Clinical Prediction Models for Treatment Outcomes in Newly-diagnosed Epilepsy ============================================================================= * Corey Ratcliffe * Vishnav Pradeep * Anthony Marson * Simon S. Keller * Laura J. Bonnett ## Abstract **Background** Up to 35% of individuals diagnosed with epilepsy proceed to develop pharmacoresistant epilepsy, leading to persistent uncontrolled seizure activity that can directly, or indirectly, significantly degrade an individual’s quality of life. The factors underlying pharmacoresistance are unclear, but it has been hypothesised that repeated ictogenic activity is conducive to the development of a more robust epileptogenic network. To ensure that the most effective treatment choices are made and ictogenic activity is minimised, accurate outcome modelling at the point of diagnosis is key. **Objectives** This review therefore aims to identify demographic, clinical, physiological (e.g. EEG), and imaging (e.g. MRI) factors that may be predictive of treatment outcomes in patients with newly diagnosed epilepsy (NDE). **Data sources, study eligibility criteria, participants, and interventions** MEDLINE and EMBASE were searched for prediction models of treatment outcomes in patients with newly diagnosed epilepsy and any non-surgical treatment plan. **Study appraisal and synthesis methods** Study characteristics were extracted and subjected to assessment of risk of bias (and applicability concerns) using the PROBAST tool. Prognostic factors associated with treatment outcomes are reported. **Results** After screening, 48 models were identified in 32 studies, which generally scored low for concerns of applicability, but universally high for susceptibility to bias. Outcomes reported were heterogenous, but fit broadly into four categories: pharmacoresistance, short-term treatment response, seizure remission, and mortality. Prognostic factors were also heterogenous, but the predictors that were commonly significantly associated with outcomes were those related to seizure characteristics (semiology), epilepsy history, and age at onset. ASM response was often included as a prognostic factor, potentially obscuring factor relationships at baseline. **Conclusions** Currently, outcome prediction models for NDE demonstrate a high risk of bias. Model development could be improved with a stronger adherence to recommended TRIPOD practices, and by avoiding including response to treatment as a prognostic factor. **Implications of key findings** This review identified semiology, epilepsy history, and age at onset as factors associated with treatment outcome prognosis, suggesting that future prediction model studies should focus on these factors in their models. Furthermore, we outline actionable changes to common practices that are intended to improve the overall quality of prediction model development in NDE. **Key Points** * **This paper presents a systematic literature search for treatment outcome prediction models in newly diagnosed epilepsy.** * **The risk of bias in the included models were evaluated using the PROBAST framework, finding a universally high risk level.** * **The relationship between semiology, epilepsy history, and age at onset with seizure remission should be examined in future prediction model studies.** * **To improve the overall quality of prediction model development in NDE, prospective authors are advised to adhere to TRIPOD guidelines, and to avoid including response to treatment as a prognostic variable.** Keywords * newly diagnosed epilepsy * seizures * prognosis * intractability * outcomes * treatment ## Introduction ### Rationale #### Clinical overview of epilepsy As one of the most common neurological diseases, epilepsy is estimated to affect over 70 million people globally.1,2 Epilepsy incidence tends to be higher in the youngest and oldest age groups, in females, and in low-middle income countries.3 Epilepsy is characterised by a predisposition to spontaneous seizure activity, which is thought to arise due to abnormalities within cortical networks.4–7 Epilepsy may refer to any subtype of a multitude of disorders that can differ in seizure type, clinical course, or prognosis; aetiology and comorbidities will also inform the specific diagnoses people with epilepsy (PWE) will receive.8,9 An accurate diagnosis is crucial for determining the appropriate first-line treatment, which will most commonly be anti-seizure medication (ASM) monotherapy. Alongside seizure activity, PWE are vulnerable to cognitive, behavioural, and neurological comorbidities, the combination of which often result in PWE experiencing a lower quality of life. #### Newly-diagnosed epilepsy To better characterise the course of epilepsy and its underlying pathomechanisms, it has been suggested that (people with) newly diagnosed epilepsy (NDE) be studied as a distinct group.10,11 Studying epilepsy at its earliest time point avoids the confounds inherent in long- standing epilepsy, including the chronic effects of seizure activity and ASM use: seizure activity in chronic epilepsy can cause injuries that encourage the development of pharmacoresistance in PWE, and successive ASM regimens are associated with a reduction in the chance of attaining seizure freedom.12–15 Predictive models for chronic epilepsy consequently include the confounding variability of seizure and ASM profiles, whereas models developed specifically for use in NDE (which can reliably prognosticate disease trajectories) do not. Models based on NDE cohorts can therefore inform treatment approaches that ameliorate the consequences of epilepsy chronicity. #### Seizure freedom and pharmacoresistance in First-line Therapies In a recent large-cohort NDE study, the rate of 1-year remission (cessation of seizure activity) following ASM mono/polytherapy was 63.7%, and the rate of pharmacoresistance (failure of two or more appropriate ASM trials to control seizure activity) was 36.3%, in line with similar studies.16–18 ASM considered efficacious for focal seizure management include oxcarbazepine, carbamazepine, and lamotrigine; for generalised seizures, sodium valproate, levetiracetam, lamotrigine, and ethosuximide are common first-line ASM.19–22 First-line monotherapy represents the best likelihood of preventing seizure recurrence and pharmacoresistance in PWE who have had two seizures or been classified as ‘high-risk’, and an optimal treatment response is most likely when the appropriate ASM is chosen for monotherapy.23–26 If a change of ASM is indicated, it is recommended that alternative monotherapy be prescribed.27 Should this prove unsuccessful, polytherapy, adjunctive therapy, or surgical therapies (i.e. two or more ASM, dietary management, and resection respectively) may be considered.20 ### Second-line Therapies The success of polytherapy is largely dependent on the choice of ASM combination. Combination ASM are often determined based on complementary mechanisms of action, but no specific selection criteria exist, due in part to difficulty in collecting randomised control trial (RCT) data for combination approaches.18 Should ASM therapy fail, the ketogenic diet is recommended as an adjunctive therapy, and has been found to be effective in encouraging seizure control in people with pharmacoresistant epilepsy, as have invasive techniques of seizure control in focal epilepsies, such as resective surgery, vagus nerve stimulation, deep brain stimulation, and responsive neuromodulation.28,29 It has been suggested that 12 months after diagnosis is the ideal time at which to evaluate the long-term efficacy of treatment, especially considering the implications for driving restrictions in Europe.30,31 Furthermore, 12 months is the earliest possible timepoint at which a PWE can be said to have experienced 12 months of continuous seizure freedom, and a referral for surgical therapy is also unlikely.32,33 ### Treatment Outcomes Early seizure control has been indicated to be crucial for ensuring optimal