Accelerating Early-Stage Clinical Research in the U.S.

Interindividual heterogeneity in Alzheimer’s disease (AD) remains poorly understood, as disparate single-cell studies leave it unclear whether findings reflect shared architecture or dataset-specific ...

Somatic mutations arise throughout life, with functional consequences tied to the cell populations in which they occur. Genome-wide studies measure somatic variations in bulk tissue, whereas single-ce...

Understanding how different cell types assemble into tissues and organs, as well as how they interact to transmit and receive biological signals, is essential for advancing biomedical and biological r...

The website URL remains www.ensembl.org. This updated website offers new scientific opportunities, as genomes and genesets are now available for a diverse array of species from across the tree of life...

The new website of the Ensembl project

Antimicrobial resistance (AMR) Portal updated with new search, genome browser and annotations

The U.S. Food and Drug Administration today issued a discussion paper on considerations for the regulation of generative artificial intelligence (GenAI)-enabled medical devices, seeking feedback from ...

Nature Biotechnology - Tech giants plunge into life sciences

Large language models embedded in autonomous agents process trusted instructions and untrusted data in one context window, leaving them open to direct and indirect prompt injection. In healthcare this is not hypothetical: a 2025 JAMA Network Open study found commercial medical LLMs followed injected instructions in 94.4% of simulated patient encounters, including life threatening recommendations . Yet the clinically decisive problem we quantify here is different. Most real clinical threats protected health information PHI exfiltration, cross patient access, bulk export, out of scope advice are fluent, legitimate looking requests that carry no attack signal, so even a state of the art injection detector passes them. Existing runtime guardrails trade safety against latency: model based auditors are accurate but add hundreds of milliseconds of Python inference, while lexical filters are fast but blind to obfuscated or semantically disguised payloads. We present QFIRE, an inline, provider

Blood-based biomarkers are increasingly used to investigate brain health, but collecting venous blood is difficult in remote and field settings. Capillary microsampling offers a practical alternative, although the ability to delay processing and its agreement with gold-standard venous blood require validation. We evaluated Tasso+, a minimally invasive upper-arm capillary blood collection system, for measuring neurological and host-response biomarkers in plasma and serum during an exercise-based protocol. Sampling occurred before, immediately after, and approximately 24-to-36 hours after exercise; Tasso+ samples were processed with or without a 72-hour room-temperature delay. Tasso+ samples were compared with matched venous blood, and Capitainer SEP10 dried plasma spots were also evaluated, using Quanterix Simoa and Alamar Biosciences NULISAseq CNS panel. Tasso+ enabled reliable measurement of several key biomarkers, including GFAP and NfL, even after delayed processing. These findings

Background: Endoscopic endonasal transsphenoidal surgery (EETS) requires navigation around neurocritical anatomy. Today, artificial intelligence clinical decision support systems (AI-CDSSs) can orientate surgeons, but clinician trust in AI remains unclear, limiting safe deployment. This study evaluates how modifiable design affects trust and performance in a real-world pituitary surgery AI-CDSS. Method: Online, 70 clinicians with pituitary surgery experience were randomised evenly to a Basic or Enhanced AI-CDSS which outline the sella on EETS operative video. The Enhanced group additionally received explanation of the model and previous publications, alongside confidence labels depicting outline reliability. Both groups annotated the sella on six video clips, first alone then with the optional AI-CDSS. Clips were ordered by declining AI performance, except for the final clip. Self-reported trust was measured using a 1-7 scale after each annotation, and performance was the DICE overlap

Background Endoscopic pituitary surgery involves navigating high-stakes anatomy where complications, such as carotid artery injury, cause devastating morbidity. While computer vision AI offers potential for real-time anatomical recognition to mitigate these risks, successful translation requires rigorous human-factors and performance evaluation. We present the iterative development and preclinical evaluation of a surgeon-controlled, real-time AI-assisted navigation system. Methods Guided by IDEAL Stage 0 and DECIDE-AI frameworks, the study was conducted in two phases. Phase 1 was an exploratory study where surgeons used the system during high-fidelity simulated surgery and provided feedback via "Think Aloud" protocols and surveys. Following prototype iteration, a Phase 2 randomized crossover comparative trial was conducted with 19 neurosurgeons (15 trainees, 4 experts) performing high-fidelity simulated tumour resections with and without AI assistance, separated by a minimum 2-week was

