Predicting seizure outcome after epilepsy surgery: Do we need more complex models, larger samples, or better data?

OBJECTIVE: The accurate prediction of seizure freedom after epilepsy surgery remains challenging. We investigated if (1) training more complex models, (2) recruiting larger sample sizes, or (3) using data-driven selection of clinical predictors would improve our ability to predict postoperative seizure outcome using clinical features. We also conducted the first substantial external validation of a machine learning model trained to predict postoperative seizure outcome. METHODS: We performed a retrospective cohort study of 797 children who had undergone resective or disconnective epilepsy surgery at a tertiary center. We extracted patient information from medical records and trained three models-a logistic regression, a multilayer perceptron, and an XGBoost model-to predict 1-year postoperative seizure outcome on our data set. We evaluated the performance of a recently published XGBoost model on the same patients. We further investigated the impact of sample size on model performance, using learning curve analysis to estimate performance at samples up to N = 2000. Finally, we examined the impact of predictor selection on model performance. RESULTS: Our logistic regression achieved an accuracy of 72% (95% confidence interval [CI] = 68%-75%, area under the curve [AUC] = .72), whereas our multilayer perceptron and XGBoost both achieved accuracies of 71% (95% CIMLP = 67%-74%, AUCMLP = .70; 95% CIXGBoost own = 68%-75%, AUCXGBoost own = .70). There was no significant difference in performance between our three models (all p > .4) and they all performed better than the external XGBoost, which achieved an accuracy of 63% (95% CI = 59%-67%, AUC = .62; pLR = .005, pMLP = .01, pXGBoost own = .01) on our data. All models showed improved performance with increasing sample size, but limited improvements beyond our current sample. The best model performance was achieved with data-driven feature selection. SIGNIFICANCE: We show that neither the deployment of complex machine learning models nor the assembly of thousands of patients alone is likely to generate significant improvements in our ability to predict postoperative seizure freedom. We instead propose that improved feature selection alongside collaboration, data standardization, and model sharing is required to advance the field.

Autism, ADHD and parent-reported behavioural difficulties in young children with epilepsy.

PURPOSE: To provide data on the prevalence of Autism Spectrum Disorder (ASD), Attention-Deficit/Hyperactivity Disorder (ADHD), and parent reported behaviour difficulties in young children with epilepsy, and to compare results with children with neurodisability (neurodevelopmental/neurological difficulties) without epilepsy. METHOD: Children with epilepsy (1-7 years, n = 48) and children with neurodisability (1-7 years, n = 48) matched for gender, chronological and developmental age underwent psychological assessment. Parents completed measures of behaviour including the Strengths and Difficulties Questionnaire (SDQ). DSM-5 diagnoses of ASD and ADHD were made at consensus case conferences. Factors associated with child behaviour were analysed using linear regression. RESULTS: Of the children with epilepsy, 18% met ASD criteria and 40% met ADHD criteria (corresponding figures in the non-epilepsy group were 41% and 27%). A large proportion (76%-78%) in both groups scored in the at-risk range on the SDQ and frequently had difficulties across multiple behavioural domains. Children with epilepsy had more concerns expressed regarding attention and mood. None of the epilepsy factors were significantly associated with scores on the behavioural measures. SIGNIFICANCE: Young children with epilepsy had a very high level of parent reported behavioural difficulties and a high risk for ADHD and ASD highlighting the need for comprehensive multidisciplinary assessment. Behavioural concerns were not greater than for other children with non-epilepsy related neurodisability with the exception of attention and mood. Epilepsy related factors were not associated with child behaviour, suggesting that seizures per se do not confer a unique risk for behavioural difficulties.

Experiences and needs of parents of young children with active epilepsy: A population-based study.

