Search

Abstract Explainable artificial intelligence (XAI) has experienced a vast increase in recognition over the last few years. While the technical developments are manifold, less focus has been placed on the clinical applicability and usability of systems. Moreover, not much attention has been given to XAI systems that can handle multimodal and longitudinal data, which we postulate are important features in many clinical workflows. In this study, we review, from a clinical perspective, the current state of XAI for multimodal and longitudinal datasets and highlight the challenges thereof. Additionally, we propose the XAI orchestrator, an instance that aims to help clinicians with the synopsis of multimodal and longitudinal data, the resulting AI predictions, and the corresponding explainability output. We propose several desirable properties of the XAI orchestrator, such as being adaptive, hierarchical, interactive, and uncertainty-aware.

NSAIDs and paracetamol are commonly used as antipyretic treatments, which may impair renal and hepatic function, respectively. Both organ systems are also negatively affected by COVID-19. In two retrospective case–control studies, we investigated whether COVID-19 is a risk factor for the development of renal or hepatic function impairment after NSAID and paracetamol use, respectively. In the NSAID study, we defined cases as patients with a decrease of ≥15% in the estimated glomerular filtration rate (eGFR). We matched them using a 1:2 ratio with controls who did not show a decrease in the eGFR. For the paracetamol study, we matched patients with ALT or ALP ≥ 3x, the upper limits of normal, using a 1:3 ratio with controls whose liver enzymes did not increase. In both studies, we selected demographic data, comorbidities, drug doses, and laboratory values as predictors in addition to SARS-CoV-2 test status. We applied different machine learning models to predict renal and hepatic function impairment. From the cohort of 12,263 unique adult inpatients, we found 288 cases of renal function impairment, which were matched with 576 controls, and 213 cases of liver function impairment, which were matched with 639 controls. In both case–control studies, testing positive for SARS-CoV-2 was not an independent risk factor for the studied adverse drug effects.

Antibody testing in inflammatory bowel disease (IBD) can add to diagnostic accuracy of the main subtypes Crohn’s disease (CD) and ulcerative colitis (UC). Whether modern modeling techniques such as supervised and unsupervised machine learning are of value for finer distinction of subtypes such as IBD-unclassified (IBD-U) is not known. We determined the antibody profile of 100 adult IBD patients from the Swiss IBD cohort study with known subtype (50 CD, 50 UC) as well as of 76 IBD-U patients. We included ASCA IgG and IgA, p-ANCA, MPO- and PR3-ANCA, and xANCA measurements for computing different antibody panels as well as machine learning models. The AUC of an optimized antibody panel was 85% (95%CI, 78–92%) to distinguish CD from UC patients. The antibody profile of IBD-U patients was closely related to UC. No specific antibody profile was predictive for IBD-U nor for re-classification. The panel diagnostic was in favor of UC reclassification prediction with a correct assignment rate of 69.2–73.1% depending on the cut-off applied. Supervised machine learning could not distinguish between CD, UC, and IBD-U. More so, unsupervised machine learning suggested only two distinct clusters as a likely number of IBD subtypes. Antibodies in IBD are supportive in confirming clinical determined subtypes CD and UC but have limited capacity to predict IBD-U and reclassification during follow-up. In terms of antibody profiles, IBD-U is not a distinct subtype of IBD.

Objective: Glucagon-like peptide 1 receptor agonists (GLP-1 RAs) lower plasma glucose through effects on insulin and glucagon secretion and by decelerating gastric emptying. GLP-1 RAs have many beneficial effects beyond glycemic control, including a protective role on the cardiovascular system. However, underlying mechanisms linking GLP-1 RAs with coronary artery disease are complex and not fully elucidated. In this mini-review, we discuss these mechanisms and subsequent clinical events.Data Sources: We searched PubMed and Google Scholar for evidence on GLP-1 RAs and coronary events. We did not apply restrictions on article type. We reviewed publications for clinical relevance.Synopsis of Content: In the first part, we review the current evidence concerning the role of GLP-1 RAs on potential mechanisms underlying the development of coronary events. Specifically, we discuss the role of GLP-1 RAs on atherosclerosis and vasospasms of epicardial coronary arteries, as well as structural/functional changes of coronary microvasculature. In the second part, we summarize the clinical evidence on the impact of GLP-1 RAs in the prevention of acute and chronic coronary syndromes and coronary revascularization. We conclude by discussing existing gaps in the literature and proposing directions for future research.

