COVID-19 and protection of vaccination in patients with systemic sclerosis-associated interstitial lung disease.

Objectives: Data on COVID-19 in patients with interstitial lung disease are scarce and whether SARS-CoV-2 may trigger interstitial lung disease progression remains unknown. We aimed to analyze outcomes of COVID-19 in patients with systemic sclerosis-associated interstitial lung disease, including possible thoracic radiographic progression. Patients and Methods: All 43 patients with systemic sclerosis-associated interstitial lung disease followed in our center (mean ± SD, 55.2 ± 11.6 years, 36 female) with confirmed SARS-CoV2 infection up to 1 September 2022 were analyzed. Individual interstitial lung disease extent on high resolution CT (HRCT) performed before (up to 3 months) and after COVID-19 (2-5 months) was compared. Results: At SARS-CoV-2 infection, 9/43 patients were unvaccinated, whereas 5, 26, and 3 had received 2, 3, or 4 doses of an mRNA vaccine, respectively. Thirty-one patients were either on monotherapy with immunosuppressives (mycophenolate, n = 7; cyclophosphamide, n = 2; methotrexate, n = 10; tocilizumab, n = 7; rituximab, n = 1; etanercept, n = 1), or their combinations (n = 3). Eight patients (20%), of whom four unvaccinated, required hospitalization for pneumonia and three (7%) died of acute respiratory failure (n = 2, both unvaccinated) or cardiac arrest. Lack of vaccination was the only independent predictor for hospitalization (OR = 7.98, 95% CI: 1.25-51.09) and marginally for death (OR = 32.7, 95% CI: 0.97-1110.98), regardless of the presence of diffuse systemic sclerosis, interstitial lung disease extent greater than 20% or immunosuppressive treatment. In 22 patients with available HRCT pairs (vaccinated = 20), the interstitial lung disease extent before COVID-19 (20.4%± 17.8%) remained unchanged (22.4% ± 18.5%) in all but one patient. Conclusion: SARS-CoV-2 vaccination is of outmost importance for every systemic sclerosis patient with interstitial lung disease. COVID-19 does not seem to promote progression of systemic sclerosis-associated interstitial lung disease in vaccinated patients, but further studies are warranted.

Long-term effectiveness and persistence of ustekinumab and TNF inhibitors in patients with psoriatic arthritis: final 3-year results from the PsABio real-world study.

OBJECTIVES: To evaluate real-world persistence and effectiveness of the IL-12/23 inhibitor, ustekinumab or a tumour necrosis factor inhibitor (TNFi) for psoriatic arthritis over 3 years. METHODS: PsABio (NCT02627768), a prospective, observational study, followed patients with PsA prescribed first-line to third-line ustekinumab or a TNFi. Persistence and effectiveness (achievement of clinical Disease Activity for PSA (cDAPSA) low disease activity (LDA)/remission and minimal disease activity/very LDA (MDA/VLDA)) were assessed every 6 months. Safety data were collected over 3 years. Analyses to compare the modes of action were adjusted on baseline differences by propensity scores (PS). RESULTS: In 895 patients (mean age 49.8 years, 44.7% males), at 3 years, the proportion of patients still on their initial treatments was similar with ustekinumab (49.9%) and TNFi (47.8%). No difference was seen in the risk of stopping/switching; PS-adjusted hazard ratio (95% CI) for stopping/switching ustekinumab versus TNFi was 0.87 (0.68 to 1.11). In the overall population, cDAPSA LDA/remission was achieved in 58.6%/31.4% ustekinumab-treated and 69.8%/45.0% TNFi-treated patients; PS-adjusted ORs (95% CI) were 0.89 (0.63 to 1.26) for cDAPSA LDA; 0.72 (0.50 to 1.05) for remission. MDA/VLDA was achieved in 41.4%/19.2% of ustekinumab-treated and 54.2%/26.9% of TNFi-treated patients with overlapping PS-adjusted ORs. A greater percentage of TNFi-treated patients achieved effectiveness outcomes. Both treatments exhibited good long-term safety profiles, although ustekinumab-treated patients had a lower rate of adverse events (AEs) versus TNFi. CONCLUSION: At 3 years, there was generally comparable persistence after ustekinumab or TNFi treatment, but AE rates were lower with ustekinumab.

Original research on Behçet's syndrome: a bibliometric analysis over 20 years (2000-2019).

