Scalable Tool Research Articles

A pilot study evaluating online training for therapist delivery of interpersonal psychotherapy for eating disorders.

Individuals with eating disorders (EDs) often do not receive evidence-based care, such as interpersonal psychotherapy (IPT), partly due to lack of accessible training in these treatments. The standard method of training (i.e., in-person workshops) is expensive and time consuming, prompting a need for more scalable training tools. The primary aim of this pilot and open trial was to examine the effects of an IPT online training platform on training outcomes (i.e., IPT fidelity, knowledge, and acceptance) and, secondarily, whether online training was different from in-person training (using a comparative sample from a separate study) in terms of training outcomes and patient symptoms. Participants were therapists (N = 60) and student patients (N = 42) at 38 college counseling centers. Therapists completed baseline questionnaires and collected data from a student patient with ED symptoms. Therapists then participated in an IPT online training program and completed post-training assessments. Following online training, acceptance of evidence-based treatments, therapist knowledge of IPT, therapist acceptance of IPT, and treatment fidelity increased; acceptance of online training was high at baseline and remained stable after training. Using the 90% confidence interval on outcome effect sizes, results suggested IPT online training was not different from in-person training on most outcomes. Results are based on 60% of therapists who originally enrolled due to high dropout rate of therapist participants. Findings from this preliminary pilot study support the use of IPT online training, which could increase access to evidence-based ED treatment and improve patient care. Lack of accessible therapist training has contributed to many therapists not delivering, and therefore many patients not receiving, evidence-based treatment. This study evaluated a highly disseminable online training and compared outcomes to traditional in-person training and found that training and patient outcomes were not different. Online training has the potential to enhance access to evidence-base care, which could in turn optimize patient outcomes.

A fully scalable homogenization method to upscale 3D elastic media

Summary Modeling seismic wavefields in complex 3D elastic media is the key in many fields of Earth Science: seismology, seismic imaging, seismic hazard assessment, earthquake source mechanism reconstruction. This modeling operation can incur significant computational cost, and its accuracy depends on the ability to take into account the scales of the subsurface heterogeneities varying. The theory of homogenization describes how the small-scale heterogeneities interact with the seismic waves and allows to upscale elastic media consistently with the wave equation. In this study, an efficient and scalable numerical homogenization tool is developed, relying on the similarity between the equations describing the propagation of elastic waves and the homogenization process. By exploiting the optimized implementation of an elastic modeling kernel based on a spectral element discretization and domain decomposition, a fully scalable homogenization process, working directly on the spectral-element mesh, is presented. Numerical experiments on the entire SEAM II foothill model and a 3D version of the Marmousi II model illustrate the efficiency and flexibility of this approach. A reduction of two orders of magnitude in terms of absolute computational cost is observed on the elastic wave modeling of the entire SEAM II model at a controlled accuracy.

Abstract LB145: Implementing timely germline genetic testing for patients with pancreatic cancer via a point-of-care digital education and health assessment tool

