Clinical Metadata Research Articles

A multimodal transformer to fuse images and metadata for skin disease classification.

Skin disease cases are rising in prevalence, and the diagnosis of skin diseases is always a challenging task in the clinic. Utilizing deep learning to diagnose skin diseases could help to meet these challenges. In this study, a novel neural network is proposed for the classification of skin diseases. Since the datasets for the research consist of skin disease images and clinical metadata, we propose a novel multimodal Transformer, which consists of two encoders for both images and metadata and one decoder to fuse the multimodal information. In the proposed network, a suitable Vision Transformer (ViT) model is utilized as the backbone to extract image deep features. As for metadata, they are regarded as labels and a new Soft Label Encoder (SLE) is designed to embed them. Furthermore, in the decoder part, a novel Mutual Attention (MA) block is proposed to better fuse image features and metadata features. To evaluate the model’s effectiveness, extensive experiments have been conducted on the private skin disease dataset and the benchmark dataset ISIC 2018. Compared with state-of-the-art methods, the proposed model shows better performance and represents an advancement in skin disease diagnosis.

The microbiome enhances transcriptional inflammatory signatures and delays clinical resolution in cutaneous leishmaniasis

Abstract Cutaneous leishmaniasis (CL) is a protozoal infection in which the immune response plays a major role in lesion development. The activation of local responses such as cytolysis by CD8+ T cells and release of IL-1β by myeloid cells is associated with poor clinical outcome in patients. We found that lesions from cutaneous leishmaniasis patients also exhibit a bacterial dysbiosis, but whether this dysbiosis affects immune responses and consequently impacts chemotherapy to heal the patients is unknown. We carried out an integrative multi-omics analysis of 64 patients, including RNA-seq and 16S-seq from lesion biopsies and clinical metadata. Staphylococcus was the dominant overrepresented genus in lesions from most patients, so we generated an in-house S. aureus pangenome to quantify transcript abundances through dual RNA-seq mapping analysis. Increased overall bacterial burden assessed by qPCR and the presence of S. aureus assessed by dual RNA-seq indicated enrichment for cytokines (e.g., IL1B), inflammatory responses (e.g., PTGS2), and chemotaxis (e.g., CXCL8) by Gene Ontology analysis. Of particular importance, we found that lesions with increased S. aureus transcript abundances exhibited high expression of inflammatory-related genes that have been associated with treatment failure, such as PRF1, GNLY, and GZMB. Correspondingly, we found that patients with a Staphylococcus spp. dysbiosis exhibited a delay in resolving their lesions following treatment. These results indicate that the microbiome influences the local immune responses in CL, affecting how patients respond to therapy by delaying the healing time. These results suggest that the use of antibiotics combined with chemotherapy may improve treatment outcome for CL. Supported by grant from NIH (R01 AI149456)

Validation of artificial intelligence prediction models for skin cancer diagnosis using dermoscopy images: the 2019 International Skin Imaging Collaboration Grand Challenge

BackgroundPrevious studies of artificial intelligence (AI) applied to dermatology have shown AI to have higher diagnostic classification accuracy than expert dermatologists; however, these studies did not adequately assess clinically realistic scenarios, such as how AI systems behave when presented with images of disease categories that are not included in the training dataset or images drawn from statistical distributions with significant shifts from training distributions. We aimed to simulate these real-world scenarios and evaluate the effects of image source institution, diagnoses outside of the training set, and other image artifacts on classification accuracy, with the goal of informing clinicians and regulatory agencies about safety and real-world accuracy.MethodsWe designed a large dermoscopic image classification challenge to quantify the performance of machine learning algorithms for the task of skin cancer classification from dermoscopic images, and how this performance is affected by shifts in statistical distributions of data, disease categories not represented in training datasets, and imaging or lesion artifacts. Factors that might be beneficial to performance, such as clinical metadata and external training data collected by challenge participants, were also evaluated. 25 331 training images collected from two datasets (in Vienna [HAM10000] and Barcelona [BCN20000]) between Jan 1, 2000, and Dec 31, 2018, across eight skin diseases, were provided to challenge participants to design appropriate algorithms. The trained algorithms were then tested for balanced accuracy against the HAM10000 and BCN20000 test datasets and data from countries not included in the training dataset (Turkey, New Zealand, Sweden, and Argentina). Test datasets contained images of all diagnostic categories available in training plus other diagnoses not included in training data (not trained category). We compared the performance of the algorithms against that of 18 dermatologists in a simulated setting that reflected intended clinical use.Findings64 teams submitted 129 state-of-the-art algorithm predictions on a test set of 8238 images. The best performing algorithm achieved 58·8% balanced accuracy on the BCN20000 data, which was designed to better reflect realistic clinical scenarios, compared with 82·0% balanced accuracy on HAM10000, which was used in a previously published benchmark. Shifted statistical distributions and disease categories not included in training data contributed to decreases in accuracy. Image artifacts, including hair, pen markings, ulceration, and imaging source institution, decreased accuracy in a complex manner that varied based on the underlying diagnosis. When comparing algorithms to expert dermatologists (2460 ratings on 1269 images), algorithms performed better than experts in most categories, except for actinic keratoses (similar accuracy on average) and images from categories not included in training data (26% correct for experts vs 6% correct for algorithms, p<0·0001). For the top 25 submitted algorithms, 47·1% of the images from categories not included in training data were misclassified as malignant diagnoses, which would lead to a substantial number of unnecessary biopsies if current state-of-the-art AI technologies were clinically deployed.InterpretationWe have identified specific deficiencies and safety issues in AI diagnostic systems for skin cancer that should be addressed in future diagnostic evaluation protocols to improve safety and reliability in clinical practice.FundingMelanoma Research Alliance and La Marató de TV3.