OBJECTIVES: To perform a bibliometric analysis of original research articles on Behçet's syndrome (BS) published over the last 20 years prior to the COVID-19 pandemic, and to systematically describe their characteristics and citation records. METHODS: PubMed database was searched for any article published on BS between 2000 and 2019. We identified all original research articles and categorised them by country of origin and type of research, i.e., clinical, translational and basic. Each article's impact was assessed using the individual citation numbers from Google Scholar search engine; we also calculated the median annual citation rates (ACRs), both per country and research type. RESULTS: Of a total of 2,381 retrieved original articles from 51 countries, the majority reported on clinical (52.6%), followed by translational (46.0%) and basic research (1.4%). Turkey had the highest number of publications (39% of articles) followed by four countries (Korea, China, Japan, Italy) where BS is also relatively prevalent. However, regarding median ACRs, France was first, followed by United Kingdom, Germany and Collaboration. Although the number of articles has almost doubled between 2010-2019 versus 2000-2009, median ACRs across either clinical or translational research had a downwards trend. CONCLUSIONS: Researchers from countries where BS is prevalent are more productive, albeit their work is of lower impact compared to countries with generally higher research budgets. A considerable increase of original research articles on BS is observed over time but further funding may be warranted for a parallel increase in the respective scientific impact.

Different Covid-19 Outcomes Among Systemic Rheumatic Diseases: A Nation-wide Cohort Study.

OBJECTIVES: To investigate Covid-19-associated risk of hospitalization and death in rheumatoid arthritis (RA), ankylosing spondylitis (AS), psoriatic arthritis (PsA), systemic lupus erythematosus (SLE) and systemic sclerosis (SSc) in comparison with the general population during pandemic's first year and compare their overall mortality with 2019. METHODS: Interlinking nation-wide electronic registries, we recorded confirmed Covid-19-associated infections, hospitalizations and deaths, and all-cause deaths between 1-March-2020 and 28-February-2021 in all adults with RA, AS, PsA, SLE, and SSc under treatment (n = 74 970, median age 67.5, 51.2, 58.1, 56.2, 62.2 years, respectively) and in matched (1:5) on age, sex, and region of domicile random comparators from the general population. Deaths from all causes during 2019 were also recorded. RESULTS: Compared with the general population incidence rates (IR) for Covid-19-associated hospitalization were higher in RA [IR ratio (IRR):1.71(1.50-1.95)], SLE [2.0(1.4-2.7)] and SSc [2.28(1.29-3.90)], while Covid-19-associated death rates were higher in RA [1.91(1.46-2.49)]. When focusing only on SARS-CoV-2 infected subjects, after adjusting for age and gender, the odds ratio for Covid-19 associated death was higher in RA [1.47(1.11- 1.94)] and SSc [2.92(1.07-7.99)] compared with the general population. All-cause mortality rate compared with the general population increased in RA during the first pandemic year (IRR : 0.71) with reference to 2019 (0.59) and decreased in SSc (IRR : 1.94 vs 4.36). CONCLUSION: Covid-19 may have more severe impact in patients with systemic rheumatic disease than the general population. Covid-19-related mortality is increased in subgroups of patients with specific rheumatic diseases, underscoring the need for priority vaccination and access to targeted treatments.

Mapping the Early Dispersal Patterns of SARS-CoV-2 Omicron BA.4 and BA.5 Subvariants in the Absence of Travel Restrictions and Testing at the Borders in Europe.

The circulation of SARS-CoV-2 omicron BA.4 and BA.5 subvariants with enhanced transmissibility and capacity for immune evasion resulted in a recent pandemic wave that began in April-May of 2022. We performed a statistical phylogeographic study that aimed to define the cross-border transmission patterns of BA.4 and BA.5 at the earliest stages of virus dispersal. Our sample included all BA.4 and BA.5 sequences that were publicly available in the GISAID database through mid-May 2022. Viral dispersal patterns were inferred using maximum likelihood phylogenetic trees with bootstrap support. We identified South Africa as the major source of both BA.4 and BA.5 that migrated to other continents. By contrast, we detected no significant export of these subvariants from Europe. Belgium was identified as a major hub for BA.4 transmission within Europe, while Portugal and Israel were identified as major sources of BA.5. Western and Northern European countries exhibited the highest rates of cross-border transmission, as did several popular tourist destinations in Southern and Central/Western Europe. Our study provides a detailed map of the early dispersal patterns of two highly transmissible SARS-CoV-2 omicron subvariants at a time when there was an overall relaxation of public health measures in Europe.

