arXiv subjects

Blanca Gallego

Publications and source records attributed to Blanca Gallego.

At least 19 recordsLinked to original sources

arXiv2026

Synthetic but Not Realistic: The Evaluation Challenge in Generative Modelling for Structured Electronic Medical Records

Synthetic healthcare data are widely proposed as privacy-preserving substitutes for real patient data, yet their evaluation remains dominated by statistical similarity and predictive performance that do not reflect clinical validity. We introduce a multi-dimensional evaluation framework grounded in epidemiology, assessing descriptive fidelity, clinical utility, and structural validity, corresponding to descriptive, predictive, and causal questions. We evaluate four representative generative paradigms - GAN-based, VAE-boosted, diffusion-based, and masked modelling - using PRIME-CVD, a 50,000-person cohort with known ground-truth structure. While all models reproduce marginal distributions, none simultaneously preserve subgroup structure, effect estimates, and dependency structure. Notably, models with strong distributional fidelity can exhibit poor calibration and distorted relationships, leading to unreliable inference. These results show that current evaluation practices can overestimate synthetic data quality and motivate domain-informed assessment based on the ability to support valid clinical and scientific conclusions.

cs.LG↗

arXiv2026

From Hazard Functions to Language Space: Cox-Supervised Distillation of Survival Risk into a Large Language Model

We investigate whether information about time-to-event risk estimated by a Cox proportional hazards model can be transferred into a generative large language model. We propose a text-based survival modelling pipeline in which structured clinical covariates are converted into text prompts and a Qwen-based large language model is fine-tuned to generate patient-specific survival risk using Cox model predictions as a training target. Across GBSG2, ACTG320, and WHAS500, the model achieves competitive held-out discrimination and calibration despite being trained as a text-generation task rather than with a conventional survival-analysis loss. We further analyse the geometry of the model's hidden states, where t-SNE visualisations reveal smooth risk gradients in latent space, suggesting that the model represents survival risk as a continuous structure rather than isolated risk categories. Together, these findings suggest that large language models can internalise survival-risk structure while supporting calibrated prediction, providing a route towards time-to-event reasoning in language models.

cs.LG↗

arXiv2026

CUICurate: A GraphRAG-based Framework for Automated Clinical Concept Curation for NLP applications

Background: Clinical named entity recognition tools commonly map free text to Unified Medical Language System (UMLS) Concept Unique Identifiers (CUIs). For many downstream tasks, however, the clinically meaningful unit is not a single CUI but a concept set comprising related synonyms, subtypes, and associated concepts. Constructing these sets is labour-intensive, inconsistently performed, and poorly supported by existing tools. Methods We present CUICurate, a graph-based retrieval-augmented generation (GraphRAG) framework for automated UMLS concept set curation. A UMLS knowledge graph (KG) was constructed and embedded for semantic retrieval. Candidate CUIs were retrieved using graph-based expansion and then filtered and classified using large language models (GPT-5 and Qwen3-32B). The framework was evaluated on five lexically heterogeneous clinical concepts against a manually curated concept sets and gold-standard concept sets. Results CUICurate produced substantially larger and more complete concept sets than the manual benchmarks. A single retrieval configuration across concepts achieved high recall of definitive concepts with manageable candidate sets. GPT-5 outperformed manual curation for all concepts and retained at least 95% of definitive gold-standard CUIs, while Qwen3-32B achieved comparable but slightly lower performance. Many missed concepts were not observed in 10,000 MIMIC-III notes. CUICurate infrastructure and end-to-end processing was inexpensive and stable across runs. Conclusions CUICurate offers a scalable, reproducible and cost-efficient approach for generating clinician-reviewable UMLS concept sets tailored to clinical natural language processing and phenotyping applications.

cs.CL↗

arXiv2026

PRIME-CVD: A Parametrically Rendered Informatics Medical Environment for Education in Cardiovascular Risk Modelling

