Two decades of the CHNRI method (2006–2025): Tracking its evolution and contribution to the emerging field of ideometrics

This paper tracks the evolution of the Child Health and Nutrition Research Initiative (CHNRI) method for setting health research priorities and situates it within a much broader, emerging field of 'ideometrics' - the quantitative study of how ideas can be generated, evaluated, and prioritised. First presented in 2006, the CHNRI method tackled three key barriers to research priority setting: an infinite number of possible research ideas, uncertainty about future payoffs of investing in research, and the need for a fair, transparent, legitimate, and broadly acceptable consensus. Its proposed solutions were based on the systematic nature of idea generation, explicit context framing, transparent criteria, and expert crowdsourcing, while its scores reflected 'collective optimism' towards many ideas that could be optionally weighted by funders and stakeholders. Early demonstrations of its usefulness were followed by the establishment of a landmark World Health Organization (WHO) programme that set priorities across the leading causes of global child mortality. The resulting publications catalysed adoption of the method by major agencies and many national governments. Within a decade, the CHNRI method became the most widely used approach to health research priority setting. The review of the first 50 exercises revealed its practical advantages: its systematic scope, transparency, inclusiveness, flexibility, simplicity, low cost, and publishable outputs. Its 'natural evolution' within the global health research community led most users to sensibly adapt its standard criteria to their specific contexts. Experiments on quantitative properties of human collective knowledge and opinion demonstrated accuracy within domains of expertise. They also showed that saturation of experts' collective opinion occurs with 45-55 scorers, achieving very stable rankings. Subsequent advances introduced bootstrapped confidence intervals, an information-theory expert agreement metric, and clustering analysis to detect scorer sub-structures, strengthening the method. Consultations with funders clarified 'funding attractiveness' as a complementary criterion, improving the method's policy traction. By the year 2025, the CHNRI method underpinned major exercises led by the leading international organisations in all of the world's regions, and supported research prioritisation in many challenging national and regional settings. A pivotal recent shift is the integration of artificial intelligence (AI)-based large language models: the CHNRI method can now accommodate AI as a partner in all steps of the priority-setting process. Moreover, years of CHNRI practice motivated a broader theoretical move: viewing the brain's 'perception of ideas' as an underappreciated human sense. These advances call for a more quantitative, testable, and replicable future developments in which the CHNRI method will contribute to 'ideometrics' - an emerging scientific field devoted to generating, evaluating and prioritising ideas that are likely to lead to human progress in health and beyond.