A Simple Prognostic Score for Critical COVID-19 Derived from Patients without Comorbidities Performs Well in Unselected Patients.

We aimed to search for laboratory predictors of critical COVID-19 in consecutive adults admitted in an academic center between 16 September 2020-20 December 2021. Patients were uniformly treated with low-molecular-weight heparin, and dexamethasone plus remdesivir when SpO2 < 94%. Among consecutive unvaccinated patients without underlying medical conditions (n = 241, 49 year-old median, 71% males), 22 (9.1%) developed critical disease and 2 died (0.8%). White-blood-cell counts, neutrophils, neutrophil-to-lymphocyte ratio, CRP, fibrinogen, ferritin, LDH and γ-GT at admission were each univariably associated with critical disease. ROC-defined cutoffs revealed that CRP > 61.8 mg/L, fibrinogen > 616.5 mg/dL and LDH > 380.5 U/L were each associated with critical disease development, independently of age, sex and days from symptom-onset. A score combining higher-than-cutoff CRP (0/2), LDH (0/1) and fibrinogen (0/1) predicted critical disease (AUC: 0.873, 95% CI: 0.820-0.926). This score performed well in an unselected patient cohort (n = 1228, 100% unvaccinated) predominantly infected by the alpha variant (AUC: 0.718, 95% CI: 0.683-0.753), as well as in a mixed cohort (n = 527, 65% unvaccinated) predominantly infected by the delta variant (AUC: 0.708, 95% CI: 0.656-0.760). Therefore, we propose that a combination of standard biomarkers of acute inflammatory response, cell death and hypercoagulability reflects the severity of COVID-19 per se independently of comorbidities, age and sex, being of value for risk stratification in unselected patients.

COVLIAS 2.0-cXAI: Cloud-Based Explainable Deep Learning System for COVID-19 Lesion Localization in Computed Tomography Scans.

BACKGROUND: The previous COVID-19 lung diagnosis system lacks both scientific validation and the role of explainable artificial intelligence (AI) for understanding lesion localization. This study presents a cloud-based explainable AI, the "COVLIAS 2.0-cXAI" system using four kinds of class activation maps (CAM) models. METHODOLOGY: Our cohort consisted of ~6000 CT slices from two sources (Croatia, 80 COVID-19 patients and Italy, 15 control patients). COVLIAS 2.0-cXAI design consisted of three stages: (i) automated lung segmentation using hybrid deep learning ResNet-UNet model by automatic adjustment of Hounsfield units, hyperparameter optimization, and parallel and distributed training, (ii) classification using three kinds of DenseNet (DN) models (DN-121, DN-169, DN-201), and (iii) validation using four kinds of CAM visualization techniques: gradient-weighted class activation mapping (Grad-CAM), Grad-CAM++, score-weighted CAM (Score-CAM), and FasterScore-CAM. The COVLIAS 2.0-cXAI was validated by three trained senior radiologists for its stability and reliability. The Friedman test was also performed on the scores of the three radiologists. RESULTS: The ResNet-UNet segmentation model resulted in dice similarity of 0.96, Jaccard index of 0.93, a correlation coefficient of 0.99, with a figure-of-merit of 95.99%, while the classifier accuracies for the three DN nets (DN-121, DN-169, and DN-201) were 98%, 98%, and 99% with a loss of ~0.003, ~0.0025, and ~0.002 using 50 epochs, respectively. The mean AUC for all three DN models was 0.99 (p < 0.0001). The COVLIAS 2.0-cXAI showed 80% scans for mean alignment index (MAI) between heatmaps and gold standard, a score of four out of five, establishing the system for clinical settings. CONCLUSIONS: The COVLIAS 2.0-cXAI successfully showed a cloud-based explainable AI system for lesion localization in lung CT scans.

Global Distribution, Dispersal Patterns, and Trend of Several Omicron Subvariants of SARS-CoV-2 across the Globe.

