Proteomic sample preparation for liquid chromatography-tandem mass spectrometry (LC-MS/MS) is increasingly addressed by automated approaches. However, in clinical settings for precision medicine, where a limited number of samples must be processed in a standardized and reproducible manner with minimal user interaction, fully automized workflows remain scarce. Here, we present the AutoPAC-disk, a centrifugal microfluidic implementation of a protein aggregation capture (PAC) sample preparation workflow for bottom-up proteomics that automates all necessary steps for on-bead proteolysis including on-disk pre-storage of buffers. Comparative evaluation of the AutoPAC-disk using HEK293 cell lysates against a manual reference workflow and a semi-automated robotic PAC implementation showed 50% and 37% more peptide identifications and 23% and 10% more protein group identifications, respectively, while maintaining high quantitative reproducibility as reflected by protein-group intensity coeffincients of variation (CVs) below 10%. Additional analysis demonstrated that the AutoPAC-disk primarily increased identifications of low-abundance proteins without introducing method specific physicochemical bias. The AutoPAC-disk was subsequently evaluated using patient-derived formalin-fixed paraffin-embedded (FFPE) prostate tumor tissue. The AutoPAC-disk yielded 8% more peptide identifications and 10% more protein groups than the manual workflow, with protein-group intensity CVs below 7% for both methods. Together, these results demonstrate that centrifugal microfluidic automation with on-disk buffer pre-storage can substantially simplify proteomic sample preparation, minimize user interaction and lower operational barriers for personnel with limited experience in proteomic sample preparation, providing a promising strategy for clinical and translational proteomics in the field of precision medicine.

The increasing demand for effective and translatable culture models in drug discovery has driven the development of advanced systems, including organ-on-a-chip (OoC), microphysiological systems (MPS), and complex in vitro models (CIVM). These technologies are recognized for their ability to model human physiological responses, particularly with the Food and Drug Administration (FDA) Modernization Act 2.0 promoting cell-based assays and computer simulations as new approach methodologies (NAMs) to reduce reliance on animal testing. However, the adoption of MPS and OoC systems in drug discovery remains limited by complex culture protocols, low throughput, and difficulties in achieving reproducible results. In this study, we demonstrate that our automated platform, "Screening Station No. 2," addresses these challenges by fully automating the end-to-end process-from cell culture to drug testing-using an angiogenesis model. The system integrates the SCALE12-MR rocking incubator with numerous instruments and dynamic scheduling to enable seamless execution of key operations, including media exchange, compound dosing, and imaging. By minimizing human intervention, it automates OoC cultivation, including gravitational stimulation via rocking culture. Screening Station No. 2 supports scalability, expandability, and automation of complex workflows that were previously difficult to implement. This study highlights the importance of versatile, scalable automated systems capable of adapting to diverse experimental conditions to enhance the efficiency and reliability of drug discovery. The advances demonstrate the potential of automated OoC systems to accelerate drug development, improve preclinical model translatability, and address the demand for innovative laboratory automation methodologies.

Attrition in the therapeutic pipeline can often be attributed to a lack of translational efficacy from pre-clinical to clinical phases. Organoids show great promise as a game-changer in disease modeling and drug screening, as they better resemble tissue structure and functionality, and show more predictive responses to drugs. However, challenges associated with the practical adoption of organoids, such as assay complexity, reproducibility, and scalability, have limited their widespread adoption as a primary screening method in drug discovery. To alleviate the bottlenecks that come with labor-intensive manual protocols, we developed a cell culture automation solution called the CellXpress.ai® Automated Cell Culture System. The instrument enables automation of the entire organoid culture for prolonged complex workflows. The CellXpress.ai system provides media exchanges, cell plating, passaging and monitoring, endpoint assays, and complex image analysis. The instrument contains functional components, including an automated imager, a liquid handler, and an incubator, connected by one unified software. Development of cell cultures is monitored by periodic imaging and AI-powered image analysis, which can trigger automatic decisions to initiate passaging, endpoint assay, or troubleshooting steps. Here we present methods and results from automation of several commonly used complex biological workflows, including maintenance and culture of iPSC, automated maintenance and expansion of 3D organoids in matrix domes, and formation, culture, and assays for 3D spheroids in the low attachment plates. Cell culture process automation powered by imaging and AI -controlled decision making has great potential to bring 3D biology to another level, allowing to increase throughput and productivity, and enabling a variety of drug discovery and precision medicine applications.