Abstract Introduction: Pancreatic ductal adenocarcinoma (PDAC) is the third leading cause of cancer-related deaths in the United States. National guidelines recommend germline genetic testing (GT) for all PDAC patients, but only 6.8-19% obtain GT. Current GT barriers include under-referral and genetic counselor shortage. Our goal was to improve uptake of GT among PDAC patients using a novel delivery model for genetic education and testing. Methods: The Genetic Information for Treatment Decisions (GIFTD) Initiative at Moffitt Cancer Center was created to streamline education and GT for PDAC patients. A Genetic Risk Education Coordinator (GREC), trained by licensed, board-certified genetic counselors, was hired to identify eligible patients and facilitate GT during clinician appointments. In partnership with Nest Genomics, a digital point-of-care tool was developed to collect personal and family history information, provide education about GT, and assess interest in GT. Patients receive a link from Nest ahead of their appointment via text and/or email to invite them to provide information and view education via mobile device. An auto-generated summary report with an indicator regarding GT interest is created in real-time for the GREC. Results: Seventy patients were invited to the Nest tool over the six-week launch period. Of those who opened the link, 83% (35/42) logged into their Nest account. Of these patients who logged into Nest, 71% (25/35) completed Nest’s GT education. The GREC met with 68% (17/25) of the patients that completed the education. 76% (13/17) of these patients were consented to GT. There were 2 patients who initially indicated that they were not interested in genetic testing in Nest but changed their mind after meeting with the GREC. Conclusion: Moffitt's collaboration with Nest through the GREC has successfully facilitated prompt germline genetic testing for PDAC patients, aiding in informed treatment decisions and family screening. Our preliminary findings suggest that the scalable Nest tool successfully captures PDAC patients who may be interested in GT, but follow-up with the GREC may further increase GT rates. Future directions for increasing uptake of GT and adhering to national guidelines include additional personalization and enhancement of the Nest tool. Citation Format: Jennifer B. Permuth, Sophie Moravec, Laura V. Barton, Revathy Suresh, Susan T. Vadaparampi, Jessica Rivera Rivera, J. Kevin Hicks, Teresa Ho, Dae Won Kim, Russel Palm, Pamela J. Hodul, Tara Schmidlen, Moran Snir, Laura Hayward. Implementing timely germline genetic testing for patients with pancreatic cancer via a point-of-care digital education and health assessment tool [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(7_Suppl):Abstract nr LB145.

Species-aware DNA language models capture regulatory elements and their evolution

BackgroundThe rise of large-scale multi-species genome sequencing projects promises to shed new light on how genomes encode gene regulatory instructions. To this end, new algorithms are needed that can leverage conservation to capture regulatory elements while accounting for their evolution.ResultsHere, we introduce species-aware DNA language models, which we trained on more than 800 species spanning over 500 million years of evolution. Investigating their ability to predict masked nucleotides from context, we show that DNA language models distinguish transcription factor and RNA-binding protein motifs from background non-coding sequence. Owing to their flexibility, DNA language models capture conserved regulatory elements over much further evolutionary distances than sequence alignment would allow. Remarkably, DNA language models reconstruct motif instances bound in vivo better than unbound ones and account for the evolution of motif sequences and their positional constraints, showing that these models capture functional high-order sequence and evolutionary context. We further show that species-aware training yields improved sequence representations for endogenous and MPRA-based gene expression prediction, as well as motif discovery.ConclusionsCollectively, these results demonstrate that species-aware DNA language models are a powerful, flexible, and scalable tool to integrate information from large compendia of highly diverged genomes.

Synchrotron‐source micro‐x‐ray computed tomography for examining butterfly eyes

AbstractComparative anatomy is an important tool for investigating evolutionary relationships among species, but the lack of scalable imaging tools and stains for rapidly mapping the microscale anatomies of related species poses a major impediment to using comparative anatomy approaches for identifying evolutionary adaptations. We describe a method using synchrotron source micro‐x‐ray computed tomography (syn‐μXCT) combined with machine learning algorithms for high‐throughput imaging of Lepidoptera (i.e., butterfly and moth) eyes. Our pipeline allows for imaging at rates of ~15 min/mm3 at 600 nm3 resolution. Image contrast is generated using standard electron microscopy labeling approaches (e.g., osmium tetroxide) that unbiasedly labels all cellular membranes in a species‐independent manner thus removing any barrier to imaging any species of interest. To demonstrate the power of the method, we analyzed the 3D morphologies of butterfly crystalline cones, a part of the visual system associated with acuity and sensitivity and found significant variation within six butterfly individuals. Despite this variation, a classic measure of optimization, the ratio of interommatidial angle to resolving power of ommatidia, largely agrees with early work on eye geometry across species. We show that this method can successfully be used to determine compound eye organization and crystalline cone morphology. Our novel pipeline provides for fast, scalable visualization and analysis of eye anatomies that can be applied to any arthropod species, enabling new questions about evolutionary adaptations of compound eyes and beyond.