Genome Profiling of SARS-CoV-2 in Indonesia, ASEAN and the Neighbouring East Asian Countries: Features, Challenges and Achievements

Whole-genome sequencing (WGS) has played a significant role in understanding the epidemiology and biology of SARS-CoV-2 virus. Here, we investigate the use of SARS-CoV-2 WGS in Southeast and East Asian countries as a genomic surveillance during the COVID-19 pandemic. Nottingham–Indonesia Collaboration for Clinical Research and Training (NICCRAT) initiative has facilitated collaboration between the University of Nottingham and a team in the Research Center for Biotechnology, National Research and Innovation Agency (BRIN), to carry out a small number of SARS-CoV-2 WGS in Indonesia using Oxford Nanopore Technology (ONT). Analyses of SARS- CoV-2 genomes deposited on GISAID reveal the importance of clinical and demographic metadata collection and the importance of open access and data sharing. Lineage and phylogenetic analyses of two periods defined by the Delta variant outbreak reveal that: (1) B.1.466.2 variants were the most predominant in Indonesia before the Delta variant outbreak, having a unique spike gene mutation N439K at more than 98% frequency, (2) Delta variants AY.23 sub-lineage took over after June 2021, and (3) the highest rate of virus transmissions between Indonesia and other countries was through interactions with Singapore and Japan, two neighbouring countries with a high degree of access and travels to and from Indonesia.

LGR5 expressing skin fibroblasts define a major cellular hub perturbed in scleroderma

Systemic sclerosis (scleroderma, SSc) is an incurable autoimmune disease with high morbidity and mortality rates. Here, we conducted a population-scale single-cell genomic analysis of skin and blood samples of 56 healthy controls and 97 SSc patients at different stages of the disease. We found immune compartment dysfunction only in a specific subtype of diffuse SSc patients but global dysregulation of the stromal compartment, particularly in a previously undefined subset of LGR5+-scleroderma-associated fibroblasts (ScAFs). ScAFs are perturbed morphologically and molecularly in SSc patients. Single-cell multiome profiling of stromal cells revealed ScAF-specific markers, pathways, regulatory elements, and transcription factors underlining disease development. Systematic analysis of these molecular features with clinical metadata associates specific ScAF targets with disease pathogenesis and SSc clinical traits. Our high-resolution atlas of the sclerodermatous skin spectrum will enable a paradigm shift in the understanding of SSc disease and facilitate the development of biomarkers and therapeutic strategies.

Next-generation sequencing: insights to advance clinical investigations of the microbiome.

Next-generation sequencing (NGS) technology has advanced our understanding of the human microbiome by allowing for the discovery and characterization of unculturable microbes with prediction of their function. Key NGS methods include 16S rRNA gene sequencing, shotgun metagenomic sequencing, and RNA sequencing. The choice of which NGS methodology to pursue for a given purpose is often unclear for clinicians and researchers. In this Review, we describe the fundamentals of NGS, with a focus on 16S rRNA and shotgun metagenomic sequencing. We also discuss pros and cons of each methodology as well as important concepts in data variability, study design, and clinical metadata collection. We further present examples of how NGS studies of the human microbiome have advanced our understanding of human disease pathophysiology across diverse clinical contexts, including the development of diagnostics and therapeutics. Finally, we share insights as to how NGS might further be integrated into and advance microbiome research and clinical care in the coming years.

