[ad_1]
Abstract
Objectives To describe the implementation of a multidisciplinary, ethically grounded hackathon as a model to develop and evaluate generative AI (GenAI) solutions for real-world clinical challenges within a hospital setting.
Methods The GenAI Health Hackathon (GAHH) organised at Hospital Clínic de Barcelona included 13 challenges were selected via an internal call based on clinical impact, feasibility and data availability. Participants accessed anonymised real-world data through a secure cloud environment. Teams employed large language models and retrieval-augmented generation to build prototypes addressing tasks such as clinical text structuring, decision support and workflow automation. Human-in-the-loop validation, explainability and regulatory safeguards were emphasised.
Results The hackathon yielded multiple AI prototypes tested on real data. Results varied: entity recognition reached 90.5% accuracy, summarisation >90% clinician concordance and nutritional models achieved F1 scores of 0.75–0.93. Lower scores (F1<0.52, Jaccard Index <0.4) were seen in complex reasoning or multilingual tasks. Bias was explored in 10 projects, with mitigations such as stratified sampling, prompt tuning, disclaimers and expert oversight. A transferable framework was proposed to replicate responsible GenAI hackathons in clinical contexts.
Discussion Interdisciplinary collaboration and real-world testing proved essential for aligning GenAI with clinical needs. The hackathon revealed challenges in bias, evaluation and integration but offered a transferable framework for responsible innovation under General Data Protection Regulation and the European Union Artificial Intelligence Act.
Conclusions The GAHH demonstrated that GenAI can be safely and effectively applied in healthcare with rigorous governance and interdisciplinary collaboration, offering a scalable model for responsible AI innovation.
Background
Generative artificial intelligence (GenAI) has the potential to transform multiple sectors, including the health sciences, through its advanced capabilities in natural language processing powered by large language models (LLMs).1 In clinical practice, GenAI offers promising applications such as reducing documentation workload, supporting clinical decision-making, improving the availability of data for research purposes, automating administrative processes and enhancing education for both patients and healthcare providers.2–5 For instance, LLMs such as Generative Pre-trained Transformer-4 (GPT-4) have been used to generate discharge summaries and progress notes, helping to alleviate the documentation burden and reduce physician burnout6; generative models have demonstrated near-passing performance on clinical decision-making tasks, including differential diagnosis and treatment planning in complex patient scenarios7; and GAN-based tools have enabled the creation of realistic synthetic health records that support research while protecting patient privacy.8
Despite these potential benefits, the adoption of GenAI in clinical settings remains limited.9 Several factors contribute to this slow integration. First, there is a lack of robust and consistent, evidence-based data demonstrating a favourable impact of GenAI tools on key clinical outcomes.10 Second, these systems present significant technical challenges, particularly regarding explainability, transparency and the risks of generating plausible but factually incorrect or fabricated information (hallucinations) and biased outputs.11 Third, the deployment of GenAI solutions is resource-intensive, creating obstacles related to their integration within the existing healthcare information systems. Fourth, in Europe, those tools need to comply with important regulatory frameworks such as General Data Protection Regulation (GDPR) and, more recently, with the new European Union Artificial Intelligence Act (EU AI Act).12 13 A sign of the prevailing caution surrounding generative AI in healthcare settings is that, as many recent surveys in the US and UK show,3 14 15 most efforts are either in the piloting or early deployment phase, with very few full-scale institutional adoptions. Furthermore, there is evidence of low adoption of GenAI due to clinician scepticism.16 17
The pathway from identifying a clinical problem suitable for a GenAI solution to its actual deployment will need the participation of multiple professional roles.18–20 Hackathons are competitive events aimed at proposing functioning engineering solutions on specific topics mainly related to data science and programming.21 22 Hackathons gather professionals from various sectors to work collaboratively and intensively in specific challenges during a predefined time frame, having the potential to unravel innovative solutions for specific challenges. To date, the only published experience involving generative AI in a healthcare setting has been reported by Small et al,23 who designed an initiative aimed at training and engaging New York University (NYU) Langone Health staff in the responsible use of GenAI. While participants interacted with real clinical data to simulate tasks such as summarisation, extraction and transformation, the activity was conducted in an educational setting, without direct application to real-time clinical workflows or patient care. As such, no prior study has reported the deployment of generative AI to perform clinically relevant tasks on authentic patient data under a production or operational framework—particularly one compliant with the GDPR and aligned with the EU AI Act’s risk evaluation framework. To our knowledge, the present work constitutes the first documented implementation that meets these criteria.3–5 24
Conclusions
The GAHH at HCB demonstrated that AI-driven innovation can be effectively applied to real-world healthcare challenges through interdisciplinary collaboration and structured evaluation. Overcoming barriers to AI adoption—such as data access, explainability and regulatory compliance—requires ongoing effort, but the hackathon serves as a key milestone toward integrating AI into clinical practice.
Future directions include implementing and scaling the most promising AI solutions from the hackathon. Efforts will focus on securing institutional support for validation, piloting selected projects within hospital workflows and fostering long-term collaborations between researchers, clinicians and AI developers. Plans are also underway to host future editions of the hackathon, expanding participation and refining the framework based on lessons from this first edition. The pioneering nature of the initiative highlights its potential for broader scalability, setting a precedent for AI-focused innovation events in healthcare worldwide. The transferable framework developed for organising responsible GenAI hackathons in healthcare, combining secure data governance, structured mentorship and built-in evaluation across all phases, serves as a blueprint for institutions aiming to explore generative AI safely and ethically in clinical settings.
[ad_2]
Source link