IntroductionThe use of digital solutions including patient-reported outcomes is limited to follow-up of patients with established diagnoses but is rarely used as first step of the diagnostic process substituting a personal contact with a health professional. We report on the diagnostic validity and cost per patient implications based on a feasibility study of a new virtual diagnostic service (VDS) for common neurological sleep disorders that, as a first step, involves the collection and automated analysis of self-reported digital patient data.MethodsThe VDS was established at the Odense University Hospital, Denmark. Assessment of diagnostic validity of the underlying algorithm was conducted independently and blinded. Estimation of effects on cost per patient was based on administrative hospital cost data comparing similar periods before and after the introduction of VDS and estimates for travel and time consumption to assess the patients' economic benefits.ResultsA questionnaire-based algorithm was developed leveraging the diagnostic criteria of the American Academy of Sleep Medicine; comprehensibility was secured and improved by initial patient involvement. Parallel use of both the questionnaire and assessment by a senior sleep specialist of the first 20 patients revealed no discernible safety concerns and resulted in additional linguistic adaptions. The final questionnaire was completed by 123 of 157 patients (78.3%) identified as suitable for VDS. The questionnaire-based algorithm resulted in correct use of additional diagnostic procedures in 84 out of 95 patients with final diagnosis at data closure (88.4%, Cohen's kappa: 0.84). The algorithm proposed a specific diagnosis in 55 patients that was correct in 49.1% of cases (Cohen's kappa: 0.39). The economic analysis revealed a 46.7% reduction of the time from referral to diagnosis of the patient (226.5 days to 120.7 days). The average number of contacts with health professionals decreased from 2.15 to 1.26, the average direct costs per patients were reduced by 39.6% from 1811 Danish Kroner (DKK) to 1093 DKK. We estimated a 40.6% reduction of the total costs per patients from 3904 DKK to 2320 DKK including time consumption and travel costs.DiscussionThis first feasibility study indicates that use of digital diagnostic solutions as first step of the diagnostic process of neurological sleep disorders combined with an essentially complete virtual work flow has high accuracy and may be associated with reduced time for diagnostics and cost reductions for health providers and patients.

Emergency department triage is commonly conceptualised as a standardised classification of patient urgency based on vital signs and presenting symptoms. However, research shows that triage is deeply shaped by the clinical context such as organizational structure, clinical uncertainty and subjective decision-making. This contextual complexity presents a challenge for the implementation of artificial intelligence decision-support in emergency care. While Artificial Intelligence-supported triage has demonstrated promising accuracy in controlled settings, these evaluations capture only a limited part of triage work and rarely account for the underlying mechanisms through which triage is sustained in everyday clinical practice. To address this gap, this study aimed to examine how triage unfolds in real-world emergency care and to understand the underlying generative mechanisms shaping triage practice. An ethnographic study was conducted across three Danish emergency departments with different organisational triage configurations. Data consisted of participant observations of triage practice and semi-structured interviews with nurses and physicians. Data were analysed using inductive qualitative content analysis. Findings were interpreted, informed by a critical realistic lens, to understand underlying generative mechanisms shaping triage practice. Triage was not a standardized classification event but a dynamic, negotiated practice across the acute admission pathway. Two core generative mechanisms were identified: 1) Coherence work sustained continuity, safety, and flow through transitional coordination, gap compensation, situated judgement, and pragmatic prioritisation. 2) Interpretive reasoning enabled healthcare professionals to navigate clinical complexity through situated interpretive filtering and experience-driven judgement. Both mechanisms were activated by managing competing demands between safety, continuity, and flow within fragmented, resource-constrained organisational setups. Both mechanisms drew on tacit and embodied knowledge not routinely documented or available to artificial intelligence systems. These findings demonstrate that safe and effective triage depends not only on accurate classification but on largely invisible adaptive work embedded in everyday practice under conditions of clinical uncertainty and organisational fragmentation. For artificial intelligence-supported triage to achieve clinical impact, system design must recognise and support the coherence- and interpretive reasoning work through which continuity, safety, and flow are maintained in real-world emergency care.

