Searcharxiv⌕ Search

arXiv subjects

Michael C. Burkhart

Publications and source records attributed to Michael C. Burkhart.

18 recordsLinked to original sources
arXiv2026

Representation Before Training: A Practical Benchmark for Generative Medical Event Model Tokenization

Generative medical event models use tokenized sequences of patient timelines as input, but practical guidance on the many decisions around tokenization is limited. We benchmark quantization granularity, reference-range anchoring, code--value fusion, numeric and temporal encodings, and native versus harmonized event representations from an expert-mapped common data model. Using both Llama and Qwen architectures, 156 models were trained on full hospitalizations from three initialization seeds, with each configuration following a shared training recipe for up to five epochs. We evaluated learned representations from the first 24 hours of hospitalization with linear probes to predict binary and continuous outcomes during hours 24-48. Fused tokens pairing codes with value deciles increased performance across all eight outcome families relative to the equivalent unfused tokenized input with area under the receiver operating characteristic curve (AUROC) gains of $+0.002$ to $+0.033$ and Spearman correlation gains of $+0.025$ to $+0.114$. Neither anchoring value bins to reference ranges nor increasing quantization granularity consistently improved performance, while xVal variants underperformed both discrete and soft encodings. Alternatives to explicit time tokens, such as event-order and admission-relative rotary position embeddings (RoPE), yielded higher family-mean point estimates across all eight families while reducing input length. When evaluating native input against input mapped to the Common Longitudinal Intensive Care Unit Data Format (CLIF), the CLIF full-hospitalization training sequences contained 28.6% as many tokens as the native sequences and improved performance across six of eight outcome families. These findings show that tokenization and event encoding are consequential design choices when learning patient representations for downstream classification and regression tasks.

cs.LG↗
arXiv2026

Federated generative event models for tokenized electronic health records

Electronic health record foundation models are limited by institutionally siloed data and substantial performance degradation under cross-site transfer. We evaluated federated training of tokenized generative event models (GEMs) across 122,251 intensive care hospitalizations from three independent health systems harmonized to the Common Longitudinal ICU Data Format. Models were assessed on 12 post-24-hour clinical prediction tasks using within-site, cross-site, centralized, and federated training configurations. GEMs achieved the highest mean within-site and cross-site ROC-AUC and were substantially more transportable than conventional supervised models: their average cross-site penalties were 0.025 ROC-AUC and 0.027 PR-AUC, compared with 0.079 and 0.089 for LightGBM. Federated Learning (FedAvg and FedAvgM) approached the performance of centralized GEM training, with most gains obtained within 5-10 communication rounds. However, centralized multi-site training provided only modest improvements over complete local training. Multi-site models were most useful when local training data were limited, with their advantage narrowing as institutional data accumulated. These findings show that federated GEM training is technically feasible and preserves most centralized performance, but that the main open challenge is learning transportable representations to translate larger, but heterogeneous data from multiple health systems into a reliable target-site benefit.

cs.LG↗
arXiv2026

Efficient Generative Prediction for EHR Foundation Models: The SCOPE and REACH Estimators

Generative foundation models trained on tokenized electronic health record (EHR) timelines show promise for clinical outcome prediction via Monte Carlo sampling of simulated future trajectories. However, this approach suffers from three coupled limitations: sparse estimate distributions that poorly differentiate patient risk levels, extreme computational cost, and high sampling variance. We propose two new estimators that leverage next-token probability distributions underutilized by standard Monte Carlo: the Sum of Conditional Outcome Probability Estimator (SCOPE) and Risk Estimation from Anticipated Conditional Hazards (REACH). We prove both are unbiased, that REACH guarantees variance reduction over Monte Carlo for any model and outcome, and that REACH is a Rao-Blackwellization of any naive importance sampling scheme that preserves the non-outcome token distribution. Empirically, across $11$ clinically important outcomes in MIMIC-IV and the UChicago health system, SCOPE and REACH match $100$-sample Monte Carlo accuracy with median token reductions of $2.5\times$ to $3.4\times$ and reductions exceeding $80\times$ for the rarest outcomes, with calibration preserved throughout. Because SCOPE reuses a single sampled pool across an arbitrary number of outcomes at no marginal generation cost while REACH provides a per-task variance guarantee, the two estimators are complementary in deployment and together meaningfully reduce the inference budget required for generative EHR foundation models, particularly for rare, high-impact outcomes in healthcare.

