Deep brain stimulation of hippocampus in treatment of refractory temporal lobe epilepsy.

INTRODUCTION: Temporal lobe epilepsy (TLE) is the most common cause of focal onset seizures, affecting 40% of adolescents and adults with epilepsy. TLE is also one of the most common drug resistant forms of epilepsy. Surgical resection remains the treatment of choice for TLE, but not all patients with TLE are suitable candidates for resective neurosurgery. For such patients, deep brain stimulation (DBS) of the hippocampus remains a reversible and efficient treatment alternative. STATE OF THE ART: We undertook a systematic review of the literature on hippocampal DBS efficacy and safety in the management of patients with TLE. A search using two electronic databases, the Medical Literature, Analysis, and Retrieval System on-line (MEDLINE) and the Cochrane Central Register of Controlled Trials (CEN-TRAL), was conducted. CLINICAL IMPLICATIONS: We found 14 articles related to hippocampal DBS for the treatment of TLE. The responder rate (defined as at least 50% reduction in seizure frequency) for all patients was 83.4%, Of 99 patients treated by hippocampal DBS, 82 were regarded as responders, and 17 as non-responders. FUTURE DIRECTIONS: Hippocampal DBS appears to be a safe and efficacious treatment alternative for patients who are not candidates for temporal lobectomy or selective amygdalohippocampectomy due to serious postoperative cognitive deficits. In selected patients with TLE, this neuromodulatory therapy may be very safe and efficacious.

Effect of repeating hydrothermal growth processes and rapid thermal annealing on CuO thin film properties.

This paper presents an investigation into the influence of repeating cycles of hydrothermal growth processes and rapid thermal annealing (HT+RTA) on the properties of CuO thin films. An innovative hydrothermal method ensures homogeneous single-phase films initially. However, their electrical instability and susceptibility to cracking under the influence of temperature have posed a challenge to their utilization in electronic devices. To address this limitation, the HT+RTA procedure has been developed, which effectively eliminated the issue. Comprehensive surface analysis confirmed the procedure's ability to yield continuous films in which the content of organic compounds responsible for the formation of cracks significantly decreases. Structural analysis underscored the achieved improvements in the crystalline quality of the films. The implementation of the HT+RTA procedure significantly enhances the potential of CuO films for electronic applications. Key findings from Kelvin probe force microscopy analysis demonstrate the possibility of modulating the work function of the material. In addition, scanning capacitance microscopy measurements provided information on the changes in the local carrier concentration with each repetition. These studies indicate the increased usefulness of CuO thin films obtained from the HT+RTA procedure, which expands the possibilities of their applications in electronic devices.

Enhancing GaN Nanowires Performance Through Partial Coverage with Oxide Shells.

Core-shell gallium nitride (GaN)-based nanowires offer noteworthy opportunities for innovation in high-frequency opto- and microelectronics. This work delves deeply into the physical properties of crystalline GaN nanowires with aluminum and hafnium oxide shells. Particular attention is paid to partial coverage of nanowires, resulting with exceptional properties. First, the crystal lattice relaxation is observed by X-ray diffraction, photoluminescence, and Raman spectroscopy measurements. A high potential of partial coverage for optoelectronic applications is revealed with photo- and cathodoluminescence spectra along with an exploration of their temperature dependency. Next, the study focuses on understanding the mechanisms behind the observed enhancement of the luminescence efficiency. It is confirmed that nanowires are effectively protected against photoadsorption using partial coatings. This research advances the frontiers of nanotechnology, investigating the benefits of partial coverage, and shedding light on its complex interaction with cores.

Deep brain stimulation of the subiculum in the treatment for refractory temporal lobe epilepsy due to unilateral mesial temporal lobe sclerosis.