In recent years, progress in medical informatics and machine learning has been accelerated by the availability of openly accessible benchmark datasets. However, patient-level electronic medical record (EMR) data are rarely available for teaching or methodological development due to privacy, governance, and re-identification risks. This has limited reproducibility, transparency, and hands-on training in cardiovascular risk modelling. Here we introduce PRIME-CVD, a parametrically rendered informatics medical environment designed explicitly for medical education. PRIME-CVD comprises two openly accessible synthetic data assets representing a cohort of 50,000 adults undergoing primary prevention for cardiovascular disease. The datasets are generated entirely from a user-specified causal directed acyclic graph parameterised using publicly available Australian population statistics and published epidemiologic effect estimates, rather than from patient-level EMR data or trained generative models. Data Asset 1 provides a clean, analysis-ready cohort suitable for exploratory analysis, stratification, and survival modelling, while Data Asset 2 restructures the same cohort into a relational, EMR-style database with realistic structural and lexical heterogeneity. Together, these assets enable instruction in data cleaning, harmonisation, causal reasoning, and policy-relevant risk modelling without exposing sensitive information. Because all individuals and events are generated de novo, PRIME-CVD preserves realistic subgroup imbalance and risk gradients while ensuring negligible disclosure risk. PRIME-CVD is released under a Creative Commons Attribution 4.0 licence to support reproducible research and scalable medical education.

cs.LG↗

arXiv2025

Limits of Generative Pre-Training in Structured EMR Trajectories with Irregular Sampling

Foundation models refer to architectures trained on vast datasets using autoregressive pre-training from natural language processing to capture intricate patterns and motifs. They were originally developed to transfer such learned knowledge to downstream predictive tasks. Recently, however, some studies repurpose these learned representations for phenotype discovery without rigorous validation, risking superficially realistic but clinically incoherent embeddings. To test this mismatch, we trained two autoregressive models -- a sequence-to-sequence LSTM and a reduced Transformer -- on longitudinal ART for HIV and Acute Hypotension datasets. Controlled irregularity was added during training via random inter-visit gaps, while test sequences stayed complete. Patient-trajectory synthesis evaluated distributional and correlational fidelity. Both reproduced feature distributions but failed to preserve cross-feature structure -- showing that generative pre-training yields local realism but limited clinical coherence. These results highlight the need for domain-specific evaluation and support trajectory synthesis as a practical probe before fine-tuning or deployment.

cs.LG↗

arXiv2025

Attention-Based Synthetic Data Generation for Calibration-Enhanced Survival Analysis: A Case Study for Chronic Kidney Disease Using Electronic Health Records

Access to real-world healthcare data is limited by stringent privacy regulations and data imbalances, hindering advancements in research and clinical applications. Synthetic data presents a promising solution, yet existing methods often fail to ensure the realism, utility, and calibration essential for robust survival analysis. Here, we introduce Masked Clinical Modelling (MCM), an attention-based framework capable of generating high-fidelity synthetic datasets that preserve critical clinical insights, such as hazard ratios, while enhancing survival model calibration. Unlike traditional statistical methods like SMOTE and machine learning models such as VAEs, MCM supports both standalone dataset synthesis for reproducibility and conditional simulation for targeted augmentation, addressing diverse research needs. Validated on a chronic kidney disease electronic health records dataset, MCM reduced the general calibration loss over the entire dataset by 15%; and MCM reduced a mean calibration loss by 9% across 10 clinically stratified subgroups, outperforming 15 alternative methods. By bridging data accessibility with translational utility, MCM advances the precision of healthcare models, promoting more efficient use of scarce healthcare resources.

cs.LG↗

arXiv2024

Masked Clinical Modelling: A Framework for Synthetic and Augmented Survival Data Generation

Access to real clinical data is often restricted due to privacy obligations, creating significant barriers for healthcare research. Synthetic datasets provide a promising solution, enabling secure data sharing and model development. However, most existing approaches focus on data realism rather than utility -- ensuring that models trained on synthetic data yield clinically meaningful insights comparable to those trained on real data. In this paper, we present Masked Clinical Modelling (MCM), a framework inspired by masked language modelling, designed for both data synthesis and conditional data augmentation. We evaluate this prototype on the WHAS500 dataset using Cox Proportional Hazards models, focusing on the preservation of hazard ratios as key clinical metrics. Our results show that data generated using the MCM framework improves both discrimination and calibration in survival analysis, outperforming existing methods. MCM demonstrates strong potential to support survival data analysis and broader healthcare applications.

cs.LG↗

arXiv2024

CK4Gen: A Knowledge Distillation Framework for Generating High-Utility Synthetic Survival Datasets in Healthcare