PhosMap: An ensemble bioinformatic platform to empower interactive analysis of quantitative phosphoproteomics

BackgroundLiquid chromatography-mass spectrometry (LC-MS)-based quantitative phosphoproteomics has been widely used to detect thousands of protein phosphorylation modifications simultaneously from the biological specimens. However, the complicated procedures for analyzing phosphoproteomics data has become a bottleneck to widening its application. MethodsHere, we develop PhosMap, a versatile and scalable tool to accomplish phosphoproteomics data analysis. A standardized phosphorylation data format was created for data analyses, from data preprocessing to downstream bioinformatic analyses such as dimension reduction, differential phosphorylation analysis, kinase activity, survival analysis, and so on. For better usability, we distribute PhosMap as a Docker image for easy local deployment upon any of Windows, Linux, and Mac system. ResultsThe source code is deposited at https://github.com/BADD-XMU/PhosMap. A free PhosMap webserver (https://huggingface.co/spaces/Bio-Add/PhosMap), with easy-to-follow fashion of dashboards, is curated for interactive data analysis. ConclusionsPhosMap fills the technical gap of large-scale phosphorylation research by empowering researchers to process their own phosphoproteomics data expediently and efficiently, and facilitates better data interpretation.

Predicting COVID-19 Cough Sounds Using Spectrogram Analysis Across Multiple Classes

The COVID-19 pandemic has underscored the need for effective diagnostic tools. One promising avenue involves analyzing cough sounds to glean insights into respiratory health. This study presents a new method for predicting COVID-19 cough sounds using spectrogram analysis across various classes. We leverage advanced deep learning models such as DenseNet121, VGG16, ResNet50, and Inception Net, alongside our novel CNN architecture, to extract pertinent features from cough sound spectrograms. We use a diverse dataset encompassing cough sounds from COVID-19 positive and negative cases, as well as other respiratory conditions, for model training and assessment. Our results demonstrate the effectiveness of our approach in accurately categorizing COVID-19 cough sounds, outperforming existing models. This methodology shows promise as a non-invasive, scalable, and economical tool for early COVID-19 detection and monitoring, aiding public health efforts during the pandemic.

End-to-End Phase Field Model Discovery Combining Experimentation, Crowdsourcing, Simulation and Learning

The availability of tera-byte scale experiment data calls for AI driven approaches which automatically discover scientific models from data. Nonetheless, significant challenges present in AI-driven scientific discovery: (i) The annotation of large scale datasets requires fundamental re-thinking in developing scalable crowdsourcing tools. (ii) The learning of scientific models from data calls for innovations beyond black-box neural nets. (iii) Novel visualization & diagnosis tools are needed for the collaboration of experimental and theoretical physicists, and computer scientists. We present Phase-Field-Lab platform for end-to-end phase field model discovery, which automatically discovers phase field physics models from experiment data, integrating experimentation, crowdsourcing, simulation and learning. Phase-Field-Lab combines (i) a streamlined annotation tool which reduces the annotation time (by ~50-75%), while increasing annotation accuracy compared to baseline; (ii) an end-to-end neural model which automatically learns phase field models from data by embedding phase field simulation and existing domain knowledge into learning; and (iii) novel interfaces and visualizations to integrate our platform into the scientific discovery cycle of domain scientists. Our platform is deployed in the analysis of nano-structure evolution in materials under extreme conditions (high temperature and irradiation). Our approach reveals new properties of nano-void defects, which otherwise cannot be detected via manual analysis.

Abstract 7410: PyVIPER: A fast and scalable toolkit for the identification of dysregulated proteins in tumor-derived single-cell RNA-seq data

