Hasso-Plattner-InstitutSDG am HPI
Hasso-Plattner-InstitutDSG am HPI
  
Login
  • de
 

Closed-loop warning system for epilepsy monitoring

Sidratul Moontaha

Digital Health - Connected Healthcare
Hasso Plattner Institute

Office: Campus III Building G2, G-2.1.21
Tel.: +49-(0)331 5509-XXXX
Email: sidratul.moontaha(at)hpi.de
Links: https://hpi.de/arnrich/people/sidratul-moontaha.html

Supervisor: Prof. Dr. Bert Arnrich

Overview

Epilepsy is the most common neurological disorder characterized by unprovoked and unexpected electrical bursts in the brain, which results in seizures. Initially, most epilepsy treatments lie on antiepileptic drugs (AEDs). However, among the 0.5-1% of the pediatric population suffering from epilepsy, about 25-30% of patients have drug-resistant epilepsy, which is defined as the failure of adequate trials of two tolerated, appropriately chosen and administered AEDs (whether as monotherapy or in combination) to achieve seizure freedom. The severity of the seizure can lead to SUDEP (sudden unexpected death in epilepsy). Additionally, the quality of life (QoL) of these patients highly depends on other comorbidities such as cognitive impairment, depression, medication side effects, sleep quality, and privacy, and importantly, the unpredictability of seizure occurrence. A closed-loop warning system is indispensable to improve the QoL of these patients, which includes continuous monitoring of patient's data, seizure prediction (before seizure onset) and detection (during seizure onset), maintaining electronic seizure diary, and providing right dosages of AEDs.

Self-prediction of seizures in drug resistance epilepsy using digital phenotyping

Introduction

Drug-resistance is a prevalent condition in children and adult patients with epilepsy. The quality of life of these patients is profoundly affected by the unpredictability of seizure ccurrence. Some of these patients are capable of reporting self-prediction of their seizures by observing their affectivity. Some patients report no signs of feeling premonitory symptoms, prodromes, or aura. In this paper, we propose a concept study that will provide objective information to self-predict seizures for both the patient groups. We will develop a model using digital phenotyping which takes both ecological momentary assessment and data from sensor technology into consideration. This method will be able to provide a feedback of their premonitory symptoms so that a pre-emptive therapy can be associated to reduce seizure frequency or eliminate seizure occurrence.

Objectives and Research Question

Within this research framework, the main objective is to quantify prodromes and premonitory symptoms in the pre-ictal and the interictal state of the brain consecutively, which leads to the following research questions:

  • Which affective states are responsible for preceding seizures in the pre-ictal and inter-ictal phase? Is it possible to extract more reliable and conclusive outcomes on affective states preceding seizures compared to the state-of-art discussed?
  • Is it possible to demonstrate the patient-reported feelings of prodromes or premonitory symptoms of seizure self-prediction by physiological measures?
  • To improve the QoL, is it possible to provide objective information of prodromes or premonitory symptoms from physiological measures to PWE who can not self-predict seizure?
  • Sensitivity: what percentage of seizures which are preceded by prodromes or premonitory symptoms can be detected using the proposed system? What are the correct PPV and correct NPV of seizures? Does it comprehend the state of the art findings?
  • False prediction rate: what percentage of specificity can be achieved?

Study Design

At the beginning of our study, we will choose a set of patients according to the criterion mentioned above and monitor them for two weeks as a baseline study. As depicted in figure 1, the PWE will be divided into two groups: group 1: who are capable of reporting premonitory symptoms and/or prodromes and group 2: who do not report of premonitory symptoms and/or prodromes. Both the groups will answer questions from the Beck Depression Inventory-II (BDI), Generalized Anxiety Disorder–7 (GAD), State-Trait Anxiety Inventory (STAI), Self-Efficacy Scale, and Positive and Negative Affect Schedule (PANAS) screening tools. The scores from these screening tools will be essential to do the psychometric evaluation of depression, anxiety, stress, and affectivity, which will be used as ground truth for PWE. Along with that, each PWE will be given a smartphone with E-diary App developed in [8]. In this App, they will log their seizures, premonitory symptoms, sleep hours, medication adherence, menstrual cycle, current illness once in a day at a randomly prompted time. The PWE also has to complete multiple mood items adapted from the circumplex models of affect using a visual analog scale. The PWE will be prompted to fill in this mood model and will be asked by the application to mention their contextual information at randomly sampled intervals daily. To measure the physiological signals such as HRV, EDA, unobtrusively, the PWE will be asked to wear smart wearable devices containing multiple sensors. These physiological signals will provide objective information about their affectivity. For group 1, we will try to recognize patient-reported premonitory symptoms preceding seizures with the physiological signals measured. For group 2, we will try to find abnormalities in the physiological signals that might have led to a seizure.

In the next two weeks, we would monitor the PWE in the hospital and observe them with video-EEG. These patients will have minimum movement restrictions to make the data more realistic to real life. To compare the emotional intensity with the previous baseline study for each patient, the emotional regulations will be evaluated by presenting five films inducing emotional responses discussed in [10]. The skin conductance will be monitored while performing the attention task and suppression task. Apart from that, the PWE will be asked to perform simple cognitive task by solving arithmetic problems with the App used in [5]. The physiological signals will be monitored for the whole period of study. Before leaving the hospital, the PWE will be asked to fill up the set of questions in the inventory mentioned earlier. After this hospital settings, the patients will be monitored for two more weeks in an uncontrolled environment. At that time, the physiological signals, the questionnaires from the inventory and E-diary will be monitored again. This data will be used for a comparison study with the baseline and the hospital study. The processing of the collected data will be used to set up a model explained in the next section and the timeline of the research is summarized in table 1.

Analyzing the effects of anti-epileptic drugs on seizure frequency

This project aims at developing and implementing a flexible methodology for modelling arbitrary count time series by employing a state space modelling approach and at designing appropriate control strategies based on the modelling results. The focus of the intended applications lies on biomedical time series; in particular, the task of monitoring and controlling the therapy of patients suffering from epilepsy by AED medication will be addressed. This application will serve as a realistic testbed for the developed methodology.

Quantitative methods from system identification and control will be introduced into an important field of clinical therapy planning. We aim at making optimal use of the available data, i.e., of the response of the patient to previously administered medication. This project is based on the hypothesis that the quantitative approach to analysing this data based on time series analysis and state space modelling will provide superior results, as compared to the current standard given by the direct least-squares regression approach or by visual inspection and subjective decision based on the experience of the physician conducting the treatment. As a result, the physician will be provided with a tool for automatic monitoring of ongoing treatment and for suggesting changes and adaptations. The minimum data set size that is required for such tool to function reliably will be investigated, possibly by a suitable simulation approach. We expect that our approach will offer a solution also for cases that are difficult to assess, e.g.,  given simultaneous administration of several drugs with time-varying dosages.

In future work, also additional data besides the daily counts of seizures may be added to the analysis, such as data on side effects of the medication.

To achieve these goals this project aims at improving and extending the available methodology for nonlinear Kalman Filtering and for estimation of model parameters by maximisation of the likelihood given count time series of limited length, as are common in many biomedical applications. For this purpose, recently proposed algorithms, like square-root Kalman Filtering and Kalman Filtering based on Singular Value Decomposition, will have to be generalised for the nonlinear case. Also available algorithms for numerical optimisation will have to be adapted to the particular situation by employing recent developments, such as square-root versions of the Expectation Maximisation algorithm.