Segmentation-based cardiomegaly detection based on semi-supervised estimation of cardiothoracic ratio.

The successful integration of neural networks in a clinical setting is still uncommon despite major successes achieved by artificial intelligence in other domains. This is mainly due to the black box characteristic of most optimized models and the undetermined generalization ability of the trained architectures. The current work tackles both issues in the radiology domain by focusing on developing an effective and interpretable cardiomegaly detection architecture based on segmentation models. The architecture consists of two distinct neural networks performing the segmentation of both cardiac and thoracic areas of a radiograph. The respective segmentation outputs are subsequently used to estimate the cardiothoracic ratio, and the corresponding radiograph is classified as a case of cardiomegaly based on a given threshold. Due to the scarcity of pixel-level labeled chest radiographs, both segmentation models are optimized in a semi-supervised manner. This results in a significant reduction in the costs of manual annotation. The resulting segmentation outputs significantly improve the interpretability of the architecture's final classification results. The generalization ability of the architecture is assessed in a cross-domain setting. The assessment shows the effectiveness of the semi-supervised optimization of the segmentation models and the robustness of the ensuing classification architecture.

INHBA is Enriched in HPV-negative Oropharyngeal Squamous Cell Carcinoma and Promotes Cancer Progression.

Patients with oropharyngeal squamous cell carcinoma (OPSCC) caused by human papilloma virus (HPV) exhibit a better prognosis than those with HPV-negative OPSCC. This study investigated the distinct molecular pathways that delineate HPV-negative from HPV-positive OPSCC to identify biologically relevant therapeutic targets. Bulk mRNA from 23 HPV-negative and 39 HPV-positive OPSCC tumors (n = 62) was sequenced to uncover the transcriptomic profiles. Differential expression followed by gene set enrichment analysis was performed to outline the top enriched biological process in the HPV-negative compared with HPV-positive entity. INHBA, the highest overexpressed gene in the HPV-negative tumor, was knocked down. Functional assays (migration, proliferation, cell death, stemness) were conducted to confirm the target's oncogenic role. Correlation analyses to reveal its impact on the tumor microenvironment were performed. We revealed that epithelial-to-mesenchymal transition (EMT) is the most enriched process in HPV-negative compared with HPV-positive OPSCC, with INHBA (inhibin beta A subunit) being the top upregulated gene. INHBA knockdown downregulated the expression of EMT transcription factors and attenuated migration, proliferation, stemness, and cell death resistance of OPSCC cells. We uncovered that INHBA associates with a pro-tumor microenvironment by negatively correlating with antitumor CD8+ T and B cells while positively correlating with pro-tumor M1 macrophages. We identified three miRNAs that are putatively involved in repressing INHBA expression. Our results indicate that the upregulation of INHBA is tumor-promoting. We propose INHBA as an attractive therapeutic target for the treatment of INHBA-enriched tumors in patients with HPV-negative OPSCC to ameliorate prognosis. SIGNIFICANCE: Patients with HPV-negative OPSCC have a poorer prognosis due to distinct molecular pathways. This study reveals significant transcriptomic differences between HPV-negative and HPV-positive OPSCC, identifying INHBA as a key upregulated gene in HPV-negative OPSCC's oncogenic pathways. INHBA is crucial in promoting EMT, cell proliferation, and an immunosuppressive tumor environment, suggesting its potential as a therapeutic target for HPV-negative OPSCC.

GatekeepR: an R Shiny application for the identification of nodes with high dynamic impact in Boolean networks.

MOTIVATION: Boolean networks can serve as straightforward models for understanding processes such as gene regulation, and employing logical rules. These rules can either be derived from existing literature or by data-driven approaches. However, in the context of large networks, the exhaustive search for intervention targets becomes challenging due to the exponential expansion of a Boolean network's state space and the multitude of potential target candidates, along with their various combinations. Instead, we can employ the logical rules and resultant interaction graph as a means to identify targets of specific interest within larger-scale models. This approach not only facilitates the screening process but also serves as a preliminary filtering step, enabling the focused investigation of candidates that hold promise for more profound dynamic analysis. However, applying this method requires a working knowledge of R, thus restricting the range of potential users. We, therefore, aim to provide an application that makes this method accessible to a broader scientific community. RESULTS: Here, we introduce GatekeepR, a graphical, web-based R Shiny application that enables scientists to screen Boolean network models for possible intervention targets whose perturbation is likely to have a large impact on the system's dynamics. This application does not require a local installation or knowledge of R and provides the suggested targets along with additional network information and visualizations in an intuitive, easy-to-use manner. The Supplementary Material describes the underlying method for identifying these nodes along with an example application in a network modeling pancreatic cancer. AVAILABILITY AND IMPLEMENTATION: https://www.github.com/sysbio-bioinf/GatekeepR https://abel.informatik.uni-ulm.de/shiny/GatekeepR/.