stat.ML↗
arXiv2025

Prototype Learning to Create Refined Interpretable Digital Phenotypes from ECGs

Prototype-based neural networks offer interpretable predictions by comparing inputs to learned, representative signal patterns anchored in training data. While such models have shown promise in the classification of physiological data, it remains unclear whether their prototypes capture an underlying structure that aligns with broader clinical phenotypes. We use a prototype-based deep learning model trained for multi-label ECG classification using the PTB-XL dataset. Then without modification we performed inference on the MIMIC-IV clinical database. We assess whether individual prototypes, trained solely for classification, are associated with hospital discharge diagnoses in the form of phecodes in this external population. Individual prototypes demonstrate significantly stronger and more specific associations with clinical outcomes compared to the classifier's class predictions, NLP-extracted concepts, or broader prototype classes across all phecode categories. Prototype classes with mixed significance patterns exhibit significantly greater intra-class distances (p $<$ 0.0001), indicating the model learned to differentiate clinically meaningful variations within diagnostic categories. The prototypes achieve strong predictive performance across diverse conditions, with AUCs ranging from 0.89 for atrial fibrillation to 0.91 for heart failure, while also showing substantial signal for non-cardiac conditions such as sepsis and renal disease. These findings suggest that prototype-based models can support interpretable digital phenotyping from physiologic time-series data, providing transferable intermediate phenotypes that capture clinically meaningful physiologic signatures beyond their original training objectives.

cs.LG↗
arXiv2025

ProtoECGNet: Case-Based Interpretable Deep Learning for Multi-Label ECG Classification with Contrastive Learning

Deep learning-based electrocardiogram (ECG) classification has shown impressive performance but clinical adoption has been slowed by the lack of transparent and faithful explanations. Post hoc methods such as saliency maps may fail to reflect a model's true decision process. Prototype-based reasoning offers a more transparent alternative by grounding decisions in similarity to learned representations of real ECG segments, enabling faithful, case-based explanations. We introduce ProtoECGNet, a prototype-based deep learning model for interpretable, multi-label ECG classification. ProtoECGNet employs a structured, multi-branch architecture that reflects clinical interpretation workflows: it integrates a 1D CNN with global prototypes for rhythm classification, a 2D CNN with time-localized prototypes for morphology-based reasoning, and a 2D CNN with global prototypes for diffuse abnormalities. Each branch is trained with a prototype loss designed for multi-label learning, combining clustering, separation, diversity, and a novel contrastive loss that encourages appropriate separation between prototypes of unrelated classes while allowing clustering for frequently co-occurring diagnoses. We evaluate ProtoECGNet on all 71 diagnostic labels from the PTB-XL dataset, demonstrating competitive performance relative to state-of-the-art black-box models while providing structured, case-based explanations. To assess prototype quality, we conduct a structured clinician review of the final model's projected prototypes, finding that they are rated as representative and clear. ProtoECGNet shows that prototype learning can be effectively scaled to complex, multi-label time-series classification, offering a practical path toward transparent and trustworthy deep learning models for clinical decision support.

cs.LG↗
arXiv2025

Quantifying surprise in clinical care: Detecting highly informative events in electronic health records with foundation models

We present a foundation model-derived method to identify highly informative tokens and events in electronic health records. Our approach considers incoming data in the entire context of a patient's hospitalization and so can flag anomalous events that rule-based approaches would consider within a normal range. We demonstrate that the events our model flags are significant for predicting downstream patient outcomes and that a fraction of events identified as carrying little information can safely be dropped. Additionally, we show how informativeness can help interpret the predictions of prognostic models trained on foundation model-derived representations.

cs.LG↗
arXiv2025

Foundation models for electronic health records: representation dynamics and transferability