Objective: Sodium glucose cotransporter 2 inhibitors (SGLT2-is) are antidiabetic drugs that improve glycemic control by limiting urinary glucose reuptake in the proximal tubule. SGLT2-is might suppress atherosclerotic processes and ameliorate the prognosis of patients with diabetes mellitus diagnosed with or at high risk of atherosclerotic cardiovascular disease (ASCVD). In this mini review, we examine the role of SGLT2-is in the development and progression of atherosclerosis throughout its spectrum, from subclinical atherosclerosis to ASCVD.Data Sources—PubMed and Google Scholar were searched for publications related to SGLT2-is and atherosclerosis. All types of articles were considered, including clinical trials, animal studies, in vitro observations, and reviews and meta-analyses. Data were examined according to their impact and clinical relevance.Synopsis of Content—We first review the underlying mechanisms of SGLT2-is on the development and progression of atherosclerosis, including favorable effects on lipid metabolism, reduction of systemic inflammation, and improvement of endothelial function. We then discuss the putative impact of SGLT2-is on the formation, composition, and stability of atherosclerotic plaque. Furthermore, we evaluate the effects of SGLT2-is in subclinical atherosclerosis assessed by carotid intima media thickness and pulse wave velocity. Subsequently, we summarize the effects of SGLT2-is in ASCVD events, including ischemic stroke, angina pectoris, myocardial infarction, revascularization, and peripheral artery disease, as well as major adverse cardiovascular events, cardiovascular mortality, heart failure, and chronic kidney disease. Moreover, we examine factors that could modify the role of SGLT2-is in atherosclerosis, including sex, age, diabetes, glycemic control, ASCVD, and SGLT2-i compounds. Additionally, we propose future directions that can improve our understanding of SGLT2-is and atherosclerosis.

ObjectivesTo identify qualitative VASARI (Visually AcceSIble Rembrandt Images) Magnetic Resonance (MR) Imaging features for differentiation of glioblastoma (GBM) and brain metastasis (BM) of different primary tumors.Materials and MethodsT1-weighted pre- and post-contrast, T2-weighted, and T2-weighted, fluid attenuated inversion recovery (FLAIR) MR images of a total of 239 lesions from 109 patients with either GBM or BM (breast cancer, non-small cell (NSCLC) adenocarcinoma, NSCLC squamous cell carcinoma, small-cell lung cancer (SCLC)) were included. A set of adapted, qualitative VASARI MR features describing tumor appearance and location was scored (binary; 1 = presence of feature, 0 = absence of feature). Exploratory data analysis was performed on binary scores using a combination of descriptive statistics (proportions with 95% binomial confidence intervals), unsupervised methods and supervised methods including multivariate feature ranking using either repeated fitting or recursive feature elimination with Support Vector Machines (SVMs).ResultsGBMs were found to involve all lobes of the cerebrum with a fronto-occipital gradient, often affected the corpus callosum (32.4%, 95% CI 19.1–49.2), and showed a strong preference for the right hemisphere (79.4%, 95% CI 63.2–89.7). BMs occurred most frequently in the frontal lobe (35.1%, 95% CI 28.9–41.9) and cerebellum (28.3%, 95% CI 22.6–34.8). The appearance of GBMs was characterized by preference for well-defined non-enhancing tumor margin (100%, 89.8–100), ependymal extension (52.9%, 36.7–68.5) and substantially less enhancing foci than BMs (44.1%, 28.9–60.6 vs. 75.1%, 68.8–80.5). Unsupervised and supervised analyses showed that GBMs are distinctively different from BMs and that this difference is driven by definition of non-enhancing tumor margin, ependymal extension and features describing laterality. Differentiation of histological subtypes of BMs was driven by the presence of well-defined enhancing and non-enhancing tumor margins and localization in the vision center. SVM models with optimal hyperparameters led to weighted F1-score of 0.865 for differentiation of GBMs from BMs and weighted F1-score of 0.326 for differentiation of BM subtypes.ConclusionVASARI MR imaging features related to definition of non-enhancing margin, ependymal extension, and tumor localization may serve as potential imaging biomarkers to differentiate GBMs from BMs.