Predicting outcomes after inpatient neurorehabilitation version 1

  • Research type

    Research Study

  • Full title

    Predicting psychosocial outcomes at the time of discharge from inpatient neurorehabilitation for acquired brain injury: Development and validation of a machine learning model.

  • IRAS ID

    266548

  • Contact name

    Breda Cullen

  • Contact email

    breda.cullen@glasgow.ac.uk

  • Sponsor organisation

    NHS Greater Glasgow and Clyde

  • Duration of Study in the UK

    0 years, 5 months, 29 days

  • Research summary

    Background

    Acquired brain injury (ABI), an injury of the brain after birth, such as stroke or traumatic brain injury, can cause significant lifestyle changes for the person affected and their family members. Rehabilitation inpatient programmes have been developed for people with severe ABI and additional complications, which aim to improve Service Users (SUs) functioning and help the person adapt to living with a brain injury. After discharge from rehabilitation, the level of functioning can vary from person to person, with differing levels of support for accommodation, work and recreational activities (known as psychosocial outcomes).

    There has been limited research for predicting how likely it is people admitted to rehabilitation will have good or poorer psychosocial outcomes. Being able to predict this will help prepare emotionally and financially for discharge.

    Machine learning (ML) is a type of data analysis for making predictions. ML methods overcome some of the limitations that traditional predictive statistics have, making ML a potentially superior method for predicting healthcare outcomes. ABI rehabilitation centres generate large volumes of data which ML is able to compute to inform us which variables are most helpful for predicting psychosocial outcomes.

    Aims and questions

    This study aims to build models to predict psychosocial outcomes at discharge from inpatient rehabilitation. In particular, the study aims to find out whether the model performs significantly better than chance to predict psychosocial outcomes using data commonly collected at admission to inpatient rehabilitation for people with ABI.

    Participants

    All SUs admitted and discharged with ABI from care of Graham Anderson House (GAH) Rehabilitation Centre, Glasgow, between 2009 and 2018 with available data will be included in the analysis. This study is using routinely collected information. No new participants need to be recruited.

  • REC name

    West of Scotland REC 1

  • REC reference

    20/WS/0026

  • Date of REC Opinion

    24 Feb 2020

  • REC opinion

    Further Information Favourable Opinion