搜索结果：vibe-coding

Building custom data analysis platforms has traditionally required extensive software engineering expertise, limiting access for many researchers. Here, I demonstrate that modern large language models (LLMs) and autonomous coding agents can dramatically lower this barrier through a process called "vibe coding", an iterative, conversational style of software creation where users describe goals in natural language and AI agents generate, test, and refine executable code in real time. Importantly, the goal here is not to introduce a new analysis platform. Instead, the example application illustrates that, in minutes, LLMs can now perform work that would normally require at least days of manual programming effort, lowering the cost and time investment by orders of magnitude. As a proof of concept, I used vibe coding to create a fully functional proteomics data analysis platform capable of performing standard tasks, including data normalization, differential expression testing, and volcano plot visualization. The entire application, including user interface, backend logic, and data upload pipeline, was developed in less than 10 min using only four natural language prompts, without writing any additional code by hand, at a model usage cost of under $2, not including hosting or personnel time. Previous works in this area have typically required substantial investment of personnel time from highly trained programmers, often amounting to tens of thousands of dollars in total research effort. I detail the step-by-step generation process and evaluate the resulting code's functionality. This demonstration highlights how vibe coding enables domain experts to rapidly prototype sophisticated analytical tools, transforming the pace and accessibility of computational biology software development.

Vibe Coding offers an accessible entry point to create interactive digital applications through human-in-the-loop conversational prompts and AI-assisted workflows. This approach not only lowers technical barriers and fosters experimentation but also raises important considerations. As educators adopt AI in their teaching, it becomes essential to address both its benefits and challenges. We present practical tips to help educators design, develop, and sustain their own educational applications, along with common pitfalls to anticipate when using AI-assisted Vibe Coding. By understanding both the opportunities and the limitations, educators can take confident steps toward creating innovative pedagogically aligned tools for health professions education. The online version contains supplementary material available at 10.1007/s40670-025-02596-1.

Nuclear medicine (NM) educators face a pivotal time of rapid scientific innovation and expansion of theranostic possibilities, rising volumes, staffing shortages, and a call for more high-quality training. In the midst of this expansion, AI technologies have disrupted nearly every aspect of society, including education. Although data on higher-level educational outcomes and patient-specific outcomes are limited, the rapid acceleration and sophistication of AI capabilities is undeniable and ubiquitous, causing many educators to express feelings of uncertainty, incompetence, distrust, and fear. Current literature on AI interventions in health education is limited, often favoring general introductory discussions over the practical application of foundational learning theory and ethical frameworks. In this 3-part article series, we began in part 1 with a foundation of traditional learning theories and ethical principles related to AI use in NM education. This article focuses on the practical application and integration of those foundational principles in the NM educator workflow. Beginning with an educator-focused theoretical framework grounded in transformative and experiential learning theories, readers are encouraged to practically engage with the activities in the article to experience the benefits and failures of AI use in education. Effective AI techniques, including prompt-engineering elements, chain-of-thought prompting, retrieval-augmented generation, custom AI tool development, multimedia content creation, and vibe coding, are introduced with practical activities. Readers should experience the activities directly and reflect on the application and integration in their own educational practice. AI-enabled NM educators who model theory-informed, ethical, and grounded AI use encourage learners to do the same, building interconnected networks of healthy human-AI interactions. These educational activities are naturally scalable and efficient, but with grounded, thoughtful approaches, they can also be equitable, accurate, and personalized. Meeting the rapid dynamics of NM practice requires a rapidly evolving and scalable educational system powered by AI technologies.

