Table 1

Preclinical evaluation of the MELD algorithm, classified by study type according to the IDEAL stage 0 framework

Study classificationEvidence
Device perspectiveSafety and efficacy have been formally assessed in the peer-reviewed literature.13–15 From the perspective of this difficult-to-localise cohort of patients undergoing SEEG, the sensitivity of the classifier was 74% with 100% specificity (no lesions detected in controls). In 3 of these 34 patients, a focal SOZ was not identified and the classifier identified additional lesions that were not implanted.13
Safety relates to identification of false positive clusters. The risk profile has been limited by sampling a maximum of three clusters per patient.
System perspectiveThere is clearly an established need for such algorithms as evidenced by the increasing evaluation of ‘MRI negative’ cases for epilepsy surgery, especially through SEEG. Not all patients undergoing epilepsy surgery have a SOZ identified and, even in those who do, a third do not achieve seizure freedom following resective surgery.20
Patient perspectiveAs part of the MAST Trial planning, a group of parents and children with epilepsy were surveyed at to assess whether the addition of up to three extra electrodes would be acceptable. Of 15 respondents (14 parents and one adolescent with epilepsy), all 15 (100%) agreed that the risk would be acceptable when balanced against the potential benefits and all 15 (100%) would enrol in the trial were they/their child to undergo SEEG implantation at our centre.
Clinician perspectiveClinicians from multiple specialties (neurosurgery, neurology, neurophysiology, neuroradiology) were involved in the initial development and retrospective evaluation studies, that showed promising performance. Specifically, these same clinicians will be involved in this prospective trial, attesting to the clinical acceptability and utility. The main users of the output will be neurosurgeons and the integration of the output into the planning software has been tested (figure 2B), although more formal learning curves will need to be assessed prior to multicentre studies.
  • IDEA, Idea, Development, Exploration, Assessment, Long-Term Follow-Up.; MELD, multicentre epilepsy lesion detection; SEEG, stereoelectroencephalography; SOZ, Seizure Onset Zone.