EconPapers    
Economics at your fingertips  
 

AI in MRI: When Faster Scans Do Not Mean Lower Emissions

Vasiliki Kostami and Nazli Sonmez Camlibel
Additional contact information
Vasiliki Kostami: HEC Paris
Nazli Sonmez Camlibel: ESMT Berlin

No 1671, HEC Research Papers Series from HEC Paris

Abstract: Problem definition: MRI plays a critical role in healthcare but is highly energy-intensive, raising concerns about its environmental impact. Recent advances in artificial intelligence (AI) can substantially reduce scan duration while maintaining diagnostic quality, creating the potential for both operational and environmental benefits. However, improvements in scan efficiency do not necessarily translate into lower total emissions. Methodology/results: We develop an analytical model of AI-enabled MRI in which physicians choose among imaging protocols based on patients' diagnostic complexity, patients decide whether to undergo imaging, and hospitals choose AI capability under capacity constraints. AI plays two distinct roles: it can enhance the diagnostic value of an existing short scan or substitute for acquisition time by allowing patients otherwise requiring a full scan to receive a shorter AI-assisted protocol. Only the latter releases scanner capacity. When AI reduces average scan duration without expanding participation, emissions cannot increase: they fall under slack capacity but remain unchanged when capacity binds, as released scanner time serves additional patients. An environmental rebound requires sufficiently large participation expansion and initially slack capacity. We also characterize hospital investment incentives and their divergence from the social optimum, showing that carbon pricing may encourage or discourage AI investment depending on its effect on active scanning time. A calibration to knee MRI in Germany shows that AI does not reduce emissions at the hospital's privately optimal capability and that hospitals underinvest relative to the social optimum. Managerial implications: The operational and environmental value of AI depends on which protocols it replaces, how scanner capacity is utilized, and whether efficiency gains translate into financial returns. Hospitals may have insufficient incentives to adopt socially beneficial AI, suggesting that policies promoting sustainable medical imaging should account for both capacity conditions and the alignment of private and social incentives.

Keywords: artificial intelligence; diagnostic imaging; carbon emissions; capacity constraints; technology adoption (search for similar items in EconPapers)
JEL-codes: I11 M11 O33 Q53 Q58 (search for similar items in EconPapers)
Pages: 48 pages
Date: 2026-11-12
References: Add references at CitEc
Citations:

Downloads: (external link)
https://ssrn.com/abstract=7682620 Full text (text/html)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:ebg:heccah:1671

DOI: 10.2139/ssrn.7682620

Access Statistics for this paper

More papers in HEC Research Papers Series from HEC Paris HEC Paris, 1 Rue de la Libération, 78350 Jouy-en-Josas, France. Contact information at EDIRC.
Bibliographic data for series maintained by David Melon ().

 
Page updated 2026-11-19
Handle: RePEc:ebg:heccah:1671