Autoregressive models can predict clinical events, but generating patient-conditioned multi-step trajectories that respond to intervention tokens and testing whether those responses preserve known pharmacological associations has received limited attention. We present FlatASCEND, a 14.5M-parameter autoregressive clinical sequence model using flat composite tokens and a zero-inflated log-normal time head. Standard distributional metrics (Jaccard 0.889-0.954) do not distinguish FlatASCEND from trivial baselines; the model's value lies in conditional generation from patient-specific prefixes. A prompt-shuffle ablation shows patient-specific conditioning amplifies mechanistic pharmacological effects (2.0-2.2x for steroid to glucose, diuretic to potassium) while leaving confounding-driven associations unchanged (0.9x for insulin to glucose). An incident-user framework assesses directional consistency against prior pharmacological knowledge on MIMIC-IV (N=500 per comparison): 4/10 recover correct mechanistic directions, 2 reproduce treatment-context associations, 4 are incorrect (9/10 significant, Wilcoxon p<0.05). This pattern - partial recovery under residual confounding - is consis
Large language models (LLMs) have shown strong empirical performance across pharmacology and drug discovery tasks, yet the internal mechanisms by which they encode pharmacological knowledge remain poorly understood. In this work, we investigate how drug-group semantics are represented and retrieved within Llama-based biomedical language models using causal and probing-based interpretability methods. We apply activation patching to localize where drug-group information is stored across model layers and token positions, and complement this analysis with linear probes trained on token-level and sum-pooled activations. Our results demonstrate that early layers play a key role in encoding drug-group knowledge, with the strongest causal effects arising from intermediate tokens within the drug-group span rather than the final drug-group token. Linear probing further reveals that pharmacological semantics are distributed across tokens and are already present in the embedding space, with token-level probes performing near chance while sum-pooled representations achieve maximal accuracy. Together, these findings suggest that drug-group semantics in LLMs are not localized to single tokens but
Recommending safe and effective medication combinations from electronic health records (EHRs) is a core clinical AI problem, yet it remains difficult because patient trajectories are long, noisy, and clinically heterogeneous. Existing methods typically excel at either temporal modeling across visits or pharmacological knowledge integration (e.g., drug-drug interactions, DDIs), but rarely achieve both while robustly suppressing noise. We present GraphDiffMed, a knowledge-constrained medication recommendation framework built on dual-scale Differential Attention v2. Differential attention is applied at both intra-visit and inter-visit levels to filter spurious signals within encounters and across longitudinal history, while pharmacological constraints are incorporated during learning. Experiments on MIMIC-III and ablation studies show that this design consistently improves recommendation quality and ranking over strong baselines while achieving a more favorable safety performance balance. We further find that the strongest-performing configuration uses only demographic auxiliary features under our experimental setting. Overall, GraphDiffMed demonstrates that combining noise-aware atte
Coronavirus disease 2019 (COVID-19), caused by SARS-CoV-2, represents a major global pandemic of the 21st century, with long-term effects termed long COVID. This systematic review and network meta-analysis (NMA) evaluated pharmacological interventions for adults with long COVID, incorporating randomized controlled trials and adjusted observational studies. Primary outcomes included all-cause mortality, hospitalization, ICU admission, and mechanical ventilation; secondary outcomes covered symptom recovery across five categories, with safety assessed via adverse events. Results from random-effects models showed that saline nasal irrigation (SMD=21.10, 95% CI [16.91, 25.30]), nitrilotriacetic acid trisodium (SMD=7.40 [5.79, 9.01]), tetra sodium pyrophosphate (SMD=3.69 [2.61, 4.77]), and sodium gluconate (SMD=3.01 [1.92, 4.09]) significantly improved anosmia versus control. For thrombosis, rivaroxaban reduced arterial (OR=0.33 [0.01, 8.19]) and venous thrombotic events (OR=0.12 [0.01, 0.97]), while therapeutic-dose anticoagulants lowered thrombotic risks but increased major bleeding events (OR=1.86 [1.19, 2.89]) compared to prophylactic dosing. This NMA provides comparative evidence to
This paper investigates sentiment classification of Steam game reviews using an attention-based Bidirectional Long Short-Term Memory (BiLSTM) model. Using a dataset of 50,000 reviews sampled from a larger Steam review corpus, the authors compare a traditional machine learning baseline based on TF-IDF and PyCaret AutoML with a deep learning approach implemented in PyTorch. The proposed BiLSTM+Attention model is trained with class-weighted cross-entropy to address class imbalance and achieves 83% accuracy and 85% weighted F1-score on the test set, with 90% recall for negative reviews. The paper also presents attention visualizations to show interpretability by highlighting sentiment-bearing words. The study concludes that the BiLSTM+Attention model is effective for analyzing user sentiment in Steam reviews and useful for helping developers understand player feedback.