treatment outcomes in NDE, putatively due to the prevention of further disruptions to seizure-related networks.5,34–37 Epilepsy treatment is individualised to ensure that: a) The risk-benefit ratio of a proposed therapy is suitable and b) The PWE is receiving the most efficacious treatment.27 The decision to begin a particular regimen is made after the consideration of several potential contraindications, such as pregnancy, medical interactions, and the risk of adverse effects.38 Importantly, the treatment choice will also be informed by the likelihood of achieving seizure freedom on a particular ASM (the efficacy) and the proportion of PWE who persist with the drug trial (the effectiveness). Predicting treatment outcomes—such as seizure remission, refractoriness, and pharmacoresistance—is non-trivial, also requiring the consideration of factors like age at onset (and the related epilepsy duration), the number of pretreatment seizures, EEG/imaging abnormalities, intellectual impairments, aetiology, and semiology to inform trajectories.39–42 ### Prediction models Prediction models are combinations of prognostic factors used to estimate the risk of a specific endpoint. Built with and validated on large cohorts, prediction models allow for individual patient outcomes to be estimated according to a formal statistical framework.43 Prognostic and diagnostic models are commonplace in epilepsy care, and the principal benefit of multivariable models (over the use of univariable factors for prediction) is accuracy, especially considering the complexity of epileptic processes.44,45 Single biomarkers (quantifiable properties indicative of normal biological processes) in epilepsy are thought to lack the granularity and robustness necessary to allow for clinical application.46 For example, several studies have investigated the relationship between EEG abnormalities and outcomes, often providing conflicting or incongruous evidence; whilst it is probable that some association exists, it is likely that EEG patterns and features influence/are mediated by external factors, and further multivariable research is required to determine how.47 To facilitate application and future evaluation (as in with systematic reviews) it is recommended that prediction models be designed and reported in a systematic manner, such as is outlined by Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) guidelines.48 Adherence to a predefined set of guidelines, such as TRIPOD, helps to ensure that the risk of bias (RoB; systematic error) and applicability concerns (AC) in the resultant study are kept to a minimum.49 Several models for the prediction of treatment outcomes in NDE have already been proposed, the latest systematic review of which was published in 2014.50,51 ### Objectives RCTs are the gold-standard source for evidence of efficacy and effectiveness for therapies, but it has previously been remarked that there is a lack of suitable high-quality evidence in the literature, with many of the existing studies suffering from methodological flaws.52 Alternatively, systematic reviews (of prediction models, for example) provide a reliable collation of evidence from which health-based interventions can be informed, without contributing to research waste.53,54 As new prediction models are developed and validated, it is crucial that they be presented in a format that allows for optimal dissemination of their actionable conclusions. Furthermore, information from previous reviews may be outdated and misleading in the context of more recent findings. The most recent comparable review, carried out by Abimbola et al. in 2014, presents several opportunities for improvement (besides being updated), namely that only studies with samples of over 100 were included and no evaluation of RoB was carried out.50 A systematic examination of multivariable prediction models for treatment outcomes in NDE was undertaken here to provide an updated and expanded review of the state of the literature, and to facilitate understanding of their conclusions. All included models were evaluated for RoB and AC using the PROBAST (Prediction model Risk Of Bias ASsessment Tool) framework.49 Between the models, common prognostic factors were identified and are presented herein, with the intention of informing future prediction model studies in NDE. ## Methods ### Protocol and registration This review is reported in adherence with the PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) guidelines, and a non-peer reviewed, publicly available, protocol was registered with PROSPERO (ID: CRD42022329936).55 ### Eligibility Criteria, Information Sources, Search, Study Selection, and Data Collection Process MEDLINE and Embase were searched for relevant publications, using PubMed/MeSH (medical subject headings) and Scopus/Boolean terms respectively. Full queries can be seen in Appendix 1. Data were screened by CR and LJB independently, with mediation of any conflicting exclusions following consensus meetings provided by SSK. Studies were included if they contained a multivariable model of treatment outcomes in a discrete sample of NDE, meeting the following criteria: * **Study design - Any primary design including (but not limited to): cohort studies, randomised control trials, quasi-randomised control trials, observational studies, and case-control studies.** * **Participants - Any person with NDE defined using the operational ILAE definition of two clinically unprovoked seizures, or one unprovoked seizure with a > 60% probability of recurrence (other definitions were evaluated for agreement with the ILAE definition *ad hoc*).56 Provoked seizures include those deemed situational or due to acute neurological insult/precipitant.57** * **Multivariable model - Prediction models developed with at least two demographic, and/or clinical, and/or neuroimaging, and/or electrophysiological factors collected and assessed as part of standard clinical practice at baseline upon a new diagnosis of epilepsy, that are associated with 12 months continuous seizure freedom (remission). Demographic factors are socioeconomic attributes that can be statistically expressed—for example age, sex, and education level. Clinical factors are signs and symptoms of disease classification or severity including aetiology, type and frequency of seizure, age of onset of epilepsy and duration of illness prior to diagnosis. The neuroimaging and neurophysiological factors include assessments of standard MRI and EEG examinations respectively, often taken upon a new diagnosis of epilepsy.** * – **Our search terms were not designed to capture machine learning/artificial intelligence-based studies, due to the complexities introduced by the structure of these models, which is very different to those of regression-based models.**58 * **Primary outcomes - 12 months (or longer) continuous seizure freedom (remission). The timeframe of 12 months was chosen in accordance with previous literature suggesting that as one seizure per year is sufficient to preclude PWE from driving, seizure freedom should be measured over the same timeframe.**17 **Furthermore, after 12 months of treatment, if seizures are not controlled it has been recommended that the PWE be referred to a specialist clinic.**59 * **Secondary outcomes - Reported seizure remission of any duration at any time point; treatment failure (adverse effects, intractability, etc.) reported in any form and at any time.