Temporal lobe epilepsy (TLE) is the most common form of drug-resistant epilepsy. The main pathological changes primarily involve hippocampal sclerosis (HS). Early resective surgery of the sclerotic hippocampus is typically associated with favorable clinical outcomes. However, not all patients are suitable candidates for resective surgery of mesial temporal lobe structures. Therefore, alternative treatment modalities should be considered. We present the case of a 50-year-old right-handed woman with left HS who underwent unilateral subiculum stimulation for drug-resistant epilepsy (DRE). Since the age of 10, the patient had been experiencing focal to bilateral tonic-clonic seizures (FBTCS). Despite multiple antiseizure medications, she experienced 12 to 17 FBTCS per month in the last two years. Due to concerns about potential memory decline and personal preferences, she refused resective surgery. As an alternative, the patient underwent left unilateral subiculum stimulation. The stimulation resulted in a nearly 67 % reduction in seizure frequency at the last follow-up (20 months after surgery). This case highlights that drug-resistant epilepsy may be effectively treated with subicular stimulation in patients with HS.

Deep brain stimulation of anterior nucleus and centromedian nucleus of thalamus in treatment for drug-resistant epilepsy.

INTRODUCTION: Drug-resistant epilepsy (DRE) remains poorly-controlled in c.33% of patients, and up to 50% of patients suffering from DRE are deemed not to be suitable candidates for resective surgery. For these patients, deep brain stimulation (DBS) may constitute the last resort in the treatment of DRE. STATE OF THE ART: We undertook a systematic review of the current literature on DBS efficacy and the safety of two thalamic nuclei-anterior nucleus of the thalamus (ANT) and the centromedian nucleus of the thalamus in the management of patients with DRE. A search using two electronic databases, the Medical Literature, Analysis, and Retrieval System on-line (MEDLINE) and the Cochrane Central Register of Controlled Trials (CEN-TRAL) was conducted. CLINICAL IMPLICATIONS: We found 30 articles related to ANT DBS and 13 articles related to CMN DBS which were further analysed. Based on the clinical research articles, we found a mean seizure frequency reduction for both thalamic nuclei. For ANT DBS, the mean seizure frequency reduction ranged from 48% to 75%, and for CMN DBS from 46.7% to 91%. The responder rate (defined as at least 50% reduction in seizure frequency) was reported to be 53.2-75% for patients after ANT DBS and 50-90% for patients after CMN DBS. FUTURE DIRECTIONS: ANT and CMN DBS appear to be safe and efficacious treatments, particularly in patients with refractory partial seizures and primary generalised seizures. ANT DBS reduces most effectively seizures originating in the temporal and frontal lobes. CMN DBS reduces mostly primary generalised tonic-clonic and atypical absences and atonic seizures. Seizures related to Lennox-Gastaut syndrome respond very favourably to CMN DBS.

Geometrical Selection of GaN Nanowires Grown by Plasma-Assisted MBE on Polycrystalline ZrN Layers.

GaN nanowires grown on metal substrates have attracted increasing interest for a wide range of applications. Herein, we report GaN nanowires grown by plasma-assisted molecular beam epitaxy on thin polycrystalline ZrN buffer layers, sputtered onto Si(111) substrates. The nanowire orientation was studied by X-ray diffraction and scanning electron microscopy, and then described within a model as a function of the Ga beam angle, nanowire tilt angle, and substrate rotation. We show that vertically aligned nanowires grow faster than inclined nanowires, which leads to an interesting effect of geometrical selection of the nanowire orientation in the directional molecular beam epitaxy technique. After a given growth time, this effect depends on the nanowire surface density. At low density, the nanowires continue to grow with random orientations as nucleated. At high density, the effect of preferential growth induced by the unidirectional supply of the material in MBE starts to dominate. Faster growing nanowires with smaller tilt angles shadow more inclined nanowires that grow slower. This helps to obtain more regular ensembles of vertically oriented GaN nanowires despite their random position induced by the metallic grains at nucleation. The obtained dense ensembles of vertically aligned GaN nanowires on ZrN/Si(111) surfaces are highly relevant for device applications. Importantly, our results are not specific for GaN nanowires on ZrN buffers, and should be relevant for any nanowires that are epitaxially linked to the randomly oriented surface grains in the directional molecular beam epitaxy.