The multi-attribute method (MAM) is an integrated peptide mapping strategy based on liquid chromatography-mass spectrometry (LC-MS) technology. It enables precise quantification and dynamic tracking of multiple site-specific modifications in a single analysis, significantly enhancing the efficiency and depth of biopharmaceutical quality control. Notably, its integrated application across process development, process monitoring, and product release has driven a paradigm shift from a "single-attribute, single-method" approach to a "multi-attribute, integrated-method" approach in quality control. This review systematically summarizes the technical principles, optimization strategies, and application progress of MAM by integrating recent research cases of complex therapeutic proteins (e.g., monoclonal antibodies and Fc fusion proteins), with a focus on specific strategies and practical paths of MAM in workflow automation, new peak detection (NPD) optimization, intact multi-attribute method (iMAM), and the integration of complementary technologies. The objective is to provide a valuable reference for the standardization and industrial application of MAM in biotechnological drug quality control. Although MAM is expected to become a core analytical tool for biopharmaceutical quality control, its widespread industrial application remains constrained by key challenges, including insufficient method robustness, incomplete standardization, and variable regulatory acceptance. Notably, a significant stride in regulatory acceptance has been made with the recent implementation of the United States Pharmacopeia (USP) General Chapter < 1060 > , which establishes the first official framework for MAM. Beyond this regulatory milestone, future efforts should focus on advancing automated platform development, creating intelligent data algorithms, and strengthening cross-disciplinary collaboration to further promote the systematic integration and standardized application of MAM throughout the full lifecycle quality management of biotechnological drugs.

Purpose: This systematic literature review investigated the impact of workflow automation — with emphasis on Robotic Process Automation (RPA) and financial technology tools — on the operational efficiency and cost structure of small and medium-sized enterprises (SMEs). The research focused on the financial sub-processes of accounts payable, accounts receivable, and bank reconciliation, analyzing how the digital transformation of these workflows affects organizational liquidity and profitability. Methodology: The PRISMA 2020 protocol was adopted for conducting and reporting the review. Searches were carried out in the Web of Science, Scopus, EBSCO Business Source Complete, and CAPES Periódicos databases, covering the period from 2019 to 2024. After applying inclusion and exclusion criteria, 52 primary studies were selected for qualitative and quantitative analysis. Findings: The literature evidences an average reduction of operational costs between 30% and 70% in automated financial processes, with productivity gains of up to 85% in repetitive tasks. The automation of accounts payable and receivable demonstrated a shortened financial cycle, improved working capital, and reduced manual errors. Barriers such as implementation costs, cultural resistance, and skill gaps were identified as predominant limiting factors for SMEs. Contributions: The study proposes an empirically validated financial digital maturity framework, structured in four stages — Diagnosis, Gradual Implementation, Optimization, and Strategic Expansion — specifically designed for SMEs seeking to initiate or deepen financial automation processes. The results expand the body of knowledge on digital transformation in smaller organizations and provide practical insights for managers, consultants, and policymakers.

In this study, we present SpineDL, an open-source deep learning (DL) approach for neuron and anatomical structure segmentation of the spinal cord in fluorescence images immunostained with NeuN and DAPI, within the context of murine models of spinal cord injury (SCI). SpineDL comprises two main modules: SpineDL-Neuron, for instance-level identification of neuronal somas; and SpineDL-Structure, for semantic segmentation of key spinal cord structures including gray matter, white matter, ependyma, and damaged tissue. To train the models, we developed the SpineDL dataset, a curated collection of 161 confocal images of mouse spinal cord, manually annotated by SCI researchers and organized into specific subsets. Both models are based on the HRNetV2-W64 architecture and were trained using state-of-the-art data augmentation and optimization techniques, implemented within the BiaPy framework, following an iterative refinement process driven by quantitative evaluation, SCI researcher feedback, and systematic error analysis. Our results demonstrate that SpineDL achieves researcher-level performance in both structural segmentation and neuron identification tasks, showing high robustness across anatomical regions and injury conditions. Overall, this work provides a reproducible and extensible platform for quantitative analysis of neuron distribution in the naïve and injured spinal cord, supporting automation, standardization, and scalability of histopathological workflows in neuroscience research and preclinical studies and translational applications.