Transcriptomic profiling of blood from autoimmune hepatitis patients reveals potential mechanisms with implications for management.

Autoimmune hepatitis (AIH) is a poorly understood, chronic disease, for which corticosteroids are still the mainstay of therapy and most patients undergo liver biopsy to obtain a diagnosis. We aimed to determine if there was a transcriptomic signature of AIH in the peripheral blood and investigate underlying biologic pathways revealed by gene expression analysis. Whole blood RNA from 75 AIH patients and 25 healthy volunteers was extracted and sequenced. Differential gene expression analysis revealed 249 genes that were significantly differentially expressed in AIH patients compared to controls. Using a random forest algorithm, we determined that less than 10 genes were sufficient to differentiate the two groups in our cohort. Interferon signaling was more active in AIH samples compared to controls, regardless of treatment status. Pegivirus sequences were detected in five AIH samples and 1 healthy sample. The gene expression data and clinical metadata were used to determine 12 genes that were significantly associated with advanced fibrosis in AIH. AIH patients with a partial response to therapy demonstrated decreased evidence of a CD8+ T cell gene expression signal. These findings represent progress in understanding a disease in need of better tests, therapies, and biomarkers.

Combined PET-CT and MRI for response evaluation in patients with squamous cell anal carcinoma treated with curative-intent chemoradiotherapy

ObjectivesTo assess the effectiveness of fluorine-18 fluorodeoxyglucose (FDG) positron-emission tomography-computed tomography (PET-CT) and magnetic resonance imaging (MRI) for response assessment post curative-intent chemoradiotherapy (CRT) in anal squamous cell carcinoma (ASCC).MethodsConsecutive ASCC patients treated with curative-intent CRT at a single centre between January 2018 and April 2020 were retrospectively identified. Clinical meta-data including progression-free survival (PFS) and overall survival (OS) outcomes were collated. Three radiologists evaluated PET-CT and MRI using qualitative response assessment criteria and agreed in consensus. Two-proportion z test was used to compare diagnostic performance metrics (sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy). Kaplan-Meier analysis (Mantel-Cox log-rank) was performed.ResultsMRI (accuracy 76%, PPV 44.8%, NPV 95.7%) and PET-CT (accuracy 69.3%, PPV 36.7%, NPV 91.1%) performance metrics were similar; when combined, there were statistically significant improvements (accuracy 94.7%, PPV 78.9%, NPV 100%). Kaplan-Meier analysis demonstrated significant differences in PFS between responders and non-responders at PET-CT (p = 0.007), MRI (p = 0.005), and consensus evaluation (p < 0.001). Cox regression analysis of PFS demonstrated a lower hazard ratio (HR) and narrower 95% confidence intervals for consensus findings (HR = 0.093, p < 0.001). Seventy-five patients, of which 52 (69.3%) were females, with median follow-up of 17.8 months (range 5–32.6) were included. Fifteen of the 75 (20%) had persistent anorectal and/or nodal disease after CRT. Three patients died, median time to death 6.2 months (range 5–18.3).ConclusionCombined PET-CT and MRI response assessment post-CRT better predicts subsequent outcome than either modality alone. This could have valuable clinical benefits by guiding personalised risk-adapted patient follow-up.Key Points• MRI and PET-CT performance metrics for assessing response following chemoradiotherapy (CRT) in patients with anal squamous cell carcinoma (ASCC) were similar.• Combined MRI and PET-CT treatment response assessment 3 months after CRT in patients with ASCC was demonstrated to be superior to either modality alone.• A combined MRI and PET-CT assessment 3 months after CRT in patients with ASCC has the potential to improve accuracy and guide optimal patient management with a greater ability to predict outcome than either modality alone

A configurable method for clinical quality measurement through electronic health records based on openEHR and CQL