Narcolepsy risk loci outline role of T cell autoimmunity and infectious triggers in narcolepsy.

Narcolepsy type 1 (NT1) is caused by a loss of hypocretin/orexin transmission. Risk factors include pandemic 2009 H1N1 influenza A infection and immunization with Pandemrix®. Here, we dissect disease mechanisms and interactions with environmental triggers in a multi-ethnic sample of 6,073 cases and 84,856 controls. We fine-mapped GWAS signals within HLA (DQ0602, DQB1*03:01 and DPB1*04:02) and discovered seven novel associations (CD207, NAB1, IKZF4-ERBB3, CTSC, DENND1B, SIRPG, PRF1). Significant signals at TRA and DQB1*06:02 loci were found in 245 vaccination-related cases, who also shared polygenic risk. T cell receptor associations in NT1 modulated TRAJ*24, TRAJ*28 and TRBV*4-2 chain-usage. Partitioned heritability and immune cell enrichment analyses found genetic signals to be driven by dendritic and helper T cells. Lastly comorbidity analysis using data from FinnGen, suggests shared effects between NT1 and other autoimmune diseases. NT1 genetic variants shape autoimmunity and response to environmental triggers, including influenza A infection and immunization with Pandemrix®.

Blood-Brain Barrier-Associated Proteins Are Elevated in Serum of Epilepsy Patients.

Blood-brain barrier (BBB) dysfunction emerges as one of the mechanisms underlying the induction of seizures and epileptogenesis. There is growing evidence that seizures also affect BBB, yet only scarce data is available regarding serum levels of BBB-associated proteins in chronic epilepsy. In this study, we aimed to assess serum levels of molecules associated with BBB in patients with epilepsy in the interictal period. Serum levels of MMP-9, MMP-2, TIMP-1, TIMP-2, S100B, CCL-2, ICAM-1, P-selectin, and TSP-2 were examined in a group of 100 patients who were seizure-free for a minimum of seven days and analyzed by ELISA. The results were compared with an age- and sex-matched control group. Serum levels of MMP-9, MMP-2, TIMP-1, TIMP-2 and S100B were higher in patients with epilepsy in comparison to control group (p < 0.0001; <0.0001; 0.001; <0.0001; <0.0001, respectively). Levels of CCL-2, ICAM-1, P-selectin and TSP-2 did not differ between the two groups. Serum levels of MMP-9, MMP-2, TIMP-1, TIMP-2 and S100B are elevated in patients with epilepsy in the interictal period, which suggests chronic processes of BBB disruption and restoration. The pathological process initiating epilepsy, in addition to seizures, is probably the factor contributing to the elevation of serum levels of the examined molecules.

Serum Proteins Associated with Blood-Brain Barrier as Potential Biomarkers for Seizure Prediction.

As 30% of epileptic patients remain drug-resistant, seizure prediction is vital. Induction of epileptic seizure is a complex process that can depend on factors such as intrinsic neuronal excitability, changes in extracellular ion concentration, glial cell activity, presence of inflammation and activation of the blood−brain barrier (BBB). In this study, we aimed to assess if levels of serum proteins associated with BBB can predict seizures. Serum levels of MMP-9, MMP-2, TIMP-1, TIMP-2, S100B, CCL-2, ICAM-1, P-selectin, and TSP-2 were examined in a group of 49 patients with epilepsy who were seizure-free for a minimum of seven days and measured by ELISA. The examination was repeated after 12 months. An extensive medical history was taken, and patients were subjected to a follow-up, including a detailed history of seizures. Serum levels of MMP-2, MMP-9, TIMP-1, CCL-2, and P-selectin differed between the two time points (p < 0.0001, p < 0.0001, p < 0.0001, p < 0.0001, p = 0.0035, respectively). General linear model analyses determined the predictors of seizures. Levels of MMP-2, MMP-9, and CCL-2 were found to influence seizure count in 1, 3, 6, and 12 months of observation. Serum levels of MMP-2, MMP-9, and CCL-2 may be considered potential biomarkers for seizure prediction and may indicate BBB activation.