Mapping BAFF/APRIL dependency across autoimmune diseases: A mechanistic framework for understanding differential therapeutic responses.

Point-of-care ultrasound (POCUS) provides real-time diagnostic capabilities at the bedside. Implementing a POCUS program in an institution is a highly complex process. Coordinating the imaging workflow of numerous clinical specialties requires meticulous planning and appropriate oversight. This white paper describes the best practices for the critical pre-deployment phase of program implementation, after having established POCUS program governance. Important considerations during the pre-deployment phase include goal setting, scaling POCUS workflow across clinical, educational, and technological domains, and addressing budgetary concerns and complexity of deployment strategies. This paper also discusses the role of the POCUS workflow manager software in encounter-based imaging workflow and the role of a system-wide clinical ultrasound director. Furthermore, it outlines the necessary approvals needed to ensure compliance and program success, including securing approvals from key departments such as imaging informatics, information technology, cybersecurity, supply chain operations, clinical engineering, infection prevention, legal, and billing departments. Finally, it presents a sample comprehensive project charter to guide this complex integration process from business case development, to clinical go-live, emphasizing best practices for sustained adoption.

Analytical characteristics of the NeuMoDx™ SARS-CoV-2 assay and clinical agreement with the BD MAX system.

Summary:Rising physician burnout, driven by increased administrative load and documentation burden, has led to the widespread adoption of artificial intelligence (AI)–powered transcription services in medical practices. These tools generate clinical notes from patient encounters and have shown the ability to significantly reduce documentation burden and streamline workflows. This article provides clinicians with a primer on the capabilities, limitations, and practical considerations for implementing AI transcription services in their practice. Across platforms, core features include encounter transcription, automated note generation, multilingual capabilities, custom templates, and Health Insurance Portability and Accountability Act compliance. Platforms with deep integration into electronic health records offer greater workflow automation, such as summarizing prior notes, drafting referral letters and after-visit summaries, and preparing orders, but require significant training and onboarding. More user-friendly options provide rapid note creation with minimal setup, though with limited electronic health record integration. Industry partnerships have facilitated the current development of novel features, including point-of-care prior authorization, a virtual assistant to schedule visits, answer patient questions related to appointments, and suggest relevant screenings. Physicians should evaluate the advantages and logistical challenges of implementing AI transcription tools while also considering liability. Each platform offers unique benefits and limitations, and although these tools may enhance physician workflow, they also present risks for both patients and providers.

Cancer staging data is vital for treatment planning, outcome prediction, clinical research, and healthcare resource allocation. Collection at the population level can improve insights, however existing manual methods are resource intensive. This study aimed to develop rules-based natural language processing (NLP) systems to: (1) extract explicit tumour, node, and metastasis (TNM) entities from data reported to the Western Australian Cancer Registry; (2) extract implicit entities translatable to individual TNM values; and (3) translate these values into cancer stages for melanoma, breast, and colorectal cancers based on the AJCC 8th edition TNM staging system. Rules-based NLP systems were developed with extensive consultation from knowledge experts to extract staging information and stage colorectal, breast, and melanoma cancers using pathology reports and a hospital inpatient morbidity dataset. Their performance was evaluated against manual collections (ground truth) created by cancer staging project officers using recall, precision, and F1-scores. After an iterative development process, the rules-based NLP systems correctly staged 87%-90% of cases compared to manual collections created by cancer staging officers. The melanoma NLP system had a weighted average precision of 0.96, recall of 0.94, and F1-score of 0.94. The colorectal and breast models had weighted average precision, recall, and F1-scores of 0.89, 0.89, 0.89; and 0.90, 0.89, and 0.89, respectively. The rules-based NLP systems demonstrated strong performance, with potential for improved accuracy using additional data sources. A rules-based NLP architecture can accurately derive TNM components and cancer stage from routinely collected clinical text. Whilst dependent on domain-expert input rather than data-driven training, the methods described demonstrate a way to support partial automation of cancer staging workflows in a setting with limited training datasets.