BackgroundArtificial intelligence (AI) and machine learning are transforming neurosurgical research and practice, yet the programming barrier has excluded most clinicians from building customized digital tools. Vibe coding - generating functional software through natural language instructions to large language models - has substantially lowered this barrier since its formalization in 2025. No study has examined its applications specifically within neurosurgery.MethodsA narrative review of the literature was conducted using PubMed, Google Scholar, and preprint repositories through April 2026, supplemented by the author's direct clinical experience developing vibe-coded tools in a tertiary neurosurgical unit.FindingsExisting literature on vibe coding in medicine is sparse and limited to non-surgical specialties; no prior publication addresses it in a neurosurgical context. Three practical domains of application are identified: (1) research data collection and multi-scale patient classification, illustrated by a personally developed integrated scoring tool for aneurysmal subarachnoid hemorrhage (Figure 1); (2) clinical workflow optimization including documentation and follow-up automation; and (3) educational tool development and literature engagement.ConclusionVibe coding represents an accessible paradigm enabling neurosurgical trainees to develop purpose-specific digital tools without programming expertise. The field lacks specialty-specific guidance on this approach. This review aims to address that gap and encourage adoption of vibe coding as a practical complement to institutional digital health infrastructure.

A major problem with reviewing the statistical methodology in published medical articles is that extracting the necessary details from large sample sets is time-consuming. This paper demonstrates how a novel automated procedure can extract information about statistical reporting from literature. To illustrate this, we searched the PubMed Central database for original research articles published in 2021 and 2023 to identify the statistical software packages used for data analysis. A key element in terms of transparency and reproducibility is the reporting of the software used for statistical analysis. A freely available Shiny App was created with the help of generative artificial intelligence, and it was used to retrieve automatically information from randomly selected samples of articles indexed in PubMed Central. We analyzed a large sample of articles (n = 1740) to determine the reporting of statistical software for nine study designs. We found that, across different study types, proprietary software such as IBM SPSS Statistics still dominates. Despite multiple calls for greater use of open-source research software, these programs are not used as frequently. In addition, a surprising number of articles did not report the software used. Furthermore, this is the first application of the recent Vibe Coding concept to statistical research methods.

Vibe coding tools enable software generation through natural language prompts powered by generative artificial intelligence, substantially lowering the technical barrier to application development. Their potential to empower health professionals without programming backgrounds is promising, yet remains largely unexplored in the literature. To analyze whether vibe coding tools effectively bridge the gap between non-technical health professionals and digital health application development, and to identify the enablers, barriers, and risks associated with their use. This study used a design science approach to create and preliminarily evaluate digital health artifacts. In a 3-hour in-person workshop, medical students used a vibe coding platform to build mHealth prototypes for dementia care challenges. The process included problem identification, ideation, prototyping, presentation, and reflection. Data came from participant observation, artifact review, and a post-workshop satisfaction survey. All three groups successfully developed functional prototypes within the allotted time. Key enablers included conversational accessibility, immediate visible results, and direct clinical applicability. Critical barriers included unfamiliarity with health data privacy regulations, absence of security measures in all prototypes, and a tendency to define excessively broad problem scopes. Notably, no group implemented user authentication or data encryption. Vibe coding effectively brings health professionals closer to digital application development, but this democratization is not sufficient on its own. It must be accompanied by training in data security, health informatics and AI fundamentals.

To present a feasible workflow for artificial intelligence (AI)-assisted software engineering in dentistry as a technical innovation report. The use of this workflow is illustrated through three self-developed open-source dental applications. Four AI-assisted development approaches were employed: chat-based interfaces of large language models, command-line interface tools, integrated development environments with AI assistance, and agent-based architectures. The dental software applications were created by a single clinician without formal programming training. Three applications were created: (1) VirtualEndo Converter, a Blender add-on for automated CBCT derived STL conversion for augmented/virtual reality (AR/VR), (2) MeshComparisonTool, a 3D Slicer extension for quantitative 3D morphology comparisons, and (3) DentalEmergencyTrainer, an application for simulating dental trauma emergency calls. All the applications are publicly available under the MIT license on GitHub. This report demonstrates that AI-assisted software development can enable dental practitioners without formal programming training to create functional prototypes of applications for research, education, and potentially clinical support. However, the reproducibility of this approach remains to be established, as the three tools were developed by a single clinician, and their clinical deployment would require thorough validation, security auditing, and regulatory assessment. AI-assisted development can help dental practitioners prototype tools that address unmet needs in clinical workflows, research, and education, but clinical use requires cautious separation from validated medical software. Before deployment, such tools require defined intended use, safety evaluation, data-protection safeguards, maintenance plans, and regulatory assessment.