Our study aims to describe the global distribution and dispersal patterns of the SARS-CoV-2 Omicron subvariants. Genomic surveillance data were extracted from the CoV-Spectrum platform, searching for BA.1*, BA.2*, BA.3*, BA.4*, and BA.5* variants by geographic region. BA.1* increased in November 2021 in South Africa, with a similar increase across all continents in early December 2021. BA.1* did not reach 100% dominance in all continents. The spread of BA.2*, first described in South Africa, differed greatly by geographic region, in contrast to BA.1*, which followed a similar global expansion, firstly occurring in Asia and subsequently in Africa, Europe, Oceania, and North and South America. BA.4* and BA.5* followed a different pattern, where BA.4* reached high proportions (maximum 60%) only in Africa. BA.5* is currently, by Mid-August 2022, the dominant strain, reaching almost 100% across Europe, which is the first continent aside from Africa to show increasing proportions, and Asia, the Americas, and Oceania are following. The emergence of new variants depends mostly on their selective advantage, translated as enhanced transmissibility and ability to invade people with existing immunity. Describing these patterns is useful for a better understanding of the epidemiology of the VOCs' transmission and for generating hypotheses about the future of emerging variants.

Protracted stress-induced hypocortisolemia may account for the clinical and immune manifestations of Long COVID.

About one out of eight people to convalesce from COVID-19 suffer from the so called Long COVID, a syndrome of non-specific symptoms with unclear pathogenesis. In a recent study published in Cell Long COVID participants reporting respiratory symptoms had low cortisol levels. In an as yet unpublished analysis from Yale University low plasma cortisol levels discriminated Long COVID from asymptomatic convalescent or healthy non-infected controls. Although various immune perturbations were present in Long COVID, low levels of cortisol were prominent and strikingly, depression and anxiety were increased. It has become clear that Long COVID features may be similar to those described in myalgic encephalomyelitis/chronic fatigue syndrome, post-SARS sickness syndrome, and various chronic stress syndromes which have been linked to hypocortisolemia. Notably, lack of response of the hypothalamic-pituitary-adrenal axis to hypocortisolemia shows a suppressed axis in Long COVID. We suggest that the inability of hypothalamic-pituitary-adrenal axis to recover after the acute illness, perhaps due to protracted stress in predisposed individuals, may represent the pathogenetic basis of the Long COVID-associated clinical and immunological manifestations.

Protection of vaccination versus hybrid immunity against infection with COVID-19 Omicron variants among Health-Care Workers.

BACKGROUND AIM: The Omicron COVID-19 variants BA.1* and BA.2* evade immune system leading to increased transmissibility and breakthrough infections. We aim to test the hypothesis that immunity achieved post COVID-19 infection combined with vaccination (hybrid immunity), is more effective against Omicron infection than vaccination alone in a health-care setting. METHODS: Data on regular pre-emptive PCR testing from all Health-Care Workers (HCWs) at Laiko University Hospital from 29th December 2020, date on which the national COVID-19 immunization program began in Greece, until 24th May 2022, were retrospectively collected and recorded. The infection rate was calculated after December 21st, 2021, when Omicron was the predominant circulating variant in Greece, as the total number of infections (positive PCR COVID-19 test regardless of symptoms) divided by the total person-months at risk. RESULTS: Of 1,305 vaccinated HCWs who were included in the analysis [median age of 47 (IQR: 36, 56) years, 66.7 % women], 13 % and 87 % had received 2 or 3 vaccine doses (full and booster vaccination), respectively. A COVID-19 infection had occurred in 135 of 1,305 of participants prior to Omicron predominance. Of those 135 HCWs with hybrid immunity only 13 (9.6 %) were re-infected. Of the 154 and 1,016 HCWs with full and booster vaccination-induced immunity, respectively, 71 (46.1 %, infection rate 13.4/100 person-months) and 448 (44.1 %, infection rate 12.2/100 person-months) were infected during the follow up period. No association between gender or age and COVID-19 infection was found and none of the participants had a severe infection or died. CONCLUSIONS: Hybrid immunity confers higher protection by almost 5-fold compared to full or booster vaccination for COVID-19 infection with the Omicron variant among HCWs who are at high risk of exposure. This may inform public health policies on how to achieve optimal immunity in terms of the timing and mode of vaccination.

Multicenter Study on COVID-19 Lung Computed Tomography Segmentation with varying Glass Ground Opacities using Unseen Deep Learning Artificial Intelligence Paradigms: COVLIAS 1.0 Validation.