** ### Risk of Bias in Individual Studies RoB was determined on a per-study basis, using 20 signalling questions over four domains (Participants, Predictors, Outcomes, Analysis); the answers to the questions indicate potential for bias, which then informs the (semi-subjective) potential for bias in that domain. If any domain is flagged as having a high potential for bias, the study is judged to have a high overall RoB.54 Similarly, three of the four domains contain an AC judgement, whereby the rater evaluates to what extent the study content matches the research question. High concern for applicability in any domain results in the study also receiving a high AC rating.49 Alongside outcome-related data extraction, data required for RoB and AC assessment were extracted by CR and VP, who independently evaluated all 32 studies in the sample. ### Summary Measures, Synthesis of Results, Risk of Bias Across Studies, and Additional Analyses Data pertinent to describing the setting, methodology, demographics, predictors, and outcomes for individual studies was synthesised into narrative form and evidence tables. Metadata for quality assessments purposes was also extracted. Sankey plots were constructed to visually present the distribution of outcomes across studies, and predictors across outcomes. Definitions for the categories proposed in this study can be found in Appendix 2. ## Results ### Study Selection After the removal of 285 duplicate entries, 878 records were excluded first based on their titles, then abstracts. The remaining 128 reports were sought for retrieval, of which 126 were obtained. The retrieved reports were then assessed for eligibility, during which 77 were excluded due to univariable modelling (44), unsuitable cohorts (20), unsuitable outcomes (12), or absence of primary analysis (1). The remaining reports underwent data extraction, during which 17 were deemed ineligible.60 Data extraction was carried out on the whole sample by CR and VP independently, using a predefined form to ensure that all relevant information was extracted systematically. ### Study Characteristics After screening, 32 studies were deemed suitable for inclusion (for a PRISMA diagram, see Figure 1), including 48 models. 12 studies used prospectively recruited PWE (37.5%), 17 used retrospective data (53.1%), and three used a combination (9.4%). Designs included one case-control study, two RCTs, and 29 cohort studies. Sample sizes ranged from 53 to 99990 PWE, with a median value of 261. Estimates of ‘events per variable’ ranged from 0.63 to 3927.38, with a median value of 9.86 (for study characteristics, see table 1). In our sample, 12 studies utilised Cox proportional hazards models (37.5%), whereas the remaining 20 employed logistic regressions (62.5%) to build their prediction models. Outcomes were evaluated at timepoints that ranged from 16 - 20 weeks, up to 32 - 36 years, with several studies assessing outcomes at the arbitrary date of the last follow-up. ![Figure 1.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/01/13/2024.01.12.24301215/F1.medium.gif) [Figure 1.](http://medrxiv.org/content/early/2024/01/13/2024.01.12.24301215/F1) Figure 1. PRISMA flow diagram of study selection. View this table: [Table 1.](http://medrxiv.org/content/early/2024/01/13/2024.01.12.24301215/T1) Table 1. Summary of included studies. Where available, ages are presented in years as mean (SD), *Median (range), or **Median (IQR). Epilepsy diagnoses are reported as in the original studies.16,39,42,61–89 Across all models, 41 unique outcomes were operationalised, which were subsequently stratified into four categories (see figure 2): Mortality; Pharmacoresistance; Seizure remission; and Short-term treatment response (a complete list of outcomes is provided in Appendix 3). In accordance with the review objectives, the seizure remission category was used for seizure outcomes of 12 months or longer, with all seizure outcomes of less than 12 months being categorised as short-term treatment response. 69 unique predictors were operationalised, which were subsequently stratified into 11 categories: Age; ASM; Comorbidity; Demographics; Diagnosis; EEG; History; Neuroimaging; Neuropsychology; Response; and Semiology (a complete list of predictors is also provided in Appendix 3). Although unavailable at baseline, response (to treatment) variables were recorded in a number of studies, and contributed significant predictors to several multivariable prediction models. ![Figure 2.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/01/13/2024.01.12.24301215/F2.medium.gif) [Figure 2.](http://medrxiv.org/content/early/2024/01/13/2024.01.12.24301215/F2) Figure 2. Sankey diagram showing model outcomes across all studies.16,39,42,61–89 ### Risk of Bias Within Studies After PROBAST assessment had reached consensus, nine (28.1%) studies presented with high AC, whilst all 32 studies demonstrated a high RoB (Table 2). AC for the participant domain were all low, but high for eight studies in the predictor domain, which was related to the inclusion of response to treatment as a prognostic factor. AC was also high for two studies in the outcome domain.49 RoB was generally low in the participant domain, but universally high in the predictor, outcome, and analysis domains. View this table: [Table 2.](http://medrxiv.org/content/early/2024/01/13/2024.01.12.24301215/T2) Table 2. Summary of the PROBAST risk of bias and applicability concern assessment of the included studies. Responses, in order of low to high risk of bias: Y = yes; PY = probably yes; NI = no information; PN = probably no; N = no. x.a. = Domain applicability; x.b. = Domain risk of bias.16,39,42,61–89 ### Results of Individual Studies In the included studies outcomes were most commonly categorised as short-term treatment response, followed by seizure remission, pharmacoresistance, and then mortality. Across the multivariable models, 112 relationships between predictors and outcomes were found to be statistically significant, with variables from the semiology category being reported as significant most frequently. Response variables were the next most frequent, followed by history and comorbidity. As shown in table 3, there were 40 cases of variables being statistically significant as predictors of seizure remission, of which 13 were categorised as response variables. View this table: [Table 3.](http://medrxiv.org/content/early/2024/01/13/2024.01.12.24301215/T3) Table 3. Summary count of predictor instances, sorted by outcome. ![Figure 3.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/01/13/2024.01.12.24301215/F3.medium.gif) [Figure 3.](http://medrxiv.org/content/early/2024/01/13/2024.01.12.24301215/F3) Figure 3. Sankey diagram showing studies and predictors, grouped by outcomes.16,39,42,61–89 ## Discussion ### Summary of Evidence Review summary The authors systematically identified 32 studies that used multivariable prediction models to assess the multifactorial prognosis of treatment outcomes in NDE. NDE is an area of research importance in the exploration of the pathomechanisms underlying the development of an epileptogenic environment, and aside from integrating recent research, this review expands on previous NDE prediction model reviews in two ways. First, the most recent review by Abimbola et al. in 2014 did not include studies with sample sizes below 100, which this review does.50 Broadening the inclusion criteria for studies facilitates iteration of the review question over time, and encourages the exploration of specific research questions within the same area. Second, quality (in the form of RoB and AC) assessment of the included studies was carried out by two independent reviewers (CR and VP) in accordance with PROBAST, the first review of epilepsy prediction models to do so.49 To best meet our primary objective—to inform the prognostic factor choices of future prediction model studies—this report has been prepared in accordance with PRISMA (where appropriate), ensuring maximum transparency, reproducibility, and clarity.60 ### Model descriptives The predictors and the outcomes of the included studies were heterogenous, so were stratified into categories to aid interpretation. The most common outcome was short-term treatment response followed by seizure remission, which aligned with our secondary and primary outcomes of interest, respectively. Models of pharmacoresistance and mortality were also reviewed, which address two of the potential treatment failure outcomes. One fifth (20%) of the studies included in this review