暂无摘要（点击查看详情）

Understanding and predicting epidemiological trends of plant viruses is essential for sustaining crop productivity and control strategies. The National Center for Biotechnology Information (NCBI) GenBank provides nucleotide sequences with metadata such as date, location, and host, offering valuable resources for research. However, GenBank lacks automated tools for visualizing temporal and spatial patterns. To address this limitation, we applied a vibe coding approach, a generative AI assisted method that enables non-programmers to process and visualize data efficiently. As a case study, we analyzed pepper (Capsicum spp.), a major East and Southeast Asian crop threatened by emerging viruses. Using vibe coding, we visualized reporting trends by country and year and mapped sequence variation and conserved regions of pepper-infecting viruses. This approach allowed rapid organization of large datasets and real-time utilization of newly deposited GenBank entries. NCBI-based plant virus analysis system provides automated analysis and visualization and is accessible at https://plantvirus-viewer.duckdns.org/.

The rise of generative artificial intelligence has begun to transform how digital tools are created, shifting development from conventional programming toward natural language prompting, iterative refinement, and AI-assisted generation. Popularly described as "vibe coding," this informal practice refers to building applications, interfaces, automations, or prototypes by describing desired outcomes conversationally and revising outputs until they function as intended. Although the term is playful, it signals a significant shift in who can participate in digital creation and what forms of expertise now matter. This paper argues that "vibe coding" provides a useful contemporary lens for examining an important issue in nursing education: the transition from nurses as users of educational technologies to potential co-designers of digital learning tools. Nursing education has increasingly incorporated simulation, learning management systems, virtual cases, electronic documentation, and AI-enabled supports, yet nurses have often remained positioned as adopters rather than designers of these systems. AI-assisted creation may begin to alter that relationship by lowering barriers to prototyping and enabling educators and students to translate pedagogical ideas into functional tools more quickly. This shift holds promise for more contextually grounded, practice-informed, and responsive educational innovation. At the same time, it raises critical concerns regarding accuracy, bias, accessibility, privacy, pedagogical quality, and the illusion of expertise. As a Contemporary Issues conceptual commentary, the paper explores the educational possibilities and ethical tensions associated with AI-assisted creation in nursing education and argues that the issue should not be approached with either uncritical enthusiasm or outright rejection. Instead, nursing education may need to cultivate critically informed digital co-design capacity. This includes moving beyond basic digital literacy toward critical AI literacy, supporting educators in responsible experimentation, and ensuring that relational, reflective, and equity-oriented commitments remain central. Ultimately, nursing education must prepare future nurses not only to use digital systems, but also to question, evaluate, and, when appropriate, help shape them.

Nuclear medicine (NM) is rapidly expanding with new radiopharmaceuticals, imaging equipment, and theranostic possibilities, necessitating a capable and competent workforce expansion. Consequently, NM education must undergo a transformation, powered in part by artificial intelligence (AI). AI is rapidly entering NM education through learner study materials, assessment design, simulation, feedback, and administrative workflows. AI's value depends less on tool novelty than on ethical alignment with effective learning principles. NM educators and trainees need practical approaches for AI tool use that preserve human judgment, accountability, critical thought, creativity, privacy, and source verification. As the first article in a 3-part series on AI-empowered NM education, this practical guide summarizes and integrates selected literature from AI in health professions education, NM education, learning theories, and ethics. Building on a framework of established learning theories (transformative learning, self-regulated learning, experiential learning, connectivism, constructivism, and cognitive load theory) and ethical concerns about AI (e.g., accuracy, bias, transparency, privacy), we describe effective AI techniques (structured prompt design, retrieval-augmented generation, meta-prompting, AI customization, vibe coding) that can be applied for the NM educator and learner in parts 2 and 3 of the series, respectively. These benefits require explicit safeguards, including AI guardrail incorporation, privacy protection, expert review, AI output transparency, and healthy skepticism. AI should be integrated into NM education through theory-informed, source-grounded, and expert-reviewed workflows. Meeting the demands of increased NM volumes and expanded professional opportunities will require AI-empowered educators and trainees that are able to self-regulate and transform their learning through healthy patterns of human-AI interaction.

