AI tools are increasingly used in college students' academic writing, raising concerns about critical thinking and academic integrity. Prior work has focused mainly on writing performance, whereas psychological and behavioral pathways remain less examined. Guided by Social Cognitive Theory, this study examined a path model linking AI tool use with critical thinking and academic integrity via self-regulated learning and innovative behavior. A cross-sectional survey was conducted with 946 undergraduates from five universities in China. Measures included AI tool use, self-regulated learning, innovative behavior, critical thinking, and academic integrity. The hypothesized model was tested using PLS-SEM with bootstrapping. AI tool use was positively associated with self-regulated learning and innovative behavior. Both intermediaries were positively associated with critical thinking and academic integrity. Bootstrapping supported significant indirect associations from AI tool use to both outcomes through self-regulated learning and innovative behavior. AI tool use in academic writing was associated with higher levels of critical thinking and academic integrity, with self-regulated learning and innovative behavior representing intermediary variables in these associations. The findings may inform AI-supported writing instruction and integrity guidance in higher education.
As generative artificial intelligence becomes integrated into higher education, teachers increasingly rely on AI text-detection reports to support judgments about authorship, writing quality, and academic integrity. Existing research has mainly examined detector accuracy, false positives, fairness, and policy; less is known about whether report design itself shapes teachers' evaluations when the judged text is unchanged. This gap matters because numerical scores and visual warnings may frame interpretation, anchor suspicion, and encourage confirmatory reading under uncertainty. Here, we tested how algorithmic warning strength and visual risk cues affect teachers' evaluations of student writing in a controlled single-stimulus experiment. In a 2 × 2 between-subjects experiment, 214 university teachers evaluated the same medium-quality Chinese social-science course paper accompanied by a fictitious AI-detection report that varied by AI detection rate (7% vs. 87%) and red-highlighting/report-presentation package (absent vs. present). A high detection rate increased perceived AI authorship likelihood and risk and lowered overall quality evaluations, percentage-based scores, originality, language expression, and logical structure. Red highlighting also influenced report perception, language-expression judgments, and self-reported intervention tendency. Significant warning × highlighting interactions emerged for percentage-based scoring, originality, language expression, logical structure, overall multidimensional quality, and intervention tendency, but not for the 1-10 overall rating or report perception. These preliminary and context-specific findings suggest that AI detection reports may function not merely as technical outputs but as socio-technical judgment environments under controlled evaluative conditions. Numerical warnings may anchor teachers' evaluations, while visual risk cues may selectively amplify suspicion and intervention-oriented responses. Responsible use of AI detection therefore requires neutral report design, independent teacher judgment, human oversight, and training on automation bias.
Modern large language models (LLMs) like ChatGPT (based on the GPT-4 architecture) and DeepSeek offer unprecedented capabilities for generating scientific text. However, their performance in replicating structured, high-quality scientific writing, especially compared to human-authored abstracts, remains insufficiently evaluated. To compare the abstract quality produced by human authors, ChatGPT/GPT4, and DeepSeek across: six evaluation criteria Clarity, Coherence, Conciseness, Accuracy, IMRaD Structure, and Language Quality, using blinded expert ratings and non-parametric statistical methods, specifically the Kruskal-Wallis test followed by pairwise Wilcoxon rank-sum tests with false discovery rate correction. We selected 23 medical and healthrelated research topics, each yielding three abstracts (human, ChatGPT, DeepSeek), for a total of 69 abstracts. Three raters scored each abstract. Kruskal-Wallis tests assessed group differences; Cliff's Delta (δ) was calculated as a nonparametric effect size for each comparison, suitable for ordinal data. Across criteria, ChatGPT and DeepSeek significantly outperformed human authors in Clarity, Coherence, IMRaD Structure, and Language Quality. In contrast, Conciseness and Accuracy showed negligible effect sizes (|δ| <0.10), suggesting parity across all three sources. ChatGPT and DeepSeek achieved significantly higher scores in clarity, coherence, structure, and language quality, while showing comparable performance in conciseness and accuracy. These findings complement recent evaluations showing competitive medical and reasoning performance of DeepSeek models compared to proprietary LLMs. While shortform abstracts, expert oversight, and domain expertise remain critical, the results suggest that LLMs-particularly GPT4 and DeepSeek-can serve as effective tools in drafting scientific abstracts.
