Enabling Recognition and Reward of Artificial Intelligence Research Objects

Le résumé fourni par la source

Researchers’ use of Artificial intelligence (AI) in biomedical research has generated a new class of research outputs for recognition. Since incentives, metrics, and impact stories remain largely human-centric, AI outputs can lack proper governance and context, potentially impacting reuse and reproducibility. This gap can be addressed through technical and social infrastructures to incentivize and reward AI research products. Make Data Count (MDC) advances data as a first-class research object, promoting responsible metrics for evaluating data use and impact. MDC supports repositories like Zenodo in producing compliant, high-quality data outputs by providing guidance and infrastructure for data metrics, metadata, and citation practices. Last year, MDC and partners produced a public Toolkit which includes: (1) a maturity model for evaluating institutional readiness; (2) a practical guide with implementation-ready resources; and (3) a set of real-world case studies. These efforts can help inform recognition and reward of AI generated outputs. Here, we extend MDC principles and the Toolkit to consider AI-enabled research outputs. We have developed a comprehensive structured taxonomy of AI research outputs to inform efforts. Other updates include an enhanced maturity model, guidance for communicating AI contributions on researcher CVs and to reviewers; and an initial set of use cases in Zenodo from several AI projects, including HF-ETIOLOGY and NUCATS, among others. This presentation will also address ethical considerations, provide context for other NIH sharing and reuse initiatives, and recommend additional activities to further support and advance recognition and reward of AI research outputs. By advancing infrastructure that captures the full spectrum of scholarly products, this work helps enable more meaningful evaluation, accelerates data-driven discovery, and supports an open and trustworthy ecosystem for AI-driven biomedical research. Acknowledgement This work was inspired and supported, in part, by the projects noted below. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH or NSF. HF-ETIOLOGY: Heart Failure Endotypes from Ethical, Multi-Modal AI driven and Molecular/Phenotypic Data Integration Enabled Discovery (NIH/OD 1OT2OD038083-01) NUCATS: Northwestern University Clinical and Translational Sciences Institute (NIH/NCATS UM1TR005121) Zenodo and the Generalist Repository Ecosystem Initiative (GREI) (NIH/OD 1OT2DB000013-01) Assigning comprehensive, standardized sample annotations to enhance the ability to discover, use, and interpret millions of omics profiles (NSF 2328140) HeartShare DeCODE-HF: Data translation center to Combine Omics, Deep phenotyping, and Electronic health records for Heart Failure subtypes and treatment targets (NIH/NHLBI 5U54HL160273-05)

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.