Mixing in 96-well plates is a fundamental operation in biology and bioengineering, yet the most common strategy, pipetting up and down to "mix well", is rarely quantified or justified. Rules of thumb such as "three mixing cycles" coexist with protocols that recommend 5, 10, or 20 pipette strokes per step, and there is limited fluid-mechanical guidance on when these choices are adequate. Here, a reduced-order compartment model of advection-diffusion in a single 96-well plate well is developed, in which the fluid is discretized into hundreds of macrocells with finite-volume-style neighbor exchange. A pipette jet is represented as a localized enhancement of exchange along a pipette-aligned "jet footprint," parameterized by an effective jet strength, effective jet footprint size as a proxy for cycled volume fraction, pipette angle, and centered versus off-center tip placement. Repeated aspirate-dispense cycles are simulated and mixing quantified by the decay of normalized concentration variance. The reduced model qualitatively reproduces the mixing hierarchy reported in high-fidelity COMSOL simulations of repetitive pipetting, with higher jet strength and jet footprint substantially accelerating homogenization, while gentle, small-footprint jets often require more than three cycles to reach near-uniform thresholds. Systematic variation of pipette angle and tip position shows that moderately off-center, moderately tilted pipetting modestly improves mixing relative to strictly vertical, centered configurations, with angle and tip position effects remaining small compared to jet strength and footprint size. In parallel, a survey of 96-well pipette mixing instructions in online research forums, commercial assay protocols, and published papers was carried out, extracting explicit "pipetted up and down N times" stroke counts. The resulting distribution varies from 2 to 20 strokes per mixing step, with no clear consensus. This combined modeling and survey analysis indicates that three cycles are not sufficient for many instances of 96-well pipette mixing, and provides a fast, interpretable tool for rationalizing and redesigning pipette-mixing protocols in high-throughput experiments.

Integrating theoretical knowledge with the practical skills essential for clinical practice remains a significant challenge in clinical education. Conventional teaching strategies often fall short in preparing clinicians to navigate the unpredictable, urgent, and multifaceted nature of clinical decision-making, while also providing limited support for the development of cognitive heuristics essential to forming independent clinical judgment. To address these challenges, we introduce vibe coding, a novel AI-assisted, no-code development approach that enables educators to create interactive, customisable learning simulations without programming expertise. By prioritising rapid prototyping and iterative refinement, vibe coding shifts the focus from technical constraints to pedagogical goals, allowing educators to generate code through intuitive, conversational prompts. We applied this approach to develop two distinct applications: the Differential Diagnosis Trainer (DDT), which enhances diagnostic reasoning through randomised clinical scenarios and AI-generated feedback, and the Insulin and Blood Sugar Simulation (IBSS), which offers real-time exploration of metabolic dynamics. Both tools were built using AI-powered no-code platforms, demonstrating significant improvements in accessibility, cost-effectiveness, and scalability. We encourage educators to transition from technology adopters to creators, leveraging AI-driven platforms to develop innovative, scalable, and personalised clinical simulations that transform learning experiences and ultimately enhance patient care.

Many software development platforms now support LLM-driven programming, or "vibe coding", a technique that allows one to specify programs in natural language and iterate from observed behavior, all without directly editing source code. While its adoption is accelerating, little is known about which skills best predict success in this workflow. We report a preregistered cross-sectional study with tertiary-level students (N = 100) who completed measures of computer-science achievement, domain-general cognitive skills, written-communication proficiency, and a vibe-coding assessment. Tasks were curated via an eight-expert consensus process and executed in a purpose-built, vibe-coding environment that mirrors commercial tools while enabling controlled evaluation. We find that both writing skill and CS achievement are significant predictors of vibe-coding performance, and that CS achievement remains a significant predictor after controlling for domain-general cognitive skills. The results may inform tool and curriculum design, including when to emphasize prompt-writing versus CS fundamentals to support future software creators.