Background: Electrolyte replacement is ubiquitous in the acute care setting, but its familiarity cannot belie that even small dosing errors with potassium can cause lethal cardiac arrhythmias. Recently, MedAgentBench offered a benchmark for agentic artificial intelligence (AI) including the ability to correctly dose potassium based on a single rule; however, this does not adequately reflect the clinical complexity or safety concerns of an agent that has been used as the lethal injection. The purpose of this analysis was to a probe leaderboard large language model (LLM) capabilities to follow basic dosing rules to safely replace potassium in a series of clinician-annotated cases. Methods: Using a clinician panel, we developed a series of dosing principles and 20 clinical cases reflective of the complexity of potassium replacement. External clinicians were surveyed to assess practice variability and agreement to clinician panel answers. We tested GPT-5-chat with each case in triplicate,

The availability of multiple commercial short-read sequencing platforms necessitates systematic cross-platform performance comparisons, particularly for challenging applications such as low-frequency somatic mutation detection. Here, a large-scale targeted sequencing dataset from five Genome in a Bottle (GIAB) human genomic DNA reference standards, HG001 to HG005, alongside Twist Biosciences cfDNA reference standards featuring 1% variant allele frequency (VAF), was generated by six platforms (NovaSeq 6000, NovaSeq X, FASTASeq 300, GenoLab M, SURFSeq 5000, and MGISEQ-T7). To build a realistic benchmark while keeping authentic sequencing backgrounds, we developed PosMix, a simulating tool that generates position-specific VAFs. To overcome the limitations of conventional variant callers (high recall with poor precision for VarScan2, higher precision with lower recall for Strelka2/Mutect2), we developed SomaticXGB, a machine learning-based caller. In this study, SURFSeq 5000 consistently e

Gray-market retatrutide use is increasing, but patient safety experiences remain poorly characterized. This cross-sectional analysis examined Reddit posts and comments from retatrutide-specific and broader peptide or weight-management communities through December 2025. A validated large language model classified self-reported retatrutide use and extracted author-attributed symptoms mapped to MedDRA Preferred Terms. Among 13,589 users reporting current use, 7,823 had at least one mapped symptom after exclusions. Unlike phase 2 trial findings dominated by gastrointestinal events, Reddit reports most often described appetite increase, fatigue, increased energy, nausea, food craving, insomnia, and elevated heart rate. Findings are hypothesis-generating and warrant pharmacovigilance attention. ### Competing Interest Statement JST reports receiving an investigator-initiated grant, on behalf of the University of Pennsylvania, from Novo Nordisk and receiving consulting fees from Currax Pharm

Background: Digital antimicrobial stewardship (AMS) interventions, such as clinical decision support systems, audit and feedback platforms, and electronic prescribing tools, have been increasingly adopted to improve antibiotic use. However, the effectiveness of these interventions across healthcare settings remains uncertain, and the certainty of the evidence has not been comprehensively evaluated. The objective of this study was to provide a comprehensive understanding of the role of digital interventions in optimizing antimicrobial use and improving clinical outcomes within a broad spectrum of healthcare settings. Methods: We conducted a systematic review and meta-analysis of randomized controlled trials evaluating digital AMS interventions that followed PRISMA 2020 guidelines and registered in PROSPERO CRD420251178854 and funded by the Wellcome Trust CAMO Net programme. Searches were performed across major databases. Primary outcomes included the appropriateness of antibiotic prescr