The Necessity of Interoperability to Uncover the Full Potential of Digital Health Devices.

Personalized health care can be optimized by including patient-reported outcomes. Standardized and disease-specific questionnaires have been developed and are routinely used. These patient-reported outcome questionnaires can be simple paper forms given to the patient to fill out with a pen or embedded in digital devices. Regardless of the format used, they provide a snapshot of the patient's feelings and indicate when therapies need to be adjusted. The advantage of digitizing these questionnaires is that they can be automatically analyzed, and patients can be monitored independently of doctor visits. Although the questions of most clinical patient-reported outcome questionnaires follow defined standards and are evaluated by clinical trials, these standards do not exist for data processing. Interoperable data formats and structures would benefit multilingual and cross-study data exchange. Linking questionnaires to standardized terminologies such as the Systematized Nomenclature of Medicine Clinical Terms (SNOMED CT) and Logical Observation Identifiers, Names, and Codes (LOINC) would improve this interoperability. However, linking clinically validated patient-reported outcome questionnaires to clinical terms available in SNOMED CT or LOINC is not as straightforward as it sounds. Here, we report our approach to link patient-reported outcomes from health applications to SNOMED CT or LOINC codes. We highlight current difficulties in this process and outline ways to minimize them.

A protocol for the use of cloud-based quantum computers for logical network analysis of biological systems.

Boolean networks are commonly used in systems biology to dynamically model gene regulatory interactions. Here, we present a protocol for implementing Boolean network dynamics as quantum circuits. We describe steps for accessing cloud-based quantum processing units offered by IBM and IonQ and downloading and parsing logic for gene regulatory networks. We then detail procedures for performing simulations of quantum circuits on local devices and visualizing measurement results. For complete details on the use and execution of this protocol, please refer to Weidner et al.1.

AMBAR - Interactive Alteration annotations for molecular tumor boards.

MOTIVATION: Personalized decision-making for cancer therapy relies on molecular profiling from sequencing data in combination with database evidence and expert knowledge. Molecular tumor boards (MTBs) bring together clinicians and scientists with diverse expertise and are increasingly established in the clinical routine for therapeutic interventions. However, the analysis and documentation of patients data are still time-consuming and difficult to manage for MTBs, especially as few tools are available for the amount of information required. RESULTS: To overcome these limitations, we developed an interactive web application AMBAR (Alteration annotations for Molecular tumor BoARds), for therapeutic decision-making support in MTBs. AMBAR is an R shiny-based application that allows customization, interactive filtering, visualization, adding expert knowledge, and export to clinical systems of annotated mutations. AVAILABILITY: AMBAR is dockerized, open source and available at https://sysbio.uni-ulm.de/?Software:Ambar Contact:hans.kestler@uni-ulm.de.

Liquid biopsy: an examination of platelet RNA obtained from head and neck squamous cell carcinoma patients for predictive molecular tumor markers.

