Cluster Analysis of Low-Dimensional Medical Concept Representations from Electronic Health Records

Jaume-Santero, Fernando; Zhang, Boya; Proios, Dimitrios; Yazdani, Anthony; Gouareb, Racha; Bjelogrlic, Mina; Teodoro, Douglas

doi:10.1007/978-3-031-20627-6_29

Cited by 4 publications

(8 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our work builds upon the concept of model-based evaluation, which involves a score derived from a learned model and has been applied in areas such as retrosynthesis and natural language processing 19,20 . For example, round-trip accuracy [10][11][12] evaluates the accuracy of predicted reactants or reaction conditions by feeding these predictions back into a product prediction model and determining the percentage of matching product predictions. The round-trip accuracy is designed to evaluate reactant or condition prediction on a system level.…”

Section: Related Workmentioning

confidence: 99%

“…We evaluate the effectiveness of CatScore as an instance-level metric for catalyst design by comparing it with experimental selectivity and DFTScore. For a baseline, we employ round-trip accuracy [10][11][12] , a metric designed initially for system-level evaluation, but here we use it at an instance level by checking if the predicted product matches the target product:…”

Section: Instance-level Evaluation Of Catscore: Correlation With Expe...mentioning

confidence: 99%

“…DFTScore In this section, we aim to assess how well CatScore can evaluate catalyst design models at the system level. We compare its performance with DFTScore and use round-trip accuracy as a baseline [10][11][12] . For this comparison, system-level scores are derived by taking the average of their respective instance-level scores.…”

Section: System-level Evaluation Of Catscore: Correlation Withmentioning

confidence: 99%

“…Importantly, while some methods, like round-trip accuracy [10][11][12] , may be adept at system-level evaluations, they are not designed for the instance level. It is at this granular level where identifying promising catalyst candidates becomes vital.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

CatScore: A High-Efficiency Evaluation Metric for Asymmetric Catalyst Design

Yan,

Cho

2023

Preprint

View full text Add to dashboard Cite

Asymmetric catalysis plays a crucial role in advancing medicine and materials science. However, the prevailing experiment-driven methods for catalyst evaluation are both resource-heavy and time-consuming. To address this challenge, we present CatScore - a learning-centric metric designed for the automatic evaluation of catalyst design models at both instance and system levels. This approach harnesses the power of deep learning to predict product selectivity as a function of reactants and the proposed catalyst. The predicted selectivity serves as a quantitative score, enabling a swift and precise assessment of a catalyst’s activity. On an instance level, CatScore’s predictions correlate closely with experimental outcomes, demonstrating a Spearman’s ρ = 0.84, which surpasses the density functional theory (DFT) with ρ = 0.54 and round-trip accuracy metrics at ρ = 0.24. Importantly, when ranking catalyst candidates, CatScore achieves a mean reciprocal ranking significantly superior to traditional DFT methods, marking a considerable reduction in labor and time investments needed to find top-performing catalysts.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Instance-level Evaluation Of Catscore: Correlation With Expe...mentioning

confidence: 99%

Section: System-level Evaluation Of Catscore: Correlation Withmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

CatScore: A High-Efficiency Evaluation Metric for Asymmetric Catalyst Design

Yan,

Cho

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…A key aspect of developing accurate patient and medical concept embeddings lies in abstracting and modeling the relevant data features. In that respect, neural language models can process EHRs, not only for clinical notes, but also by considering events in the patient trajectory as tokens, and entire trajectories as a sequence of tokens, similar to word sentences [75][76][77][78][79][80]. Medical concepts relate to each other, either with causal relationships, e.g., diagnosis and medication, or through similar meanings, e.g., related, or synonymous diagnoses or compounds with similar effects or indications.…”

Section: Introductionmentioning

confidence: 99%

Comparing neural language models for medical concept representation and patient trajectory prediction

Bornet

Proios

Yazdani

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

Effective representation of medical concepts is crucial for secondary analyses of electronic health records. Neural language models have shown promise in automatically deriving medical concept representations from clinical data. However, the comparative performance of different language models for creating these empirical representations, and the extent to which they encode medical semantics, has not been extensively studied. This study aims to address this gap by evaluating the effectiveness of three popular language models - word2vec, fastText, and GloVe - in creating medical concept embeddings. By using a large dataset of digital health records, we created patient trajectories and used them to train the language models. We then assessed the ability of the learned embeddings to encode semantics through an explicit comparison with biomedical terminologies, and implicitly by predicting patient outcomes and trajectories with different degrees of information. Our qualitative analysis shows that empirical clusters of embeddings learned by fastText exhibit the highest similarity with theoretical clustering patterns obtained from biomedical terminologies, with a similarity score between empirical and theoretical clusters of 0.88, 0.80, and 0.92 for diagnosis, procedures, and medication codes, respectively. Conversely, for outcome prediction, word2vec and GloVe tend to outperform fastText, with the former achieving AUROC as high as 0.80, 0.63, and 0.88 for length-of-stay, readmission, and mortality prediction, respectively. In predicting the next steps in patient trajectories, GloVe achieves the highest performance for diagnostic and medication codes (AUPRC of 0.46 and of 0.82, respectively) at the highest level of the semantic hierarchy, while fastText outperforms the other models for procedure codes (AUPRC of 0.67). Our study demonstrates that subword information is crucial for learning medical concept representations, but global embedding vectors are better suited for downstream tasks, such as trajectory prediction. Thus, these models can be harnessed to learn representations that convey clinical meaning, and our insights highlight the potential of using machine learning techniques to semantically encode medical data.

show abstract

EAPR: explainable and augmented patient representation learning for disease prediction

Zhang,

Xu,

et al. 2023

Health Inf Sci Syst

View full text Add to dashboard Cite

Cluster Analysis of Low-Dimensional Medical Concept Representations from Electronic Health Records

Abstract: Archive ouverte UNIGEhttps://archive-ouverte.unige.ch Chapitre d'actes 2022 Accepted version Open Access This is an author manuscript post-peer-reviewing (accepted version) of the original publication. The layout of the published version may differ .

Cited by 4 publications

References 20 publications

CatScore: A High-Efficiency Evaluation Metric for Asymmetric Catalyst Design

CatScore: A High-Efficiency Evaluation Metric for Asymmetric Catalyst Design

Comparing neural language models for medical concept representation and patient trajectory prediction

EAPR: explainable and augmented patient representation learning for disease prediction

Contact Info

Product

Resources

About