None listed
Conditions
Brief summary
To establish the (1) effectiveness, (2) cost-effectiveness and (3) acceptability of our machine learning (ML) model compared to usual care in selecting the first anti-seizure medication (ASM). Hypothesis: (1) seizure-free rate at 1 year of treatment with the first ASM will be higher in the machine learning group. (2) quality of life, depression and anxiety will improve more in the ML group, and ML model will be more cost-effective. (3) ML model will be acceptable to patients and clinicians. This is a multicentre randomised controlled trial (RCT) across all six states of Australia. Adults will be randomised 1:1 to ML Group (ASM recommended by the ML model) or the UC Group (ASM selected by the neurologist) and followed for 12 months. A sample size of 234 (including 10% dropout) participants will allow for measure of a minimum absolute difference of 20% in 1-year seizure-free rate on the first ASM between the study group (55% ML vs. 35% UC).
Interventions
A machine learning based clinical decision support software (CDSS) to assist the neurologist in recommending the initial anti-seizure medication prescription for newly diagnosed epilepsy.
The CDSS has been designed specifically for use in this study. The CDSS functions through input of 16 baseline clinical factors by the neurologist including patient demographics, medical history, family history, epilepsy risk factors, number of pre-treatment seizures, EEG results and MRI results. It also uses t
A machine learning based clinical decision support software (CDSS) to assist the neurologist in recommending the initial anti-seizure medication prescription for newly diagnosed epilepsy. The CDSS has been designed specifically for use in this study. The CDSS functions through input of 16 baseline clinical factors by the neurologist including patient demographics, medical history, family history, epilepsy risk factors, number of pre-treatment seizures, EEG results and MRI results. It also uses the input of deidentified report text from a routine EEG and brain MRI report. The model will provide a recommendation of a single anti-seizure medication with which the participant has the highest probability of attaining seizure freedom. Training required to utilise the model will be provided by the main site study coordinator in a single remote session taking approximately 30 minutes. This training will detail what is required to access the data entry point for the model, the format of the data required and how the recommendation will be displayed. It is anticipated that it will take approximately 15 minutes per participant to input the data and receive the recommendation from the program. Data entry for the model and the recommendation produced are intrinsically tied to the study REDCap database which is updated anytime the model is utilised. This database has auditing tools for tracking changes made to entries and as such can be used to monitor model utilisation.
Sponsors
Monash University
Study design
Allocation
Randomised controlled trial
Intervention model
Parallel
Primary purpose
Treatment
Masking
Blinded (masking used) (Subject, Outcomes Assessor)
Eligibility
Sex/Gender
All
Age
18 Years to 67 Years
Healthy volunteers
No
Inclusion criteria
- Diagnosis of epilepsy consistent with International League against Epilepsy (ILAE) criteria, defined as either: (1) at least two seizures within the past 12 months, or (2) one seizure within the past 6 months and epileptiform discharges on electroencephalography (EEG), or presence of epileptogenic lesion on computed tomography (CT) or magnetic resonance imaging (MRI) - EEG and MRI brain done within 2 years prior to randomisation (MRI can be substituted with epileptogenic lesion shown on CT) - Either naive to anti-seizure medication (ASM) treatment or has been treated with a single ASM for 14 days or less before randomisation
Exclusion criteria
- Previously treated epilepsy or current use of ASM for any reason for more than 14 days - Pregnant or breast-feeding - Current substance abuse disorder that may affect treatment adherence or response - Unable/failed to undergo EEG or brain CT or MRI due to contraindication - Patient with psychogenic non-epileptic seizures - Patients with progressive central nervous system disease, series hepatic or renal disease, or terminal cancer that would affect the assessment of response to ASMs
Outcome results
None listed