Abstract Single-cell RNA sequencing (scRNA-seq) enabled the characterization of tumor microenvironment heterogeneity with unprecedented resolution. However, its low signal-to-noise ratio due to data sparsity hinders biomarkers quantitative measurements. To perform a more robust assessments of protein markers, we have previously introduced the use of the VIPER algorithm that estimates regulatory proteins activity using the expression of their downstream targets by leveraging lineage specific regulatory networks. We have shown that this approach compares favorably with both gene expression and protein abundance measurements to identify novel cell states corresponding to tumor subpopulations that are undetectable at the gene expression level, by inferring the mechanistic contribution of regulatory and signaling proteins to cellular phenotypes. VIPER-inferred cell states have been linked to critical functional roles, including immune-evasion and progression, disease heterogeneity and drug sensitivity in cancer. However, the ever-growing size of scRNA-seq datasets demands for scalable tools that simultaneously analyze tens or hundreds of thousands of cells in tumor contexts, such as tumor atlases and large clinical cohorts. We present PyVIPER, a fast, memory-efficient, and highly scalable Python framework that allows protein activity inference of large scRNA-seq datasets. PyVIPER is fully integrated with both AnnData and Scanpy frameworks for single-cell analysis and enables a seamless interface with cutting-edge machine learning libraries (e.g., sckit-learn and TensorFlow). PyVIPER provides multiple algorithms for enrichment analysis, novel modules for postprocessing and pathway analysis, and a set of procedures tailored for precision oncology studies in single cells, including OncoMatch, a validated platform to assess the fidelity of models (in vitro or animal) with human tumors for drug sensitivity elucidation. We benchmarked PyVIPER against several scRNA-seq cancer datasets obtaining runtimes reduction of several orders of magnitude (from hours to minutes) when compared to previous VIPER implementations. Taken together, PyVIPER scalability and its embedded toolkit for cancer data analysis address the need to efficiently extend protein activity analyses to large single-cell datasets, with the goal of providing cancer researchers with an effective platform to identify aberrant molecular mechanisms in cancer cells. Citation Format: Luca Zanella, Alexander L. Wang, Zizhao J. Lin, Lukas J. Vlahos, Miquel Anglada Girotto, Aziz Zafar, Andrea Califano, Alessandro Vasciaveo. PyVIPER: A fast and scalable toolkit for the identification of dysregulated proteins in tumor-derived single-cell RNA-seq data [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 7410.

Deep learning to detect left ventricular structural abnormalities in chest X-rays.

Early identification of cardiac structural abnormalities indicative of heart failure is crucial to improving patient outcomes. Chest X-rays (CXRs) are routinely conducted on a broad population of patients, presenting an opportunity to build scalable screening tools for structural abnormalities indicative of Stage B or worse heart failure with deep learning methods. In this study, a model was developed to identify severe left ventricular hypertrophy (SLVH) and dilated left ventricle (DLV) using CXRs. A total of 71 589 unique CXRs from 24 689 different patients completed within 1 year of echocardiograms were identified. Labels for SLVH, DLV, and a composite label indicating the presence of either were extracted from echocardiograms. A deep learning model was developed and evaluated using area under the receiver operating characteristic curve (AUROC). Performance was additionally validated on 8003 CXRs from an external site and compared against visual assessment by 15 board-certified radiologists. The model yielded an AUROC of 0.79 (0.76-0.81) for SLVH, 0.80 (0.77-0.84) for DLV, and 0.80 (0.78-0.83) for the composite label, with similar performance on an external data set. The model outperformed all 15 individual radiologists for predicting the composite label and achieved a sensitivity of 71% vs. 66% against the consensus vote across all radiologists at a fixed specificity of 73%. Deep learning analysis of CXRs can accurately detect the presence of certain structural abnormalities and may be useful in early identification of patients with LV hypertrophy and dilation. As a resource to promote further innovation, 71 589 CXRs with adjoining echocardiographic labels have been made publicly available.

SpyMask enables combinatorial assembly of bispecific binders

Bispecific antibodies are a successful and expanding therapeutic class. Standard approaches to generate bispecifics are complicated by the need for disulfide reduction/oxidation or specialized formats. Here we present SpyMask, a modular approach to bispecifics using SpyTag/SpyCatcher spontaneous amidation. Two SpyTag-fused antigen-binding modules can be precisely conjugated onto DoubleCatcher, a tandem SpyCatcher where the second SpyCatcher is protease-activatable. We engineer a panel of structurally-distinct DoubleCatchers, from which binders project in different directions. We establish a generalized methodology for one-pot assembly and purification of bispecifics in 96-well plates. A panel of binders recognizing different HER2 epitopes were coupled to DoubleCatcher, revealing unexpected combinations with anti-proliferative or pro-proliferative activity on HER2-addicted cancer cells. Bispecific activity depended sensitively on both binder orientation and DoubleCatcher scaffold geometry. These findings support the need for straightforward assembly in different formats. SpyMask provides a scalable tool to discover synergy in bispecific activity, through modulating receptor organization and geometry.