Idling for Decades: A European Study on Risk Factors Associated with the Delay Before a Narcolepsy Diagnosis.

Purpose: Narcolepsy type-1 (NT1) is a rare chronic neurological sleep disorder with excessive daytime sleepiness (EDS) as usual first and cataplexy as pathognomonic symptom. Shortening the NT1 diagnostic delay is the key to reduce disease burden and related low quality of life. Here we investigated the changes of diagnostic delay over the diagnostic years (1990-2018) and the factors associated with the delay in Europe. Patients and Methods: We analyzed 580 NT1 patients (male: 325, female: 255) from 12 European countries using the European Narcolepsy Network database. We combined machine learning and linear mixed-effect regression to identify factors associated with the delay. Results: The mean age at EDS onset and diagnosis of our patients was 20.9±11.8 (mean ± standard deviation) and 30.5±14.9 years old, respectively. Their mean and median diagnostic delay was 9.7±11.5 and 5.3 (interquartile range: 1.7-13.2 years) years, respectively. We did not find significant differences in the diagnostic delay over years in either the whole dataset or in individual countries, although the delay showed significant differences in various countries. The number of patients with short (≤2-year) and long (≥13-year) diagnostic delay equally increased over decades, suggesting that subgroups of NT1 patients with variable disease progression may co-exist. Younger age at cataplexy onset, longer interval between EDS and cataplexy onsets, lower cataplexy frequency, shorter duration of irresistible daytime sleep, lower daytime REM sleep propensity, and being female are associated with longer diagnostic delay. Conclusion: Our findings contrast the results of previous studies reporting shorter delay over time which is confounded by calendar year, because they characterized the changes in diagnostic delay over the symptom onset year. Our study indicates that new strategies such as increasing media attention/awareness and developing new biomarkers are needed to better detect EDS, cataplexy, and changes of nocturnal sleep in narcolepsy, in order to shorten the diagnostic interval.

The Role of the Built-In Electric Field in Recombination Processes of GaN/AlGaN Quantum Wells: Temperature- and Pressure-Dependent Study of Polar and Non-Polar Structures.

In this paper, we present a comparative analysis of the optical properties of non-polar and polar GaN/AlGaN multi-quantum well (MQW) structures by time-resolved photoluminescence (TRPL) and pressure-dependent studies. The lack of internal electric fields across the non-polar structures results in an improved electron and hole wavefunction overlap with respect to the polar structures. Therefore, the radiative recombination presents shorter decay times, independent of the well width. On the contrary, the presence of electric fields in the polar structures reduces the emission energy and the wavefunction overlap, which leads to a strong decrease in the recombination rate when increasing the well width. Taking into account the different energy dependences of radiative recombination in non-polar and polar structures of the same geometry, and assuming that non-radiative processes are energy independent, we attempted to explain the 'S-shape' behavior of the PL energy observed in polar GaN/AlGaN QWs, and its absence in non-polar structures. This approach has been applied previously to InGaN/GaN structures, showing that the interplay of radiative and non-radiative recombination processes can justify the 'S-shape' in polar InGaN/GaN MQWs. Our results show that the differences in the energy dependences of radiative and non-radiative recombination processes cannot explain the 'S-shape' behavior by itself, and localization effects due to the QW width fluctuation are also important. Additionally, the influence of the electric field on the pressure behavior of the investigated structures was studied, revealing different pressure dependences of the PL energy in non-polar and polar MQWs. Non-polar MQWs generally follow the pressure dependence of the GaN bandgap. In contrast, the pressure coefficients of the PL energy in polar QWs are highly reduced with respect to those of the bulk GaN, which is due to the hydrostatic-pressure-induced increase in the piezoelectric field in quantum structures and the nonlinear behavior of the piezoelectric constant.

