arXiv subjects

Daniel S. Herman

Publications and source records attributed to Daniel S. Herman.

4 recordsLinked to original sources

arXiv2025

Towards symbolic regression for interpretable clinical decision scores

Medical decision-making makes frequent use of algorithms that combine risk equations with rules, providing clear and standardized treatment pathways. Symbolic regression (SR) traditionally limits its search space to continuous function forms and their parameters, making it difficult to model this decision-making. However, due to its ability to derive data-driven, interpretable models, SR holds promise for developing data-driven clinical risk scores. To that end we introduce Brush, an SR algorithm that combines decision-tree-like splitting algorithms with non-linear constant optimization, allowing for seamless integration of rule-based logic into symbolic regression and classification models. Brush achieves Pareto-optimal performance on SRBench, and was applied to recapitulate two widely used clinical scoring systems, achieving high accuracy and interpretable models. Compared to decision trees, random forests, and other SR methods, Brush achieves comparable or superior predictive performance while producing simpler models.

cs.LG↗

arXiv2025

Iterative Learning of Computable Phenotypes for Treatment Resistant Hypertension using Large Language Models

Large language models (LLMs) have demonstrated remarkable capabilities for medical question answering and programming, but their potential for generating interpretable computable phenotypes (CPs) is under-explored. In this work, we investigate whether LLMs can generate accurate and concise CPs for six clinical phenotypes of varying complexity, which could be leveraged to enable scalable clinical decision support to improve care for patients with hypertension. In addition to evaluating zero-short performance, we propose and test a synthesize, execute, debug, instruct strategy that uses LLMs to generate and iteratively refine CPs using data-driven feedback. Our results show that LLMs, coupled with iterative learning, can generate interpretable and reasonably accurate programs that approach the performance of state-of-the-art ML methods while requiring significantly fewer training examples.

cs.LG↗

arXiv2021

Inference for the Case Probability in High-dimensional Logistic Regression

Labeling patients in electronic health records with respect to their statuses of having a disease or condition, i.e. case or control statuses, has increasingly relied on prediction models using high-dimensional variables derived from structured and unstructured electronic health record data. A major hurdle currently is a lack of valid statistical inference methods for the case probability. In this paper, considering high-dimensional sparse logistic regression models for prediction, we propose a novel bias-corrected estimator for the case probability through the development of linearization and variance enhancement techniques. We establish asymptotic normality of the proposed estimator for any loading vector in high dimensions. We construct a confidence interval for the case probability and propose a hypothesis testing procedure for patient case-control labelling. We demonstrate the proposed method via extensive simulation studies and application to real-world electronic health record data.

stat.ME↗

arXiv2019

Electronic Health Record Phenotyping with Internally Assessable Performance (PhIAP) using Anchor-Positive and Unlabeled Patients

Building phenotype models using electronic health record (EHR) data conventionally requires manually labeled cases and controls. Assigning labels is labor intensive and, for some phenotypes, identifying gold-standard controls is prohibitive. To facilitate comprehensive clinical decision support and research, we sought to develop an accurate EHR phenotyping approach that assesses its performance without a validation set. Our framework relies on specifying a random subset of cases, potentially using an anchor variable that has excellent positive predictive value and sensitivity that is independent of predictors. We developed a novel maximum likelihood approach that efficiently leverages data from anchor-positive and unlabeled patients to develop logistic regression phenotyping models. Additionally, we described novel statistical methods for estimating phenotyping prevalence and assessing model calibration and predictive performance measures. Theoretical and simulation studies indicated our method generates accurate predicted probabilities, leading to excellent discrimination and calibration, and consistent estimates of phenotype prevalence and anchor sensitivity. The method appeared robust to minor lack-of-fit and the proposed calibration assessment detected major lack-of-fit. We applied our method to EHR data to develop a preliminary model for identifying patients with primary aldosteronism, which achieved an AUC of 0.99 and PPV of 0.8. We developed novel statistical methods for accurate model development and validation with minimal manual labeling, facilitating development of scalable, transferable, semi-automated case labeling and practice-specific models. Our EHR phenotyping approach decreases labor-intensive manual phenotyping and annotation, which should enable broader model development and dissemination for EHR clinical decision support and research.

stat.AP↗