included treatment response variables as predictors, which limits the applicability of the resultant models. Although statistical significance does not always confer clinical importance, prediction models are at their most informative when being used to inform treatment initiation, i.e. at baseline/diagnosis. Unsurprisingly, treatment response was often a statistically significant prognostic factor of treatment outcome, and potentially obscured predictive relationships that are interrogatable at baseline - such as treatment decision. In the included studies, seizure characteristics and epilepsy history were frequent statistically significant prognostic factors for seizure remission. For research and prediction model purposes, detailed seizure descriptions and family/personal history factors should be collected at baseline. Furthermore, the consistency of prediction models in epilepsy research would be improved by avoiding convenience-based decisions when designing studies, i.e. using a homogenous, predetermined timepoint, instead of the last follow-up. This could be achieved with the development of a Core Outcome Set for NDE.90,91 ### Bias The models were found to contain a moderate amount of AC, but universally high RoB. Whilst it seems unlikely that models created in a clinical context can be free of RoB, there are steps that can be taken to minimise it - thus increasing the applicability of any resultant models. For example, adherence to the modelling guidelines of TRIPOD by journal editors (similar to CONSORT) and researchers, and nominal acknowledgement of best practices, such as clearly reporting study and model characteristics, would facilitate research communication and uptake.92 Whilst AC were much less common in the included studies, simple steps such as ensuring consistency between study objectives and methods can be taken to help alleviate them in future studies. ### Limitations Whilst all individuals with new-onset epilepsy have (by definition) NDE, the inverse is not true—some PWE may have an undisclosed or unreported history of seizures, extending beyond the recommended 12 month cut-off.10 In consideration of its distinction from NDE, this review has purposefully avoided misattributing any samples as ‘new-onset epilepsy’, instead opting for the more verifiable NDE label. With this omission comes a potential loss of specificity that may hamper the accuracy of the presented model to certain PWE; guidelines for reporting seizure histories have been suggested, which should help to prevent this necessity in future reports.93 By including only studies involving a discrete sample of ILAE compliant NDE, this review addresses a sample who are not vulnerable to the common confounds of epilepsy research (such as ASM use and chronicity), or the heterogeneity of broader seizure research.10 However, this specificity comes at the cost of generalisability to provoked seizure research. The exploration of febrile, traumatic, and other acute seizure activity also has the potential to elucidate the pathomechanisms of ictogenesis, with ostensible benefit to unprovoked seizure research. Indeed, the two categories of predictors that were the strongest prognostic factors of epileptic seizure remission in this review, history and semiology, allude to pathomechanistic vulnerabilities (respectively, a predisposition to ictogenesis and the seizure insult) that could potentially describe provocation once fully understood. The conclusions of this review should be weighed against those of reviews into early and first seizures of mixed aetiology, to fully understand the influence of precipitation on ictogenesis.57,94 Machine learning is a rapidly expanding field in the health data sciences, which has demonstrated widespread potential utility.58 Machine learning prediction models were omitted from this review to ensure the comparability of the included regression model studies, and should not be taken as a dismissal of their increasing value to prognosis and diagnosis. Traditional multivariable models have the potential to inform the design of machine learning models, however, further exploration is necessary to evaluate the current state of machine learning prediction modelling in epilepsy. To facilitate the systematic comparison of artificial intelligence prediction model studies, reporting guidelines (TRIPOD- AI and PROBAST-AI) are in development.95 ### Conclusions The studies included in this review are heterogenous in both predictor and outcome selection, which is a hindrance to systematic comparison. To evaluate their effectiveness, prediction modelling of treatment decisions should be encouraged, whilst the inclusion of response to treatment as a prognostic factor should be avoided. Authors should also attempt to ensure that their studies adhere to reporting guidelines, to reduce RoB and AC. ## Supporting information Appendix 1; Appendix 2; Appendix 3 [[supplements/301215_file09.docx]](pending:yes) ## Data Availability All data produced in the present study are available upon reasonable request to the authors ## Postscript ### Acknowledgements Nothing to acknowledge. ### Contributions CR: Conceptualisation, investigation, visualisation, writing - original draft preparation, writing - review and editing. ORCID ID: https://orcid.org/0000-0002-3824-1681 VP: Investigation. AM: Writing - review and editing. ORCID ID: https://orcid.org/0000-0002-6861-8806 SSK: Writing - review and editing, supervision. ORCID ID: https://orcid.org/0000-0001-5247-9795 LJB: Conceptualisation, investigation, writing - review and editing. ORCID ID: https://orcid.org/0000-0002-6981-9212 ### Conflict of interest disclosure None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. ### Ethics and Integrity Nothing to declare. ## Footnotes * Vishnav Pradeep * Anthony Marson * Simon S. Keller * Laura J. Bonnett ## Abbreviations NDE : Newly Diagnosed Epilepsy PWE : People With Epilepsy ASM : Anti-seizure Medication(s) RCT : Randomised Control Trial RoB : Risk of Bias AC : Applicability Concerns TRIPOD(-AI) : Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (Artificial Intelligence) PROBAST(-AI) : Prediction model Risk Of Bias ASsessment Tool (Artificial Intelligence) PRISMA : Preferred Reporting Items for Systematic reviews and Meta-Analyses * Received January 12, 2024. * Revision received January 12, 2024. * Accepted January 13, 2024. * © 2024, Posted by Cold Spring Harbor Laboratory This pre-print is available under a Creative Commons License (Attribution-NonCommercial-NoDerivs 4.0 International), CC BY-NC-ND 4.0, as described at [http://creativecommons.org/licenses/by-nc-nd/4.0/](http://creativecommons.org/licenses/by-nc-nd/4.0/) ## References 1. 1.Beghi E. The Epidemiology of Epilepsy. Neuroepidemiology. 2020;54(2):185–191. doi:10.1159/000503831 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1159/000503831&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 2. 2.Ngugi AK, Bottomley C, Kleinschmidt I, Sander JW, Newton CR. Estimation of the burden of active and life-time epilepsy: A meta-analytic approach. Epilepsia. 2010;51(5):883–890. doi:10.1111/j.1528-1167.2009.02481.x [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/j.1528-1167.2009.02481.x&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=20067507&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000277000900022&link_type=ISI) 3. 3.Fiest KM, Sauro KM, Wiebe S, et al. Prevalence and incidence of epilepsy: A systematic review and meta-analysis of international studies. Neurology. 2017;88(3):296–303. doi:10.1212/WNL.0000000000003509 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1212/WNL.0000000000003509&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=27986877&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 4. 4.Fisher RS, Boas W van E, Blume W, et al. Epileptic Seizures and Epilepsy: Definitions Proposed by the International League Against Epilepsy (ILAE) and the International Bureau for Epilepsy (IBE). Epilepsia. 