Patient portal messaging has become a primary channel for asynchronous clinical communication, it spans a wide range of content, from symptom reports and medication concerns to administrative requests. Despite this volume and diversity, there is no formal representation for what a portal message contains: no vocabulary for the clinical and administrative events it describes, or for the attributes of those events that the patient has actually disclosed. Without such a representation, it is difficult to systematically analyze portal communication, assess message completeness, or build downstream tools that depend on structured input, such as automated triage, response drafting, and follow-up question generation. A clinical event schema, grounded in real portal messages and reviewed by clinicians, would provide this missing foundation. We introduce a clinical event ontology for patient portal messages, containing 8 event types and 70 roles that span clinical content (symptoms, medications

npj Digital Medicine - AI for predicting exacerbations in KIDs with asthma (AIRE-KIDS)

npj Digital Medicine - A clinical neuroimaging platform for rapid, automated lesion detection and personalized post-stroke outcome prediction

npj Digital Medicine - Automated evaluation can distinguish the good and bad AI responses to patient questions about hospitalization

npj Digital Medicine - Detecting dataset bias in medical AI using a generalized and modality agnostic auditing approach

npj Digital Medicine - Feasibility and impact of a large language model pipeline for surgical trial abstracts

Background Prediction of long-term neurodevelopmental outcomes remains challenging after perinatal asphyxia. Here, we studied whether computational metrics of brain function derived from neonatal EEG are associated with long-term neurodevelopment in infants with perinatal asphyxia. Methods Total of 36 term-born infants with perinatal asphyxia with or without hypoxic-ischemic encephalopathy were studied with neonatal multichannel electroencephalography (EEG). We computed local EEG amplitudes and phase-amplitude coupling (PAC), as well as large-scale functional cortical networks estimated using amplitude-amplitude correlations (AAC) and phase-phase correlations (PPC). These EEG-derived markers were tested for associations with neurodevelopmental outcomes at two years, assessed using the Griffiths Scales of Child Development, 3rd edition (GMDS-III). Results EEG amplitudes showed positive associations with GMDS-III Foundations of Learning and General Development scores across most electr

Myelin forms an insulating sheath around axons enabling both rapid and energy-efficient conduction of action potentials and myelin abnormalities or loss can lead to severe motor, sensory, and cognitive impairment. While electron microscopy can resolve multiple axonal components that are affected myelin, their large-scale quantitative analysis is both difficult and time consuming. To overcome such limitations, we developed a machine learning framework that automatically recognizes and quantifies multiple features of axons and myelin including axonal mitochondrial density and periaxonal area. Applying that framework to fibers in the spinal cord of variably hypomyelinated mice, we show here that reduction in the thickness and length of myelin sheaths results in correlating changes in mitochondrial density and periaxonal area. The machine learning framework introduced here should contribute to future insight into the axon, myelin, and mitochondrial relationships that change during neurolog

Spousal caregivers of individuals with Alzheimer’s disease and related dementias frequently experience elevated perceived stress, caregiver burden, and loneliness, which are associated with adverse health outcomes. Early identification is therefore critical for timely intervention. Existing approaches commonly rely on wearable sensor data and standardized psychological questionnaires, while recent multimodal methods aim to improve prediction by integrating behavioral and linguistic information. In this study, we explored three modality configurations, wearable-derived features, interview-based text, and their combination, to classify caregiver psychological risk using the Perceived Stress Scale (PSS), Zarit Burden Interview, and UCLA Loneliness Scale. We compared traditional machine learning models and large language models (LLMs) (Gemini 2.0, Llama 4, and GPT-4o) under psychometrician-centered and caregiver-centered prompting strategies. Traditional machine learning models performed

Purpose To predict retinal nerve fiber layer thickness (RNFLT) norms from fundus images. Methods We selected 18,000 OCT scans and visual fields (VF) from the Massachusetts Eye and Ear Glaucoma Service. A U-Net-based deep learning model was developed to predict RNFLT norms from OCT en face fundus images. A total of 10,000 OCT scans with normal VFs (mean deviation [MD] ≥ -1 dB, glaucoma hemifield test within normal limits, and pattern standard deviation probability > 5%) tested within 30 days were used for training, while the remaining 8,000 OCT scans (mean VF MD: −3.3 ± 4.9 dB), including 2,419 scans with normal VFs, were used for evaluation. Structure-function correlations between RNFLT maps and VFs were assessed using linear regression and VGG-16 across original RNFLT maps, deviation maps, and their combination. Performance was evaluated using correlation coefficients, mean absolute error (MAE), and R2. Results Predicted RNFLT norm maps showed agreement with baseline RNFLT maps in e