Marine diesel propulsion engines are essential to naval platforms, enabling maneuvering, navigation readiness, and training operations. However, maintenance of propulsion consumables—particularly fuel filtration elements—often remains time-based and corrective despite the growing availability of onboard operational records. This paper presents the development and validation of a Remaining Useful Life (RUL) prediction model for the propulsion engine filtration system of the Colombian Navy (ARC) training ship, aiming to estimate time to replacement for cartridge-based filters. The proposed approach handles imperfect manual operational data and scarce, non-uniform maintenance labels through a Prognostics and Health Management (PHM) workflow guided by the Cross-Industry Standard Process for Data Mining (CRISP-DM). It combines physics-informed data quality control using plausibility bounds, outlier mitigation, and time-series reconstruction; expert validation of representative operating cycles using a Delphi protocol; and event logging to align filter-replacement actions with gap-aware approximations. It was trained supervised regression models using an automated machine learning (AutoML) strategy implemented in PyCaret and refined through hyperparameter optimization in Optuna. A Random Forest model achieved the best performance, reaching a test root mean squared error (RMSE) of 52.92 hours with a coefficient of determination of 0.921.

This paper describes the development of an integrated assay for the purification of adeno-associated virus (AAV) particles from crude lysate using a microfluidic cartridge and a centrifugal platform that enables pneumatic actuation of liquids during rotation. The cartridge features POROS™ CaptureSelect™ AAVX Affinity Resin as a solid-phase extraction (SPE) matrix and a layout that is compatible with workflow automation and low-volume sample processing. The integrated process takes ∼30 to 60 min depending on the input volume and comprises 22 consecutive steps, starting with crude lysate and ending with an AAV sample conditioned for downstream analysis or long-term storage. Numerical simulations, conducted as part of the design validation, reveal uniform flow across the SPE matrix for both in-plane and out-of-plane orientations. The performance of the system is demonstrated using AAV8 with volumes ranging from 50 to 500 μL. The on-chip purification process generally shows higher extraction efficiency than a benchmark procedure performed manually using spin columns. The capacity to remove plasmid DNA is validated using samples with calibrated spike-in concentration that were analyzed by qPCR, suggesting depletion of >99.99%. Biological activity of purified AAV particles was assessed in a transduction assay involving human embryonic kidney cells and recombinant AAV2 expressing green fluorescent protein. The platform's compatibility with low-volume production addresses gaps in current gene therapy manufacturing pipelines, underscoring its potential to streamline quality assurance protocols, reduce manual intervention, and accelerate scalable AAV production.