2005;46(4):470–472. doi:10.1111/j.0013-9580.2005.66104.x [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/j.0013-9580.2005.66104.x&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=15816939&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000227837200001&link_type=ISI) 5. 5.Terry JR, Benjamin O, Richardson MP. Seizure generation: The role of nodes and networks. Epilepsia. 2012;53(9):e166–e169. doi:10.1111/j.1528-1167.2012.03560.x [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/j.1528-1167.2012.03560.x&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=22709380&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 6. 6.Srinivas HV, Shah U. Comorbidities of epilepsy. Neurology India. 2017;65(Supplement):S18-S24. doi:10.4103/neuroindia.NI\_922\_16 [CrossRef](http://medrxiv.org/lookup/external-ref?access\_num=10.4103/neuroindia.NI_922_16&link_type=DOI) 7. 7.Ridsdale L, Wojewodka G, Robinson E, et al. Characteristics associated with quality of life among people with drug-resistant epilepsy. Journal of Neurology. 2017;264(6):1174–1184. doi:10.1007/s00415-017-8512-1 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1007/s00415-017-8512-1&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=28550480&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 8. 8.Scheffer IE, Berkovic S, Capovilla G, et al. ILAE classification of the epilepsies: Position paper of the ILAE Commission for Classification and Terminology. Epilepsia. 2017;58(4):512–521. doi:10.1111/epi.13709 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/epi.13709&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 9. 9.Fisher RS, Cross JH, French JA, et al. Operational classification of seizure types by the International League Against Epilepsy: Position Paper of the ILAE Commission for Classification and Terminology. Epilepsia. 2017;58(4):522–530. doi:10.1111/epi.13670 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/epi.13670&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=28276060&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 10. 10.Pohlmann-Eden B. Conceptual relevance of new-onset epilepsy. Epilepsia. 2011;52 Suppl 4:1–6. doi:10.1111/j.1528-1167.2011.03142.x [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/j.1528-1167.2011.03142.x&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=21219301&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 11. 11.Leek NJ, Neason M, Kreilkamp BaK, et al. Thalamohippocampal atrophy in focal epilepsy of unknown cause at the time of diagnosis. European Journal of Neurology. 2021;28(2):367–376. doi:10.1111/ene.14565 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/ene.14565&link_type=DOI) 12. 12.Kreilkamp BAK, McKavanagh A, Alonazi B, et al. Altered structural connectome in non-lesional newly diagnosed focal epilepsy: Relation to pharmacoresistance. NeuroImage : Clinical. 2021;29. doi:10.1016/j.nicl.2021.102564 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.nicl.2021.102564&link_type=DOI) 13. 13.Alonazi BK, Keller SS, Fallon N, et al. Resting-state functional brain networks in adults with a new diagnosis of focal epilepsy. Brain and Behavior. 2019;9(1):e01168. doi:10.1002/brb3.1168 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1002/brb3.1168&link_type=DOI) 14. 14.Pohlmann-Eden B, Crocker CE, Schmidt MH. A conceptual framework for the use of neuroimaging to study and predict pharmacoresistance in epilepsy. Epilepsia. 2013;54 Suppl 2:75–79. doi:10.1111/epi.12190 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/epi.12190&link_type=DOI) 15. 15.Brodie MJ, Barry SJE, Bamagous GA, Norrie JD, Kwan P. Patterns of treatment response in newly diagnosed epilepsy. Neurology. 2012;78(20):1548–1554. doi:10.1212/WNL.0b013e3182563b19 [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6OToibmV1cm9sb2d5IjtzOjU6InJlc2lkIjtzOjEwOiI3OC8yMC8xNTQ4IjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjQvMDEvMTMvMjAyNC4wMS4xMi4yNDMwMTIxNS5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=) 16. 16.Chen Z, Brodie MJ, Liew D, Kwan P. Treatment Outcomes in Patients With Newly Diagnosed Epilepsy Treated With Established and New Antiepileptic Drugs: A 30-Year Longitudinal Cohort Study. JAMA neurology. 2017;75(3):279–286. doi:10.1001/jamaneurol.2017.3949 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1001/jamaneurol.2017.3949&link_type=DOI) 17. 17.Kwan P, Arzimanoglou A, Berg AT, et al. Definition of drug resistant epilepsy: Consensus proposal by the ad hoc Task Force of the ILAE Commission on Therapeutic Strategies. Epilepsia. 2010;51(6):1069–1077. doi:10.1111/j.1528-1167.2009.02397.x [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/j.1528-1167.2009.02397.x&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=19889013&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000278307900017&link_type=ISI) 18. 18.Verrotti A, Tambucci R, Di Francesco L, et al. The role of polytherapy in the management of epilepsy: Suggestions for rational antiepileptic drug selection. Expert Review of Neurotherapeutics. 2020;20(2):167–173. doi:10.1080/14737175.2020.1707668 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1080/14737175.2020.1707668&link_type=DOI) 19. 19.Marson AG, Williamson PR, Clough H, Hutton JL, Chadwick DW, Group OBOTEMT. Carbamazepine versus Valproate Monotherapy for Epilepsy: A Meta-analysis. Epilepsia. 2002;43(5):505–513. doi:10.1046/j.1528-1157.2002.20801.x [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1046/j.1528-1157.2002.20801.x&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=12027911&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 20. 20.Goldenberg MM. Overview of drugs used for epilepsy and seizures: Etiology, diagnosis, and treatment. P & T: A Peer-Reviewed Journal for Formulary Management. 2010;35(7):392–415. 21. 21.Nevitt SJ, Sudell M, Weston J, Tudur Smith C, Marson AG. Antiepileptic drug monotherapy for epilepsy: A network meta-analysis of individual participant data. The Cochrane Database of Systematic Reviews. 2017;12:CD011412. doi:10.1002/14651858.CD011412.pub3 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1002/14651858.CD011412.pub3&link_type=DOI) 22. 22.Kanner AM, Bicchi MM. Antiseizure Medications for Adults With Epilepsy: A Review. JAMA. 2022;327(13):1269–1281. doi:10.1001/jama.2022.3880 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1001/jama.2022.3880&link_type=DOI) 23. 23.FIRST STG(FIRST). Randomized clinical trial on the efficacy of antiepileptic drugs in reducing the risk of relapse after a first unprovoked tonic-clonic seizure. First Seizure Trial Group (FIR.S.T. Group). Neurology. 1993;43(3 Pt 1):478–483. doi:10.1212/wnl.43.3\_part\_1.478 [CrossRef](http://medrxiv.org/lookup/external-ref?access\_num=10.1212/wnl.43.3_part_1.478&link_type=DOI) 24. 24.Marson A, Jacoby A, Johnson A, et al. Immediate versus deferred antiepileptic drug treatment for early epilepsy and single seizures: A randomised controlled trial. *Lancet (London*, England*)*. 2005;365(9476):2007–2013. doi:10.1016/S0140-6736(05)66694-9 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/S0140-6736(05)66694-9&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=15950714&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000229701400022&link_type=ISI) 25. 25.Kim LG, Johnson TL, Marson AG, Chadwick DW. Prediction of risk of seizure recurrence after a single seizure and early epilepsy: Further results from the MESS trial. The Lancet Neurology. 2006;5(4):317–322. doi:10.1016/S1474-4422(06)70383-0 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/S1474-4422(06)70383-0&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=16545748&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000236535200021&link_type=ISI) 26. 26.Bonnett LJ, Kim L, Johnson A, et al. Risk of seizure recurrence in people with single seizures and early epilepsy Model development and external validation. Seizure. 2022;94:26–32. doi:10.1016/j.seizure.2021.11.007 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.seizure.2021.11.007&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=34852983&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 27. 