BackgroundOne of the primary obstacles to measure clinical quality is the lack of configurable solutions to make computers understand and compute clinical quality indicators. The paper presents a solution that can help clinical staff develop clinical quality measurement more easily and generate the corresponding data reports and visualization by a configurable method based on openEHR and Clinical Quality Language (CQL).MethodsFirst, expression logic adopted from CQL was combined with openEHR to express clinical quality indicators. Archetype binding provides the clinical information models used in expression logic, terminology binding makes the medical concepts consistent used in clinical quality artifacts and metadata is regarded as the essential component for sharing and management. Then, a systematic approach was put forward to facilitate the development of clinical quality indicators and the generation of corresponding data reports and visualization. Finally, clinical physicians were invited to test our system and give their opinions.ResultsWith the combination of openEHR and CQL, 64 indicators from Centers for Medicare & Medicaid Services (CMS) were expressed for verification and a complicated indicator was shown as an example. 68 indicators from 17 different scenes in the local environment were also expressed and computed in our system. A platform was built to support the development of indicators in a unified way. Also, an execution engine can parse and compute these indicators. Based on a clinical data repository (CDR), indicators were used to generate data reports and visualization and shown in a dashboard.ConclusionOur method is capable of expressing clinical quality indicators formally. With the computer-interpretable indicators, a systematic approach can make it more easily to define clinical indicators and generate medical data reports and visualization, and facilitate the adoption of clinical quality measurements.

Effective deep learning approaches for predicting COVID-19 outcomes from chest computed tomography volumes

The rapid evolution of the novel coronavirus disease (COVID-19) pandemic has resulted in an urgent need for effective clinical tools to reduce transmission and manage severe illness. Numerous teams are quickly developing artificial intelligence approaches to these problems, including using deep learning to predict COVID-19 diagnosis and prognosis from chest computed tomography (CT) imaging data. In this work, we assess the value of aggregated chest CT data for COVID-19 prognosis compared to clinical metadata alone. We develop a novel patient-level algorithm to aggregate the chest CT volume into a 2D representation that can be easily integrated with clinical metadata to distinguish COVID-19 pneumonia from chest CT volumes from healthy participants and participants with other viral pneumonia. Furthermore, we present a multitask model for joint segmentation of different classes of pulmonary lesions present in COVID-19 infected lungs that can outperform individual segmentation models for each task. We directly compare this multitask segmentation approach to combining feature-agnostic volumetric CT classification feature maps with clinical metadata for predicting mortality. We show that the combination of features derived from the chest CT volumes improve the AUC performance to 0.80 from the 0.52 obtained by using patients’ clinical data alone. These approaches enable the automated extraction of clinically relevant features from chest CT volumes for risk stratification of COVID-19 patients.

Association of Exposure to High-risk Antibiotics in Acute Care Hospitals With Multidrug-Resistant Organism Burden in Nursing Homes

Little is known about the contribution of hospital antibiotic prescribing to multidrug-resistant organism (MDRO) burden in nursing homes (NHs). To characterize antibiotic exposures across the NH patient's health care continuum (preceding health care exposure and NH stay) and to investigate whether recent antibiotic exposure is associated with MDRO colonization and room environment contamination at NH study enrollment. This is a secondary analysis of a prospective cohort study (conducted from 2013-2016) that enrolled NH patients and followed them up for as long as 6 months. The study was conducted in 6 NHs in Michigan among NH patients who were enrolled within 14 days of admission. Clinical metadata abstraction, multi-anatomical site screening, and room environment surveillance for MDROs were conducted at each study visit. Data were analyzed between May 2019 and November 2021. Antibiotic data were abstracted from NH electronic medical records by trained research staff and characterized by class, route, indication, location of therapy initiation, risk for Clostridioides difficile infection (C diffogenic agents), and 2019 World Health Organization Access, Watch, and Reserve (AWARE) antibiotic stewardship framework categories. The primary outcomes were MDRO colonization and MDRO room environment contamination at NH study enrollment, measured using standard microbiology methods. Multivariable logistic regression was used to identify whether antibiotic exposure within 60 days was associated with MDRO burden at NH study enrollment. Additionally, antibiotic exposure data were characterized using descriptive statistics. A total of 642 patients were included (mean [SD] age, 74.7 [12.2] years; 369 [57.5%] women; 402 [62.6%] White; median [IQR] NH days to enrollment, 6.0 [3.0-7.0]). Of these, 422 (65.7%) received 1191 antibiotic exposures: 368 (57.3%) received 971 hospital-associated prescriptions, and 119 (18.5%) received 198 NH-associated prescriptions. Overall, 283 patients (44.1%) received at least 1 C diffogenic agent, and 322 (50.2%) received at least 1 high-risk WHO AWARE antibiotic (watch or reserve agent). More than half of NH patients (364 [56.7%]) and room environments (437 [68.1%]) had MDRO-positive results at enrollment. In multivariable analysis, recent antibiotic exposure was positively associated with baseline MDRO colonization (odds ratio [OR], 1.70; 95% CI, 1.22-2.38) and MDRO environmental contamination (OR, 1.67; 95% CI, 1.17-2.39). Exploratory stratification by C diffogenic agent exposure increased the effect size (MDRO colonization: OR, 1.99; 95% CI, 1.33-2.96; MDRO environmental contamination: OR, 1.86; 95% CI, 1.24-2.79). Likewise, exploratory stratification by exposure to high-risk WHO AWARE antibiotics increased the effect size (MDRO colonization: OR, 2.32; 95% CI, 1.61-3.36; MDRO environmental contamination: OR, 1.86; 95% CI, 1.26-2.75). The findings of this study suggest that high-risk, hospital-based antibiotics are a potentially high-value target to reduce MDROs in postacute care NHs. This study underscores the potential utility of integrated hospital and NH stewardship programming on regional MDRO epidemiology.