The aim of the study was to gain a comprehensive understanding of the experiences and needs of parents of young children with epilepsy from a total population sample. The parents (mothers (n = 38), fathers (n = 9)) of 40/53 (75% of total population) young children (1-7 years; 23 males, 17 females) with 'active' epilepsy (had a seizure in the last year or taking Anti-epileptic drugs (AEDs)) were interviewed either in person or over the telephone using a semistructured interview schedule. The families were resident in the south of the UK. The interviews were audio-recorded, transcribed, and coded using thematic analysis. Thematic analysis revealed six main themes: diagnostic journey, parental perception of epilepsy management, awareness and impact of associated neurobehavioral difficulties, inconsistent availability of therapeutic and educational supports, impact on family functioning, and need for parental support. Parents reported often having difficulty accessing a professional knowledgeable about epilepsy. While parents were generally satisfied with the initial information they received about seizures and their management, they reported that the association between epilepsy and neurobehavioral issues was often not broached. These developmental/behavioral difficulties often had a bigger impact on child wellbeing and family functioning, but provision of therapeutic and educational supports for the difficulties was often very patchy. Parents noted that early onset epilepsy and associated neurobehavioral difficulties often have a very significant impact on family functioning including increased restrictions on family activities and increased financial burden. Parents would like informational and emotional support to extend beyond the time of epilepsy diagnosis. There is a clear need for comprehensive childhood epilepsy services to include provision for identification and management of child neurobehavioral needs and a focus on family-centered care.

Global development and adaptive behaviour in children with early-onset epilepsy: a population-based case-control study.

AIM: There are limited population-based data on global development and adaptive behaviour in children with early-onset epilepsy. The aims of this study were: (1) to identify the prevalence of deficits in global development and adaptive behaviour experienced by children with early-onset epilepsy; (2) to identify factors associated with such deficits; and (3) to compare the relationship between measures of neurodevelopment in the group with epilepsy to a group without epilepsy who had other neurological or neurodevelopmental difficulties. METHOD: The Sussex Early Epilepsy and Neurobehaviour study is a prospective, community-based study involving children (1-7y) with epilepsy. We undertook comprehensive psychological assessment with participants, including measures of global development and adaptive behaviour. We compared the children with epilepsy with a sex, age, and developmentally-matched group of children without epilepsy who had neurodevelopmental or neurological difficulties using correlation matrices. RESULTS: Forty-eight children (91% of the eligible population) with epilepsy underwent assessment. Seventy-one per cent of children displayed delayed global development (<2SD) and 56% showed significant deficits (<2SD) in adaptive behaviour. Our analysis revealed that non-white ethnicity and use of polytherapy were independently associated with decreased scores on measures of global development and adaptive behaviour. The correlations between measures of developmental functioning were higher in children with epilepsy than in those without. INTERPRETATION: Children with early-onset epilepsy frequently have difficulties with global development and adaptive behaviour. The higher correlations between neurodevelopmental measures in children with epilepsy suggest that the profile in children with epilepsy is different. This may have significant implications for both neuropathology and interventions. WHAT THIS PAPER ADDS: Children with early-onset epilepsy are at significant risk of intellectual disability. Developmental impairment is associated with use of polytherapy but not with any seizure parameters. Developmental profiles in young children with epilepsy differ from other conditions.

Parenting stress and perceived stigma in mothers of young children with epilepsy: A case-control study.

The aim was to provide data on parenting stress and perceived stigma in mothers (n = 47) of young children with epilepsy, and to compare findings with those of mothers (n = 48) of developmental, age- and gender-matched children with nonepilepsy-related neurodisability (neurological and/or neurodevelopmental concerns). The mothers of young children (1-7 years) with epilepsy and mothers of children with neurodisability in a defined geographical area of the UK, completed the Parenting Stress Index-4th Edition (PSI-4) and a measure of perceived stigma. Factors associated with parenting stress and stigma were analyzed using linear regression. Thirty-eight percent of mothers of children with epilepsy scored in the at-risk range (>85th percentile) on the Total Stress score of the PSI-4 (Neurodisability 21%) (p = 0.06). Significantly more mothers of children with epilepsy scored in the at-risk range on the Parent-Child Dysfunctional Interaction subscale than mothers of children with neurodisability (Epilepsy 45% vs. Neurodisability 21%; p = 0.01), but not on the Parental Distress subscale (Epilepsy 32% vs. Neurodisability 23%; p = 0.33) or Difficult Child (Epilepsy 57% vs. Neurodisability 46%; p = 0.26) subscales. There was no statistically significant difference in perceived stigma between mothers in both groups (p = 0.51). Factors significantly associated with increased parenting stress in the group with epilepsy were child behavior difficulties (p < 0.001) and maternal sleep difficulties (p = 0.02). Lower child developmental level was the only factor independently associated with increased stigma in the group with epilepsy (p = 0.08). Mothers of young children with epilepsy report high levels of parenting stress and higher levels of difficulties with parent-child interaction compared with that of mothers of children with nonepilepsy-related neurodisability. Parenting stress and stigma in epilepsy were not associated with epilepsy factors. Efforts at reducing parenting stress and stigma should focus on interventions targeting child development and maternal sleep.