27.Perucca E, Tomson T. The pharmacological treatment of epilepsy in adults. The Lancet Neurology. 2011;10(5):446–456. doi:10.1016/S1474-4422(11)70047-3 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/S1474-4422(11)70047-3&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=21511198&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000290424800014&link_type=ISI) 28. 28.Imdad K, Abualait T, Kanwal A, et al. The Metabolic Role of Ketogenic Diets in Treating Epilepsy. Nutrients. 2022;14(23):5074. doi:10.3390/nu14235074 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.3390/nu14235074&link_type=DOI) 29. 29.Janmohamed M, Brodie MJ, Kwan P. Pharmacoresistance - Epidemiology, mechanisms, and impact on epilepsy treatment. Neuropharmacology. 2020;168:107790. doi:10.1016/j.neuropharm.2019.107790 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.neuropharm.2019.107790&link_type=DOI) 30. 30.Bonnett LJ, Marson AG, Johnson A, et al. External validation of a prognostic model for seizure recurrence following a first unprovoked seizure and implications for driving. PloS One. 2014;9(6):e99063. doi:10.1371/journal.pone.0099063 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1371/journal.pone.0099063&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 31. 31.Epilepsy Action. Driving rules for epilepsy. Epilepsy Action. Published online 2023. 32. 32.Mohanraj R, Brodie MJ. Measuring the efficacy of antiepileptic drugs. Seizure. 2003;12(7):413–443. doi:10.1016/S1059-1311(03)00047-5 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/S1059-1311(03)00047-5&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=12967570&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 33. 33.Berg AT. Understanding the delay before epilepsy surgery: Who develops intractable focal epilepsy and when? CNS spectrums. 2004;9(2):136–144. doi:10.1017/s109285290000849x [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1017/s109285290000849x&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=14999169&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000226890700016&link_type=ISI) 34. 34.Shorvon S, Luciano AL. Prognosis of chronic and newly diagnosed epilepsy: Revisiting temporal aspects. Current Opinion in Neurology. 2007;20(2):208–212. doi:10.1097/WCO.0b013e3280555175 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1097/WCO.0b013e3280555175&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=17351493&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 35. 35.Berg AT. Risk of recurrence after a first unprovoked seizure. Epilepsia. 2008;49(s1):13–18. doi:10.1111/j.1528-1167.2008.01444.x [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/j.1528-1167.2008.01444.x&link_type=DOI) 36. 36.Caciagli L, Bernhardt BC, Hong SJ, Bernasconi A, Bernasconi N. Functional network alterations and their structural substrate in drug-resistant epilepsy. Frontiers in Neuroscience. 2014;8. doi:10.3389/fnins.2014.00411 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.3389/fnins.2014.00411&link_type=DOI) 37. 37.Engel J, Thompson PM, Stern JM, Staba RJ, Bragin A, Mody I. Connectomics and epilepsy. Current opinion in neurology. 2013;26(2):186–194. doi:10.1097/WCO.0b013e32835ee5b8 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1097/WCO.0b013e32835ee5b8&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=23406911&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 38. 38.Sharma S, Chen Z, Rychkova M, et al. Treatment initiation decisions in newly diagnosed epilepsyA longitudinal cohort study. Epilepsia. 2020;61(3):445–454. doi:10.1111/epi.16439 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/epi.16439&link_type=DOI) 39. 39.Hitiris N, Mohanraj R, Norrie J, Sills GJ, Brodie MJ. Predictors of pharmacoresistant epilepsy. Epilepsy Research. 2007;75(2-3):192–196. doi:10.1016/j.eplepsyres.2007.06.003 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.eplepsyres.2007.06.003&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=17628429&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000248904200015&link_type=ISI) 40. 40.Li Z, Cao W, Sun H, et al. Potential clinical and biochemical markers for the prediction of drug-resistant epilepsy: A literature review. Neurobiology of Disease. Published online September 2022:105872. doi:10.1016/j.nbd.2022.105872 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.nbd.2022.105872&link_type=DOI) 41. 41.Niriayo YL, Mamo A, Kassa TD, et al. Treatment outcome and associated factors among patients with epilepsy. Scientific Reports. 2018;8(1):17354. doi:10.1038/s41598-018-35906-2 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/s41598-018-35906-2&link_type=DOI) 42. 42.Zhang Y, Yu N, Su L, Di Q. A prospective cohort study of prognosis for newly diagnosed epilepsy in east China. BMC Neurology. 2013;13(1):116. doi:10.1186/1471-2377-13-116 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1186/1471-2377-13-116&link_type=DOI) 43. 43.Steyerberg EW, Moons KGM, van der Windt DA, et al. Prognosis Research Strategy (PROGRESS) 3: Prognostic Model Research. PLoS Medicine. 2013;10(2):e1001381. doi:10.1371/journal.pmed.1001381 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1371/journal.pmed.1001381&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=23393430&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 44. 44.Sansevere AJ, Kapur K, Peters JM, Fernández IS, Loddenkemper T, Soul JS. Seizure Prediction Models in the Neonatal Intensive Care Unit. Journal of Clinical Neurophysiology: Official Publication of the American Electroencephalographic Society. 2019;36(3):186–194. doi:10.1097/WNP.0000000000000574 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1097/WNP.0000000000000574&link_type=DOI) 45. 45.Engel J, Pitkänen A. Biomarkers for epileptogenesis and its treatment. Neuropharmacology. 2020;167:107735. doi:10.1016/j.neuropharm.2019.107735 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.neuropharm.2019.107735&link_type=DOI) 46. 46.FDA-NIH Biomarker Working Group. BEST (Biomarkers, EndpointS, and Other Tools) Resource. Food and Drug Administration (US); 2016. 47. 47. Cerulli Irelli E, Leodori G, Morano A, Di Bonaventura C. EEG Markers of Treatment Resistance in Idiopathic Generalized Epilepsy: From Standard EEG Findings to Advanced Signal Analysis. Biomedicines. 2022;10(10):2428. doi:10.3390/biomedicines10102428 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.3390/biomedicines10102428&link_type=DOI) 48. 48.Collins GS, Reitsma JB, Altman DG, Moons KGM. Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD): The TRIPOD Statement. Annals of Internal Medicine. 2015;162(1):55–63. doi:10.7326/M14-0697 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.7326/M14-0697&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=25560714&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 49. 49.Moons KGM, Wolff RF, Riley RD, et al. PROBAST: A Tool to Assess Risk of Bias and Applicability of Prediction Model Studies: Explanation and Elaboration. Annals of Internal Medicine. 2019;170(1):W1–W33. doi:10.7326/M18-1377 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.7326/M18-1377&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=30596876&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 50. 50.Abimbola S, Martiniuk ALC, Hackett ML, Glozier N, Mohamed A, Anderson CS. Early predictors of remission in newly diagnosed epilepsy: A systematic approach to reviewing prognostic factor studies. Neurological Research. 2014;36(1):1–12. doi:10.1179/1743132813Y.0000000257 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1179/1743132813Y.0000000257&link_type=DOI) 51. 51.Mohanraj R, Brodie MJ. Early predictors of outcome in newly diagnosed epilepsy. Seizure. 2013;22(5):333–344. doi:10.1016/j.seizure.2013.02.002 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.seizure.2013.02.002&link_type=DOI) 52. 