Foundation models (FMs) trained on electronic health records (EHRs) have shown strong performance on a range of clinical prediction tasks. However, adapting these models to local health systems remains challenging due to limited data availability and resource constraints. In this study, we investigated what these models learn and evaluated the transferability of an FM trained on MIMIC-IV to an institutional EHR dataset at the University of Chicago Medical Center. We assessed their ability to identify outlier patients and examined representation-space patient trajectories in relation to future clinical outcomes. We also evaluated the performance of supervised fine-tuned classifiers on both source and target datasets. Our findings offer insights into the adaptability of FMs across different healthcare systems, highlight considerations for their effective implementation, and provide an empirical analysis of the underlying factors that contribute to their predictive performance.

cs.LG↗
arXiv2025

Actions of nilpotent groups on nilpotent groups

For finite nilpotent groups $J$ and $N$, suppose $J$ acts on $N$ via automorphisms. We exhibit a decomposition of the first cohomology set in terms of the first cohomologies of the Sylow $p$-subgroups of $J$ that mirrors the primary decomposition of $H^1(J,N)$ for abelian $N$. We then show that if $N \rtimes J$ acts on some non-empty set $Ω$, where the action of $N$ is transitive and for each prime $p$ a Sylow $p$-subgroup of $J$ fixes an element of $Ω$, then $J$ fixes an element of $Ω$.

math.GR↗
arXiv2025

Conjugacy conditions for supersoluble complements of an abelian base and a fixed point result for non-coprime actions

We demonstrate that two supersoluble complements of an abelian base in a finite split extension are conjugate if and only if, for each prime $p$, a Sylow $p$-subgroup of one complement is conjugate to a Sylow $p$-subgroup of the other. As a corollary, we find that any two supersoluble complements of an abelian subgroup $N$ in a finite split extension $G$ are conjugate if and only if, for each prime $p$, there exists a Sylow $p$-subgroup $S$ of $G$ such that any two complements of $S\cap N$ in $S$ are conjugate in $G$. In particular, restricting to supersoluble groups allows us to ease D. G. Higman's stipulation that the complements of $S\cap N$ in $S$ be conjugate within $S$. We then consider group actions and obtain a fixed point result for non-coprime actions analogous to Glauberman's lemma.

math.GR↗
arXiv2023

Fixed point conditions for non-coprime actions

In the setting of finite groups, suppose $J$ acts on $N$ via automorphisms so that the induced semidirect product $N\rtimes J$ acts on some non-empty set $Ω$, with $N$ acting transitively. Glauberman proved that if the orders of $J$ and $N$ are coprime, then $J$ fixes a point in $Ω$. We consider the non-coprime case and show that if $N$ is abelian and a Sylow $p$-subgroup of $J$ fixes a point in $Ω$ for each prime $p$, then $J$ fixes a point in $Ω$. We also show that if $N$ is nilpotent, $N\rtimes J$ is supersoluble, and a Sylow $p$-subgroup of $J$ fixes a point in $Ω$ for each prime $p$, then $J$ fixes a point in $Ω$.

math.GR↗
arXiv2023

Discriminative Bayesian filtering lends momentum to the stochastic Newton method for minimizing log-convex functions

To minimize the average of a set of log-convex functions, the stochastic Newton method iteratively updates its estimate using subsampled versions of the full objective's gradient and Hessian. We contextualize this optimization problem as sequential Bayesian inference on a latent state-space model with a discriminatively-specified observation process. Applying Bayesian filtering then yields a novel optimization algorithm that considers the entire history of gradients and Hessians when forming an update. We establish matrix-based conditions under which the effect of older observations diminishes over time, in a manner analogous to Polyak's heavy ball momentum. We illustrate various aspects of our approach with an example and review other relevant innovations for the stochastic Newton method.

stat.ML↗
arXiv2022

Deep Low-Density Separation for Semi-Supervised Classification

Given a small set of labeled data and a large set of unlabeled data, semi-supervised learning (SSL) attempts to leverage the location of the unlabeled datapoints in order to create a better classifier than could be obtained from supervised methods applied to the labeled training set alone. Effective SSL imposes structural assumptions on the data, e.g. that neighbors are more likely to share a classification or that the decision boundary lies in an area of low density. For complex and high-dimensional data, neural networks can learn feature embeddings to which traditional SSL methods can then be applied in what we call hybrid methods. Previously-developed hybrid methods iterate between refining a latent representation and performing graph-based SSL on this representation. In this paper, we introduce a novel hybrid method that instead applies low-density separation to the embedded features. We describe it in detail and discuss why low-density separation may be better suited for SSL on neural network-based embeddings than graph-based algorithms. We validate our method using in-house customer survey data and compare it to other state-of-the-art learning methods. Our approach effectively classifies thousands of unlabeled users from a relatively small number of hand-classified examples.