Data-Driven Phenotyping of Central Disorders of Hypersomnolence With Unsupervised Clustering.

BACKGROUND AND OBJECTIVES: Recent studies fueled doubts as to whether all currently defined central disorders of hypersomnolence are stable entities, especially narcolepsy type 2 and idiopathic hypersomnia. New reliable biomarkers are needed, and the question arises of whether current diagnostic criteria of hypersomnolence disorders should be reassessed. The main aim of this data-driven observational study was to see whether data-driven algorithms would segregate narcolepsy type 1 and identify more reliable subgrouping of individuals without cataplexy with new clinical biomarkers. METHODS: We used agglomerative hierarchical clustering, an unsupervised machine learning algorithm, to identify distinct hypersomnolence clusters in the large-scale European Narcolepsy Network database. We included 97 variables, covering all aspects of central hypersomnolence disorders such as symptoms, demographics, objective and subjective sleep measures, and laboratory biomarkers. We specifically focused on subgrouping of patients without cataplexy. The number of clusters was chosen to be the minimal number for which patients without cataplexy were put in distinct groups. RESULTS: We included 1,078 unmedicated adolescents and adults. Seven clusters were identified, of which 4 clusters included predominantly individuals with cataplexy. The 2 most distinct clusters consisted of 158 and 157 patients, were dominated by those without cataplexy, and among other variables, significantly differed in presence of sleep drunkenness, subjective difficulty awakening, and weekend-week sleep length difference. Patients formally diagnosed as having narcolepsy type 2 and idiopathic hypersomnia were evenly mixed in these 2 clusters. DISCUSSION: Using a data-driven approach in the largest study on central disorders of hypersomnolence to date, our study identified distinct patient subgroups within the central disorders of hypersomnolence population. Our results contest inclusion of sleep-onset REM periods in diagnostic criteria for people without cataplexy and provide promising new variables for reliable diagnostic categories that better resemble different patient phenotypes. Cluster-guided classification will result in a more solid hypersomnolence classification system that is less vulnerable to instability of single features.

Critical Evaluation of Various Spontaneous Polarization Models and Induced Electric Fields in III-Nitride Multi-Quantum Wells.

In this paper, ab initio calculations are used to determine polarization difference in zinc blende (ZB), hexagonal (H) and wurtzite (WZ) AlN-GaN and GaN-InN superlattices. It is shown that a polarization difference exists between WZ nitride compounds, while for H and ZB lattices the results are consistent with zero polarization difference. It is therefore proven that the difference in Berry phase spontaneous polarization for bulk nitrides (AlN, GaN and InN) obtained by Bernardini et al. and Dreyer et al. was not caused by the different reference phase. These models provided absolute values of the polarization that differed by more than one order of magnitude for the same material, but they provided similar polarization differences between binary compounds, which agree also with our ab initio calculations. In multi-quantum wells (MQWs), the electric fields are generated by the well-barrier polarization difference; hence, the calculated electric fields are similar for the three models, both for GaN/AlN and InN/GaN structures. Including piezoelectric effect, which can account for 50% of the total polarization difference, these theoretical data are in satisfactory agreement with photoluminescence measurements in GaN/AlN MQWs. Therefore, the three models considered above are equivalent in the treatment of III-nitride MQWs and can be equally used for the description of the electric properties of active layers in nitride-based optoelectronic devices.

Treatment of insomnia - effect of trazodone and hypnotics on sleep. / Leczenie bezsennosci ­ wplyw trazodonu i leków nasennych na sen.