52.Glauser T, Ben-Menachem E, Bourgeois B, et al. ILAE Treatment Guidelines: Evidence-based Analysis of Antiepileptic Drug Efficacy and Effectiveness as Initial Monotherapy for Epileptic Seizures and Syndromes. Epilepsia. 2006;47(7):1094–1120. doi:10.1111/j.1528-1167.2006.00585.x [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/j.1528-1167.2006.00585.x&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=16886973&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000239119700002&link_type=ISI) 53. 53.Higgins JPT, Thomas J, Chandler J, et al., eds. Cochrane Handbook for Systematic Reviews of Interventions. 1st ed. Wiley; 2019. doi:10.1002/9781119536604 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1002/9781119536604&link_type=DOI) 54. 54.Wolff RF, Moons KGM, Riley RD, et al. PROBAST: A Tool to Assess the Risk of Bias and Applicability of Prediction Model Studies. Annals of Internal Medicine. 2019;170(1):51–58. doi:10.7326/M18-1376 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.7326/M18-1376&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=30596875&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 55. 55.Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred Reporting Items for Systematic Reviews and Meta-Analyses: The PRISMA Statement. PLoS Medicine. 2009;6(7):e1000097. doi:10.1371/journal.pmed.1000097 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1371/journal.pmed.1000097&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=19621072&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 56. 56.Fisher RS, Acevedo C, Arzimanoglou A, et al. ILAE Official Report: A practical clinical definition of epilepsy. Epilepsia. 2014;55(4):475–482. doi:10.1111/epi.12550 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/epi.12550&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=24730690&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000334657000005&link_type=ISI) 57. 57.Neligan A, Adan G, Nevitt SJ, et al. Prognosis of adults and children following a first unprovoked seizure. Cochrane Database of Systematic Reviews. 2023;(1). doi:10.1002/14651858.CD013847.pub2 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1002/14651858.CD013847.pub2&link_type=DOI) 58. 58.Dhiman P, Ma J, Andaur Navarro CL, et al. Methodological conduct of prognostic prediction models developed using machine learning in oncology: A systematic review. BMC Medical Research Methodology. 2022;22(1):101. doi:10.1186/s12874-022-01577-x [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1186/s12874-022-01577-x&link_type=DOI) 59. 59.Labiner DM, Bagic AI, Herman ST, et al. Essential services, personnel, and facilities in specialized epilepsy centers–revised 2010 guidelines. Epilepsia. 2010;51(11):2322–2333. doi:10.1111/j.1528-1167.2010.02648.x [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/j.1528-1167.2010.02648.x&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=20561026&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 60. 60.Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA Statement for Reporting Systematic Reviews and Meta-Analyses of Studies That Evaluate Health Care Interventions: Explanation and Elaboration. PLoS Medicine. 2009;6(7):e1000100. doi:10.1371/journal.pmed.1000100 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1371/journal.pmed.1000100&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=19621070&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 61. 61.Aikiä M, Kälviäinen R, Mervaala E, Riekkinen PJ. Predictors of seizure outcome in newly diagnosed partial epilepsy: Memory performance as a prognostic factor. Epilepsy Research. 1999;37(2):159–167. doi:10.1016/s0920-1211(99)00059-5 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/S0920-1211(99)00059-5&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=10510982&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000082792200007&link_type=ISI) 62. 62.Arya R, Gillespie CW, Cnaan A, et al. Obesity and overweight as CAE comorbidities and differential drug response modifiers. Neurology. 2016;86(17):1613–1621. doi:10.1212/WNL.0000000000002611 [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6OToibmV1cm9sb2d5IjtzOjU6InJlc2lkIjtzOjEwOiI4Ni8xNy8xNjEzIjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjQvMDEvMTMvMjAyNC4wMS4xMi4yNDMwMTIxNS5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=) 63. 63.Ashmawi A, Hosny H, Abdelalim A, Bianchi E, Beghi E. The long-term prognosis of newly diagnosed epilepsy in Egypt: A retrospective cohort study from an epilepsy center in Greater Cairo. Seizure - European Journal of Epilepsy. 2016;41:86–95. doi:10.1016/j.seizure.2016.07.016 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.seizure.2016.07.016&link_type=DOI) 64. 64.Beydoun A, Sawaya MT, Alam E, Hmaimess G, Ezzeddine K, Younes F. Treatment registry in focal epilepsy (TRIP): Multicenter observational study in Lebanon. Seizure - European Journal of Epilepsy. 2015;27:54–59. doi:10.1016/j.seizure.2015.03.001 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.seizure.2015.03.001&link_type=DOI) 65. 65.Blank LJ, Acton EK, Willis AW. Predictors of Mortality in Older Adults With Epilepsy: Implications for Learning Health Systems. Neurology. 2021;96(1):e93–e101. doi:10.1212/WNL.0000000000011079 [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6OToibmV1cm9sb2d5IjtzOjU6InJlc2lkIjtzOjg6Ijk2LzEvZTkzIjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjQvMDEvMTMvMjAyNC4wMS4xMi4yNDMwMTIxNS5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=) 66. 66.Bruun E, Kälviäinen R, Keränen T. Outcome of initial antiepileptic drug treatment in elderly patients with newly diagnosed epilepsy. Epilepsy Research. 2016;127:60–65. doi:10.1016/j.eplepsyres.2016.08.023 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.eplepsyres.2016.08.023&link_type=DOI) 67. 67. Cerulli Irelli E, Morano A, Orlando B, et al. Seizure outcome trajectories in a well- defined cohort of newly diagnosed juvenile myoclonic epilepsy patients. Acta Neurologica Scandinavica. 2022;145(3):314–321. doi:10.1111/ane.13556 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/ane.13556&link_type=DOI) 68. 68.Chen RH, Li BF, Wen JH, Zhong CL, Ji MM. Clinical and electroencephalogram characteristics and treatment outcomes in children with benign epilepsy and centrotemporal spikes. World Journal of Clinical Cases. 2021;9(33):10116–10125. doi:10.12998/wjcc.v9.i33.10116 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.12998/wjcc.v9.i33.10116&link_type=DOI) 69. 69.Dlugos DJ, Buono RJ. Predicting outcome of initial treatment with carbamazepine in childhood focal epilepsy. Pediatric Neurology. 2004;30(5):311–315. doi:10.1016/j.pediatrneurol.2003.10.009 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.pediatrneurol.2003.10.009&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=15165631&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 70. 70.Dlugos D, Shinnar S, Cnaan A, et al. Pretreatment EEG in childhood absence epilepsy: Associations with attention and treatment outcome. Neurology. 2013;81(2):150–156. doi:10.1212/WNL.0b013e31829a3373 [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6OToibmV1cm9sb2d5IjtzOjU6InJlc2lkIjtzOjg6IjgxLzIvMTUwIjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjQvMDEvMTMvMjAyNC4wMS4xMi4yNDMwMTIxNS5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=) 71. 71.Dragoumi P, Tzetzi O, Vargiami E, et al. Clinical course and seizure outcome of idiopathic childhood epilepsy: Determinants of early and long-term prognosis. BMC Neurology. 2013;13. doi:10.1186/1471-2377-13-206 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1186/1471-2377-13-206&link_type=DOI) 72. 72.Gasparini S, Ferlazzo E, Beghi E, et al. Family history and frontal lobe seizures predict long-term remission in newly diagnosed cryptogenic focal epilepsy. Epilepsy Research. 2013;107(1):101–108. doi:10.1016/j.eplepsyres.2013.07.004 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.eplepsyres.2013.07.004&link_type=DOI) 73. 