THE FUNCTION OF CLOUD COMPUTING TECHNOLOGY: AN INVALUABLE WEAPON AGAINST COVID-19

This study looks into how important cloud computing is to reducing the difficulties brought on by the COVID-19 epidemic. As the workforce becomes more reliant on digital solutions and works remotely more frequently, cloud computing becomes a vital tool for maintaining flexibility and company continuity. The study examines how cloud-based technologies, like scalable infrastructure and tools for remote collaboration, are being used to meet the extraordinary needs of distant work environments.

DeGenPrime provides robust primer design and optimization unlocking the biosphere.

Polymerase chain reaction (PCR) is the world's most important molecular diagnostic with applications ranging from medicine to ecology. PCR can fail because of poor primer design. The nearest-neighbor thermodynamic properties, picking conserved regions, and filtration via penalty of oligonucleotides form the basis for good primer design. DeGenPrime is a console-based high-quality PCR primer design tool that can utilize MSA formats and degenerate bases expanding the target range for a single primer set. Our software utilizes thermodynamic properties, filtration metrics, penalty scoring, and conserved region finding of any proposed primer. It has degeneracy, repeated k-mers, relative GC content, and temperature range filters. Minimal penalty scoring is included according to secondary structure self-dimerization metrics, GC clamping, tri- and tetra-loop hairpins, and internal repetition. We compared PrimerDesign-M, DegePrime, ConsensusPrimer, and DeGenPrime on acceptable primer yield. PrimerDesign-M, DegePrime, and ConsensusPrimer provided 0%, 11%, and 17% yield, respectively, for the alternative iron nitrogenase (anfD) gene target. DeGenPrime successfully identified quality primers within the conserved regions of the T4-like phage major capsid protein (g23), conserved regions of molybdenum-based nitrogenase (nif), and its alternatives vanadium (vnf) and iron (anf) nitrogenase. DeGenPrime provides a universal and scalable primer design tool for the entire tree of life. DeGenPrime is written in C++ and distributed under a BSD-3-Clause license. The source code for DeGenPrime is freely available on www.github.com/raw-lab/degenprime.

From manual to machine: assessing the efficacy of large language models in content analysis

ABSTRACT This study compares the performance of Large Language Models (LLMs) and human coders in predicting relational uncertainty from textual data. Employing various LLMs (gpt-4.0-turbo, gpt-3.5-turbo, Claude 2, llama7b-v2-chat, and llama13b-v2-chat), we found that these models perform comparably to human coders, with only minor differences in Mean Squared Error (MSE) values. However, not all LLMs performed equally, underscoring the importance of model selection. Our findings highlight the potential of LLMs as a scalable tool for content analysis, but also emphasize their nuanced application based on the specific research context. The study advances the discourse on the use of LLMs in content analysis and provides insights for future research in this rapidly evolving field.

Comparison of two plasma p-tau217 assays to detect and monitor Alzheimer’s pathology

Blood-based biomarkers of Alzheimer's disease (AD) have become increasingly important as scalable tools for diagnosis and determining clinical trial eligibility. P-tau217 is the most promising due to its excellent sensitivity and specificity for AD-related pathological changes. We compared the performance of two commercially available plasma p-tau217 assays (ALZpath p-tau217 and Janssen p-tau217+) in 294 individuals cross-sectionally. Correlations with amyloid PET and tau PET were assessed, and Receiver Operating Characteristic (ROC) analyses evaluated both p-tau217 assays for identifying AD pathology. Both plasma p-tau217 assays were strongly associated with amyloid and tau PET. Furthermore, both plasma p-tau217 assays identified individuals with AD vs other neurodegenerative diseases (ALZpath AUC=0.95; Janssen AUC=0.96). Additionally, plasma p-tau217 concentrations rose with AD severity and their annual changes correlated with tau PET annual change. Both p-tau217 assays had excellent diagnostic performance for AD. Our study supports the future clinical use of commercially-available assays for p-tau217. This research is supported by the Weston Brain Institute, Canadian Institutes of Health Research (CIHR), Canadian Consortium on Neurodegeneration in Aging, the Alzheimer's Association, Brain Canada Foundation, the Fonds de Recherche du Québec - Santé and the Colin J. Adair Charitable Foundation.