As generative AI tools become integrated into educational settings, foreign language teaching and learning are gradually adapting. This cross-sectional mixed-methods study examines the relationships between college students' perceptions of intelligent learning environments and their multimodal L2 willingness to communicate (WTC), while analyzing L2 self-efficacy as a potential mediator and foreign language anxiety (FLA) as a moderator. Survey and qualitative data were collected from 960 college students. Quantitative analyses were conducted to examine the associations between perceptions of intelligent environments, L2 self-efficacy, FLA, and WTC across both receptive (listening, reading) and productive (speaking, writing) modalities. Qualitative feedback was analyzed to contextualize these relationships. The survey data indicated a significant positive association between the perception of intelligent environments and overall WTC. Although baseline WTC for productive skills was lower than for receptive tasks, favorable perceptions of AI showed comparable positive associations with communication willingness across both modalities. L2 self-efficacy served as an indirect link, accounting for 56.19% of the total association. Exploratory analyses indicated that high self-efficacy exhibited a potential buffering tendency against the negative correlation between FLA and WTC, though this interaction did not reach statistical significance (p = 0.068). Multi-group analyses demonstrated that these positive associations were consistent across different language proficiency levels. Qualitatively, AI platforms were characterized as perceived "psychological safety buffers" with lower social-evaluative risks. These cross-sectional insights suggest that favorable perceptions of intelligent environments are positively related to L2 communication willingness. Rather than treating these environments merely as technical aids, educators might leverage their low-pressure attributes as supportive digital scaffolding while monitoring cognitive dependence to encourage eventual autonomous communication.
Aphasia is a neurological condition characterized by partial or total loss of verbal communication, most commonly caused by stroke. In hospital settings, brief and reliable tools are essential to identify and quantify aphasia early, enabling prompt diagnosis and appropriate rehabilitation planning. However, Brazil lacks language assessment instruments adapted for bedside use. To perform the cross-cultural adaptation and validation of the Bedside Language test for Brazilian Portuguese. The BL underwent translation, back-translation, expert review, and cultural adaptation. The final version was administered to 70 participants. Two evaluators applied the instrument on the same day to assess inter-rater reliability, and the same examiner reapplied it within two days to assess test-retest reliability. The Bedside Evaluation Screening Test (BEST-2) language test was also used for concurrent validity. The Brazilian version of the BL demonstrated high internal consistency (Cronbach's alpha: spontaneous reading - LE=0.80; comprehension - CO=0.86; repetition - RE=0.95; writing - ES=0.73; overall=0.87), as well as satisfactory intra-rater (p=0.22) and inter-rater (p=0.77) reliability. It showed a strong correlation with the BEST-2 across all linguistic domains except writing. The Brazilian Portuguese version of the BL is a brief, valid, and reliable bedside screening tool for identifying aphasia in patients with ischemic stroke, supporting high-quality care in specialized stroke centers. A afasia é uma condição neurológica caracterizada pela perda parcial ou total da capacidade de comunicação verbal, cuja causa mais comum é o acidente vascular cerebral (AVC). No contexto hospitalar, são essenciais instrumentos breves e confiáveis para identificar e quantificar a afasia precocemente, favorecendo o diagnóstico rápido e o planejamento adequado da reabilitação. No Brasil, há escassez de instrumentos de avaliação linguística adaptados para uso à beira-leito. Realizar a adaptação transcultural e validação do Bedside Language test para o português do Brasil. O BL passou por etapas de tradução, retrotradução, análise por especialistas e adaptação cultural. A versão final foi aplicada a 70 participantes. Dois avaliadores aplicaram o instrumento no mesmo dia (para avaliar a confiabilidade interavaliadores), e o mesmo pesquisador o reaplicou após dois dias (para avaliar a confiabilidade teste-reteste). O teste Bedside Evaluation Screening Test (BEST-2) foi utilizado para análise de validade concorrente. A versão brasileira do BL apresentou alta consistência interna (alfa de Cronbach: leitura espontânea — LE=0,80; compreensão — CO=0,86; repetição — RE=0,95; escrita — ES=0,73; total=0,87), bem como confiabilidade intra-avaliador (p=0,22) e interavaliadores (p=0,77) satisfatória. Observou-se forte correlação com o BEST-2 em todos os domínios linguísticos, exceto escrita. A versão em português brasileiro do BL mostrou-se um instrumento breve, válido e confiável para o rastreio de afasia em pacientes com AVC isquêmico, apoiando o cuidado de qualidade em centros especializados em AVC.