73.Gidey K, Chelkeba L, Gemechu TD, Daba FB. Treatment response and predictors in patients with newly diagnosed epilepsy in Ethiopia: A retrospective cohort study. Scientific Reports. 2020;10(1):2465. doi:10.1038/s41598-020-59359-8 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/s41598-020-59359-8&link_type=DOI) 74. 74.Hersi H, Saarinen JT, Raitanen J, Peltola J. Response to first antiseizure medication in patients diagnosed with epilepsy. Acta Neurologica Scandinavica. 2021;144(1):67–75. doi:10.1111/ane.13426 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/ane.13426&link_type=DOI) 75. 75.Huang C, Feng L, Li Y, et al. Clinical features and prognosis of epilepsy in the elderly in western China. Seizure - European Journal of Epilepsy. 2016;38:26–31. doi:10.1016/j.seizure.2016.03.011 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.seizure.2016.03.011&link_type=DOI) 76. 76.Jiang Y, Yuan F, Yang F, et al. Prognostic analysis for short- and long-term outcomes of newly diagnosed epilepsy. Seizure - European Journal of Epilepsy. 2017;47:92–98. doi:10.1016/j.seizure.2017.02.018 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.seizure.2017.02.018&link_type=DOI) 77. 77.Kessler SK, Shinnar S, Cnaan A, et al. Pretreatment seizure semiology in childhood absence epilepsy. Neurology. 2017;89(7):673–679. doi:10.1212/WNL.0000000000004226 [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6OToibmV1cm9sb2d5IjtzOjU6InJlc2lkIjtzOjg6Ijg5LzcvNjczIjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjQvMDEvMTMvMjAyNC4wMS4xMi4yNDMwMTIxNS5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=) 78. 78.Kim HC, Kim SE, Lee BI, Park KM. Can we predict drug response by volumes of the corpus callosum in newly diagnosed focal epilepsy? Brain and Behavior. 2017;7(8):e00751. doi:10.1002/brb3.751 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1002/brb3.751&link_type=DOI) 79. 79.Kwong KL, Ting YW, Wong SN, So KT. Acute Seizure-Related Hospitalizations in Children with Newly Diagnosed Epilepsy. Pediatric Neurology. 2007;36(5):318–323. doi:10.1016/j.pediatrneurol.2007.01.016 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.pediatrneurol.2007.01.016&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=17509464&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 80. 80.Li Y, Xia L, Wang Y, Li R, Li J, Pan S. Long-term response and response patterns to antiepileptic drugs in patients with newly diagnosed epilepsy. Epilepsy and Behavior. 2021;124. doi:10.1016/j.yebeh.2021.108309 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.yebeh.2021.108309&link_type=DOI) 81. 81.Mangunatmadja I, Ismael S, Sastroasmoro S, Suyatna FD, van Nieuwenhuizen O, Cornelis van Huffelen A. Risk factors predicting intractability in focal epilepsy in children under 3 years of age: A cohort study. Epilepsy and Behavior. 2021;123:108234:108239. doi:10.1016/j.yebeh.2021.108234 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.yebeh.2021.108234&link_type=DOI) 82. 82.Ollivier ML, Dubois MF, Krajinovic M, Cossette P, Carmant L. Risk factors for valproic acid resistance in childhood absence epilepsy. Seizure. 2009;18(10):690–694. doi:10.1016/j.seizure.2009.09.007 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.seizure.2009.09.007&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=19836978&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000273067100006&link_type=ISI) 83. 83.Oskoui M, Webster RI, Zhang X, Shevell MI. Factors predictive of outcome in childhood epilepsy. Journal of Child Neurology. 2005;20(11):898–904. doi:10.1177/08830738050200110701 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1177/08830738050200110701&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=16417860&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000234727700007&link_type=ISI) 84. 84.Park KM, Hur Y, Kim HY, et al. Initial response to antiepileptic drugs in patients with newly diagnosed epilepsy. Journal of Clinical Neuroscience. 2014;21(6):923–926. doi:10.1016/j.jocn.2013.10.031 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.jocn.2013.10.031&link_type=DOI) 85. 85.Quintana M, Sánchez-López J, Mazuela G, et al. Incidence and mortality in adults with epilepsy in northern Spain. Acta Neurologica Scandinavica. 2021;143(1):27–33. doi:10.1111/ane.13349 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/ane.13349&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 86. 86.Sharma S, Chen Z, Rychkova M, et al. Short- and long-term outcomes of immediate and delayed treatment in epilepsy diagnosed after one or multiple seizures. Epilepsy and Behavior. 2021;117. doi:10.1016/j.yebeh.2021.107880 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.yebeh.2021.107880&link_type=DOI) 87. 87.Sillanpää M, Shinnar S. Status epilepticus in a population-based cohort with childhood-onset epilepsy in Finland. Annals of Neurology. 2002;52(3):303–310. doi:10.1002/ana.10286 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1002/ana.10286&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=12205642&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000177820100007&link_type=ISI) 88. 88.Tartara E, Micalizzi E, Scanziani S, Ballante E, Paoletti M, Galimberti CA. Late- Onset Focal Epilepsy: Electroclinical Features and Prognostic Role of Leukoaraiosis. Frontiers in Neurology. 2022;13. 89. 89.Yang S, Han X, Wang N, et al. Predicting seizure freedom with AED treatment in newly diagnosed patients with MRI-negative epilepsy: A large cohort and multicenter study. Epilepsy & Behavior. 2020;106:107022. doi:10.1016/j.yebeh.2020.107022 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.yebeh.2020.107022&link_type=DOI) 90. 90.Mitchell JW, Noble A, Baker G, et al. Protocol for the development of an international Core Outcome Set for treatment trials in adults with epilepsy: The EPilepsy outcome Set for Effectiveness Trials Project (EPSET). Trials. 2022;23(1):943. doi:10.1186/s13063-022-06729-4 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1186/s13063-022-06729-4&link_type=DOI) 91. 91.Prinsen CAC, Vohra S, Rose MR, et al. Core Outcome Measures in Effectiveness Trials (COMET) initiative: Protocol for an international Delphi study to achieve consensus on how to select outcome measurement instruments for outcomes included in a ’core outcome set’. Trials. 2014;15:247. doi:10.1186/1745-6215-15-247 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1186/1745-6215-15-247&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=24962012&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 92. 92.Schulz KF, Altman DG, Moher D, CONSORT Group. CONSORT 2010 statement: Updated guidelines for reporting parallel group randomised trials. PLoS medicine. 2010;7(3):e1000251. doi:10.1371/journal.pmed.1000251 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1371/journal.pmed.1000251&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=20352064&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F01%2F13%2F2024.01.12.24301215.atom) 93. 93.Wirrell E. Evaluation of First Seizure and Newly Diagnosed Epilepsy. CONTINUUM: Lifelong Learning in Neurology. 2022;28(2):230. doi:10.1212/CON.0000000000001074 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1212/CON.0000000000001074&link_type=DOI) 94. 94.Ratcliffe C, Adan G, Marson A, et al. Neurocysticercosis-related Seizures: Imaging Biomarkers. Seizure. Published online April 2023. doi:10.1016/j.seizure.2023.04.005 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.seizure.2023.04.005&link_type=DOI) 95. 95.Collins GS, Dhiman P, Navarro CLA, et al. Protocol for development of a reporting guideline (TRIPOD-AI) and risk of bias tool (PROBAST-AI) for diagnostic and prognostic prediction model studies based on artificial intelligence. BMJ Open. 2021;11(7):e048008. doi:10.1136/bmjopen-2020-048008 [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NzoiYm1qb3BlbiI7czo1OiJyZXNpZCI7czoxMjoiMTEvNy9lMDQ4MDA4IjtzOjQ6ImF0b20iO3M6NTA6Ii9tZWRyeGl2L2Vhcmx5LzIwMjQvMDEvMTMvMjAyNC4wMS4xMi4yNDMwMTIxNS5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=)