Variations in COVID-19 lesions such as glass ground opacities (GGO), consolidations, and crazy paving can compromise the ability of solo-deep learning (SDL) or hybrid-deep learning (HDL) artificial intelligence (AI) models in predicting automated COVID-19 lung segmentation in Computed Tomography (CT) from unseen data leading to poor clinical manifestations. As the first study of its kind, "COVLIAS 1.0-Unseen" proves two hypotheses, (i) contrast adjustment is vital for AI, and (ii) HDL is superior to SDL. In a multicenter study, 10,000 CT slices were collected from 72 Italian (ITA) patients with low-GGO, and 80 Croatian (CRO) patients with high-GGO. Hounsfield Units (HU) were automatically adjusted to train the AI models and predict from test data, leading to four combinations-two Unseen sets: (i) train-CRO:test-ITA, (ii) train-ITA:test-CRO, and two Seen sets: (iii) train-CRO:test-CRO, (iv) train-ITA:test-ITA. COVILAS used three SDL models: PSPNet, SegNet, UNet and six HDL models: VGG-PSPNet, VGG-SegNet, VGG-UNet, ResNet-PSPNet, ResNet-SegNet, and ResNet-UNet. Two trained, blinded senior radiologists conducted ground truth annotations. Five types of performance metrics were used to validate COVLIAS 1.0-Unseen which was further benchmarked against MedSeg, an open-source web-based system. After HU adjustment for DS and JI, HDL (Unseen AI) > SDL (Unseen AI) by 4% and 5%, respectively. For CC, HDL (Unseen AI) > SDL (Unseen AI) by 6%. The COVLIAS-MedSeg difference was < 5%, meeting regulatory guidelines.Unseen AI was successfully demonstrated using automated HU adjustment. HDL was found to be superior to SDL.

COVLIAS 1.0Lesion vs. MedSeg: An Artificial Intelligence Framework for Automated Lesion Segmentation in COVID-19 Lung Computed Tomography Scans.

BACKGROUND: COVID-19 is a disease with multiple variants, and is quickly spreading throughout the world. It is crucial to identify patients who are suspected of having COVID-19 early, because the vaccine is not readily available in certain parts of the world. METHODOLOGY: Lung computed tomography (CT) imaging can be used to diagnose COVID-19 as an alternative to the RT-PCR test in some cases. The occurrence of ground-glass opacities in the lung region is a characteristic of COVID-19 in chest CT scans, and these are daunting to locate and segment manually. The proposed study consists of a combination of solo deep learning (DL) and hybrid DL (HDL) models to tackle the lesion location and segmentation more quickly. One DL and four HDL models-namely, PSPNet, VGG-SegNet, ResNet-SegNet, VGG-UNet, and ResNet-UNet-were trained by an expert radiologist. The training scheme adopted a fivefold cross-validation strategy on a cohort of 3000 images selected from a set of 40 COVID-19-positive individuals. RESULTS: The proposed variability study uses tracings from two trained radiologists as part of the validation. Five artificial intelligence (AI) models were benchmarked against MedSeg. The best AI model, ResNet-UNet, was superior to MedSeg by 9% and 15% for Dice and Jaccard, respectively, when compared against MD 1, and by 4% and 8%, respectively, when compared against MD 2. Statistical tests-namely, the Mann-Whitney test, paired t-test, and Wilcoxon test-demonstrated its stability and reliability, with p < 0.0001. The online system for each slice was <1 s. CONCLUSIONS: The AI models reliably located and segmented COVID-19 lesions in CT scans. The COVLIAS 1.0Lesion lesion locator passed the intervariability test.

The second decade of anti-TNF-a therapy in clinical practice: new lessons and future directions in the COVID-19 era.

