Related Experiment Video
Updated: Oct 28, 2026
Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System
Published on: July 11, 2025
Assessment of comprehensiveness of national cancer control plans using a multiagent artificial intelligence model: a
Caleb J Kumar1, Che L Reddy1, Hugh Pearson1
1Health Systems Innovation Lab, Department of Global Health and Population, Harvard T H Chan School of Public Health, Harvard University, Boston, MA, USA.
Background:
National cancer control plans (NCCPs) vary in scope and comprehensiveness, yet expert-led review is resource intensive and limits cross-country comparability. We aimed to assess NCCP comprehensiveness across 50 countries using the policy reasoning integrated sequential model (PRISM), a validated multiagent artificial intelligence (AI) system.
Methods:
We identified the most recent national NCCP for each country from the International Cancer Control Partnership repository, the WHO Noncommunicable Disease Document Repository, and government websites; sub-national plans, broader non-communicable disease or health-sector strategies, drafts, and superseded versions were excluded. We developed an analytical framework of 12 elements and 76 sub-elements, aligned with WHO, International Cancer Control Partnership, and Union for International Cancer Control guidance and validated it through a modified two-stage Delphi process (online survey of 67 experts; triangulation workshop with 31 participants). PRISM comprised document ingestion (Qwen2.5-VL-72B), policy analysis with retrieval-augmented generation (Llama 4 Scout, 109B), and semantic quality assurance. PRISM was validated against masked dual-expert ratings for six purposely selected NCCPs, then applied to all NCCPs. Scores were normalised to 0-5 and medians compared across World Bank income group and WHO region.
Findings:
We identified 50 NCCPs that met the inclusion criteria. Median overall comprehensiveness was 2·78 (IQR 2·25-3·09), with high comprehensiveness classified as 4·0-5·0, moderate as 2·0-3·9, and low as 0-1·9. Europe (2·96; 2·78-3·35) and lower-middle-income countries (2·97; 2·48-3·15) had the highest medians; the western Pacific (2·21; 1·78-2·87) and low-income countries (2·05; 1·94-2·22) had the lowest. Element-level medians were highest for implementation (4·17; 3·54-4·38), health services (3·89; 2·63-4·19), and governance and organisation (3·67; 3·35-3·98), and lowest for financing (1·88; 1·00-2·69), outcomes (1·83; 1·33-2·33), and threats (2·19; 1·72-2·34). Financing showed the largest gaps-only five of eleven financing sub-elements had median scores of 1·0 or less.
Interpretation:
Our analysis suggests that NCCP comprehensiveness varies substantially between countries with consistent weaknesses in financing, outcome measurement, and contextual analysis. A shared health systems framework and scalable AI appraisal tools, such as PRISM, could support more complete, comparable, and actionable plans to strengthen cancer control globally by providing the possibility for rapid analysis of comprehensiveness of NCCPs and generation of timely, achievable, and evidence-informed, context-specific plans that use the latest data.
Funding:
Harvard University.