A dermatologist may choose to become an autobiographical author. Once that decision has been made, there are many potential writing presentations that have been used by these individuals to share their information. Writing a comprehensive autobiography and presenting the content in a book is the classical approach. A variation on the traditional presentation of providing a complete autobiography has been adopted by two dermatologists; these individuals elected to only discuss their transition from being a physician to becoming a patient and to focus the novel that they wrote on a single life threatening condition. However, several alternative methods have been embraced by other dermatologists. For example, a dermatologist selected varous dermatologists and asked them to answer a series of questions that would allow them to share autobiographic information in their replies; he subsequently edited two volumes of a book that included the content he received in short chapters from 100 dermatologists; thereafter, using a similar format, he created and edited a monthly journal feature that succinctly summarizes the professional acheivements and personal insights of a dermatologist each issue. Monthly features that are published in dermatology magazines provides two dermatologists an opportunity to share insights into not only their personality and perspective toward dermatology but also their interactions with patients. In addition, one dermatologist has decided to publish individual case reports that discuss not only his benign but also his more serious medical conditions. In conclusion, dermatologists who publish as autobiographical authors are a unique group of highly motivated individuals who have a sincere passion for sharing personal aspects of their life journey.
Artificial intelligence (AI) promises to rapidly transform healthcare, but as of 2026, the direct implications for clinical care supporting individuals with cerebral palsy (CP) are unclear. In this narrative review, several publications were identified evaluating AI-based tools to aid in early diagnosis of CP or in functional assessment of individuals with CP. As of the time of writing, one application (a computer-adapted version of the Gross Motor Function Measure developed using AI) has been prospectively validated. Many additional clinical applications had Oxford Centre Level 2b Evidence (exploratory diagnostic studies), employing cross-validation using existing data sets to demonstrate plausibility without prospectively evaluating performance within intended contexts of use. Based on this review, we summarize recurring challenges and review potential barriers to implementation including scientific, regulatory, and ethical considerations. We anticipate that the performance of AI-based tools will continue to evolve after regulatory approval, requiring ongoing vigilance for clinicians who employ AI-based tools. We believe that meaningfully including people with lived experience throughout the research lifecycle is critical for ensuring that AI-based tools are developed in a responsible and patient-centered fashion while augmenting, not replacing, clinicians' ability to provide high-quality care.
Generative AI systems increasingly support users in coding, data analysis, and creative tasks through natural-language interaction. However, user prompts are often underspecified or ambiguous, and current LLM-based assistants typically proceed with a plausible interpretation, leaving misalignments to be discovered only after the output is inspected or executed. This behavior can trigger costly trial-and-error prompt revisions and may be amplified by sycophantic tendencies that reinforce incorrect assumptions. We present a progressive disambiguation framework that resolves prompt uncertainty before producing a final answer. The framework separates intent clarification from solution synthesis by guiding users through a structured pre-generation dialogue that elicits missing constraints, compares alternative interpretations, and illustrates consequences via targeted input-output examples and representative edge cases. In addition, it performs incremental constraint consistency checks that flag implausible or conflicting user-provided assumptions and request explicit confirmation before generation. After incompatible interpretations are pruned and constraints are validated, the system generates a single, intent-aligned solution. Experiments on a diverse benchmark covering coding, data analysis, and creative writing show that our approach improves output accuracy, reduces corrective iterations and overall user effort, and achieves competitive end-to-end resolution time and higher user satisfaction compared to one-shot and reactive clarification baselines.