Access to real clinical data is heavily restricted by privacy regulations, hindering both healthcare research and education. These constraints slow progress in developing new treatments and data-driven healthcare solutions, while also limiting students' access to real-world datasets, leaving them without essential practical skills. High-utility synthetic datasets are therefore critical for advancing research and providing meaningful training material. However, current generative models -- such as Variational Autoencoders (VAEs) and Generative Adversarial Networks (GANs) -- produce surface-level realism at the expense of healthcare utility, blending distinct patient profiles and producing synthetic data of limited practical relevance. To overcome these limitations, we introduce CK4Gen (Cox Knowledge for Generation), a novel framework that leverages knowledge distillation from Cox Proportional Hazards (CoxPH) models to create synthetic survival datasets that preserve key clinical characteristics, including hazard ratios and survival curves. CK4Gen avoids the interpolation issues seen in VAEs and GANs by maintaining distinct patient risk profiles, ensuring realistic and reliable outputs for research and educational use. Validated across four benchmark datasets -- GBSG2, ACTG320, WHAS500, and FLChain -- CK4Gen outperforms competing techniques by better aligning real and synthetic data, enhancing survival model performance in both discrimination and calibration via data augmentation. As CK4Gen is scalable across clinical conditions, and with code to be made publicly available, future researchers can apply it to their own datasets to generate synthetic versions suitable for open sharing.

cs.LG↗

arXiv2023

Continuous Treatment Recommendation with Deep Survival Dose Response Function

We propose a general formulation for continuous treatment recommendation problems in settings with clinical survival data, which we call the Deep Survival Dose Response Function (DeepSDRF). That is, we consider the problem of learning the conditional average dose response (CADR) function solely from historical data in which observed factors (confounders) affect both observed treatment and time-to-event outcomes. The estimated treatment effect from DeepSDRF enables us to develop recommender algorithms with the correction for selection bias. We compared two recommender approaches based on random search and reinforcement learning and found similar performance in terms of patient outcome. We tested the DeepSDRF and the corresponding recommender on extensive simulation studies and the eICU Research Institute (eRI) database. To the best of our knowledge, this is the first time that causal models are used to address the continuous treatment effect with observational data in a medical context.

stat.ML↗

arXiv2023

Predicting adverse outcomes following catheter ablation treatment for atrial fibrillation

Objective: To develop prognostic survival models for predicting adverse outcomes after catheter ablation treatment for non-valvular atrial fibrillation (AF). Methods: We used a linked dataset including hospital administrative data, prescription medicine claims, emergency department presentations, and death registrations of patients in New South Wales, Australia. The cohort included patients who received catheter ablation for AF. Traditional and deep survival models were trained to predict major bleeding events and a composite of heart failure, stroke, cardiac arrest, and death. Results: Out of a total of 3285 patients in the cohort, 177 (5.3%) experienced the composite outcome (heart failure, stroke, cardiac arrest, death) and 167 (5.1%) experienced major bleeding events after catheter ablation treatment. Models predicting the composite outcome had high risk discrimination accuracy, with the best model having a concordance index > 0.79 at the evaluated time horizons. Models for predicting major bleeding events had poor risk discrimination performance, with all models having a concordance index < 0.66. The most impactful features for the models predicting higher risk were comorbidities indicative of poor health, older age, and therapies commonly used in sicker patients to treat heart failure and AF. Conclusions: Diagnosis and medication history did not contain sufficient information for precise risk prediction of experiencing major bleeding events. The models for predicting the composite outcome have the potential to enable clinicians to identify and manage high-risk patients following catheter ablation proactively. Future research is needed to validate the usefulness of these models in clinical practice.

cs.LG↗

arXiv2023

The Cardiac Analytics and Innovation (CardiacAI) Data Repository: An Australian data resource for translational cardiovascular research

In Australia, cardiovascular diseases (CVD) are managed in a complex and fragmented healthcare system across multiple providers. A data repository that links data sources, and enables advanced analytics and big data technologies, will generate novel insights, and allow development of translational tools that can improve patient care and outcomes. The Cardiac Analytics and Innovation (CardiacAI) project has established a research-ready electronic medical records data resource to enable collaborative and translational cardiovascular research. The CardiacAI data repository prospectively extracts de-identified electronic medical record (EMR) data from two local health districts (LHD) in New South Wales (NSW), Australia. These data are linked with Australian population health data to ascertain longitudinal hospitalisation and death outcomes. The data are stored within a secure, cloud-based storage and analytics platform. The CardiacAI data repository is a not-for-profit data resource that promotes collaboration and responsible sharing of data. The CardiacAI data repository is a resource for Australian healthcare providers, clinicians and researchers seeking to improve cardiovascular care. The project is expanding to include data from stroke hospitalisations and two additional NSW LHDs, and is actively exploring linkage with ECG signal data, medical imaging data and community-based healthcare. The CardiacAI project has the potential to unlock a wealth of novel insights and translational tools that improve secondary prevention and treatment of CVD.