Aim: Recently, a tumor cell-platelet interaction was identified in different tumor entities, resulting in a transfer of tumor-derived RNA into platelets, named further "tumor-educated platelets (TEP)". The present pilot study aims to investigate whether such a tumor-platelet transfer of RNA occurs also in patients suffering from head and neck squamous cell carcinoma (HNSCC). Methods: Sequencing analysis of RNA derived from platelets of tumor patients (TPs) and healthy donors (HDs) were performed. Subsequently, quantitative reverse transcription-polymerase chain reaction (qRT-PCR) was used for verification of differentially expressed genes in platelets from TPs and HDs in a second cohort of patients and HDs. Data were analyzed by applying bioinformatic tools. Results: Sequencing of RNA derived from the tumor as well as from platelets of TPs and HDs revealed 426 significantly differentially existing RNA, at which 406 RNA were more and 20 RNA less abundant in platelets from TPs in comparison to that of HDs. In TPs' platelets, abundantly existing RNA coding for 49 genes were detected, characteristically expressed in epithelial cells and RNA, the products of which are involved in tumor progression. Applying bioinformatic tools and verification on a second TP/HD cohort, collagen type I alpha 1 chain (COL1A1) and zinc finger protein 750 (ZNF750) were identified as the strongest potentially platelet-RNA-sequencing (RNA-seq)-based biomarkers for HNSCC. Conclusions: These results indicate a transfer of tumor-derived messenger RNA (mRNA) into platelets of HNSCC patients. Therefore, analyses of a patient's platelet RNA could be an efficient option for liquid biopsy in order to diagnose HNSCC or to monitor tumorigenesis as well as therapeutic responses at any time and in real time.

Ribosomal Dysfunction Is a Common Pathomechanism in Different Forms of Trichothiodystrophy.

Mutations in a broad variety of genes can provoke the severe childhood disorder trichothiodystrophy (TTD) that is classified as a DNA repair disease or a transcription syndrome of RNA polymerase II. In an attempt to identify the common underlying pathomechanism of TTD we performed a knockout/knockdown of the two unrelated TTD factors TTDN1 and RNF113A and investigated the consequences on ribosomal biogenesis and performance. Interestingly, interference with these TTD factors created a nearly uniform impact on RNA polymerase I transcription with downregulation of UBF, disturbed rRNA processing and reduction of the backbone of the small ribosomal subunit rRNA 18S. This was accompanied by a reduced quality of decoding in protein translation and the accumulation of misfolded and carbonylated proteins, indicating a loss of protein homeostasis (proteostasis). As the loss of proteostasis by the ribosome has been identified in the other forms of TTD, here we postulate that ribosomal dysfunction is a common underlying pathomechanism of TTD.

Supporting SURgery with GEriatric Co-Management and AI (SURGE-Ahead): A study protocol for the development of a digital geriatrician.

INTRODUCTION: Geriatric co-management is known to improve treatment of older adults in various clinical settings, however, widespread application of the concept is limited due to restricted resources. Digitalization may offer options to overcome these shortages by providing structured, relevant information and decision support tools for medical professionals. We present the SURGE-Ahead project (Supporting SURgery with GEriatric co-management and Artificial Intelligence) addressing this challenge. METHODS: A digital application with a dashboard-style user interface will be developed, displaying 1) evidence-based recommendations for geriatric co-management and 2) artificial intelligence-enhanced suggestions for continuity of care (COC) decisions. The development and implementation of the SURGE-Ahead application (SAA) will follow the Medical research council framework for complex medical interventions. In the development phase a minimum geriatric data set (MGDS) will be defined that combines parametrized information from the hospital information system with a concise assessment battery and sensor data. Two literature reviews will be conducted to create an evidence base for co-management and COC suggestions that will be used to display guideline-compliant recommendations. Principles of machine learning will be used for further data processing and COC proposals for the postoperative course. In an observational and AI-development study, data will be collected in three surgical departments of a University Hospital (trauma surgery, general and visceral surgery, urology) for AI-training, feasibility testing of the MGDS and identification of co-management needs. Usability will be tested in a workshop with potential users. During a subsequent project phase, the SAA will be tested and evaluated in clinical routine, allowing its further improvement through an iterative process. DISCUSSION: The outline offers insights into a novel and comprehensive project that combines geriatric co-management with digital support tools to improve inpatient surgical care and continuity of care of older adults. TRIAL REGISTRATION: German clinical trials registry (Deutsches Register für klinische Studien, DRKS00030684), registered on 21st November 2022.

A systems biology approach to define mechanisms, phenotypes, and drivers in PanNETs with a personalized perspective.

