Research Appraisals
Evidence-based critical appraisals of the latest medical research, systematically evaluated using Oxford CEBM methodology.
Showing 2 appraisals
Journal of chemical information and modeling
A Unified Molecular Graph and Protein Language Model Framework for Predicting Human Drug-Hormone Receptor Interactions with Structure-Aware Validation
Hormones regulate many essential biological processes by interacting with specific receptors that control gene expression, metabolism, growth, and immune function. Because numerous therapeutic compounds can influence or disrupt hormone signaling pathways, understanding drug-hormone receptor interactions (DHRI) is crucial for ensuring both therapeutic efficacy and endocrine safety. However, computational approaches for predicting DHRI remain limited, and most existing models do not explicitly incorporate hormone receptor-specific information. In this study, we propose a receptor-aware deep learning framework for DHRI prediction that integrates structural drug features with contextually embedded hormone receptor information. Drug molecules are represented using a hybrid encoding strategy that combines Morgan fingerprints and graph transformer-based molecular features to capture both global chemical properties and local structural information, while hormone receptor sequences are encoded using the pretrained ESM2 protein language model to obtain biologically meaningful sequence representations. The fused drug-hormone receptor features are then processed through a multilayer neural network to predict interaction probabilities. Model performance was evaluated using three splitting strategies, including random split, cold-drug split, and scaffold-based split, to assess both predictive accuracy and generalization ability. Under the random split setting, the model achieved an accuracy of 0.93, sensitivity of 0.94, specificity and precision of 0.92, F1-score of 0.93, and MCC of 0.86 on an independent data set, while also maintaining comparable performance under the more stringent cold-drug and scaffold-based settings. Feature importance analysis showed that atomic identity, hybridization, atom degree, and bond type were key shared determinants, while receptor-stratified results revealed receptor-dependent contributions from secondary features such as aromaticity, chirality, etc. In addition, t-SNE visualization showed clear class separation after training, and molecular docking across estrogen, androgen, and glucocorticoid receptors further supported the biological relevance of the predictions. Together, these findings demonstrate that incorporating hormone receptor-specific sequence information enables more reliable and biologically meaningful prediction of drug-hormone receptor interactions.
28 July 2026
Read appraisal →Journal of medical Internet research
Extracting and Classifying Drug Discontinuations From Estonian Electronic Health Records: Development and Validation Study
BACKGROUND: Drug adherence is crucial for chronic disease management, yet treatment discontinuation remains common due to factors such as side effects, inefficacy, or cost. These reasons are often recorded only in free-text clinical notes, making large-scale analysis difficult. While large language models (LLMs) can interpret such unstructured data more effectively than traditional natural language processing methods, few studies have systematically categorized reasons for discontinuation or identified whether the decision was initiated by the patient or the clinician, especially in low-resource languages such as Estonian. OBJECTIVE: This study aimed to assess the ability of LLMs to extract and classify reasons for drug discontinuation and identify who initiated it using Estonian electronic health records and characterize the observed discontinuation patterns and initiators for statins and antidiabetic medications. METHODS: We combined prescription data with free-text anamneses from a 10% sample of the Estonian population (2012-2019). LLMs (Llama 3.1-70B and GPT-4o) were applied to extract discontinuation phrases and reasons, classify them into a clinician-developed taxonomy, and identify who discontinued the treatment. Performance was evaluated on 100 randomly chosen cases per drug group. RESULTS: Extraction yielded 625 antidiabetic drug and 233 statin discontinuation cases. Validation confirmed a precision of 0.93 to 0.98 for extracting phrases and 0.95 to 0.96 for extracting reasons. Classification of discontinuation reasons achieved weighted F1-scores of 0.81 to 0.84, whereas classification of who initiated discontinuation achieved weighted F1-scores of 0.64 to 0.78. Adverse reactions were the most frequent reason overall, accounting for 70% (163/233) of statin discontinuations and 44.8% (280/625) of antidiabetic drug discontinuations. Regarding antidiabetic drugs, treatment inefficacy and contraindications were more common. Patients more often stopped due to adverse reactions or nonmedical reasons, whereas physicians more often initiated discontinuation for contraindications. CONCLUSIONS: LLMs can accurately extract and classify medication discontinuation reasons and show variable performance in identifying discontinuation initiators in Estonian clinical narratives. Both local and proprietary models showed promising results, enabling scalable analyses that complement structured health records. This demonstrates the potential of LLMs to unlock information from clinical notes, turning this underused electronic health record component into a valuable resource for monitoring treatment patterns and detecting adverse event signals.
19 June 2026
Read appraisal →