Amyotrophic lateral sclerosis (ALS) is a progressive neurodegenerative disease affecting more than 450,000 individuals worldwide and is frequently diagnosed more than 12 months after symptom onset, delaying intervention during a critical early window. Because up to 80% of patients develop dysarthria within two years, subtle changes in speech provide a signal of early bulbar motor neuron degeneration. However, existing speech-based systems rely on supervised classification trained on limited datasets, achieving moderate sensitivity and depending heavily on labeled disease examples, which restrict scalability and early detection. This study introduces SPEAK-NORM, the first-ever normative speech modeling framework for early ALS diagnosis, which learns age- and sex-conditioned motor-speech distributions exclusively from healthy individuals. A conditional variational autoencoder models coordination of hypoglossal, laryngeal, and respiratory motor pathways, and deviation from this healthy ma

Objective: Hispanic/Latinx populations in the U.S. experience higher rates of chronic disease linked to physical inactivity, yet digital health interventions remain largely inaccessible to more than 16 million Hispanic/Latinx adults with limited English proficiency. While large language models (LLMs) offer scalable personalization, their use in non-English behavioral coaching is unexplored. This study introduces MHC-Coach-ES, a Spanish-language LLM fine-tuned on the Transtheoretical Model (TTM) of behavior change. Materials and Methods: We fine-tuned Llama 3-70B-Instruct using a two-stage pipeline. First, the model was adapted to Spanish health and motivational language using a 2.21-million-token corpus. Second, it was instruction-tuned on 3,268 translated human written messages to align the model with the Transtheoretical Model (TTM) of Behavioral Change. We compared MHC-Coach-ES with Llama 3-70B-Instruct and translated human-expert messages using a forced-choice preference survey (N

Objective. Biomedical knowledge graphs (KGs) such as PrimeKG, Hetionet, UMLS, and PharmGKB are increasingly used as the substrate for downstream machine-learning, retrieval-augmented generation, drug-repurposing, and electronic health record (EHR) augmentation pipelines. The dominant assumption in published work is that integrating two or more such KGs is a tractable engineering step solved by identifier (ID) matching. This paper interrogates that assumption empirically. We quantify how much concept overlap survives realistic alignment, and we characterize the new failure modes introduced by the methods that practitioners reach for when ID matching is insufficient. Materials and Methods. We compared four widely used biomedical KGs (PrimeKG, Hetionet v1.0, the full UMLS Metathesaurus, and PharmGKB) across eleven node types using a tiered alignment pipeline: (1) direct ID matching for nodes sharing a primary vocabulary; (2) cross-ontology bridging using standard mappings (e.g., MONDO-DOI

The Epic Sepsis Model version 2 (ESMv2) is a prediction model embedded into the electronic medical record used to warn clinicians which hospitalized patients are at risk for sepsis. We conducted a retrospective cohort study of 31,951 hospitalizations of 25,760 patients to compare analyses conducted at the commonly used patient-level (where a maximum prediction prior to the onset of sepsis is used to measure performance) vs novel prediction-level (where each prediction is used to measure performance). Sepsis, defined by the Sepsis 3 criteria occurred during 1,049 hospitalizations (3.3%). Patient-level analyses suggested excellent discrimination AUC 0.86; [IQR 0.85, 0.87], whereas prediction-level analyses demonstrated lower performance AUC 0.62; [IQR 0.57, 0.65]. Low estimates of the positive predictive value (14.5% at the patient level vs 4% at the prediction level) imply a high number of false alerts. Common evaluation approaches may overstate the performance of dynamic prediction mod