Evaluating LLMs is challenging, as benchmark scores often fail to capture models' real-world usefulness. Instead, users often rely on ``vibe-testing'': informal experience-based evaluation, such as comparing models on coding tasks related to their own workflow. While prevalent, vibe-testing is often too ad hoc and unstructured to analyze or reproduce at scale. In this work, we study how vibe-testing works in practice and then formalize it to support systematic analysis. We first analyze two empirical resources: (1) a survey of user evaluation practices, and (2) a collection of in-the-wild model comparison reports from blogs and social media. Based on these resources, we formalize vibe-testing as a two-part process: users personalize both what they test and how they judge responses. We then introduce a proof-of-concept evaluation pipeline that follows this formulation by generating personalized prompts and comparing model outputs using user-aware subjective criteria. In experiments on coding benchmarks, we find that combining personalized prompts and user-aware evaluation can change which model is preferred, reflecting the role of vibe-testing in practice. These findings suggest tha

Clinicians often face workflow problems that are perceived as either too bespoke or low stakes to attract commercial attention. Historically, most do not have the technical knowledge to address these problems, but the recent emergence of "vibe coding" presents a transformative opportunity. Vibe coding refers to the co-development of software using natural language prompts to large language models. It offers a pathway to create simple tools that address these real-world pain points, or to prototype more complex ideas. In this review, written by a group of early adopter clinicians with a range of programming expertise, we introduce vibe coding for clinicians (especially those with no or minimal coding experience) as a way of democratising innovation from the front lines. We discuss foundational skills, outline some common challenges, provide a practical step-by-step playbook, and illustrate this approach with some case examples, taking care to consider caveats and guardrails for deployment. We propose that vibe coding is more than a technical shortcut for beginners and is not a replacement for professional software developers. Instead, it can bridge the gap between clinical insight a

Code-generating tools are increasingly used in software development, yet experience reports on conversational "vibe coding" under production constraints remain limited. This paper presents an experience report from a small full-stack team that applied contextual prompting and explicit architectural constraints to build (i) a multi-project agent learning platform designed for sustained, production-oriented use and (ii) an academic retrieval-augmented generation system. The agent platform supports multiple isolated projects, each with structured memory and background processing, thereby enforcing project-level isolation. The RAG system provides citation-grounded answers, role-based access control, and evaluation tracking. Across both systems, vibe coding accelerated scaffolding and integration. However, the generated code often under-specified isolation rules and infrastructure constraints when these were not explicitly defined. Consequently, aspects such as multi-tenancy, access control, memory policies, and asynchronous processing required deliberate architectural design and verification. We observe a shift in engineering effort from boilerplate implementation toward constraint spe

Data visualization is essential for data analysis and communication, yet creating expressive visualizations remains labor-intensive. Recent AI-driven ``vibe coding'' tools enable users to generate visualizations through natural language interaction, lowering the barrier to entry. However, visualization implementation requires precise alignment between user intent and visual representation, which may differ from general software development practices. We present an empirical study with 16 participants of varying expertise to examine how users employ vibe coding tools for visualization implementation. Participants completed two visualization tasks and a semi-structured interview. Our findings characterize the diverse practices users adopt across prompting, evaluation, and iteration, and surface the challenges they encounter throughout the process.

While large language models (LLMs) have accelerated 2D software development through intent-driven "vibe coding", prototyping intelligent Extended Reality (XR) experiences remains a major challenge. The fundamental barrier is not just the steep learning curve for human creators, but that low-level sensor APIs and complex game engine hierarchies are ill-suited for LLM reasoning, routinely exceeding context windows and inducing syntax hallucinations. To bridge this gap, we contribute XR Blocks, an open-source, LLM-native WebXR framework. Unlike traditional engines, XR Blocks introduces a semantic "Reality Model" that aligns spatial computing primitives (users, physical environments, and agents) with natural language, providing a robust, concise vocabulary optimized for generative AI. Building upon this foundation, we present Vibe Coding XR, an end-to-end prototyping workflow that leverages LLMs to translate high-level prompts (e.g., "create a dandelion that reacts to my hand") directly into functional, physics-aware mixed-reality applications. To minimize the friction of on-device testing, the workflow introduces a seamless desktop "simulated reality" to headset deployment loop. Final