Child and parental sleep in young children with epilepsy: A population-based case-control study.

OBJECTIVE: To determine the prevalence of parent-reported sleep problems in young children with epilepsy and their parents, and to compare findings with those in a non-epilepsy-related neurodisability (neurodevelopmental/neurological difficulties) group. METHOD: Parents of young children (1-7 years) with epilepsy (n = 48 [91% ascertainment]) completed the Child Sleep Habits Questionnaire (CSHQ). Parents (mothers and fathers) also completed the Pittsburgh Sleep Quality Index (PSQI) and the Iowa Fatigue Scale (IFS) in relation to their own functioning. The responses of parents of children with epilepsy were compared with parents of developmental-, age-, and gender-matched children with nonepilepsy-related neurodisability (n = 48). RESULTS: There was not a significant difference in the proportion of children with epilepsy and the children with neurodisability scoring in the at-risk range on the CSHQ (81% vs. 71% respectively) (p = 0.232). 62% of mothers and 44% of fathers of children with epilepsy had 'poor quality sleep' on the PSQI; there was not a significant difference between mothers of children with epilepsy and those of children with neurodisability (p = 0.526) or IFS (p = 0.245) total scores. However, mothers of children with epilepsy had significantly more difficulties on the productivity subscale of the IFS (p = 0.004). There were no significant differences between fathers' scores on either measure. In the epilepsy group, child behavioral problems (p = 0.001) were independently associated with child sleep difficulties and maternal mental health problems were associated with parental sleep difficulties (p = 0.04) and fatigue (p = 0.018). SIGNIFICANCE: Young children with epilepsy and their parents have a high rate of sleep difficulties. There is a need to develop effective interventions for this population, taking into consideration of the role of child behavioral problems and parental mental health difficulties.

Symptoms of depression, anxiety, and stress in parents of young children with epilepsy: A case controlled population-based study.

The objective was to provide population-based data on depression, anxiety, and stress in parents of young children with epilepsy and to compare findings with those of parents of developmental-, age-, and gender-matched children with nonepilepsy-related neurodisability (neurological and/or neurodevelopmental concerns). The parents (mothers and fathers) of 47 (89% ascertainment) young children (1-7years) with epilepsy in a defined geographical area of the UK completed the Depression Anxiety Stress Scales - Short Form (DASS-21), a screening measure for depression, anxiety, and stress. The responses of parents of children with epilepsy were compared with parents of developmental-, age-, and gender-matched children with nonepilepsy-related neurodisability (n=48). Factors associated with parental symptoms were analyzed using regression. In the group with epilepsy, 47 mothers and 39 fathers completed the DASS-21. Seventy-two percent of mothers scored in the at-risk range on at least one DASS-21 subscale (Fathers 49%). Mothers of children with epilepsy were significantly more likely to score in the at risk range than fathers on depression (55% vs. 33%), anxiety (47% vs. 26%), and stress (55% vs. 31%) subscales (all p<0.05). Mothers of children with epilepsy were also significantly more likely to score in the at-risk range than mothers of children with neurodisability on measures of depression (p=0.005) and stress (p=0.03). There was not a significant difference between fathers in both groups on any measures. In the group with epilepsy, increased child emotional-behavioral difficulties were associated with increased DASS-21 scores on multivariable analysis (p=0.04). Mothers of young children with epilepsy are at high risk for mental health difficulties, and all should be screened for such difficulties. There is a need to explore what parent and/or child focused interventions might be useful to reduce the mental health difficulties reported by mothers of young children with epilepsy.