A novel ACE2-Based electrochemical biosensor for sensitive detection of SARS-CoV-2

SARS-CoV-2 emerged in late 2019 and quickly spread globally, resulting in significant morbidity, mortality, and socio-economic disruptions. As of now, collaborative global efforts in vaccination and the advent of novel diagnostic tools have considerably curbed the spread and impact of the virus in many regions. Despite this progress, the demand remains for low-cost, accurate, rapid and scalable diagnostic tools to reduce the influence of SARS-CoV-2. Herein, the angiotensin-converting enzyme 2 (ACE2), a receptor for SARS-CoV-2, was immobilized on two types of electrodes, a screen-printed gold electrode (SPGE) and a screen-printed carbon electrode (SPCE), to develop electrochemical biosensors for detecting SARS-CoV-2 with high sensitivity and selectivity. This was achieved by using 1H, 1H, 2H, 2H-perfluorodecanethiol (PFDT) and aryl diazonium salt serving as linkers for SPGEs and SPCEs, respectively. Once SARS-CoV-2 was anchored onto the ACE2, the interaction of the virus with the redox probe was analyzed using electrochemical impedance spectroscopy (EIS) and cyclic voltammetry (CV). Aryl diazonium salt was observed as a superior linker compared to PFDT due to its consistent performance in the modification of the SPCEs and effective ACE2 enzyme immobilization. A distinct pair of redox peaks in the cyclic voltammogram of the biosensor modified with aryl diazonium salt highlighted the redox reaction between the functional groups of SARS-CoV-2 and the redox probe. The sensor presented a linear relationship between the redox response and the logarithm of SARS-CoV-2 concentration, with a detection limit of 1.02 × 106 TCID50/mL (50% tissue culture infectious dose). Furthermore, the biosensor showed remarkable selectivity towards SARS-CoV-2 over H1N1virus.

MetaCerberus: distributed highly parallelized HMM-based processing for robust functional annotation across the tree of life.

MetaCerberus is a massively parallel, fast, low memory, scalable annotation tool for inference gene function across genomes to metacommunities. MetaCerberus provides an elusive HMM/HMMER-based tool at a rapid scale with low memory. It offers scalable gene elucidation to major public databases, including KEGG (KO), COGs, CAZy, FOAM, and specific databases for viruses, including VOGs and PHROGs, from single genomes to metacommunities. MetaCerberus is 1.3× as fast on a single node than eggNOG-mapper v2 on 5× less memory using an exclusively HMM/HMMER mode. In a direct comparison, MetaCerberus provides better annotation of viruses, phages, and archaeal viruses than DRAM, Prokka, or InterProScan. MetaCerberus annotates more KOs across domains when compared to DRAM, with a 186× smaller database, and with 63× less memory. MetaCerberus is fully integrated for automatic analysis of statistics and pathways using differential statistic tools (i.e. DESeq2 and edgeR), pathway enrichment (GAGE R), and pathview R. MetaCerberus provides a novel tool for unlocking the biosphere across the tree of life at scale. MetaCerberus is written in Python and distributed under a BSD-3 license. The source code of MetaCerberus is freely available at https://github.com/raw-lab/metacerberus compatible with Python 3 and works on both Mac OS X and Linux. MetaCerberus can also be easily installed using bioconda: mamba create -n metacerberus -c bioconda -c conda-forge metacerberus.