Cuminum cyminum L. (cumin) is a medicinal and edible plant widely used in traditional Chinese medicine (TCM) for treating various ailments, including diarrhea, abdominal pain, inflammation, asthma, and diabetes. While previous research has primarily focused on its essential oils, studies on its protein-derived bioactive peptides remain limited. In this study, we employed an innovative extraction method to isolate peptides from cumin seeds for the first time and screened their biological activities, revealing significant antimicrobial, antioxidant, and hypoglycemic properties. Guided by bioactivity, we utilized advanced separation and structural identification techniques, including Matrix-Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry (MALDI-TOF/TOF MS/MS), to systematically purify and characterize cumin-derived peptides. A total of 479 unique peptide sequences were identified using Mascot software and the SwissProt/UniProt_Bos databases. Among these, 15 highly bioactive peptides were selected for further analysis based on bioactivity and toxicity predictions using PeptideRanker and ToxinPred. Structural characterization revealed key features, such as α-helice
In this study, we investigate how supporting serendipitous discovery and analysis of online product reviews can encourage readers to explore reviews more comprehensively prior to making purchase decisions. We propose two interventions -- Exploration Metrics that can help readers understand and track their exploration patterns through visual indicators and a Bias Mitigation Model that intends to maximize knowledge discovery by suggesting sentiment and semantically diverse reviews. We designed, developed, and evaluated a text analytics system called Serendyze, where we integrated these interventions. We asked 100 crowd workers to use Serendyze to make purchase decisions based on product reviews. Our evaluation suggests that exploration metrics enabled readers to efficiently cover more reviews in a balanced way, and suggestions from the bias mitigation model influenced readers to make confident data-driven decisions. We discuss the role of user agency and trust in text-level analysis systems and their applicability in domains beyond review exploration.
Book reviews play important roles in scholarly communication especially in arts and humanities disciplines. By using Web of Science's Science Citation Index Expanded, Social Sciences Citation Index, and Arts & Humanities Citation Index, this study probed the patterns and dynamics of book reviews within these three indexes empirically during the past decade (2006-2015). We found that the absolute numbers of book reviews among all the three indexes were relatively stable but the relative shares were decreasing. Book reviews were very common in arts and humanities, common in social sciences, but rare in natural sciences. Book reviews are mainly contributed by authors from developed economies such as the USA and the UK. Oppositely, scholars from China and Japan are unlikely to contribute to book reviews.
Mild cognitive impairment (MCI) leading to dementia results in a constellation of psychiatric disorders including depression, mood disorders, schizophrenia and others. With increasing age, mild cognitive impairment leads to increased disability-adjusted life-years and healthcare burden. A huge number of drug trials for the treatment of MCI associated with Alzheimer's disease have undergone failure leading to the development of drugs that could avert the progression of the disease. However, some novel non-drug-based therapies like ultrasound ablation of amyloid plaques have influenced researchers to explore the non-pharmacological modalities for the treatment of mild cognitive impairment. To compensate for neurodegenerative loss resulting in coexisting psychiatric disorders, neurofeedback therapy has also been proven to improve behavioural outcomes by inducing neuroplasticity. The aim of the current review is to highlight the pathophysiological aspects of mild cognitive impairment leading to dementia that could be addressed with no pharmacological interventions and to understand the mechanisms behind the effects of these interventions.