Sedative antidepressants are commonly used drugs in the treatment of insomnia. However, some recommendations claim that only hypnotics have been proven effective in the treatment of sleep initiation and maintenance disorders. The aim of this article is to compare the effect of hypnotics and trazodone on sleep, and to analyse the evidence for the use of trazodone in the treatment of insomnia. Three studies investigated the effects of trazodone on sleep in primary insomnia, 5 studies on insomnia in the course of affective disorders and 6 studies on insomnia in other indications (PTSD, Alzheimer's disease, alcohol and opiate dependence, somatoform disorder, and insomnia during pregnancy). In the treatment of insomnia, trazodone is less effective than hypnotics in the treatment of sleep onset insomnia (i.e., disorders of falling asleep). For this indication it needs to be administered earlier than hypnotics, at least 1 hour before bedtime. It is, however, very effective in the treatment of sleep-maintenance insomnia, especially in patients with comorbid mental disorders or patients treated with activating antidepressants. Hypnotics and trazodone have the opposite effect on deep sleep. Trazodone increases the duration of deep sleep, which is associated with better sleep quality as assessed by patients. In contrast, hypnotics decrease slow-wave activityin sleep EEG, which is the biomarker of deep sleep. The main mechanism through which trazodone promotes sleep is its antagonistic effect on 5-HT2 serotonin receptors, while hypnotics are agonists of gamma-aminobutyric acid GABAA receptors, and other sedative antidepressants block H1 histamine receptors. This is associated with a low risk of weight gain, which is rare with trazodone treatment.

Sleep among primary care physicians-Association with overtime, night duties and strategies to counteract poor sleep quality.

Physicians belong to the risk group for sleep disorders as a result of work related stress, excessive working time, large amount of on-call duties and shift work. Poor sleep quality of healthcare providers threatens not only their safety, but also the safety of their patients. This study examines if there are any differences in how working either long hours or at night duties relates to self-reported sleep time, sleep quality and daytime sleepiness in primary care physicians. We analyzed data from a survey research of 807 participants. Respondents were divided into four groups based on their reported work hours per week and number of nights on duty per month. Overtime was associated with shorter sleep time and worse subjective sleep quality even when compared with that of participants who work on night duties. All investigated groups of participants reported short sleep latencies which suggests that sleep debt is a common problem in this population. Surprisingly in the case of poor sleep quality participants rarely used recommended methods like regular physical activity or specialist advice. The most frequently reported answer was that they do nothing to improve sleep and every third participants uses hypnotics to do that. These results suggest that primary care physicians despite being at high risk for sleep disorders due to working overtime and at night, pay less attention to their sleep quality and do not use recommended strategies to improve it.

New 2013 incidence peak in childhood narcolepsy: more than vaccination?

Increased incidence rates of narcolepsy type-1 (NT1) have been reported worldwide after the 2009-2010 H1N1 influenza pandemic (pH1N1). While some European countries found an association between the NT1 incidence increase and the H1N1 vaccination Pandemrix, reports from Asian countries suggested the H1N1 virus itself to be linked to the increased NT1 incidence. Using robust data-driven modeling approaches, that is, locally estimated scatterplot smoothing methods, we analyzed the number of de novo NT1 cases (n = 508) in the last two decades using the European Narcolepsy Network database. We confirmed the peak of NT1 incidence in 2010, that is, 2.54-fold (95% confidence interval [CI]: [2.11, 3.19]) increase in NT1 onset following 2009-2010 pH1N1. This peak in 2010 was found in both childhood NT1 (2.75-fold increase, 95% CI: [1.95, 4.69]) and adulthood NT1 (2.43-fold increase, 95% CI: [2.05, 2.97]). In addition, we identified a new peak in 2013 that is age-specific for children/adolescents (i.e. 2.09-fold increase, 95% CI: [1.52, 3.32]). Most of these children/adolescents were HLA DQB1*06:02 positive and showed a subacute disease onset consistent with an immune-mediated type of narcolepsy. The new 2013 incidence peak is likely not related to Pandemrix as it was not used after 2010. Our results suggest that the increased NT1 incidence after 2009-2010 pH1N1 is not unique and our study provides an opportunity to develop new hypotheses, for example, considering other (influenza) viruses or epidemiological events to further investigate the pathophysiology of immune-mediated narcolepsy.

