Browsing by Subject "Support vector machine"

Sort by: Order: Results:

Now showing items 1-4 of 4
  • Koolen, Ninah; Oberdorfer, Lisa; Rona, Zsofia; Giordano, Vito; Werther, Tobias; Klebermass-Schrehof, Katrin; Stevenson, Nathan; Vanhatalo, Sampsa (2017)
    Objective: To develop a method for automated neonatal sleep state classification based on EEG that can be applied over a wide range of age. Methods: We collected 231 EEG recordings from 67 infants between 24 and 45 weeks of postmenstrual age. Ten minute epochs of 8 channel polysomnography (N = 323) from active and quiet sleep were used as a training dataset. We extracted a set of 57 EEG features from the time, frequency, and spatial domains. A greedy algorithm was used to define a reduced feature set to be used in a support vector machine classifier. Results: Performance tests showed that our algorithm was able to classify quiet and active sleep epochs with 85% accuracy, 83% sensitivity, and 87% specificity. The performance was not substantially lowered by reducing the epoch length or EEG channel number. The classifier output was used to construct a novel trend, the sleep state probability index, that improves the visualisation of brain state fluctuations. Conclusions: A robust EEG-based sleep state classifier was developed. It performs consistently well across a large span of postmenstrual ages. Significance: This method enables the visualisation of sleep state in preterm infants which can assist clinical management in the neonatal intensive care unit. (C) 2017 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved.
  • O'Toole, John M.; Boylan, Geraldine B.; Lloyd, Rhodri O.; Goulding, Robert M.; Vanhatalo, Sampsa; Stevenson, Nathan J. (2017)
    Aim: To develop a method that segments preterm EEG into bursts and inter-bursts by extracting and combining multiple EEG features. Methods: Two EEG experts annotated bursts in individual EEG channels for 36 preterm infants with gestational age <30 weeks. The feature set included spectral, amplitude, and frequency-weighted energy features. Using a consensus annotation, feature selection removed redundant features and a support vector machine combined features. Area under the receiver operator characteristic (AUC) and Cohen's kappa (K) evaluated performance within a cross-validation procedure. Results: The proposed channel-independent method improves AUC by 4-5% over existing methods (p <0.001, n = 36), with median (95% confidence interval) AUC of 0.989 (0.973-0.997) and sensitivity -specificity of 95.8-94.4%. Agreement rates between the detector and experts' annotations, K = 0.72 (0.36-0.83) and K = 0.65 (0.32-0.81), are comparable to inter-rater agreement, K = 0.60 (0.21-0.74). Conclusions: Automating the visual identification of bursts in preterm EEG is achievable with a high level of accuracy. Multiple features, combined using a data-driven approach, improves on existing single-feature methods. (C) 2017 The Authors. Published by Elsevier Ltd on behalf of IPEM.
  • Longato, Enrico; Acciaroli, Giada; Facchinetti, Andrea; Hakaste, Liisa; Tuomi, Tiinamaija; Maran, Alberto; Sparacino, Giovanni (2018)
    Many glycaemic variability (GV) indices extracted from continuous glucose monitoring systems data have been proposed for the characterisation of various aspects of glucose concentration profile dynamics in both healthy and non-healthy individuals. However, the inter-index correlations have made it difficult to reach a consensus regarding the best applications or a subset of indices for clinical scenarios, such as distinguishing subjects according to diabetes progression stage. Recently, a logistic regression-based method was used to address the basic problem of differentiating between healthy subjects and those affected by impaired glucose tolerance (IGT) or type 2 diabetes (T2D) in a pool of 25 GV-based indices. Whereas healthy subjects were classified accurately, the distinction between patients with IGT and T2D remained critical. In the present work, by using a dataset of CGM time-series collected in 62 subjects, we developed a polynomial-kernel support vector machine-based approach and demonstrated the ability to distinguish between subjects affected by IGT and T2D based on a pool of 37 GV indices complemented by four basic parameters—age, sex, BMI, and waist circumference—with an accuracy of 87.1%.
  • Tapani, Karoliina T.; Nevalainen, Päivi; Vanhatalo, Sampsa; Stevenson, Nathan J. (2022)
    Neonatal seizure detection algorithms (SDA) are approaching the benchmark of human expert annotation. Measures of algorithm generalizability and non-inferiority as well as measures of clinical efficacy are needed to assess the full scope of neonatal SDA performance. We validated our neonatal SDA on an independent data set of 28 neonates. Generalizability was tested by comparing the performance of the original training set (cross -validation) to its performance on the validation set. Non-inferiority was tested by assessing inter-observer agreement between combinations of SDA and two human expert annotations. Clinical efficacy was tested by comparing how the SDA and human experts quantified seizure burden and identified clinically significant periods of seizure activity in the EEG. Algorithm performance was consistent between training and validation sets with no significant worsening in AUC (p > 0.05, n = 28). SDA output was inferior to the annotation of the human expert, however, re-training with an increased diversity of data resulted in non-inferior performance (delta kappa = 0.077, 95% CI:-0.002-0.232, n = 18). The SDA assessment of seizure burden had an accuracy ranging from 89 to 93%, and 87% for identifying periods of clinical interest. The proposed SDA is approaching human equivalence and provides a clinically relevant interpretation of the EEG.