This paper describes a rapid feasibility study of using GPT-4, a large language model (LLM), to (semi)automate data extraction in systematic reviews. Despite the recent surge of interest in LLMs there is still a lack of understanding of how to design LLM-based automation tools and how to robustly evaluate their performance. During the 2023 Evidence Synthesis Hackathon we conducted two feasibility studies. Firstly, to automatically extract study characteristics from human clinical, animal, and social science domain studies. We used two studies from each category for prompt-development; and ten for evaluation. Secondly, we used the LLM to predict Participants, Interventions, Controls and Outcomes (PICOs) labelled within 100 abstracts in the EBM-NLP dataset. Overall, results indicated an accuracy of around 80%, with some variability between domains (82% for human clinical, 80% for animal, and 72% for studies of human social sciences). Causal inference methods and study design were the data extraction items with the most errors. In the PICO study, participants and intervention/control showed high accuracy (>80%), outcomes were more challenging. Evaluation was done manually; scoring
Drug-drug interactions (DDI) can cause severe adverse drug reactions and pose a major challenge to medication therapy. Recently, informatics-based approaches are emerging for DDI studies. In this paper, we aim to identify key pharmacological components in DDI based on large-scale data from DrugBank, a comprehensive DDI database. With pharmacological components as features, logistic regression is used to perform DDI classification with a focus on searching for most predictive features, a process of identifying key pharmacological components. Using univariate feature selection with chi-squared statistic as the ranking criteria, our study reveals that top 10% features can achieve comparable classification performance compared to that using all features. The top 10% features are identified to be key pharmacological components. Furthermore, their importance is quantified by feature coefficients in the classifier, which measures the DDI potential and provides a novel perspective to evaluate pharmacological components.
The morphological form of a word can often give cues to its meaning, but purely relying on these mappings can lead to overgeneralization in high-stakes domains. In the medical domain, for instance, LLMs can confidently reason about fictitious drugs from their affixes alone (e.g., wugcillin) and generate plausible-looking clinical content. We present a behavioral and mechanistic study of LLM "affix heuristics" in pharmacology. Using fictitious drug names built from real affixes, we show that affix signals alone elicit class-level pharmacological responses. We introduce a framework for identifying whether a model's drug semantics are driven mainly by the affix, the stem, or the drug name as a whole. Applied across 653 drugs, our framework reveals that models often induce drug meaning primarily through affix cues, yet rarely explicitly indicate this reliance, and sometimes incorrectly conflate properties among affix-sharing drugs. Activation patching across models further localizes this behavior to early-mid layers. These findings show that morphological shortcuts pose a subtle but measurable risk to safety.
The integration of power electronics-based energy storage systems (PEESs) into power systems introduces potential instabilities. This study reviews efforts in dynamic analysis of both AC and DC power systems integrated with PEESs, covering dynamic modeling, analysis methods, and potential instability risks. Major conclusions are drawn as: 1) Simplified models of PEESs have been widely used for dynamic analysis of power systems. However, it may cause "error aggregation" as the scale of PEESs increases, leading to mistakes in results, which induces significant concerns. 2) Traditional stability mechanism analysis methods remain effective for single grid-connected PEES and large-scale PEESs with parallel and series connections. However, they are inadequate for PEESs with distributed connections. To fill in this gap, an idea of mechanism analysis based on "dynamic reconstruction" is proposed. 3) Potential instability risks caused by PEESs integration may differ from those caused by renewable energy integration due to differences in functional controls and bidirectional power flow. However, comprehensive investigations in this regard are lacking and require significant attention. To ens
AI coding agents increasingly accept assigned software tasks, modify repositories under bounded authority, and return work packages for review. Prior work proposed the software delegation contract, covering the task, authority, returned work package, and acceptance context, as the unit of analysis for delegated coding work, but did not measure its effects. This paper reports a controlled pilot study of explicit delegation contracts for coding agents. We built a dependency-free TypeScript API task environment with seeded defects and documentation gaps, authored ten tasks across five families, and ran 64 agent executions across two model tiers under three conditions: a realistic issue-style prompt, an explicit delegation contract, and a contract with a required evidence bundle. Each run was scored with hidden acceptance tests, mutation checks, and scope analysis, then reviewed by three independent condition-blinded model-based reviewers using a fixed rubric, for 192 reviews. Explicit contracts did not improve objective task outcomes: all 64 runs passed hidden acceptance checks, with zero scope violations. They did improve reviewability. Evidence sufficiency improved in 22 of 30 paire