Direct laser writing via multiphoton polymerization (MPP) lithography has significantly advanced micro- and nanofabrication, yet its applicability remains largely confined to radical polymerization of acrylates and methacrylates. Here, we demonstrate that the ketocoumarin 7-diethylamino-3-thenoylcoumarin (DETC), widely used as a photoinitiator in subdiffractional radical polymerization lithography inspired by stimulated emission depletion (STED) microscopy, also effectively initiates oxidative step-growth polymerization of EDOT, enabling direct lithographic formation of sub-100 nm PEDOT nanostructures using visible light. Two approaches for subdiffraction MPP are pursued: (i) slow scanning with low excitation power, allowing for the influx of oxygen into the illumination point spread function (PSF), leading to chemical quenching of the DETC triplet state, and (ii) STED-inspired transient state absorption depletion lithography, where the DETC triplet states are optically quenched in the outer rim of the excitation PSF. Approach (i) yields 65 nm linewidths at the cost of an ultra-slow scan speed of 2.5 μm/s, while approach (ii) yields 105 nm wide lines at 100 μm/s. Raman spectroscopy suggests that conductivity in MPP-written PEDOT wires arises only when graphitic phases are formed.
This study examines how three large language models (LLMs), ChatGPT, Claude, and Gemini, assign grades to undergraduate-level essays in a biology course using a standardized rubric. Each LLM evaluated a data set of 200 essays under two prompting conditions: zero-shot (uncalibrated) and few-shot (calibrated using a small set of exemplar essays). LLM-assigned scores were directly compared with instructor-assigned scores, showing only moderate alignment with instructor grading, with variability observed across models and prompting strategies. Differences in grading behavior were also evident with different items on the rubric, with higher alignment for structural writing components and lower alignment for content- and reasoning-based criteria. Additionally, LLMs showed greater agreement with instructor scores than with one another, indicating substantial inter-model variability under identical grading conditions. These findings suggest that LLM grading outputs vary meaningfully across models, prompting strategies, and rubric components. In this context, LLMs may be best understood as tools that can support specific aspects of structured grading rather than as interchangeable evaluators.
Lithographic patterning of semiconductor materials is essential for most modern optoelectronic devices. However, traditional inorganic and nanocrystal derived resists exhibit low electron-dose sensitivity, weak solubility contrast, and limited chemical compatibility, restricting high resolution functional lithography. Here, we present a molecular complex platform that converts ordinary metal halides (MXn; nine compositions) into intrinsically electron beam responsive, solution processable resists for direct write electron beam lithography. Coordination of MXn with oleylamine yields metal-ligand complexes with comparatively low dose sensitivity among additive-free inorganic resists (0.81 mC cm-2), high contrast (γ = 3.1), and sub-30 nm resolution. Across the tested metal halide library, resist sensitivity shows an exponential dependence on molecular weight, establishing the first universal scaling relationship for molecular resist energetics. Mechanistic studies reveal that electron irradiation induces bond cleavage and coordination network collapse, generating metal halide domains with high structural fidelity. The patterned nanostructures retain optical functionality, nanodots displaying super linear PL excitation (α > 1) characteristics. Furthermore, sequential multilayer writing enables deterministic RGB nano-pixel architectures, exemplified by registered 3.9 × 104 pixel full color parrot micrograph. This additive free, tunable molecular resist system provides a high-resolution lithography route for scalable quantum photonic and optoelectronic fabrication.
The current study examined whether imagining a future secure relationship can increase state attachment security to a degree comparable to recalling a secure past or current relationship. In a within-subjects design, 88 participants completed three weekly tasks: future thinking, episodic recall, and a neutral shopping control. Each session involved a brief imagery-and-writing task followed immediately by state attachment measures. Compared with the neutral control, both relational priming tasks were associated with higher state attachment security, and the future and recall tasks did not differ. Effects on attachment anxiety and avoidance were small and did not reach conventional levels of statistical significance after correction. These findings provide preliminary evidence that episodic future thinking can prime felt security in the short term, while also indicating that stronger evidence is needed before drawing conclusions about functional or intervention-related significance.
