What this means in practice
For AI technologies that can be used during the evidence generation period
The 6 AI technologies listed in recommendation 1.1 can be used in the NHS as options to help detect colorectal polyps for people who do not have diagnosed IBD or Lynch syndrome during the evidence generation period (4 years) and paid for using core NHS funding. During this time, more evidence will be collected to address any uncertainties. Companies are responsible for organising funding for evidence generation activities.
After this, NICE will review this guidance and the recommendations may change. Take this into account when negotiating the length of contracts and licence costs.
Potential benefits of use in the NHS during the evidence generation period
-
Clinical benefit: Clinical evidence shows that AI technologies can help endoscopists detect adenomatous polyps (which can develop into colorectal cancer) during colonoscopy which might have otherwise been missed. Early detection means earlier access to appropriate treatment or surveillance pathways as needed. This could reduce the number of future colorectal cancer cases and improve clinical outcomes.
-
Resources: AI technologies could reduce the need for more intensive healthcare resources later, such as for emergency presentations or more advanced stages of cancer.
Managing the risk of use in the NHS during the evidence generation period
-
Costs: Results from economic analyses suggest that the AI technologies could be cost effective. But these results are uncertain. This uncertainty could be reduced by collecting more evidence on how effective the AI technologies are, and the numbers of colorectal cancers missed and prevented in the future. NHS trusts should consider the costs of the AI technologies, particularly upfront costs, when implementing them.
-
Clinical risk: The AI technologies should only be used as an aid to detect colorectal polyps. The technologies cannot replace endoscopist review. AI technologies for characterising colorectal polyps should only be used in research because evidence for characterisation is more limited.
-
Equality: Evidence generation could help address equality by collecting data on differences in how effective the AI technologies are for some groups of people compared with others.
For AI technologies that need more research
There is not enough evidence to support NHS funding of the:
-
4 AI technologies to help detect colorectal polyps for people who do not have diagnosed IBD or Lynch syndrome (see recommendation 1.4)
-
4 AI technologies to help characterise colorectal polyps for people who do not have diagnosed IBD or Lynch syndrome (see recommendation 1.5)
-
10 AI technologies to help detect or characterise colorectal polyps for people with diagnosed IBD or Lynch syndrome (see recommendation 1.6).
When more research is needed, access to the AI technologies should be through company, research or non-core NHS funding, and clinical or financial risks should be managed appropriately.
What evidence generation is needed
More evidence needs to be generated on the effect of the AI technologies on:
The evidence generation plan gives further information on the prioritised evidence gaps and outcomes, ongoing studies and potential real-world data sources. It includes how the evidence gaps could be resolved through real-world evidence studies.
Why the committee made these recommendations
Clinical evidence shows that 6 AI technologies (CADDIE, CAD EYE, ENDO-AID, EndoScreener, GI Genius and MAGENTIQ-COLO) increase the percentage of colonoscopies in which 1 or more of a type of potentially cancerous polyp, called an adenoma, is detected. But evidence suggests that the technologies mostly increase the number of smaller adenomas found, which are less likely than large polyps to develop into bowel cancer. It is uncertain whether the technologies increase the number of large adenomas found. There is also not enough evidence to show how effective the technologies are at helping to detect SSLs, which are another type of potentially cancerous polyp. So, it is uncertain in practice whether using AI technologies increases the number of cancer cases detected during colonoscopy. This means that their cost effectiveness is also uncertain. These 6 AI technologies can be used while evidence is generated.
There is limited evidence on the effectiveness of Argus, Discovery and EMIS in detecting colorectal polyps. So, it is uncertain whether they increase the percentage of colonoscopies in which 1 or more adenoma is found. Also, it was not possible to estimate the cost effectiveness for ENDOANGEL. So, more research is needed on these 4 AI technologies.
People with IBD or Lynch syndrome were excluded from many of the clinical trials. Their bowels may look different or they may have different types of polyps. It is uncertain whether the AI technologies work as well in people with these conditions. So, more research is needed for this population.
Some AI technologies include a function to help the endoscopist characterise the polyp and decide if it is potentially cancerous without removing it or examining it in a laboratory. There is less evidence for this function, and the research is of lower quality. So, more research is needed on AI technologies to help characterise colorectal polyps.