Background Artificial intelligence-enhanced electrocardiography (AI-ECG) enables scalable, low-cost cardiac dysfunction screening, but existing models are annotation-intensive and predominantly adult-derived, leaving paediatric generalizability uncertain. Paediatric cohorts exhibit highly variable cardiac morphology and function compared to adults, which may be useful for learning generalizable AI-ECG models. Methods We pretrained ECG-Fyler on a predominantly paediatric, all-age cohort at Boston Children's Hospital (1992-2023), annotated with a cardiology-specific coding system (Fyler codes), and evaluated it on assessments from echocardiography (echo) and cardiac magnetic resonance (CMR) studies. We validated on an external adult cohort from Columbia University Irving Medical Center. Performance was benchmarked against several AI-ECG foundation models by AUROC across age groups, lesion types, and limited-data scenarios. Findings The pretraining cohort comprised 782,138 ECGs from 255,2

The surge in medical imaging has spurred the development of vision-language models (VLMs) to alleviate radiologist workloads. However, clinical deployment is hindered by the lack of meaningful evaluation frameworks. Current metrics - ranging from semantic similarity to large language model (LLM) based judges - often fail to distinguish between clinically trivial and critical discrepancies, poorly reflecting real-world clinical judgment. To address this, we introduce DISCERN (Discordance and Significance-aware Entity-level Radiology Report Comparison). DISCERN is a significance-aware framework that weighs report errors based on their potential impact on patient care. Our results demonstrate that DISCERN powered by closed source LLMs aligns more closely with expert radiologist assessments than traditional metrics or current LLM evaluators, providing a more interpretable and clinically relevant benchmark. By modeling radiologist prioritization and entity-level feedback, DISCERN facilitate

Neurodegenerative diseases such as Parkinson disease (PD) result from complex interactions between neuronal stress and the surrounding tissue environment, yet the determinants that govern this interplay remain incompletely understood. While neuronal responses to mitochondrial dysfunction and proteotoxic stress have been extensively characterized, the contribution of microglial state to disease progression remains unclear. Here, we identify the transcription factor Foxo3 as a key regulator of dopaminergic vulnerability acting predominantly through microglia-, rather than neuron-intrinsic, mechanisms. Foxo3 was rapidly induced and translocated to the nucleus in dopaminergic-like cells in response to mitochondrial complex I inhibition and alpha-synuclein aggregation, indicating activation of a conserved neuronal stress response. However, neuron-specific deletion of Foxo3 attenuated early Parkinson-like transcriptional signatures but did not confer sustained in vivo neuroprotection. In con

Mitochondria are essential for cellular energy production and are particularly critical for brain development and function. Neurons rely predominantly on oxidative phosphorylation for energy production, rendering the brain highly vulnerable to mitochondrial dysfunction. Consequently, impaired mitochondrial function contributes to a broad spectrum of neurological and systemic disorders, making mitochondria attractive therapeutic targets. Despite this importance, there is currently no non-invasive, spatially resolved method to assess mitochondrial function in the intact living brain. Here, we establish a non-invasive functional MRI approach--4D Oxy-wavelet MRI--to probe in vivo mitochondrial electron transport chain (ETC) function in a spatially specific manner across the lifespan, from fetal to adult brains. This method employs a low-rank k-t sub-Nyquist acquisition strategy to achieve simultaneous structural and functional imaging with high spatial (78 m) and temporal ([~]14 ms) resol

CRISPR screens with single-cell RNA-seq readouts provide a powerful tool for characterizing the functions of noncoding elements and genes. However, designing these experiments to balance statistical power and cost is challenging, given the large number of design parameters. The only available tool for this purpose is a simulation-based power calculator, but it is computationally costly and requires high-performance computing to run. We derive a novel analytical formula for the power to detect perturbation-expression associations, recapitulating power estimates from the simulation-based tool while reducing runtime by up to seven orders of magnitude. This acceleration unlocks the possibility of interactive single-cell CRISPR screen design. [A]ccordingly, we develop PerturbPlan, an interactive web application built on the analytical power formula. PerturbPlan helps users address 11 design questions for two types of single-cell CRISPR screens, Perturb-seq and targeted Perturb-seq (TAP-seq)