Dynamic 3D radiomics analysis using artificial intelligence to assess the stage of COVID-19 on CT images.

ObjectiveTo develop a dynamic 3D radiomics analysis method using artificial intelligence technique for automatically assessing four disease stages (i.e., early, progressive, peak, and absorption stages) of COVID-19 patients on CT images.MethodsThe dynamic 3D radiomics analysis method was composed of three AI algorithms (the lung segmentation, lesion segmentation, and stage-assessing AI algorithms) that were trained and tested on 313,767 CT images from 520 COVID-19 patients. This proposed method used 3D lung lesion that was segmented by the lung and lesion segmentation algorithms to extract radiomics features, and then combined with clinical metadata to assess the possible stage of COVID-19 patients using stage-assessing algorithm. Area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, and specificity were used to evaluate diagnostic performance.ResultsOf 520 patients, 66 patients (mean age, 57 years ± 15 [standard deviation]; 35 women), including 203 CT scans, were tested. The dynamic 3D radiomics analysis method used 30 features, including 27 radiomics features and 3 clinical features to assess the possible disease stage of COVID-19 with an accuracy of 90%. For the prediction of each stage, the AUC of stage 1 was 0.965 (95% CI: 0.934, 0.997), AUC of stage 2 was 0.958 (95% CI: 0.931, 0.984), AUC of stage 3 was 0.998 (95% CI: 0.994, 1.000), and AUC of stage 4 was 0.975 (95% CI: 0.956, 0.994).ConclusionWith high diagnostic performance, the dynamic 3D radiomics analysis using artificial intelligence could represent a potential tool for helping hospitals make appropriate resource allocations and follow-up of treatment response.Key Points • The AI segmentation algorithms were able to accurately segment the lung and lesion of COVID-19 patients of different stages. • The dynamic 3D radiomics analysis method successfully extracted the radiomics features from the 3D lung lesion. • The stage-assessing AI algorithm combining with clinical metadata was able to assess the four stages with an accuracy of 90%, a macro-average AUC of 0.975. Supplementary InformationThe online version contains supplementary material available at 10.1007/s00330-021-08533-1.

P691 Influence of ethnicity on the phenotype and clinical outcomes in Inflammatory Bowel Disease; comparison of a South American with a North American population

Abstract Background Inflammatory Bowel Diseases (IBD) are chronic systemic inflammatory disorders, which affect the gastrointestinal tract. An increased incidence of IBD has been vastly reported in Europe, and North America, however, IBD has now emerged as a global disease affecting people of different genetic backgrounds and geographic locations, with increasing incidence in developing and newly industrialized regions such as South America, offering the opportunity to explore differences and similarities in disease presentation and outcomes across these variables. Methods Our study includes 265 patients with a diagnosis of Ulcerative Colitis (UC) (n=173) or Crohn′s Disease (CD) (n=92). We carried out an exploratory analysis of databases of Chilean and North American IBD patients to compare the clinical phenotype between both cohorts. We used an unsupervised machine learning approach using Principal Component Analysis (PCA), t- Stochastic Neighbor Embedding (t-SNE), among others, for each disease (CD and UC). Finally, we predicted the center (North American versus Chilean) using random forest, among other supervised machine learning methods to confirm these results. Results Several unsupervised machine learning methods separated two main groups supporting differences between North American and Chilean patients in each disease. The variables that explained the loadings of the clinical metadata on the principal components were related to therapies and disease extension/location at diagnose. Our Random Forest (RF) models were trained for cohort classification (Chilean or North American) based on clinical characteristics, obtaining a high accuracy of 0.86 (UC) and 0.89 (CD). Similarly, variables related to therapies and disease extension/location had a higher Gini index. In the same line, univariate analysis showed a later CD age at diagnoses in Chilean IBD patients (37 vs 24; p=0.005) Conclusion Our study suggests a clinical difference between North American and Chilean IBD patients; later CD age at diagnoses with a predominate less aggressive phenotype (39% vs. 54% B1), a more localized disease despite less biological therapies used in Chile for both diseases. The next research steps to confirm these results with more patients and explore more factors between different populations that could determine a less aggressive phenotype. Understanding the causes of these differences could unravel new pathways or protective factors, which could improve treatments.