PurposeRapid and accurate diagnosis of Mycobacterium tuberculosis (MTB) in the patients with pulmonary tuberculosis (PTB) is essential for the patients management and infection control, particularly in cases with negative results from conventional bacteriological tests, as early detection enables timely initiation of treatment and interruption of disease transmission.Patients and MethodsWe evaluated the diagnostic performance of a dual-target droplet digital PCR (ddPCR) assay, targeting IS6110 and IS1081, using bronchoalveolar lavage fluid (BALF) samples collected from 506 hospitalized patients including 397 PTB and 109 non-PTB from August 2024 to December 2024. The performance of ddPCR was compared with that of MGIT960 culture, real-time quantitative PCR (qPCR), multi-color melting curve analysis (MMCA), and GeneXpert MTB/RIF (Xpert).ResultsThe ddPCR assay demonstrated a sensitivity of 89.7% (95% CI: 86.3–92.3%), specificity of 89.0% (95% CI: 81.5–94.0%), positive predictive value (PPV) of 96.7% (95% CI: 94.5–98.1%), and negative predictive value (NPV) of 70.3% (95% CI: 61.8–77.6%). The overall concordance rate was 89.5% (95% CI: 86.5–91.9%), with a Kappa value of 0.768 (95% CI: 0.712–0.824). The superior diagnostic performance of ddPCR was further supported by an area under the receiver operating characteristic (ROC) curve (AUC) of 0.893 (95% CI: 0.855–0.931). In a subgroup analysis of 216 bacteriologically negative PTB cases, ddPCR exhibited a sensitivity of 81.9%, which was significantly higher than that of Xpert (34.3%), MMCA (17.6%), and qPCR (31.0%) (all P < 0.001).ConclusionThe dual-target ddPCR assay targeting IS6110 and IS1081 offers a rapid, sensitive, and accurate method for the diagnosis of pulmonary tuberculosis. Its cost-competitiveness and manageable workflow make it a promising tool for clinical practice.

The goals of this project were to determine (1) what proportion of compounded sterile product investigational products (CSP IPs) stocked by our institution's investigational drug service have barcodes, and (2) whether CSP IPs that do have barcodes contain a number that is usable by our institution's IV workflow management system. A secondary goal was to determine sponsor willingness to embed a drug identifier in the barcodes. A survey of the site's inventory was conducted to identify CSP IPs. The vials were examined to determine if they contained a barcode, and if so, scanning was performed to determine what number was embedded in each barcode. In addition, sponsors of vials with barcodes were queried about whether they would be willing to embed a drug identifier, while sponsors with no barcodes were queried about their willingness to add barcodes. The study found that 66.9% of the CSP IP vials contained barcodes. However, the numbers embedded in the barcodes on most of these IPs varied from vial to vial. The structure of the numbers embedded in each barcode did not match the GS1-128 structure seen on FDA-approved products, and there was no usable drug identifier in any barcode in the opinion of the team's informatics pharmacist.

Cyclic peptide represents an attractive drug modality with its advantages in targeting proteins that were traditionally considered "undruggable". Molecular dynamics (MD) simulation remains a powerful tool for understanding peptide conformations and their interactions with the corresponding receptors in complex form. However, the preparation of macrocyclic peptide parameter files consisting of nonproteinogenic amino acids can be time-consuming and error prone. Here, we introduce an automatic workflow for the preparation of topology and force field parameter files for cyclic peptide systems. Partial charges and force field parameters of nonproteinogenic amino acids are automatically processed in a widely adopted rigorous way. Peptides spanning from mono- to tetra-cyclization were tested in bound and unbound states, which yielded valid parameter files and similar simulation results to those generated by different MD engines. Overall, the tool chain streamlines macrocyclic peptide MD setup while maintaining rigor and transferability across diverse cyclization patterns and simulation engines.

Human-in-the-loop governance of artificial intelligence in cardiology: From ethical principles to operational paradigms.

Modern organizations face increasing operational complexity due to fragmented tooling, manual task allocation, and reactive workflow management. Traditional systems rely on rigid rule-based automation, lacking predictive foresight and adaptive decision-making capabilities. This paper presents an AIBased Workflow &amp; Task Optimization System thatintegrates Machine Learning (ML), Natural Language Processing (NLP), and Predictive Process Monitoring (PPM) into a unified microservices architecture. The platform features an intelligent task assignment engine powered by XGBoost, a real-time Service Level Agreement (SLA) breach prediction module using gradient boosting classifiers, and a conversational NLP interface leveraging GPT-4. Developed with Django REST Framework, React.js, PostgreSQL, and Celery, the system supports horizontal scalability, eventdriven processing, and seamless third-party integration. Experimental evaluation demonstrates a weighted F1-score of 0.887 for task assignment, an AUC-ROC of 0.921 for SLA prediction, and a 52% reduction in workflow bottlenecks. User acceptance testing yielded an overall satisfaction score of 4.4/5. The system delivers enterprise-grade intelligent automation while maintaining costefficiency through open-source technologies.