In biomedical scientific discovery, synthesizing prior knowledge from the literature is an essential component of interpreting numerical omics data analyses for disease target identification and drug discovery. Large language models (LLMs) alone can rapidly retrieve disease mechanisms from biomedical text, but text-only outputs are general and unreliable for target and drug prioritization without cohort-specific quantitative evidence. Herein, we propose a provenance-aware Text-to-Target framework that couples schema-constrained multi-model LLM retrieval with numeric omics data analysis. The key design is a modality-aware fusion step: candidates are partitioned into overlap-supported anchors, retrieval-only hidden hubs, and network-emergent novelty nodes, then propagated into staged hypothesis and strategy generation under topology constraints. We evaluate the model in Alzheimer's disease (AD) and pancreatic ductal adenocarcinoma (PDAC). In PDAC, the workflow produced a balanced 75-gene

Predicting cognitive processes from brain activation maps has remained an open question within the neuroscience community for many years. Meta-analytic functional decoding methods aim to tackle this issue by providing a quantitative estimation of behavioral profiles associated with specific brain regions. Existing methods face intrinsic challenges in neuroimaging meta-analysis, particularly in consolidating textual information from publications, as they rely on limited metrics that do not capture the semantic context of the text. The combination of large language models (LLMs) with advanced deep contrastive learning models (e.g., CLIP) for aligning text with images has revolutionized neuroimaging meta-analysis, potentially offering solutions to functional decoding challenges. In this work, we present NiCLIP, a contrastive language-image pretrained model that predicts cognitive tasks, concepts, and domains from brain activation patterns. We leveraged over 23,000 neuroscientific articles

Adaptive learning requires distinguishing volatility, changes in the latent state of the environment, from moment-to-moment stochasticity of observations. The two demand opposite adjustments to the learning rate: volatility calls for faster updating, stochasticity for slower. Disentangling them is computationally difficult because both inflate experienced variance, leaving the inference prone to systematic individual differences with potential consequences for psychopathology. Three computational phenotypes capture this variation: intact learners; stochasticity-blind learners, who over-update by treating noise as change; and volatility-blind learners, who under-update by treating change as noise. In two large online samples and across three tasks, we found a double dissociation between these phenotypes and transdiagnostic psychiatric dimensions: stochasticity-blind learners scored higher on Internalizing (anxiety, depression), volatility-blind learners on Externalizing (behavioral addi

Predictive modelling is important for health data analysis and data-driven clinical decision-making. However, predictive studies are challenging to design optimally by hand when tens or even hundreds of features require selection, transformation, or interaction modelling. While complex machine learning models offer high performance, their "black-box" nature limits the clinical trust, transparency, and interpretability required for decision-making. We developed and evaluated an Exploratory AI Recommender that provides data-driven recommendations to improve predictive performance of existing interpretable statistical models. The developed framework uses flexible AI modelling to capture complex data patterns and explainable AI techniques to translate the patterns into three recommendation types: feature exclusion, non-linear terms, and feature interactions. We evaluated the framework by comparing predictive performance of a baseline (i.e., no interactions or non-linear terms) Cox Proporti

AI-driven chatbots have been utilized in healthcare to automate administrative tasks, improve patient education, and expand access to medical information; however, their role in genetic counseling remains underexplored. To investigate the adoption, perceptions, and potential utility of AI-based chatbots in genetic counseling practice, 217 genetic counselors and genetic counseling students from across North America were surveyed regarding chatbot usage, confidence in their application, and perceived benefits and limitations. While most participants (166/217; 76.5%) reported using general AI chatbots outside of clinical settings, far fewer (18/204; 8.8%) reported using or recommending clinical genetics chatbots in clinical practice. For those that used clinical genetics chatbots, the primary purpose was for communication with at-risk family members (11/18; 61.1%) and patient education (10/18; 55.6%). Confidence in chatbot technology varied, with highest confidence in gathering family his