Ultralight dark matter refers to the lightest potential dark matter candidates. We will focus on the mass range that has been studied using astrophysical and cosmological observations, corresponding to a mass $10^{-24} \, \mathrm{eV} \lesssim m \lesssim 10^{-18} \, \mathrm{eV}$. We will discuss the motivations for this mass range. The most studied model in this range corresponds to a minimally coupled, single, classical, spin-0 field comprising all dark matter. However, the work exploring extensions of this model (for example, higher spin, self-coupled, multiple field, and mixed models) will be one of the focuses of this review. The phenomenology associated with ultralight dark matter is rich and includes linear effects on the primordial power spectrum, core structures forming at the center of halos, nonlinear effects resulting in heating of stellar distributions, and non-relativistic effects relating to pulsar signals and black hole superradiance, to name a few. This set of effects has been studied using an equally extensive set of numerical tools. We will summarize the most common ones and discuss their applications and limitations. Ultralight dark matter also has a wide variety
Hemostasis and thrombosis are often thought as two sides of the same clotting mechanism whereas hemostasis is a natural protective mechanism to prevent bleeding and thrombosis is a blood clot abnormally formulated inside a blood vessel, blocking the normal blood flow. The evidence to date suggests that at least arterial thrombosis results from the same critical pathways of hemostasis. Analysis of these complex processes and pathways using quantitative systems pharmacological model-based approach can facilitate the delineation of the causal pathways that lead to the emergence of thrombosis. In this paper, we provide an overview of the main molecular and physiological mechanisms associated with hemostasis and thrombosis, and review the models and quantitative system pharmacological modeling approaches that are relevant in characterizing the interplay among the multiple factors and pathways of thrombosis. An emphasis is given to computational models for drug development. Future trends are discussed.
Increasing demands on medical imaging departments are taking a toll on the radiologist's ability to deliver timely and accurate reports. Recent technological advances in artificial intelligence have demonstrated great potential for automatic radiology report generation (ARRG), sparking an explosion of research. This survey paper conducts a methodological review of contemporary ARRG approaches by way of (i) assessing datasets based on characteristics, such as availability, size, and adoption rate, (ii) examining deep learning training methods, such as contrastive learning and reinforcement learning, (iii) exploring state-of-the-art model architectures, including variations of CNN and transformer models, (iv) outlining techniques integrating clinical knowledge through multimodal inputs and knowledge graphs, and (v) scrutinising current model evaluation techniques, including commonly applied NLP metrics and qualitative clinical reviews. Furthermore, the quantitative results of the reviewed models are analysed, where the top performing models are examined to seek further insights. Finally, potential new directions are highlighted, with the adoption of additional datasets from other rad
There are very different statistical methods for demonstrating a trend in pharmacological experiments. Here, the focus is on sparse models with only one parameter to be estimated and interpreted: the increase in the regression model and the difference to control in the contrast model. This provides both p-values and confidence intervals for an appropriate effect size. A combined test consisting of the Tukey regression approach and the multiple contrast test according to Williams is recommended, which can be generalized to the generalized linear (mixed effect) model. Thus numerous variable types occurring in pharmacology/toxicology can be adequately evaluated. Software is available through CRAN packages. The most significant limitation of this approach is for designs with very small sample sizes, often in pharmacology/toxicology.
This paper reviews the work done on black hole interior volume, entropy, and evaporation. An insight into the basics for understanding the interior volume is presented. A general analogy to investigate the interior volume of a black hole, the associated quantum mode's entropy, and the evolution relation between the interior and exterior entropy is explained. Using this analogy, we predicted the future of information stored in a BH, its radiation, and evaporation. The results are noted in tables (\ref{tab:1}) and (\ref{tab:2}). To apply this analogy in BH space-time, we investigated the interior volume, entropy, and evaluation relation for different types of BHs. Finally, we also investigated the nature of BH radiation and the probability of particle emission during the evaporation process.
A fundamental mistake in receptor theory has led to an enduring misunderstanding of how to estimate the affinity and efficacy of an agonist. These properties are inextricably linked and cannot be easily separated in any case where the binding of a ligand induces a conformation change in its receptor. Consequently, binding curves and concentration-response relationships for receptor agonists have no straightforward interpretation. This problem, the affinity-efficacy problem, remains overlooked and misunderstood despite it being recognised in 1987. To avoid the further propagation of this misunderstanding, we propose that the affinity-efficacy problem should be included in the core curricula for pharmacology undergraduates proposed by the British Pharmacological Society and IUPHAR.