Lessons Learnt From Using the Machine Learning Random Forest Algorithm to Predict Virulence in Streptococcus pyogenes.

Group A Streptococcus is a globally significant human pathogen. The extensive variability of the GAS genome, virulence phenotypes and clinical outcomes, render it an excellent candidate for the application of genotype-phenotype association studies in the era of whole-genome sequencing. We have catalogued the distribution and diversity of the transcription regulators of GAS, and employed phylogenetics, concordance metrics and machine learning (ML) to test for associations. In this review, we communicate the lessons learnt in the context of the recent bacteria genotype-phenotype association studies of others that have utilised both genome-wide association studies (GWAS) and ML. We envisage a promising future for the application GWAS in bacteria genotype-phenotype association studies and foresee the increasing use of ML. However, progress in this field is hindered by several outstanding bottlenecks. These include the shortcomings that are observed when GWAS techniques that have been fine-tuned on human genomes, are applied to bacterial genomes. Furthermore, there is a deficit of easy-to-use end-to-end workflows, and a lag in the collection of detailed phenotype and clinical genomic metadata. We propose a novel quality control protocol for the collection of high-quality GAS virulence phenotype coupled to clinical outcome data. Finally, we incorporate this protocol into a workflow for testing genotype-phenotype associations using ML and ‘linked’ patient-microbe genome sets that better represent the infection event.

Intestinal function and transit associate with gut microbiota dysbiosis in cystic fibrosis

BackgroundMost people with cystic fibrosis (pwCF) suffer from gastrointestinal symptoms and are at risk of gut complications. Gut microbiota dysbiosis is apparent within the CF population across all age groups, with evidence linking dysbiosis to intestinal inflammation and other markers of health. This pilot study aimed to investigate the potential relationships between the gut microbiota and gastrointestinal physiology, transit, and health. Study designFaecal samples from 10 pwCF and matched controls were subject to 16S rRNA sequencing. Results were combined with clinical metadata and MRI metrics of gut function to investigate relationships. ResultspwCF had significantly reduced microbiota diversity compared to controls. Microbiota compositions were significantly different, suggesting remodelling of core and rarer satellite taxa in CF. Dissimilarity between groups was driven by a variety of taxa, including Escherichia coli, Bacteroides spp., Clostridium spp., and Faecalibacterium prausnitzii. The core taxa were explained primarily by CF disease, whilst the satellite taxa were associated with pulmonary antibiotic usage, CF disease, and gut function metrics. Species-specific ordination biplots revealed relationships between taxa and the clinical or MRI-based variables observed. ConclusionsAlterations in gut function and transit resultant of CF disease are associated with the gut microbiota composition, notably the satellite taxa. Delayed transit in the small intestine might allow for the expansion of satellite taxa resulting in potential downstream consequences for core community function in the colon.

An Open-Source, Standard-Compliant, and Mobile Electronic Data Capture System for Medical Research (OpenEDC): Design and Evaluation Study.

BackgroundMedical research and machine learning for health care depend on high-quality data. Electronic data capture (EDC) systems have been widely adopted for metadata-driven digital data collection. However, many systems use proprietary and incompatible formats that inhibit clinical data exchange and metadata reuse. In addition, the configuration and financial requirements of typical EDC systems frequently prevent small-scale studies from benefiting from their inherent advantages.ObjectiveThe aim of this study is to develop and publish an open-source EDC system that addresses these issues. We aim to plan a system that is applicable to a wide range of research projects.MethodsWe conducted a literature-based requirements analysis to identify the academic and regulatory demands for digital data collection. After designing and implementing OpenEDC, we performed a usability evaluation to obtain feedback from users.ResultsWe identified 20 frequently stated requirements for EDC. According to the International Organization for Standardization/International Electrotechnical Commission (ISO/IEC) 25010 norm, we categorized the requirements into functional suitability, availability, compatibility, usability, and security. We developed OpenEDC based on the regulatory-compliant Clinical Data Interchange Standards Consortium Operational Data Model (CDISC ODM) standard. Mobile device support enables the collection of patient-reported outcomes. OpenEDC is publicly available and released under the MIT open-source license.ConclusionsAdopting an established standard without modifications supports metadata reuse and clinical data exchange, but it limits item layouts. OpenEDC is a stand-alone web app that can be used without a setup or configuration. This should foster compatibility between medical research and open science. OpenEDC is targeted at observational and translational research studies by clinicians.

