How are you taking part in this consultation?

You will not be able to change how you comment later.

You must be signed in to answer questions

  • Question on Consultation

    Are the summaries of clinical and cost effectiveness reasonable interpretations of the evidence?

  • Question on Consultation

    Are the recommendations sound and a suitable basis for guidance to the NHS?

  • Question on Consultation

    Are there any equality issues that need special consideration and are not covered in the medical technology consultation document?

  • Question on Consultation

    Has all of the relevant evidence been taken into account?

Guidance assessment consultation document for HTG10865 Software with artificial intelligence (AI)-derived algorithms to help interpret electroencephalograms (EEGs) for suspected epilepsy: early-use assessment

Open for comments Open until   Request commenting lead permission

We have updated this service so that members of the same organisation can now collaborate on a joint online response.

Read our blog to learn more.

1 Recommendation

1.1

More research is needed on the following software technologies with artificial intelligence (AI)-derived algorithms to help interpret electroencephalograms (EEGs) for suspected epilepsy before they can be funded by the NHS:

  • BioEP

  • encevis

  • NeuroCenter EEG

  • NeuroWorks

  • Persyst 15.

What this means in practice

There is not enough evidence to support funding for software with AI-derived algorithms to help interpret EEGs for suspected epilepsy in the NHS.

Access to the technologies should be through company, research or non-core NHS funding, and clinical or financial risks should be managed appropriately.

What research is needed

For software that helps review visual epileptiform activity on EEG

For encevis, NeuroCenter EEG, NeuroWorks and Persyst 15, more research, ideally from the UK, is needed on:

  • diagnostic accuracy of healthcare professional EEG review with and without the software

  • time to review and interpret EEGs with and without the software

  • time to diagnosis and treatment with and without the software

  • using the software for longer-term ambulatory EEGs

  • how the software helps in prioritising EEG review

  • the proportion of people who have an EEG that is either inconclusive or shows an epileptic abnormality and who are then diagnosed with epilepsy

  • healthcare resource use for people having an EEG (including the type of EEG for first and follow-up EEGs).

For software that estimates the likelihood of epilepsy on visually inconclusive EEGs

For BioEP, more research, ideally from the UK, is needed on:

  • diagnostic accuracy of epilepsy diagnosis made by healthcare professionals with and without the software

  • time to diagnosis and treatment with and without the software

  • how the software helps in prioritising EEG follow up

  • healthcare resource use for people having an EEG (including the type of EEG for first and follow-up EEGs).

Why the committee made this recommendation

These software technologies could support the diagnosis of epilepsy by:

  • enabling healthcare professionals to review epileptic abnormalities on EEGs more quickly (encevis, NeuroCenter EEG, NeuroWorks, Persyst 15)

  • providing healthcare professionals with supporting information about the likelihood of epilepsy on visually inconclusive EEGs (BioEP).

The results from 1 small study suggest that using the software to help detect epileptiform activity reduces healthcare professionals' review time of EEGs compared with no use of the software. But it is uncertain whether this time saving would be meaningful in NHS clinical practice. This is because the unmet need in the NHS is associated more with the varying availability of specialist healthcare professionals who diagnose and treat epilepsy than with the speed of interpreting EEGs or having additional information about the EEG.

There is no evidence on how having supporting information about the likelihood of epilepsy on visually inconclusive EEGs affects healthcare professionals' decision making on diagnosis, treatment or follow up. There is also no evidence on the effect of using the software on the time to diagnosis and treatment.

The results of the economic model suggest the software could be cost effective in some situations. But the plausibility of cost effectiveness is uncertain because there is not enough clinical evidence to support the model's assumptions.

So, more research is needed on all 5 software technologies. Ideally, research would be done in the UK. This is because current practice for investigating suspected epilepsy varies across countries. So, evidence from other countries may not be generalisable to NHS practice.