cs.DL↗

arXiv2022

Causal inference for observational longitudinal studies using deep survival models

Causal inference for observational longitudinal studies often requires the accurate estimation of treatment effects on time-to-event outcomes in the presence of time-dependent patient history and time-dependent covariates. To tackle this longitudinal treatment effect estimation problem, we have developed a time-variant causal survival (TCS) model that uses the potential outcomes framework with an ensemble of recurrent subnetworks to estimate the difference in survival probabilities and its confidence interval over time as a function of time-dependent covariates and treatments. Using simulated survival datasets, the TCS model showed good causal effect estimation performance across scenarios of varying sample dimensions, event rates, confounding and overlapping. However, increasing the sample size was not effective in alleviating the adverse impact of a high level of confounding. In a large clinical cohort study, TCS identified the expected conditional average treatment effect and detected individual treatment effect heterogeneity over time. TCS provides an efficient way to estimate and update individualized treatment effects over time, in order to improve clinical decisions. The use of a propensity score layer and potential outcome subnetworks helps correcting for selection bias. However, the proposed model is limited in its ability to correct the bias from unmeasured confounding, and more extensive testing of TCS under extreme scenarios such as low overlapping and the presence of unmeasured confounders is desired and left for future work.

stat.ML↗

arXiv2022

Assessing the effectiveness of empirical calibration under different bias scenarios

Background: Estimations of causal effects from observational data are subject to various sources of bias. One method of adjusting for the residual biases in the estimation of a treatment effect is through negative control outcomes, where the treatment does not affect the outcome. The empirical calibration procedure is a technique that uses negative controls to calibrate p-values. An extension of empirical calibration calibrates the coverage of the 95% confidence interval of a treatment effect estimate by using negative control outcomes as well as positive control outcomes (where treatment affects the outcome). Methods: The effect of empirical calibration of confidence intervals was analyzed using simulated datasets with known treatment effects. The simulations consisted of binary treatment and binary outcome, with biases resulting from unmeasured confounder, model misspecification, measurement error, and lack of positivity. The performance of the empirical calibration was evaluated by determining the change in the coverage of the confidence interval and the bias in the treatment effect estimate. Results: Empirical calibration increased coverage of the 95% confidence interval of the treatment effect estimate under most bias scenarios but was inconsistent in adjusting the bias in the treatment effect estimate. Empirical calibration of confidence intervals was most effective when adjusting for the unmeasured confounding bias. Suitable negative controls had a large impact on the adjustment made by empirical calibration, but small improvements in the coverage of the outcome of interest were also observable when using unsuitable negative controls.

stat.ME↗

arXiv2022

TreatmentEstimatoR: a Dashboard for Estimating Treatment Effects from Observational Health Data

Observational health data can be leveraged to measure the real-world use and potential benefits or risks of existing medical interventions. However, lack of programming proficiency and advanced knowledge of causal inference methods excludes some clinicians and non-computational researchers from performing such analyses. Code-free dashboard tools provide accessible means to estimate and visualize treatment effects from observational health data. We present TreatmentEstimatoR, an R Shiny dashboard that facilitates the estimation of treatment effects from observational data without any programming knowledge required. The dashboard provides effect estimates from multiple algorithms simultaneously and accommodates binary, continuous, and time-to-event outcomes. TreatmentEstimatoR allows for flexible covariate selection for treatment and outcome models, comprehensive model performance metrics, and an exploratory data analysis tool. TreatmentEstimatoR is available at https://github.com/CollinSakal/TreatmentEstimatoR. We provide full installation instructions and detailed vignettes for how to best use the dashboard.

stat.ME↗

arXiv2021

Incorporating Uncertainty in Learning to Defer Algorithms for Safe Computer-Aided Diagnosis