VirusLab: A Tool for Customized SARS-CoV-2 Data Analysis.

Since the beginning of 2020, the COVID-19 pandemic has posed unprecedented challenges to viral data analysis and connected host disease diagnostic methods. We propose VirusLab, a flexible system for analysing SARS-CoV-2 viral sequences and relating them to metadata or clinical information about the host. VirusLab capitalizes on two existing resources: ViruSurf, a database of public SARS-CoV-2 sequences supporting metadata-driven search, and VirusViz, a tool for visual analysis of search results. VirusLab is designed for taking advantage of these resources within a server-side architecture that: (i) covers pipelines based on approaches already in use (ARTIC, Galaxy) but entirely cutomizable upon user request; (ii) predigests analysis of raw sequencing data from different platforms (Oxford Nanopore and Illumina); (iii) gives access to public archives datasets; (iv) supplies user-friendly reporting – making it a tool that can also be integrated into a business environment. VirusLab can be installed and hosted within the premises of any organization where information about SARS-CoV-2 sequences can be safely integrated with information about hosts (e.g., clinical metadata). A system such as VirusLab is not currently available in the landscape of similar providers: our results show that VirusLab is a powerful tool to generate tabular/graphical and machine readable reports that can be integrated in more complex pipelines. We foresee that the proposed system can support many research-oriented and therapeutic scenarios within hospitals or the tracing of viral sequences and their mutational processes within organizations for viral surveillance.

Longitudinal Analysis of Gut Bacteriome, Mycobiome, Virome and Metabolome in Allogeneic Stem Cell Transplantation Reveals Susceptibility for Acute Graft-Versus-Host Disease

The intestinal microbiome directly effects outcomes in patients undergoing allogeneic stem cell transplantation (ASCT) (Peled et al., N Engl J Med, 2020). Recent developments in microbiome research allow us to look beyond bacterial communities at the composition of the intestinal mycobiome and virome. However, only few studies have addressed these fungal (Zhang et al., Nat Commun, 2021; Legoff et al., Nat Med, 2017) and viral communities (Van Der Velden et al., Biol Blood Marrow Transplant, 2013) in the ASCT setting.Loss of bacterial diversity and domination of single taxa was linked to increased incidence of graft-versus-host-disease (aGvHD) and mortality. However, the functional consequences of intestinal dysbiosis remain poorly understood. Recently, microbial-derived metabolites such as short-chain fatty acids (SCFAs), indoles and secondary bile acids (BA) have been proposed to explain how microbiota exert effects on epithelial and immune responses and shown to protect from GVHD in animal models.To assemble a complete picture of the phylogenetic and metabolic changes which occur as patients undergo ASCT we designed a multiomics approach and identified patterns in bacteriome, mycobiome, virome and metabolome dynamics in relation to clinical factors and GvHD.353 stool samples from 80 patients undergoing ASCT were obtained at two different transplantation centers (Munich and Regensburg) at defined timepoints before and after transplantation. Changes in metabolite composition were evaluated via mass spectrometry using a targeted metabolomic profiling approach, accompanied by bacterial 16S rDNA-, fungal 18S-ITS- and deep virome sequencing. Results were linked among each other and with clinical metadata such as GvHD, response to steroids, antimicrobial therapy and mortality.Longitudinal profiling of ASCT patients revealed a significant loss of bacterial, fungal and viral diversity following ASCT. At both centers, the effect was most pronounced in the early post-transplant period. Patients that developed GI-GVHD had the lowest diversity scores (Panel 1). GVHD patient samples were dominated by Enteroccocus and Candida, and loss of Caudovirales species, indicative of deep dysbiosis effecting all intestinal communities.ASCT had a strong impact on microbiome function and their ability to produce metabolites. We assayed each sample for production of metabolites with reported protective effects: SCFAs (acetate, butyrate and propionate), indoles (indole-3-carboxylaldehyde, ICA) and BAs (deoxycholic and lithocholic acid). When admitted to the transplantation unit (timepoint “conditioning”) most patients had rich metabolite expression profiles. In the early post-transplant period, metabolite levels became progressively more depleted. GVHD patients displayed a nearly complete loss of metabolites (Panel 2a). Importantly, levels of the metabolite ICA at early timepoints could predict the occurrence of GVHD, leading us to consider prophylactic administration in future studies.We performed beta diversity analysis to identify clinical factors that had the highest impact on microbial communities. As expected, we noted that administration of broad-spectrum antibiotics had a profound impact of bacteria, fungi and viruses; but also had a significant negative effect on metabolites levels.Lastly, we report on a case of successful treatment of severe GI-GVHD with fecal microbiota transplant. Response to treatment was characterized by the recipient adopting the FMT donor's rich metabolite expression profile (Panel 2b) .To our knowledge, this is the first multicenter study that combines bacteriome, mycobiome, virome and metabolome analyses a longitudinal fashion in matched patient cohorts. Our results are consistent with previous findings that there are major alterations in gut microbiome and metabolite composition after ASCT. We expand upon those findings by characterizing the time course of phylogenetic alterations and their functional consequences. Those alterations play an important role in GvHD pathogenesis and might serve as potential biomarkers to identify patients at high risk for developing lethal GvHD early on. Our results highlight two strategies for targeted microbiome-based interventions: (1) preventing loss of microbial communities by restrictive use of antibiotics and (2) prophylactic administration of metabolite cocktails. [Display omitted] DisclosuresNo relevant conflicts of interest to declare.