Crossing the lines between epilepsy syndromes: a myoclonic epilepsy variant with prominent eyelid myoclonia and atonic components.

Accurate diagnosis of a distinct epilepsy syndrome is based on well-defined electroclinical features that differentiate separate nosological entities. In clinical practice, however, syndromes may overlap and cases may present with unusual manifestations posing a diagnostic challenge. This heterogeneity has been documented in several cases presenting with eyelid myoclonia with or without absences (EMA) diagnosed either as Jeavons syndrome (JS) variants or as genetic generalised epilepsies defined by the presence of this unique clinical entity. The hallmark of JS is the triad: (1) eyelid myoclonia with or without absences, (2) eye closure-induced paroxysms, and (3) photosensitivity. The presence of massive myoclonus, intellectual disability, or slowing of the EEG background are not typical features of the syndrome and may cause delay in making the correct diagnosis. Adding to the variability of clinical features, we describe two female paediatric patients with probable genetic epilepsy who presented with EMA but demonstrated clear atypical features, such as prominent myoclonic seizures, atonic components on video-EEG, and cognitive impairment. We also note the presence of interictal and ictal posterior discharges during eyelid myoclonia in one, supporting similar previous observations leading to consideration of EMA as an occipital cortex-initiated seizure activity. [Published with video sequences on www.epilepticdisorders.com].

Parent- and Teacher-Reported Symptoms of ADHD in School-Aged Children With Active Epilepsy: A Population-Based Study.

OBJECTIVE: Provide data on the distribution of parent- and teacher-reported symptoms of ADHD in childhood epilepsy and describe coexisting cognitive and behavioral disorders in children with both epilepsy and ADHD. METHOD: Eighty-five (74% of those eligible) children (5-15 years) in a population-based sample with active epilepsy underwent psychological assessment. The ADHD Rating Scale-IV (ADHD-RS-IV) scale was completed by parents ( n = 69) and teachers ( n = 67) of participating children with an IQ > 34. ADHD was diagnosed with respect to Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.). RESULTS: Parents reported significantly more symptoms of ADHD than teachers ( p < .001). Symptoms of inattention were more commonly reported than symptoms of hyperactivity-impulsivity ( p < .001). Neurobehavioral comorbidity was similar in those with ADHD and non-ADHD with the exception of oppositional defiant disorder (ODD) and developmental coordination disorder (DCD), which were more common in those with both epilepsy and ADHD. CONCLUSION: Symptoms of ADHD are very common in childhood epilepsy but prevalence is influenced by informant.

Epilepsy with myoclonic-atonic seizures (Doose syndrome): When video-EEG polygraphy holds the key to syndrome diagnosis.

An electroclinical epilepsy syndrome diagnosis enables physicians to predict outcomes as well as select appropriate treatment options. We report a child who presented with reflex myoclonus at the age of 9 months and was initially diagnosed with myoclonic epilepsy in infancy. After 9 years of medically resistant myoclonic seizures, extensive investigations, and emerging learning difficulties, she was referred for video-telemetry to characterize her seizures in an attempt to make a syndromic diagnosis. A three-day video-telemetry assessment was performed to document seizures. Neck and deltoid EMG channels were applied from the onset of the recording. Frequent generalized bursts of 3- to 5-Hz spike/polyspike and slow wave discharges, associated with clinical manifestations, mostly myoclonic seizures, were noted. In addition, definite atonic components were noted on the neck EMG as well as the deltoids associated with the slow component of the ictal discharges. The EEG and polygraphy findings are suggestive of a generalized epilepsy characterized by predominantly myoclonic seizures with atonic components. This raises the possibility whether a variant of epilepsy with myoclonic-atonic seizures (Doose syndrome) may be the underlying diagnosis for this girl. A trial of the ketogenic diet would therefore be considered as an option in her future management in view of its beneficial effect in this condition.