Today, advancements in our understanding of cancer biology are increasingly attributed to large-scale clinical-molecular datasets. The case in point for multiple myeloma, the second-most prevalent haematological malignancy, is the CoMMpass study, a dataset with the paired clinical and sequencing data of 1,143 patients. Given its complexity, the multi-omics data of CoMMpass demands programming skills which imposes a hurdle for experimental myeloma researchers who want to validate their hypotheses on population data. The rise of agentic AI over the past few years presents unparalleled opportunities to bridge this technical gap. We propose MyeGPT (Myeloma Generative Pretrained Transformer), an AI bioinformatician for multiple myeloma that relies on the CoMMpass dataset as its ground truth. MyeGPT converts natural language queries such as 'What are the characteristics of patients who relapse after induction therapy' or 'Compare the overall survival of high vs normal NSD2 expression' into d

Earlier ICU mortality prediction is more clinically useful because it can identify high-risk patients while treatment decisions can still change. Yet most models are trained on data from a fixed time window, so it is unclear whether a model trained on the first 48 hours of ICU data remains reliable when used earlier in the ICU stay. We evaluated a multimodal ICU mortality model trained once at 48 hours and then applied unchanged at 6, 12, 24, and 48 hours on MIMIC-III. The model combines an LSTM for physiological time-series data, a finetuned ClinicalModernBERT model for clinical notes, and a logistic regression fusion layer. Performance remained strong at earlier time points, suggesting that useful mortality prediction is possible earlier in the ICU stay even without retraining. At 6 hours, the model achieved AUROC 0.777 and remained well-calibrated (expected calibration error, ECE 0.038) without any recalibration, and it outperformed both single-modality models at every horizon. The

Background Structuring oncology clinical notes into registry-grade variables is essential for research and care but remains labour-intensive and error-prone. Objective To develop and evaluate a privacy-preserving large language model pipeline for oncology registry abstraction in a real-world clinical setting. Methods We deployed an open-source Meta Llama 3.3 70B–based pipeline to extract over 50 variables from 6,700 oncology notes at a cancer centre in Singapore. Data were de-identified locally using a Hide-In-Plain-Sight approach, ensuring no identifiable data left hospital infrastructure. Performance was assessed on 200 randomly sampled notes with adjudicated ground truth. A structure-aware framework classified outputs as correct, missing, spurious, or incorrect. Results F1 scores were high across variables, including diagnosis (97.2%), histology (95.8%), stage (92.6%), biomarkers (91.4%), and treatments (88.1%). Transferability testing on 50 external notes showed strong performan

Background: Structured phenotypic annotations linked to genetic data can drive diagnostic insight and therapeutic discovery in complex diseases. However, poor research access to the rich clinical data trapped in unstructured clinical records remains a significant barrier to phenotype-genotype integration. Here, we present Human Phenotype Ontology (HPO) Mapper, a scalable AI-assisted tool designed to ingest semantically structured clinical findings paired with anatomical region and accurately map them to HPO terms and associated genes. Results: We applied HPO Mapper to two forms of standardised clinical input extracted from inflammatory bowel disease (IBD) patient records. The first data type consisted of paired 'clinical findings + anatomical regions' derived from unstructured clinical reports and the second was standardised ICD-10 code-derived phenotypes. HPO Mapper achieved high semantic alignment and mapping accuracy for both data types (F1 = 0.85 ± 0.05 and 0.84 ± 0.03, respectivel

Therapeutics fused to brain shuttles that exploit endogenous receptor-mediated transport at the blood brain barrier (BBB) offer a promising strategy to deliver large molecule drugs and biologics to the CNS. A fundamental but untested assumption underlying their clinical development is that their endothelial receptor targets are consistently expressed between individuals and between patient populations. Here, we analyzed gene and protein expression of eleven canonical brain shuttle targets in isolated human brain microvascular endothelial cells and brain microvessels from 11 large cohorts, using single-cell and single-nucleus transcriptomics and quantitative proteomics. Expression was remarkably stable between brain regions, sexes, ages, and normal health versus four major neurodegenerative conditions, Alzheimer's disease, Parkinson's disease, Huntington's disease, and amyotrophic lateral sclerosis, with 612 of 631 comparisons (97%) showing no significant difference. Regional heterogene

npj Digital Medicine - Machine learning based hepatic safety score predicts decompensation in hepatocellular carcinoma systemic therapy