Deep neural networks are increasingly being used for computer-aided diagnosis, but erroneous diagnoses can be extremely costly for patients. We propose a learning to defer with uncertainty (LDU) algorithm which identifies patients for whom diagnostic uncertainty is high and defers them for evaluation by human experts. LDU was evaluated on the diagnosis of myocardial infarction (using discharge summaries), the diagnosis of any comorbidities (using structured data), and the diagnosis of pleural effusion and pneumothorax (using chest x-rays), and compared with 'learning to defer without uncertainty information' (LD) and 'direct triage by uncertainty' (DT) methods. LDU achieved the same F1 score as LD but deferred considerably fewer patients (e.g. 36% vs. 69% deferral rate for diagnosing pleural effusion with an F1 score of 0.96). Furthermore, even when many patients were assigned the wrong diagnosis with high confidence (e.g. for the diagnosis of any comorbidities) LDU achieved a 17% increase in F1 score, whereas DT was not applicable. Importantly, the weight of the defer loss in LDU can be easily adjusted to obtain the desired trade-off between diagnostic accuracy and deferral rate. In conclusion, LDU can readily augment any existing diagnostic network to reduce the risk of erroneous diagnoses in clinical practice.

cs.LG↗

arXiv2021

Dynamic prediction of time to event with survival curves

With the ever-growing complexity of primary health care system, proactive patient failure management is an effective way to enhancing the availability of health care resource. One key enabler is the dynamic prediction of time-to-event outcomes. Conventional explanatory statistical approach lacks the capability of making precise individual level prediction, while the data adaptive binary predictors does not provide nominal survival curves for biologically plausible survival analysis. The purpose of this article is to elucidate that the knowledge of explanatory survival analysis can significantly enhance the current black-box data adaptive prediction models. We apply our recently developed counterfactual dynamic survival model (CDSM) to static and longitudinal observational data and testify that the inflection point of its estimated individual survival curves provides reliable prediction of the patient failure time.

stat.ML↗

arXiv2021

Comparing Broadband ISP Performance using Big Data from M-Lab

Comparing ISPs on broadband speed is challenging, since measurements can vary due to subscriber attributes such as operation system and test conditions such as access capacity, server distance, TCP window size, time-of-day, and network segment size. In this paper, we draw inspiration from observational studies in medicine, which face a similar challenge in comparing the effect of treatments on patients with diverse characteristics, and have successfully tackled this using "causal inference" techniques for {\em post facto} analysis of medical records. Our first contribution is to develop a tool to pre-process and visualize the millions of data points in M-Lab at various time- and space-granularities to get preliminary insights on factors affecting broadband performance. Next, we analyze 24 months of data pertaining to twelve ISPs across three countries, and demonstrate that there is observational bias in the data due to disparities amongst ISPs in their attribute distributions. For our third contribution, we apply a multi-variate matching method to identify suitable cohorts that can be compared without bias, which reveals that ISPs are closer in performance than thought before. Our final contribution is to refine our model by developing a method for estimating speed-tier and re-apply matching for comparison of ISP performance. Our results challenge conventional rankings of ISPs, and pave the way towards data-driven approaches for unbiased comparisons of ISPs world-wide.

cs.PF↗

arXiv2019

Targeted Estimation of Heterogeneous Treatment Effect in Observational Survival Analysis

The aim of clinical effectiveness research using repositories of electronic health records is to identify what health interventions 'work best' in real-world settings. Since there are several reasons why the net benefit of intervention may differ across patients, current comparative effectiveness literature focuses on investigating heterogeneous treatment effect and predicting whether an individual might benefit from an intervention. The majority of this literature has concentrated on the estimation of the effect of treatment on binary outcomes. However, many medical interventions are evaluated in terms of their effect on future events, which are subject to loss to follow-up. In this study, we describe a framework for the estimation of heterogeneous treatment effect in terms of differences in time-to-event (survival) probabilities. We divide the problem into three phases: (1) estimation of treatment effect conditioned on unique sets of the covariate vector; (2) identification of features important for heterogeneity using an ensemble of non-parametric variable importance methods; and (3) estimation of treatment effect on the reference classes defined by the previously selected features, using one-step Targeted Maximum Likelihood Estimation. We conducted a series of simulation studies and found that this method performs well when either sample size or event rate is high enough and the number of covariates contributing to the effect heterogeneity is moderate. An application of this method to a clinical case study was conducted by estimating the effect of oral anticoagulants on newly diagnosed non-valvular atrial fibrillation patients using data from the UK Clinical Practice Research Datalink.

stat.ME↗