BrainApp: Using near-patient sensing through a mobile app and machine learning in brain tumour patients

Abstract

Abstract Aims 1. To assess the feasibility, acceptability, and performance of a mobile app, 'BRIAN', developed by The Brain Tumour Charity (BTC), in collecting data on quality of life (QOL), activity and sleep, for predicting disease progression in adult brain tumour patients. 2. To generate a prospectively collected dataset of patient measures obtained through mobile devices in brain tumour patients and healthy volunteers. 3. To assess compliance and performance of micro-challenges (hand coordination, visual memory, speech and facial features) in study participants using a mobile application. 4. To assess differences and systematic variation in micro-challenge performance between healthy volunteers and brain tumour patients. 5. To assess factors associated with micro-challenge performance in brain tumour patients, the relationship between micro-challenges and standard measures of QoL and disease progression. 6. To assess the diagnostic performance of different machine learning models in detecting brain tumour progression. Method This abstract describes the protocol for a multi-centre observational non-randomised phase II trial for adult brain tumour patients and healthy volunteers in the UK. Participants will use the BRIAN mobile app, developed by BTC to help individuals cope with a brain tumour, and share their journey with both researchers and clinicians. Participants will be required to enter information on their medical background, mood, and QOL; have the option to link fitness trackers to the app; as well as perform mini-games which assess speech, coordination, facial features and reaction time. Patients will have their brain imaging and histopathology report submitted to the sponsor. We will then investigate the correlation and temporal relation of these multimodal data with conventional measures of disease progression. We will use traditional statistical methods initially (i.e. descriptive statistics and multilevel modelling), which will then inform the development of a machine learning model in predicting brain tumour progression. Results The BRIAN app is currently being beta-tested by healthy volunteers from the Computational Oncology Lab, Imperial College London. We are in the process of obtaining ethical approval for the trial. Conclusion This study may enable the development of a tool that allows us to detect earlier signs of disease progression, and so offer earlier treatment and preservation of quality of life; and hence the best course of action. Such a tool would also be non-invasive, cheap, quick, and can be used by patients in the comfort of their own homes.