cs.LG↗
arXiv2022

Neuroevolutionary Feature Representations for Causal Inference

Within the field of causal inference, we consider the problem of estimating heterogeneous treatment effects from data. We propose and validate a novel approach for learning feature representations to aid the estimation of the conditional average treatment effect or CATE. Our method focuses on an intermediate layer in a neural network trained to predict the outcome from the features. In contrast to previous approaches that encourage the distribution of representations to be treatment-invariant, we leverage a genetic algorithm that optimizes over representations useful for predicting the outcome to select those less useful for predicting the treatment. This allows us to retain information within the features useful for predicting outcome even if that information may be related to treatment assignment. We validate our method on synthetic examples and illustrate its use on a real life dataset.

stat.ML↗
arXiv2018

A Discriminative Approach to Bayesian Filtering with Applications to Human Neural Decoding

Given a stationary state-space model that relates a sequence of hidden states and corresponding measurements or observations, Bayesian filtering provides a principled statistical framework for inferring the posterior distribution of the current state given all measurements up to the present time. For example, the Apollo lunar module implemented a Kalman filter to infer its location from a sequence of earth-based radar measurements and land safely on the moon. To perform Bayesian filtering, we require a measurement model that describes the conditional distribution of each observation given state. The Kalman filter takes this measurement model to be linear, Gaussian. Here we show how a nonlinear, Gaussian approximation to the distribution of state given observation can be used in conjunction with Bayes' rule to build a nonlinear, non-Gaussian measurement model. The resulting approach, called the Discriminative Kalman Filter (DKF), retains fast closed-form updates for the posterior. We argue there are many cases where the distribution of state given measurement is better-approximated as Gaussian, especially when the dimensionality of measurements far exceeds that of states and the Bernstein-von Mises theorem applies. Online neural decoding for brain-computer interfaces provides a motivating example, where filtering incorporates increasingly detailed measurements of neural activity to provide users control over external devices. Within the BrainGate2 clinical trial, the DKF successfully enabled three volunteers with quadriplegia to control an on-screen cursor in real-time using mental imagery alone. Participant "T9" used the DKF to type out messages on a tablet PC.

stat.ML↗
arXiv2016

The discriminative Kalman filter for nonlinear and non-Gaussian sequential Bayesian filtering

The Kalman filter (KF) is used in a variety of applications for computing the posterior distribution of latent states in a state space model. The model requires a linear relationship between states and observations. Extensions to the Kalman filter have been proposed that incorporate linear approximations to nonlinear models, such as the extended Kalman filter (EKF) and the unscented Kalman filter (UKF). However, we argue that in cases where the dimensionality of observed variables greatly exceeds the dimensionality of state variables, a model for $p(\text{state}|\text{observation})$ proves both easier to learn and more accurate for latent space estimation. We derive and validate what we call the discriminative Kalman filter (DKF): a closed-form discriminative version of Bayesian filtering that readily incorporates off-the-shelf discriminative learning techniques. Further, we demonstrate that given mild assumptions, highly non-linear models for $p(\text{state}|\text{observation})$ can be specified. We motivate and validate on synthetic datasets and in neural decoding from non-human primates, showing substantial increases in decoding performance versus the standard Kalman filter.

stat.ML↗
arXiv2012

Conditional Probabilities of Multivariate Poisson Distributions

Multivariate Poisson distributions have numerous applications. Fast computation of these distributions, holding constant a fixed set of linear combinations of these variables, has been explored by Sontag and Zeilberger. This elaborates on their work.

math.PR↗
arXiv2012

Linear Transformations & the Multivariate Generating Function

This note examines linear combinations of multi-indexed sequences and derives the multivariate generating function of such a linear combination in terms of the original sequence's m.g.f. Applications include finding distributions and moments of non-negative discrete random variables conditioned on non-negative linear combinations of the original variables. Examples include independent Poisson r.v.'s and a $d$-variate multinomial distribution.

math.CO↗