Pancreatic neuroendocrine tumors (PanNETs) are a rare tumor entity with largely unpredictable progression and increasing incidence in developed countries. Molecular pathways involved in PanNETs development are still not elucidated, and specific biomarkers are missing. Moreover, the heterogeneity of PanNETs makes their treatment challenging and most approved targeted therapeutic options for PanNETs lack objective responses. Here, we applied a systems biology approach integrating dynamic modeling strategies, foreign classifier tailored approaches, and patient expression profiles to predict PanNETs progression as well as resistance mechanisms to clinically approved treatments such as the mammalian target of rapamycin complex 1 (mTORC1) inhibitors. We set up a model able to represent frequently reported PanNETs drivers in patient cohorts, such as Menin-1 (MEN1), Death domain associated protein (DAXX), Tuberous Sclerosis (TSC), as well as wild-type tumors. Model-based simulations suggested drivers of cancer progression as both first and second hits after MEN1 loss. In addition, we could predict the benefit of mTORC1 inhibitors on differentially mutated cohorts and hypothesize resistance mechanisms. Our approach sheds light on a more personalized prediction and treatment of PanNET mutant phenotypes.

Self-Assessment of Having COVID-19 With the Corona Check mHealth App.

At the beginning of the COVID-19 pandemic, with a lack of knowledge about the novel virus and a lack of widely available tests, getting first feedback about being infected was not easy. To support all citizens in this respect, we developed the mobile health app Corona Check. Based on a self-reported questionnaire about symptoms and contact history, users get first feedback about a possible corona infection and advice on what to do. We developed Corona Check based on our existing software framework and released the app on Google Play and the Apple App Store on April 4, 2020. Until October 30, 2021, we collected 51,323 assessments from 35,118 users with explicit agreement of the users that their anonymized data may be used for research purposes. For 70.6% of the assessments, the users additionally shared their coarse geolocation with us. To the best of our knowledge, we are the first to report about such a large-scale study in this context of COVID-19 mHealth systems. Although users from some countries reported more symptoms on average than users from other countries, we did not find any statistically significant differences between symptom distributions (regarding country, age, and sex). Overall, the Corona Check app provided easily accessible information on corona symptoms and showed the potential to help overburdened corona telephone hotlines, especially during the beginning of the pandemic. Corona Check thus was able to support fighting the spread of the novel coronavirus. mHealth apps further prove to be valuable tools for longitudinal health data collection.

Leveraging quantum computing for dynamic analyses of logical networks in systems biology.

The dynamics of cellular mechanisms can be investigated through the analysis of networks. One of the simplest but most popular modeling strategies involves logic-based models. However, these models still face exponential growth in simulation complexity compared with a linear increase in nodes. We transfer this modeling approach to quantum computing and use the upcoming technique in the field to simulate the resulting networks. Leveraging logic modeling in quantum computing has many benefits, including complexity reduction and quantum algorithms for systems biology tasks. To showcase the applicability of our approach to systems biology tasks, we implemented a model of mammalian cortical development. Here, we applied a quantum algorithm to estimate the tendency of the model to reach particular stable conditions and further revert dynamics. Results from two actual quantum processing units and a noisy simulator are presented, and current technical challenges are discussed.

Unsupervised domain adaptation for the detection of cardiomegaly in cross-domain chest X-ray images.

In recent years, several deep learning approaches have been successfully applied in the field of medical image analysis. More specifically, different deep neural network architectures have been proposed and assessed for the detection of various pathologies based on chest X-ray images. While the performed assessments have shown very promising results, most of them consist in training and evaluating the performance of the proposed approaches on a single data set. However, the generalization of such models is quite limited in a cross-domain setting, since a significant performance degradation can be observed when these models are evaluated on data sets stemming from different medical centers or recorded under different protocols. The performance degradation is mostly caused by the domain shift between the training set and the evaluation set. To alleviate this problem, different unsupervised domain adaptation approaches are proposed and evaluated in the current work, for the detection of cardiomegaly based on chest X-ray images, in a cross-domain setting. The proposed approaches generate domain invariant feature representations by adapting the parameters of a model optimized on a large set of labeled samples, to a set of unlabeled images stemming from a different data set. The performed evaluation points to the effectiveness of the proposed approaches, since the adapted models outperform optimized models which are directly applied to the evaluation sets without any form of domain adaptation.

The HLA ligandome of oropharyngeal squamous cell carcinomas reveals shared tumour-exclusive peptides for semi-personalised vaccination.

