Automated analysis of transient loss of consciousness history

  • Research type

    Research Study

  • Full title

    Improving the diagnosis of transient loss of consciousness through automated analysis of the history

  • IRAS ID

    274410

  • Contact name

    Nathan Darnley

  • Contact email

    ndarnley1@sheffield.ac.uk

  • Sponsor organisation

    Sheffield Teaching Hospital

  • Duration of Study in the UK

    2 years, 4 months, 27 days

  • Research summary

    Over 90% of patients presenting to health services with a Transient Loss of Consciousness (TLOC) are diagnosed with epilepsy, syncope (fainting) or dissociation (nonepileptic seizures). Patients are often asymptomatic on presentation to a health professional, and investigations are of limited diagnostic value in the differentiation of the common causes of TLOC. This means that the act of an expert taking and analysing the history available from patients and witnesses (if available) continues to represent the cornerstone of the diagnostic pathway. Unfortunately, expertise in the medical settings where most patients with TLOC initially present (Primary Care or Emergency Departments) is often limited, and many patients receive an inaccurate initial diagnosis or are referred to the wrong secondary care service. A recent study modelling the diagnostic performance of a procedure called the i-Paroxysmal Event Profile (iPEP), consisting of a 36-item questionnaire analysed using the machine learning method ‘Random Forest’, suggested that 86% of all patients could be classified accurately into one of the three common diagnostic groups based on their responses. Separately, other research has shown that the particularly challenging diagnostic differentiation between epileptic and dissociative seizures may be enhanced by the detailed analysis of the conversational behaviour of patients with epilepsy or dissociative seizures. Previous studies in patients presenting with memory disorders have demonstrated that it is feasible to automate such analyses of conversational behaviour. The present research aims to create a ‘digital doctor’, i.e. a fully-automated, computer-based diagnostic method for TLOC presentations combining the iPEP with an automated analysis of patients’ free descriptions of their seizure experiences. The 'digital doctor' diagnosis will be compared to best possible medical diagnoses. A subgroup of participants will be invited to provide feedback on the 'iPEP procedure' to assess the acceptability of the automated history-taking procedure from the patient's perspective.

  • REC name

    East Midlands - Leicester South Research Ethics Committee

  • REC reference

    20/EM/0106

  • Date of REC Opinion

    7 May 2020

  • REC opinion

    Further Information Favourable Opinion