Generative artificial intelligence (AI), particularly large language models, is rapidly becoming embedded in academic research and scholarly publishing. These systems assist with drafting, literature synthesis, and analytical writing, increasingly contributing to the production of academic text. This shift raises a central question: how should higher education institutions evaluate scholarly contribution when parts of research production become technologically mediated? This paper examines governance misalignment between publication standards and institutional evaluation systems in higher education. Drawing on an exploratory qualitative survey of 18 journal editors and associate editors across business-related disciplines, we analyze editorial perspectives on AI-assisted manuscript preparation, authorship, accountability, productivity, and academic evaluation. We identify three recurring concerns: policy fragmentation, ambiguity surrounding authorship and accountability, and apprehension about AI-enabled productivity acceleration. We introduce the concept of metric distortion to describe the weakening relationship between measurable scholarly outputs and the intellectual labor those outputs are assumed to represent under conditions of AI-assisted production. As drafting and textual production become more technologically scalable, publication-based indicators may become less reliable proxies for conceptual contribution and interpretive effort. Building on scholarship on academic capitalism, audit culture, and digital governance, the paper argues that generative AI functions as a stress test for existing evaluation regimes. We propose a process-based framework emphasizing disclosure, documented intellectual contribution, and institutional alignment between editorial governance and tenure evaluation systems.
Failure to Rescue (FTR) refers to patient death following a complication despite opportunities for timely recognition and intervention. Residents are particularly vulnerable during FTR events because they provide frontline care while training within hierarchical systems. Although FTR is widely used as a quality indicator, its impact on residents remains underexplored. This study examined residents' emotional responses, coping strategies, and the influence of supervisory behavior and institutional culture on recovery and learning. We conducted a qualitative interview study following constructivist grounded theory (CGT). Semi-structured interviews were performed with residents from multiple Swiss hospitals between October 2022 and May 2023. Interviews explored emotional experiences related to FTR events, coping mechanisms, supervisory interactions, and perceptions of institutional culture. Data was analyzed iteratively using constant comparison, memo writing, and reflexive team discussions. Fifteen residents (4 males, 11 females), primarily from surgery and internal medicine, participated. Mean age was 30.6 years, with an average of 2.6 years of residency experience. Four interrelated themes emerged: (1) temporal layering of emotional responses, from shock to insecurity and rumination; (2) problem- and emotion-focused coping strategies, including vigilance, peer support, and counseling; (3) supervisory responses, where guidance, debrief, or silence shaped emotional outcomes; and (4) institutional error culture, which influenced expectations around coping and responsibility. Participants described self-doubt, avoidance, and hypervigilance that evolved into either constructive learning or prolonged distress depending on team and organizational support. FTR events can leave emotional and professional effects on residents. Supportive leadership, structured debriefings, and accessible psychological support facilitated recovery and learning, whereas silence and blame intensified distress and avoidance. These findings informed a model of contextually mediated recovery in which supervisors serve as key interpreters of failure, while institutional culture facilitates or constrains this role. Coping with FTR is a collective responsibility embedded within team and institutional culture.
