Hugely grateful to NHS Education for Scotland for funding the NHS Clinical Fellowship in AI which facilitated this work, and to the Digital Health Validation Lab in
@nhsggc.bsky.social and @uofglasgow.bsky.social
for hosting me! #radiology #AI #digitalhealth #sustainability [5/5]

🖥️ However, there was little consistency in how outcomes were reported, and no studies compared climate outcomes of clinical pathways with and without AI. We argue for better integration of environmental impact assessment into AI health research and evaluation [4/5]

🌳 All studies reporting environmental measures evaluated AI algorithms designed to be resource efficient, comparing them with state-of-the-art alternatives. They found large reductions in emissions (2-17x) and energy use (2-752x) could be achieved without reducing clinical performance [3/5]

🩻 Despite increasing concern about AI's climate impacts, environmental outcomes are rarely measured in evaluations of AI-enabled clinical pathways: only 18/4449 studies mentioned sustainability (typically in a single sentence) and only 6 quantitatively measured any environmental outcomes [2/5]

Thanks Claire, this is really helpful! Could you add me?