Fibrosis-Net: A Tailored Deep Convolutional Neural Network Design for Prediction of Pulmonary Fibrosis Progression From Chest CT Images.

Pulmonary fibrosis is a devastating chronic lung disease that causes irreparable lung tissue scarring and damage, resulting in progressive loss in lung capacity and has no known cure. A critical step in the treatment and management of pulmonary fibrosis is the assessment of lung function decline, with computed tomography (CT) imaging being a particularly effective method for determining the extent of lung damage caused by pulmonary fibrosis. Motivated by this, we introduce Fibrosis-Net, a deep convolutional neural network design tailored for the prediction of pulmonary fibrosis progression from chest CT images. More specifically, machine-driven design exploration was leveraged to determine a strong architectural design for CT lung analysis, upon which we build a customized network design tailored for predicting forced vital capacity (FVC) based on a patient’s CT scan, initial spirometry measurement, and clinical metadata. Finally, we leverage an explainability-driven performance validation strategy to study the decision-making behavior of Fibrosis-Net as to verify that predictions are based on relevant visual indicators in CT images. Experiments using a patient cohort from the OSIC Pulmonary Fibrosis Progression Challenge showed that the proposed Fibrosis-Net is able to achieve a significantly higher modified Laplace Log Likelihood score than the winning solutions on the challenge. Furthermore, explainability-driven performance validation demonstrated that the proposed Fibrosis-Net exhibits correct decision-making behavior by leveraging clinically-relevant visual indicators in CT images when making predictions on pulmonary fibrosis progress. Fibrosis-Net is able to achieve a significantly higher modified Laplace Log Likelihood score than the winning solutions on the OSIC Pulmonary Fibrosis Progression Challenge, and has been shown to exhibit correct decision-making behavior when making predictions. Fibrosis-Net is available to the general public in an open-source and open access manner as part of the OpenMedAI initiative. While Fibrosis-Net is not yet a production-ready clinical assessment solution, we hope that its release will encourage researchers, clinicians, and citizen data scientists alike to leverage and build upon it.

Statistical Enrichment Analysis of Samples: A General-Purpose Tool to Annotate Metadata Neighborhoods of Biological Samples.

Unsupervised learning techniques, such as clustering and embedding, have been increasingly popular to cluster biomedical samples from high-dimensional biomedical data. Extracting clinical data or sample meta-data shared in common among biomedical samples of a given biological condition remains a major challenge. Here, we describe a powerful analytical method called Statistical Enrichment Analysis of Samples (SEAS) for interpreting clustered or embedded sample data from omics studies. The method derives its power by focusing on sample sets, i.e., groups of biological samples that were constructed for various purposes, e.g., manual curation of samples sharing specific characteristics or automated clusters generated by embedding sample omic profiles from multi-dimensional omics space. The samples in the sample set share common clinical measurements, which we refer to as “clinotypes,” such as age group, gender, treatment status, or survival days. We demonstrate how SEAS yields insights into biological data sets using glioblastoma (GBM) samples. Notably, when analyzing the combined The Cancer Genome Atlas (TCGA)—patient-derived xenograft (PDX) data, SEAS allows approximating the different clinical outcomes of radiotherapy-treated PDX samples, which has not been solved by other tools. The result shows that SEAS may support the clinical decision. The SEAS tool is publicly available as a freely available software package at https://aimed-lab.shinyapps.io/SEAS/.