Assessing professionalism in a valid, objective, and cost-effective manner remains a persistent challenge in clinical education. In Indonesia, this issue is particularly relevant in surgical training, where residents must integrate technical competence with professional behaviour in high-stakes, resource-limited settings. This study aimed to develop and evaluate the educational suitability of a Situational Judgement Test (SJT) combined with guided written reflection to assess and promote professionalism among general surgery residents. A mixed-methods sequential design was used. An SJT was developed through blueprinting, item writing, and expert validation, resulting in 26 scenarios across six professionalism domains. Residents (n = 64 at baseline; n = 44 completers) completed a proctored pre-test SJT, participated in an 8-week guided reflection program (four submissions), and completed an asynchronous post-test SJT. Educational suitability evidence was examined through content validity (CVI) and internal consistency (Kendall's W). No construct or criterion validity testing was performed. Reflections were scored using Kember's four-level reflective framework by three independent assessors with Fleiss' Kappa for inter-rater agreement. Changes in SJT and reflection depth were analysed using paired tests and correlation analysis. Data were analysed using SPSS version 23. The mean CVI across items was 0.97 (range 0.90-1.00). The mean SJT score increased significantly from 386.59 ± 19.31 to 420.45 ± 12.56 among completers, with a mean difference of 33.86 ± 16.03 (p < 0.001). Reliability improved from Kendall's W = 0.38 to 0.52. Sensitivity analysis comparing baseline pre-test scores between completers (386.6 ± 19.3) and dropouts (391.0 ± 15.5) showed no significant difference (Mann-Whitney U, p = 0.258), indicating limited attrition bias. Reflective depth progressed from median level 1 to 4 across four submissions (median Δ = +3, p < 0.001), with substantial inter-rater agreement (κ = 0.68). There was no significant correlation between Δ reflection and Δ SJT (r = 0.059, p = 0.703). Integrating SJT with guided reflection demonstrated strong content validity, internal consistency, and educational benefit, supporting its suitability for formative educational use in surgical professionalism training.
Extensive bone defects complicated by infection, malignancy, or metabolic disorders remain a critical clinical challenge, as conventional calcium phosphate bioceramics provide only passive osteoconductive support. The integration of additive manufacturing with external physical stimuli, such as mechanical, piezoelectric, photothermal, magnetothermal, and ultrasonic, has catalyzed a paradigm shift from static scaffolding to responsive therapeutic platforms. This review examines how advanced techniques, including digital light processing (DLP), direct ink writing (DIW), and two-photon lithography (TPL), enable precise architectural programming of porosity, topology, and compositional gradients, establishing the physicochemical foundation for efficient field coupling. We dissect the mechanisms by which field-active bioceramics transduce external stimuli into bioelectrical, thermal, and mechanical cues, activating the mechanotransduction pathway that orchestrates osteogenic differentiation, immunomodulation, angiogenesis, and antibacterial activity. Particular emphasis is placed on multifunctional strategies, including tumor ablation-to-regeneration transitions, antibacterial-to-osteogenic modality switching, and 4D-printed shape memory architectures, alongside emerging self-powered systems harvesting endogenous mechanical energy. By elucidating the synergistic interplay among scaffold structure, material composition, and external field stimulation, this review establishes design principles for next-generation biomaterials that adaptively respond to complex bone-defect microenvironments.
Arginine methylation is a common post-translational modification that exists in three distinct forms-monomethylation, asymmetric dimethylation, and symmetric dimethylation-through which it regulates precursor RNA splicing and maintains cellular homeostasis. Dysregulation of the writing, reading, or erasure of arginine methylation promotes cancer development. Recent studies have identified PRMTs as key regulators of alternative splicing, and aberrant PRMT-driven splicing directly impacts multiple biological processes, including tumor proliferation, apoptosis resistance, metastasis, and immune evasion. This review focuses on the molecular mechanisms by which PRMTs regulate alternative splicing, their connections to oncogenic processes, and the therapeutic implications and challenges of targeting the PRMT-splicing axis in cancer.
Automatic Essay Scoring (AES) aims to evaluate the quality of written essays automatically, providing fast, consistent, and objective assessments of students' writing ability. Existing deep learning approaches-including recurrent, convolutional, and transformer-based models-primarily focus on textual semantics, yet they often overlook the spatio-temporal nature of essay composition, where meaning evolves across sentences and paragraphs through discourse progression. To address this gap, this study presents a prompt-aware Spatio-Temporal Graph Neural Network (PaSTO-GNN) for AES. In this framework, each essay is first segmented into sentences, and each sentence is represented as a node in a spatio-temporal graph. The feature representation of each node is constructed by combining contextual sentence embeddings extracted from a RoBERTa encoder adapted via Low-Rank Adaptation (LoRA), semantic embeddings obtained from Sentence-BERT (SBERT), and a learned prompt embedding that conditions scoring on the essay prompt. Spatial edges capture semantic relationships between sentences, while temporal edges encode the sequential progression of ideas throughout the essay. The resulting node representations are processed through a spatio-temporal message passing network, followed by a BiGRU layer and temporal attention pooling to obtain a global essay representation. To model the ordered nature of essay scores, prompt-specific ordinal prediction heads based on CORAL are employed, together with a per-prompt calibration step that better aligns predicted scores with human scoring distributions. Experimental results on the AES 2.0 benchmark dataset show that PaSTO-GNN achieves a Quadratic Weighted Kappa (QWK) of 0.8329 on the validation set after prompt calibration and a Pearson correlation of 0.8168, highlighting the effectiveness of combining spatio-temporal discourse modeling with prompt-aware representations for automated essay evaluation.