Since the late 1990s, tumor necrosis factor alpha (TNF-α) inhibitors (anti-TNFs) have revolutionized the therapy of immune-mediated inflammatory diseases (IMIDs) affecting the gut, joints, skin and eyes. Although the therapeutic armamentarium in IMIDs is being constantly expanded, anti-TNFs remain the cornerstone of their treatment. During the second decade of their application in clinical practice, a large body of additional knowledge has accumulated regarding various aspects of anti-TNF-α therapy, whereas new indications have been added. Recent experimental studies have shown that anti-TNFs exert their beneficial effects not only by restoring aberrant TNF-mediated immune mechanisms, but also by de-activating pathogenic fibroblast-like mesenchymal cells. Real-world data on millions of patients further confirmed the remarkable efficacy of anti-TNFs. It is now clear that anti-TNFs alter the physical course of inflammatory arthritis and inflammatory bowel disease, leading to inhibition of local and systemic bone loss and to a decline in the number of surgeries for disease-related complications, while anti-TNFs improve morbidity and mortality, acting beneficially also on cardiovascular comorbidities. On the other hand, no new safety signals emerged, whereas anti-TNF-α safety in pregnancy and amid the COVID-19 pandemic was confirmed. The use of biosimilars was associated with cost reductions making anti-TNFs more widely available. Moreover, the current implementation of the "treat-to-target" approach and treatment de-escalation strategies of IMIDs were based on anti-TNFs. An intensive search to discover biomarkers to optimize response to anti-TNF-α treatment is currently ongoing. Finally, selective targeting of TNF-α receptors, new forms of anti-TNFs and combinations with other agents, are being tested in clinical trials and will probably expand the spectrum of TNF-α inhibition as a therapeutic strategy for IMIDs.

Distinct type I interferon responses between younger women and older men contribute to the variability of COVID-19 outcomes: Hypothesis generating insights from COVID-19 convalescent individuals.

BACKGROUND/OBJECTIVE: Older age and male sex have been consistently found to be associated with dismal outcomes among COVID-19 infected patients. In contrast, premenopausal females present the lowest mortality among adults infected by SARS-CoV-2. The goal of the present study was to investigate whether peripheral blood type I interferon (IFN) signature and interleukin (IL)-6 serum levels -previously shown to contribute to COVID-19-related outcomes in hospitalized patients- is shaped by demographic contributors among COVID-19 convalescent individuals. PATIENTS AND METHODS: Type I IFN-inducible genes in peripheral blood, as well as serum IL-6 levels were quantified in 61 COVID-19 convalescent healthy individuals (34 females, 27 males; age range 18-70 years, mean 35.7 ± 15.9 years) who recovered from COVID-19 without requiring hospitalization within a median of 3 months prior to inclusion in the present study. Among those, 17 were older than 50 years (11 males, 6 females) and 44 equal to or less than 50 years (16 males, 28 females). Expression analysis of type I IFN-inducible genes (MX-1, IFIT-1, IFI44) was performed by real time PCR and a type I IFN score, reflecting type I IFN peripheral activity, was calculated. IL-6 and C-reactive protein levels were determined by a commercially available ELISA. RESULTS: COVID-19 convalescent individuals older than 50 years exhibited significantly decreased peripheral blood type I IFN scores along with significantly increased IL-6 serum levels compared to their younger counterparts less than 50 years old (5.4 ± 4.3 vs 16.8 ± 24.7, p = 0.02 and 10.6 ± 16.9 vs 2.9 ± 8.0 ng/L, p = 0.03, respectively). Following sex stratification, peripheral blood type I IFN score was found to be significantly higher in younger females compared to both younger and older males (22.9 ± 29.2 vs 6.3 ± 4.6 vs 4.5 ± 3.7, p = 0.01 and p = 0.002, respectively). Regarding IL-6, an opposite pattern was observed, with the highest levels being detected among older males and the lowest levels among younger females (11.6 ± 18.9 vs 2.5 ± 7.8 ng/L, p = 0.03). CONCLUSION: Constitutive higher type I IFN responses and dampened IL-6 production observed in younger women of premenopausal age, along with lower type I IFN responses and increased IL-6 levels in older males, could account for the discrete clinical outcomes seen in the two population groups, as consistently revealed in COVID-19 epidemiological studies.

Deep Learning Paradigm for Cardiovascular Disease/Stroke Risk Stratification in Parkinson's Disease Affected by COVID-19: A Narrative Review.