BACKGROUND: The immune peptidome of OPSCC has not previously been studied. Cancer-antigen specific vaccination may improve clinical outcome and efficacy of immune checkpoint inhibitors such as PD1/PD-L1 antibodies. METHODS: Mapping of the OPSCC HLA ligandome was performed by mass spectrometry (MS) based analysis of naturally presented HLA ligands isolated from tumour tissue samples (n = 40) using immunoaffinity purification. The cohort included 22 HPV-positive (primarily HPV-16) and 18 HPV-negative samples. A benign reference dataset comprised of the HLA ligandomes of benign haematological and tissue datasets was used to identify tumour-associated antigens. RESULTS: MS analysis led to the identification of naturally HLA-presented peptides in OPSCC tumour tissue. In total, 22,769 peptides from 9485 source proteins were detected on HLA class I. For HLA class II, 15,203 peptides from 4634 source proteins were discovered. By comparative profiling against the benign HLA ligandomic datasets, 29 OPSCC-associated HLA class I ligands covering 11 different HLA allotypes and nine HLA class II ligands were selected to create a peptide warehouse. CONCLUSION: Tumour-associated peptides are HLA-presented on the cell surfaces of OPSCCs. The established warehouse of OPSCC-associated peptides can be used for downstream immunogenicity testing and peptide-based immunotherapy in (semi)personalised strategies.

Federated electronic data capture (fEDC): Architecture and prototype.

In clinical research as well as patient care, structured documentation of findings is an important task. In many cases, this is achieved by means of electronic case report forms (eCRF) using corresponding information technology systems. To avoid double data entry, eCRF systems can be integrated with electronic health records (EHR). However, when researchers from different institutions collaborate in collecting data, they often use a single joint eCRF system on the Internet. In this case, integration with EHR systems is not possible in most cases due to information security and data protection restrictions. To overcome this shortcoming, we propose a novel architecture for a federated electronic data capture system (fEDC). Four key requirements were identified for fEDC: Definitions of forms have to be available in a reliable and controlled fashion, integration with electronic health record systems must be possible, patient data should be under full local control until they are explicitly transferred for joint analysis, and the system must support data sharing principles accepted by the scientific community for both data model and data captured. With our approach, sites participating in a joint study can run their own instance of an fEDC system that complies with local standards (such as being behind a network firewall) while also being able to benefit from using identical form definitions by sharing metadata in the Operational Data Model (ODM) format published by the Clinical Data Interchange Standards Consortium (CDISC) throughout the collaboration. The fEDC architecture was validated with a working open-source prototype at five German university hospitals. The fEDC architecture provides a novel approach with the potential to significantly improve collaborative data capture: Efforts for data entry are reduced and at the same time, data quality is increased since barriers for integrating with local electronic health record systems are lowered. Further, metadata are shared and patient privacy is ensured at a high level.

PREDICT-juvenile-stroke: PRospective evaluation of a prediction score determining individual clinical outcome three months after ischemic stroke in young adults - a study protocol.

BACKGROUND: Although of high individual and socioeconomic relevance, a reliable prediction model for the prognosis of juvenile stroke (18-55 years) is missing. Therefore, the study presented in this protocol aims to prospectively validate the discriminatory power of a prediction score for the 3 months functional outcome after juvenile stroke or transient ischemic attack (TIA) that has been derived from an independent retrospective study using standard clinical workup data. METHODS: PREDICT-Juvenile-Stroke is a multi-centre (n = 4) prospective observational cohort study collecting standard clinical workup data and data on treatment success at 3 months after acute ischemic stroke or TIA that aims to validate a new prediction score for juvenile stroke. The prediction score has been developed upon single center retrospective analysis of 340 juvenile stroke patients. The score determines the patient's individual probability for treatment success defined by a modified Rankin Scale (mRS) 0-2 or return to pre-stroke baseline mRS 3 months after stroke or TIA. This probability will be compared to the observed clinical outcome at 3 months using the area under the receiver operating characteristic curve. The primary endpoint is to validate the clinical potential of the new prediction score for a favourable outcome 3 months after juvenile stroke or TIA. Secondary outcomes are to determine to what extent predictive factors in juvenile stroke or TIA patients differ from those in older patients and to determine the predictive accuracy of the juvenile stroke prediction score on other clinical and paraclinical endpoints. A minimum of 430 juvenile patients (< 55 years) with acute ischemic stroke or TIA, and the same number of older patients will be enrolled for the prospective validation study. DISCUSSION: The juvenile stroke prediction score has the potential to enable personalisation of counselling, provision of appropriate information regarding the prognosis and identification of patients who benefit from specific treatments. TRIAL REGISTRATION: The study has been registered at https://drks.de on March 31, 2022 ( DRKS00024407 ).

