A Systematic Review of the Methods and Quality of Economic Evaluations for AI-Assisted Cancer Screening or Diagnosis

Author(s)

Yuyanzi Zhang, PhD1, Lei Wang2, Yifang Liang, BS3, Annushiah Vasan Thakumar, BSc, PhD4, Hong-fei Hu5, leah lee6, Hongchao Li, MSc, PhD3, Luying Wang, BS, MS, PhD3.
1student, China pharmaceutical university, Nanjing, China, 2China Pharmaceutical University, Jiangning District, Nanjing, China, 3China Pharmaceutical University, Nanjing, China, 4Taylor's University, Subang Jaya, Malaysia, 5Nanjing, China, 6China Pharmaceutical University, NanJing, China.
OBJECTIVES: This study aimed to systematically review existing health economic evaluations of Artificial Intelligence (AI) -assisted cancer screening or diagnosis, summarizing the modeling methods, outcomes, and quality of research.
METHODS: A comprehensive electronic literature search was conducted in Medline, Embase, Web of Science, and Cochrane Library databases from inception through December 2024. Two researchers independently screened and extracted data, and performed a descriptive analysis for the basic characteristics and modeling methods of the included studies. The reporting quality of the selected studies was assessed using the Consolidated Health Economic Evaluation Reporting Standards for Interventions That Use Artificial Intelligence (CHEERS-AI).
RESULTS: A total of 16 studies were included, evaluating AI-assisted screening or diagnosis across multiple cancer types, including prostate, lung, gastric, colorectal, breast, cervical, and melanoma skin cancer. AI was primarily applied to enhance the sensitivity or specificity of the cancer screening or diagnosis process. Cost-utility analysis was the predominant economic evaluation method (n = 13), followed by two studies using cost-effectiveness analysis and one study applying both approaches. All studies compared AI-assisted interventions with non-AI alternatives, with six also assessing cost-effectiveness across varying screening intervals. The Markov model (n=14, with 4 in combination with decision tree models) was the most frequently used, with cohort simulation remaining the most popular simulation method (n=10). All studies concluded that AI-assisted interventions were cost-effective. However, some CHEERS-AI items were not adequately addressed, particularly relating to the measurement and modeling AI learning over time, accounting for population differences, detailing analytics and assumptions, and considering implementation aspects.
CONCLUSIONS: Although modeling methods varied across the studies, all demonstrated that AI-assisted interventions were cost-effective. However, AI-specific elements in economic evaluation should be more comprehensively addressed in line with the CHEERS-AI checklist.

Conference/Value in Health Info

2025-11, ISPOR Europe 2025, Glasgow, Scotland

Value in Health, Volume 28, Issue S2

Code

EE35

Topic

Economic Evaluation, Methodological & Statistical Research

Disease

Oncology

Your browser is out-of-date

ISPOR recommends that you update your browser for more security, speed and the best experience on ispor.org. Update my browser now

×