Generative AI is becoming widely used in medical education, but what drives medical students to continue using domestic tools such as DeepSeek after initial adoption remains poorly understood. This study examined the psychological mechanisms underlying medical students continued use of DeepSeek, with particular attention to how cognitive appraisals, satisfaction, and task fit shaped continuance in real learning contexts. An explanatory sequential mixed-methods design was used. In the quantitative phase, 630 valid questionnaires were analyzed using structural equation modeling to test a continuance pathway centered on cognitive appraisal, satisfaction, and behavioral intention, while interview data were used to explain unexpected and nonsignificant quantitative findings. System quality and subjective norm positively affected perceived ease of use, while subjective norm and expectation confirmation positively affected perceived usefulness. Perceived ease of use and perceived usefulness both increased satisfaction, and satisfaction was the strongest predictor of continuance intention. Task-technology fit also positively influenced continuance intention, which strongly predicted actual continued use. Technology characteristics and task characteristics both improved task-technology fit. By contrast, information quality negatively affected perceived ease of use, subjective norm negatively affected satisfaction, and privacy concerns and expectation confirmation did not significantly affect continuance intention or satisfaction. Students mainly continued using DeepSeek because it was easy to access, helpful for academic writing and exam preparation, and suited to some specialized tasks; their primary concerns were unstable performance, inaccurate outputs, future pricing, and data security. Continued DeepSeek use followed a cognitive-affective-behavioral sequence: perceived ease of use and usefulness drove satisfaction, which in turn predicted continuance intention (β = 0.769), while task-technology fit provided an independent behavioral pathway (β = 0.157), together accounting for actual continued use (β = 0.732).
The Little Book of Bad Wounds by Yvette Godwin is a concise field manual intended to guide civilian surgeons managing complex traumatic and wartime wounds in resource-limited environments. Written from the perspective of frontline surgical experience rather than tertiary academic practice, the text emphasizes practical wound stabilization strategies over definitive reconstruction and introduces the concept of the "delayed acute wound" (DAW), a wound that may present days after injury yet still requires acute management principles. This editorial evaluates the book's practicality, readability, and consistency with current wound care literature. Godwin work succeeds in presenting a highly accessible framework for the management of blast and ballistic injuries, particularly its "blueprint recipe" structure, which allows for rapid bedside reference. The text appropriately emphasizes repeated debridement, antimicrobial stewardship, preservation of reconstructive options, and the clinical significance of biofilm in delayed traumatic wounds. Additional strengths include pragmatic discussion of austere-resource solutions, realistic commentary on limb salvage versus amputation, and a writing style that remains engaging without sacrificing clinical utility. Several recommendations, however, differ from current guideline-supported practice and warrant contextualization. These include the use of high-concentration povidone-iodine solutions, broad rejection of pulse lavage, omission of negative pressure wound therapy, and selective recommendations regarding topical antimicrobials and antibiotic timing. While many of these viewpoints are grounded in extensive field experience, they should be interpreted as practice-based recommendations rather than universal standards of care. Despite these limitations, The Little Book of Bad Wounds provides a valuable and timely contribution to modern reconstructive and wartime wound literature. As civilian health care systems increasingly encounter injuries associated with armed conflict and mass trauma, this work offers surgeons a practical framework for stabilizing complex wounds in compromised circumstances. When used with appropriate clinical discernment, the book serves as an effective field guide for surgeons managing devastating injuries under austere conditions.