ScCircle-seq unveils the diversity and complexity of extrachromosomal circular DNAs in single cells

Extrachromosomal circular DNAs (eccDNAs) have emerged as important intra-cellular mobile genetic elements that affect gene copy number and exert in trans regulatory roles within the cell nucleus. Here, we describe scCircle-seq, a method for profiling eccDNAs and unraveling their diversity and complexity in single cells. We implement and validate scCircle-seq in normal and cancer cell lines, demonstrating that most eccDNAs vary largely between cells and are stochastically inherited during cell division, although their genomic landscape is cell type-specific and can be used to accurately cluster cells of the same origin. eccDNAs are preferentially produced from chromatin regions enriched in H3K9me3 and H3K27me3 histone marks and are induced during replication stress conditions. Concomitant sequencing of eccDNAs and RNA from the same cell uncovers the absence of correlation between eccDNA copy number and gene expression levels, except for a few oncogenes, including MYC, contained within a large eccDNA in colorectal cancer cells. Lastly, we apply scCircle-seq to one prostate cancer and two breast cancer specimens, revealing cancer-specific eccDNA landscapes and a higher propensity of eccDNAs to form in amplified genomic regions. scCircle-seq is a scalable tool that can be used to dissect the complexity of eccDNAs across different cell and tissue types, and further expands the potential of eccDNAs for cancer diagnostics.

Deciphering the Efficacy of No-Attention Architectures in Computed Tomography Image Classification: A Paradigm Shift

The burgeoning domain of medical imaging has witnessed a paradigm shift with the integration of AI, particularly deep learning, enhancing diagnostic precision and expediting the analysis of Computed Tomography (CT) images. This study introduces an innovative Multilayer Perceptron-driven model, DiagnosticMLP, which sidesteps the computational intensity of attention-based mechanisms, favoring a no-attention architecture that leverages Fourier Transforms for global information capture and spatial gating units for local feature emphasis. This study’s methodology encompasses a sophisticated augmentation and patching strategy at the input level, followed by a series of MLP blocks designed to extract hierarchical features and spatial relationships, culminating in a global average pooling layer before classification. Evaluated against state-of-the-art MLP-based models including MLP-Mixer, FNet, gMLP, and ResMLP across diverse and extensive CT datasets, including abdominal, and chest scans, DiagnosticMLP demonstrated a remarkable ability to converge efficiently, with competitive accuracy, F1 scores, and AUC metrics. Notably, in datasets featuring kidney and abdomen disorders, the model showcased superior generalization capabilities, underpinned by its unique design that addresses the complexity inherent in CT imaging. The findings in terms of accuracy and precision-recall balance posit DiagnosticMLP as an exceptional outperforming alternative to attention-reliant models, paving the way for streamlined, efficient, and scalable AI tools in medical diagnostics, reinforcing the potential for AI-augmented precision medicine without the dependency on attention-based architectures.

Evaluation of the internet-based intervention "Selfapy" in participants with unipolar depression and the impact on quality of life: a randomized, parallel group study.

Depressive disorders cause a major burden of disease worldwide and often lead to a loss of social functioning. Patients suffering from depressive disorders report a lower quality of life (QOL) than people without a history of mental health issues. Internet-based interventions (IBIs) based on cognitive behavioral therapy (CBT) are effective in reducing symptom severity but data on their impact on quality of life in clinically depressed patients so far is scarce. Selfapy is a CBT-based IBI for depressive disorders. 401 participants (332 female, mean age 37 (SD = 11) with a diagnosis of major depressive disorder (MDD) or dysthymia were enrolled in a randomized, parallel, three-arm trial comparing a therapist-guided Selfapy intervention with an unguided Selfapy intervention and a waiting list control. QOL was measured using the WHOQOL-BREF at baseline, post-treatment (12weeks) and at 24-week follow-up. The effects of the interventions on QOL were calculated using linear mixed effects models. At post-treatment (12weeks) the guided and unguided intervention groups reported an increase in QOL on physical and psychological health domains compared to controls (significant group*time interaction). The gain in QOL was maintained over the follow-up period only for psychological health. QOL decreased in the social relationships and environment domains over the course of treatment and during the follow-up treatment for all participants. There were no differences between the guided and the unguided intervention. Selfapy proved to positively affect psychological and physical QOL in a sample of participants suffering from depressive disorders and can therefore be considered an effective and highly scalable therapeutic tool. The pattern of results might partly be attributable to effects of the COVID-19 pandemic and public health measures that coincided with the trial. German Clinical Trials Register (DRKS): DRKS00017191. Registered June 14th, 2019, https://www.drks.de/drks_web/navigate.do?navigationId=trial.HTML&TRIAL_ID=DRKS00017191 .