancies exhibited the highest IgG seroprevalence. Risk factor analysis revealed the highest association between HEV exposure and the consumption of undercooked pork or crop-based agriculture. Conclusions: Significant variability among HEV serological assays highlights the challenges of reliable HEV diagnostics in immunosuppressed oncology patients. Assay selection and improved testing strategies are critical for this high-risk group.

Objective: To establish organotypic human brain slice cultures (hBSCs) as a translational screening platform for evaluating novel neuromodulation devices and to demonstrate the feasibility of the model using magnetoelectric nanoparticles (MENPs) as a representative neurostimulation modality. Approach: Viable hBSCs were prepared from resected cortical tissue of epilepsy surgery patients and a GCaMP-based calcium imaging, multi-electrode array (MEA) recordings, and immunohistochemical staining for c-Fos was conducted. The MENPs were injected into the hBSCs and stimulated with an alternating magnetic field to assess their neuromodulatory effects. Main Results: GCaMP transduction enables the real-time visualization of MENP-induced neuronal activity. Electrophysiological signals, including spiking and local field potentials, were observed in fresh, but not cultured, slices. c-Fos immunostaining revealed a significant increase in c-Fos expression in stimulated MENP-injected cultures compared to sham-treated controls. This protocol yielded reproducible tissue viability and consistent results across patient-derived samples. Significance: This technical note demonstrates that hBSCs represent a reproducible and ethically preferable translational model suitable for screening applications in neurotechnology research. The platform enables early-stage functional evaluation of neuromodulatory devices, particularly those with a higher risk of failure in vivo or curiosity-driven early-phase concepts in a setting superior to traditional in vitro approaches. This platform may help reduce reliance on animal models in neurotechnology development.