Background and Motivation: Parkinson's disease (PD) is one of the most serious, non-curable, and expensive to treat. Recently, machine learning (ML) has shown to be able to predict cardiovascular/stroke risk in PD patients. The presence of COVID-19 causes the ML systems to become severely non-linear and poses challenges in cardiovascular/stroke risk stratification. Further, due to comorbidity, sample size constraints, and poor scientific and clinical validation techniques, there have been no well-explained ML paradigms. Deep neural networks are powerful learning machines that generalize non-linear conditions. This study presents a novel investigation of deep learning (DL) solutions for CVD/stroke risk prediction in PD patients affected by the COVID-19 framework. Method: The PRISMA search strategy was used for the selection of 292 studies closely associated with the effect of PD on CVD risk in the COVID-19 framework. We study the hypothesis that PD in the presence of COVID-19 can cause more harm to the heart and brain than in non-COVID-19 conditions. COVID-19 lung damage severity can be used as a covariate during DL training model designs. We, therefore, propose a DL model for the estimation of, (i) COVID-19 lesions in computed tomography (CT) scans and (ii) combining the covariates of PD, COVID-19 lesions, office and laboratory arterial atherosclerotic image-based biomarkers, and medicine usage for the PD patients for the design of DL point-based models for CVD/stroke risk stratification. Results: We validated the feasibility of CVD/stroke risk stratification in PD patients in the presence of a COVID-19 environment and this was also verified. DL architectures like long short-term memory (LSTM), and recurrent neural network (RNN) were studied for CVD/stroke risk stratification showing powerful designs. Lastly, we examined the artificial intelligence bias and provided recommendations for early detection of CVD/stroke in PD patients in the presence of COVID-19. Conclusion: The DL is a very powerful tool for predicting CVD/stroke risk in PD patients affected by COVID-19.

COVLIAS 2.0-cXAI: Cloud-Based Explainable Deep Learning System for COVID-19 Lesion Localization in Computed Tomography Scans

Background: The previous COVID-19 lung diagnosis system lacks both scientific validation and the role of explainable artificial intelligence (AI) for understanding lesion localization. This study presents a cloud-based explainable AI, the 'COVLIAS 2.0-cXAI';system using four kinds of class activation maps (CAM) models. Methodology: Our cohort consisted of ~6000 CT slices from two sources (Croatia, 80 COVID-19 patients and Italy, 15 control patients). COVLIAS 2.0-cXAI design consisted of three stages: (i) automated lung segmentation using hybrid deep learning ResNet-UNet model by automatic adjustment of Hounsfield units, hyperparameter optimization, and parallel and distributed training, (ii) classification using three kinds of DenseNet (DN) models (DN-121, DN-169, DN-201), and (iii) validation using four kinds of CAM visualization techniques: gradient-weighted class activation mapping (Grad-CAM), Grad-CAM++, score-weighted CAM (Score-CAM), and FasterScore-CAM. The COVLIAS 2.0-cXAI was validated by three trained senior radiologists for its stability and reliability. The Friedman test was also performed on the scores of the three radiologists. Results: The ResNet-UNet segmentation model resulted in dice similarity of 0.96, Jaccard index of 0.93, a correlation coefficient of 0.99, with a figure-of-merit of 95.99%, while the classifier accuracies for the three DN nets (DN-121, DN-169, and DN-201) were 98%, 98%, and 99% with a loss of ~0.003, ~0.0025, and ~0.002 using 50 epochs, respectively. The mean AUC for all three DN models was 0.99 (p < 0.0001). The COVLIAS 2.0-cXAI showed 80% scans for mean alignment index (MAI) between heatmaps and gold standard, a score of four out of five, establishing the system for clinical settings. Conclusions: The COVLIAS 2.0-cXAI successfully showed a cloud-based explainable AI system for lesion localization in lung CT scans.

Different systemic rheumatic diseases as risk factors for COVID-19-related mortality.

COVID-19 has been associated with increased morbidity and mortality, globally. Whether COVID-19-related mortality is increased in patients with systemic rheumatic diseases (SRDs) is still debatable. Although results are somewhat conflicting, there are a handful of nationwide studies published indicating that, in individuals with SRD, there is signal for increased adverse COVID-19-related outcomes and higher mortality. It appears that there are differences in COVID-19-related mortality across various SRDs. Besides, certain disease-specific (disease activity, disease duration, medication received) and/or other features (e.g. comorbidities) seem to also affect COVID-19-related mortality in SRD patients. Herein, we wanted to highlight that a more individualized approach taking into consideration the effect of the aforementioned factors into the risk calculation for COVID-19 adverse outcomes, including mortality, in SRD patients is warranted. A multinational study based on nationwide data, examining all common SRDs and stratifying accordingly, would be of interest, toward this direction. Key Points • It is still debatable whether Covid-19-related mortality is increased in patients with sytemic rheumatic diseases (SRD). • Disease-specific risk factors (e.g. type of SRD, disease activity) should be taken into account in risk assessment for Covid-19-releted outcomes in SRD patients.