Abstract
Common variable immune deficiency (CVID) is a primary immunodeficiency disorder, with hypogammaglobulinemia and increased susceptibility to recurrent infections, autoimmune disorders, granulomatous diseases and malignancy. Among the solid organ transplant (SOT) recipient population, those with primary immunodeficiency disorders under chronic immunosuppression therapy can theoretically be at higher risk of atypical infections, autoimmune complications and disease recurrence with suboptimal long term graft survival, but literature is scarce. Here, we report a 27-year-old female with type 1 diabetes mellitus, complicated with nephropathy that progressed to end-stage renal disease (ESRD), who had a history of a chronic inflammatory response dysregulation, with chronic monoarthritis, persistent elevation of inflammation markers, recurrent infections, low immunoglobulin G (IgG) and A (IgA) serum levels, a slightly decreased population of memory B cells at flow cytometric immunophenotyping, and a confirmed pathological heterozygous mutation in the tumor necrosis factor receptor superfamily 13B (TNFRSF13B), with a suspected diagnosis of CVID. Whilst on hemodialysis, she received a simultaneous kidney and pancreas transplant from a standard criteria donor (SCD), and our induction and maintenance immunosuppression protocol and prophylaxis regimen allowed for a successful transplant with immediate pancreatic function, with no evidence of renal graft rejection upon biopsy in the early post-transplant period, and no novel episodes of serious infectious complications were recorded during a follow-up period of six months.

Abstract
This work considers the problem of segmenting heart sounds into their fundamental components. We unify statistical and data-driven solutions by introducing Markov-based Neural Networks (MNNs), a hybrid end-to-end framework that exploits Markov models as statistical inductive biases for an Artificial Neural Network (ANN) discriminator. We show that an MNN leveraging a simple one-dimensional Convolutional ANN significantly outperforms two recent purely data-driven solutions for this task in two publicly available datasets: PhysioNet 2016 (Sensitivity: 0.947 +/- 0.02; Positive Predictive Value : 0.937 +/- 0.025) and the CirCor DigiScope 2022 (Sensitivity: 0.950 +/- 0.008; Positive Predictive Value: 0.943 +/- 0.012). We also propose a novel gradient-based unsupervised learning algorithm that effectively makes the MNN adaptive to unseen datum sampled from unknown distributions. We perform a cross dataset analysis and show that an MNN pre-trained in the CirCor DigiScope 2022 can benefit from an average improvement of 3.90% Positive Predictive Value on unseen observations from the PhysioNet 2016 dataset using this method.

Abstract
The COVID-19 pandemic increased morbidity and mortality worldwide, particularly in the Kidney and Kidney-Pancreas Transplant Recipient (KTR/KPTR) population. Aiming at assessing the absolute and relative excess mortality (EM) in a Portuguese KTR/KPTR cohort, we conducted a retrospective observational study of two KTR/KPTRs cohorts: cohort 1 (P1; n = 2,179) between September/2012 and March/2020; cohort 2 (P2; n = 2067) between March/2020, and August/2022. A correlation between relative and absolute EM and age, sex, time from transplantation and cause of death was explored. A total of 145 and 84 deaths by all causes were observed in P1 and P2, respectively. The absolute EM in P2 versus P1 was 19.2 deaths (observed/expected mortality ratio 1.30, p = 0.006), and the relative EM was 1.47/1,000 person-months (95% CI 1.11-1.93, p = 0.006). Compared to the same period in the general population, the standardized mortality rate by age in P2 was 3.86 (95% CI 2.40-5.31), with a peak at 9.00 (95% CI 4.84-13.16) in P2C. The higher EM identified in this population was associated, mainly, with COVID-19 infection, with much higher values during the second seasonal COVID-19 peak when compared to the general population, despite generalized vaccination. These highlight the need for further preventive measures and improved therapies in these patients.

Abstract
Persistent left superior vena cava (PLSVC) is the most frequent thoracic venous anatomical variant in the general population. Isolated PLSVC, without formation of the right superior vena cava, is described in 10% of cases of PLSVC only. While it can be associated with congenital heart disease, arrhythmias, and premature death, adult patients with PLSVC are mostly asymptomatic, and the diagnosis is usually accidental. We present the case of a 72-year-old male with end-stage renal disease who was started on urgent hemodialysis through a temporary non-tunneled femoral central venous catheter (CVC) in the SLED (slow low-efficiency dialysis) modality and later remained dependent on hemodialysis. At this stage, placement of a tunneled CVC in the right internal jugular vein was necessary and fluoroscopy guidance was not available. There were no complications during the procedure, but postoperative conventional chest radiography revealed an inadequate positioning of the CVC tip in the left hemithorax, crossing the midline. Subsequently, the diagnosis of PLSVC was obtained by performing a thoracic angio-CT scan, confirming CVC tip positioning inside the PLSVC, and also excluded the presence of cardiac defects or additional anatomical variations of the great vessels of the thorax. Early evaluation for the creation of autologous vascular access was started under our care, and there were no mechanical or other complications associated with hemodialysis sessions during early follow-up after discharge.

Abstract
Gastric Intestinal Metaplasia (GIM) is a precancerous gastric lesion and its early detection facilitates patient followup, thus lowering significantly the risk of death by gastric cancer. However, effective screening of this condition is a very challenging task, resulting low intra and inter-observer concordance. Computer assisted diagnosis systems leveraging deep neural networks (DNNs) have emerged as a way to mitigate these ailments. Notwithstanding, these approaches typically require large datasets in order to learn invariance to the extreme variations typically present in Esophagogastroduodenoscopy (EGD) still frames, such as perspective, illumination, and scale. Hence, we propose to combine a priori information regarding texture characteristics of GIM with data-driven DNN solutions. In particular, we define two different models that treat pre-trained DNNs as general features extractors, whose pairwise interactions with a collection of highly invariant local texture descriptors grounded on fractal geometry are computed by means of an outer product in the embedding space. Our experiments show that these models outperform a baseline DNN by a significant margin over several metrics (e.g., area under the curve (AUC) 0.792 vs. 0.705) in a dataset comprised of EGD narrow-band images. Our best model measures double the positive likelihood ratio when compared to a baseline GIM detector.

Abstract