Automatic artefact detection in single-channel sleep EEG recordings.

Quantitative electroencephalogram analysis (e.g. spectral analysis) has become an important tool in sleep research and sleep medicine. However, reliable results are only obtained if artefacts are removed or excluded. Artefact detection is often performed manually during sleep stage scoring, which is time consuming and prevents application to large datasets. We aimed to test the performance of mostly simple algorithms of artefact detection in polysomnographic recordings, derive optimal parameters and test their generalization capacity. We implemented 14 different artefact detection methods, optimized parameters for derivation C3A2 using receiver operator characteristic curves of 32 recordings, and validated them on 21 recordings of healthy participants and 10 recordings of patients (different laboratory) and considered the methods as generalizable. We also compared average power density spectra with artefacts excluded based on algorithms and expert scoring. Analyses were performed retrospectively. We could reliably identify artefact contaminated epochs in sleep electroencephalogram recordings of two laboratories (healthy participants and patients) reaching good sensitivity (specificity 0.9) with most algorithms. The best performance was obtained using fixed thresholds of the electroencephalogram slope, high-frequency power (25-90 Hz or 45-90 Hz) and residuals of adaptive autoregressive models. Artefacts in electroencephalogram data can be reliably excluded by simple algorithms with good performance, and average electroencephalogram power density spectra with artefact exclusion based on algorithms and manual scoring are very similar in the frequency range relevant for most applications in sleep research and sleep medicine, allowing application to large datasets as needed to address questions related to genetics, epidemiology or precision medicine.

Automatic Human Sleep Stage Scoring Using Deep Neural Networks.

The classification of sleep stages is the first and an important step in the quantitative analysis of polysomnographic recordings. Sleep stage scoring relies heavily on visual pattern recognition by a human expert and is time consuming and subjective. Thus, there is a need for automatic classification. In this work we developed machine learning algorithms for sleep classification: random forest (RF) classification based on features and artificial neural networks (ANNs) working both with features and raw data. We tested our methods in healthy subjects and in patients. Most algorithms yielded good results comparable to human interrater agreement. Our study revealed that deep neural networks (DNNs) working with raw data performed better than feature-based methods. We also demonstrated that taking the local temporal structure of sleep into account a priori is important. Our results demonstrate the utility of neural network architectures for the classification of sleep.

Exploring the clinical features of narcolepsy type 1 versus narcolepsy type 2 from European Narcolepsy Network database with machine learning.

Narcolepsy is a rare life-long disease that exists in two forms, narcolepsy type-1 (NT1) or type-2 (NT2), but only NT1 is accepted as clearly defined entity. Both types of narcolepsies belong to the group of central hypersomnias (CH), a spectrum of poorly defined diseases with excessive daytime sleepiness as a core feature. Due to the considerable overlap of symptoms and the rarity of the diseases, it is difficult to identify distinct phenotypes of CH. Machine learning (ML) can help to identify phenotypes as it learns to recognize clinical features invisible for humans. Here we apply ML to data from the huge European Narcolepsy Network (EU-NN) that contains hundreds of mixed features of narcolepsy making it difficult to analyze with classical statistics. Stochastic gradient boosting, a supervised learning model with built-in feature selection, results in high performances in testing set. While cataplexy features are recognized as the most influential predictors, machine find additional features, e.g. mean rapid-eye-movement sleep latency of multiple sleep latency test contributes to classify NT1 and NT2 as confirmed by classical statistical analysis. Our results suggest ML can identify features of CH on machine scale from complex databases, thus providing 'ideas' and promising candidates for future diagnostic classifications.