Efficient cross-validation traversals in feature subset selection.

Sparse and robust classification models have the potential for revealing common predictive patterns that not only allow for categorizing objects into classes but also for generating mechanistic hypotheses. Identifying a small and informative subset of features is their main ingredient. However, the exponential search space of feature subsets and the heuristic nature of selection algorithms limit the coverage of these analyses, even for low-dimensional datasets. We present methods for reducing the computational complexity of feature selection criteria allowing for higher efficiency and coverage of screenings. We achieve this by reducing the preparation costs of high-dimensional subsets [Formula: see text] to those of one-dimensional ones [Formula: see text]. Our methods are based on a tight interaction between a parallelizable cross-validation traversal strategy and distance-based classification algorithms and can be used with any product distance or kernel. We evaluate the traversal strategy exemplarily in exhaustive feature subset selection experiments (perfect coverage). Its runtime, fitness landscape, and predictive performance are analyzed on publicly available datasets. Even in low-dimensional settings, we achieve approximately a 15-fold increase in exhaustively generating distance matrices for feature combinations bringing a new level of evaluations into reach.

Enhancing mHealth data collection applications with sensing capabilities.

Smart mobile devices such as smartphones or tablets have become an important factor for collecting data in complex health scenarios (e.g., psychological studies, medical trials), and are more and more replacing traditional pen-and-paper instruments. However, simply digitizing such instruments does not yet realize the full potential of mobile devices: most modern smartphones have a variety of different sensor technologies (e.g., microphone, GPS data, camera, ...) that can also provide valuable data and potentially valuable insights for the medical purpose or the researcher. In this context, a significant development effort is required to integrate sensing capabilities into (existing) data collection applications. Developers may have to deal with platform-specific peculiarities (e.g., Android vs. iOS) or proprietary sensor data formats, resulting in unnecessary development effort to support researchers with such digital solutions. Therefore, a cross-platform mobile data collection framework has been developed to extend existing data collection applications with sensor capabilities and address the aforementioned challenges in the process. This framework will enable researchers to collect additional information from participants and environment, increasing the amount of data collected and drawing new insights from existing data.

CANTATA-prediction of missing links in Boolean networks using genetic programming.

MOTIVATION: Biological processes are complex systems with distinct behaviour. Despite the growing amount of available data, knowledge is sparse and often insufficient to investigate the complex regulatory behaviour of these systems. Moreover, different cellular phenotypes are possible under varying conditions. Mathematical models attempt to unravel these mechanisms by investigating the dynamics of regulatory networks. Therefore, a major challenge is to combine regulations and phenotypical information as well as the underlying mechanisms. To predict regulatory links in these models, we established an approach called CANTATA to support the integration of information into regulatory networks and retrieve potential underlying regulations. This is achieved by optimizing both static and dynamic properties of these networks. RESULTS: Initial results show that the algorithm predicts missing interactions by recapitulating the known phenotypes while preserving the original topology and optimizing the robustness of the model. The resulting models allow for hypothesizing about the biological impact of certain regulatory dependencies. AVAILABILITY AND IMPLEMENTATION: Source code of the application, example files and results are available at https://github.com/sysbio-bioinf/Cantata. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Weighted average ensemble-based semantic segmentation in biological electron microscopy images.

Semantic segmentation of electron microscopy images using deep learning methods is a valuable tool for the detailed analysis of organelles and cell structures. However, these methods require a large amount of labeled ground truth data that is often unavailable. To address this limitation, we present a weighted average ensemble model that can automatically segment biological structures in electron microscopy images when trained with only a small dataset. Thus, we exploit the fact that a combination of diverse base-learners is able to outperform one single segmentation model. Our experiments with seven different biological electron microscopy datasets demonstrate quantitative and qualitative improvements. We show that the Grad-CAM method can be used to interpret and verify the prediction of our model. Compared with a standard U-Net, the performance of our method is superior for all tested datasets. Furthermore, our model leverages a limited number of labeled training data to segment the electron microscopy images and therefore has a high potential for automated biological applications.