Detecting early signs of relapse in psychosis using remote monitoring technology

Psychotic disorders, which may include hearing voices or having delusional beliefs, affect around 1% of the UK population. People living with psychosis experience fluctuations in their mental wellbeing. More severe and longer-lasting deterioration, called relapse, is distressing and disabling, and often results in admission to hospital. Identifying changes that may be early signs of impending relapse is an important goal, as it provides the opportunity to intervene at an early stage and prevent relapse.

Patients, carers and clinicians often find that natural rhythms such as sleep and levels of activity become disturbed during relapse. It may be possible to detect these changes using everyday technology including smartphones and activity trackers that monitor these rhythms passively, with minimal burden to the user. Information is sent continuously and securely to the clinical team, and in the future will also fed back to the patient, using the mobile network, so that any deterioration can be acted upon by patients, their clinicians, or both. In a pilot study, we developed a system together with service users that looks at sleep-wake patterns, movement, and variation in heart-beat. We learnt that people with serious mental illness were not only very enthusiastic about the idea, but were willing and able to engage with the technology over a period of two months.

During this fellowship at the Institute of Psychiatry, Psychology and Neuroscience (IoPPN), I will use this system for up to a year, in people for whom it would be most valuable - those who have recently been discharged from hospital, and are therefore at higher risk of further relapse. Fifty adults with a diagnosis of psychosis over a range of ages will be asked to take part. They will be provided with a smartphone containing specially designed software, an activity tracker worn on the wrist - which they will be asked to wear day and night - and a mobile phone contract. They will be asked to keep the devices charged, and complete a twice-weekly questionnaire about their thoughts and feelings. They will have the option of keeping the phone after successful completion of the study, as compensation for the time and effort involved in taking part.

Once the information has been collected, I will explore the extent to which changes in sleep patterns, movements and heart beat are related to deterioration in symptoms, and how soon before deterioration these can be detected. I will also specifically look at how disrupted sleep is related to the re-emergence of psychosis and mood symptoms. This will help us to understand when and in whom preventative interventions should be applied, and also what kind of intervention would be most useful.

The results of the study will be shared in scientific conferences, journals, as well as with national and local mental health charities, groups and the media. This will be an important step forward in developing a tool that might in the future detect relapse, and will put patients at the centre of their care by providing them with information that will assist them to better understand and manage their disorder.

AIM: To investigate the association between disturbances in sleep and circadian rhythm, motor activity, and heart rate variability (predictor variables), and symptomatic deterioration and relapse (outcome variables) in psychosis, using remote monitoring technology (RMT).

OBJECTIVES
1) To use RMT to collect predictor and outcome variables in 50 patients with psychosis for up to a year.
2) To evaluate the predictive value and temporal relationship of the bio-signatures for relapse, individually and in combination.
3) To identify evidence for sleep and circadian rhythm disturbance playing a causal role in the development of psychopathology.
4) To gain specific skills in the use and development of RMT in psychiatry, that will allow me to play a role in the future development and evaluation of clinical prediction tools.

METHODOLOGY
Objective 1: A sample of 50 individuals with psychosis will use the RMT for 12 months. Predictor variables will be gathered using a wrist-worn device and smartphone sensors. Outcome variables are obtained from a twice-weekly, validated smartphone symptom questionnaire, and from fortnightly review of the electronic patient record.

Objectives 2 and 3: The strength and temporal association between predictors and outcomes will be modelled using multilevel modelling (continuous symptom fluctuation outcomes) and penalised functional regression (binary relapse outcomes). Additionally, predictive modelling (machine learning) approaches will be used to explore these associations.

Objective 4: I will fulfil this training objective through the running of this study, and in visits to project supervisors and collaborators at this and other centres of expertise.

SCIENTIFIC AND MEDICAL OPPORTUNITIES
Results may have far-reaching implications for the development of a relapse prediction tool for psychosis, and for understanding the relationship between sleep disruption and psychopathology.

The overarching ambition of this proposal is to improve outcomes in psychosis by using widely available wearable and mobile technologies to remotely, continuously and passively collect objective signals that reflect early changes in clinical state. Realising the enormous potential of digital interventions stands to benefit patients, care providers, the healthcare system, industry, and wider society. This research is timely, relevant and necessary, and resonates with several national priorities. A realistic time course for early clinical translation would be in the next 5-10 years.

Patients and care-providers

Relapse is a major clinical problem in psychosis. Around 80% of patients undergo relapse within five years of a first episode of illness, and each episode is associated with substantial disability, distress, and negative impact on illness trajectory. When prospective behavioural and physiological data are made available in real-time to clinicians, they will be able to detect subtle fluctuations in clinical status that herald relapse, respond more quickly and efficiently, and potentially intervene in life-saving ways.

For patients, they promise to provide deeper, personalised insights into the course of illness. Helping patients to identify connections between behaviours, thoughts and feelings places them at the centre of their care by, leading to improved self-management and relapse prevention. This will promote a shift in the relationship between patients and their care-providers from being passive recipients of care, to one that encourages them to take an active role in their health.

Healthcare system

In the face of unprecedented demand and cost pressures, health services are required to do more for less. At the same time, significant sections of society, including those with mental disorders, are becoming increasingly connected and familiar with using mobile technologies, while the cost of devices and access to high-speed internet are decreasing throughout the world. The urgent need to harness the explosion in ubiquitous and scaleable digital technologies for healthcare delivery has been emphasised at policy level: for example the national mental health strategy 'No health without mental health' recommends greater use of information technology to improve care across services. This proposal responds to this need by harnessing cost-effective, scalable digital technologies to develop interventions that are accessible to a wide range of users.

Relapse is a particularly costly clinical outcome. Hospitalisation associated with relapse in psychosis accounts for 38% of the £3.9 billion annual cost of psychosis to the NHS. By developing tools for the early detection of relapse, timely preventative intervention can be instituted, saving resources and clinician time, with significant economic benefits to the healthcare system.

The model of care for chronic diseases, of which psychosis is an archetypal example, is shifting from a 'diagnose and treat' to a 'pre-empt and predict' model, with significant potential benefits for disease progression and healthcare costs. Using remote, non-invasive monitoring to predict changes in dynamic disease states in psychosis is an ideal test-bed for this paradigm.

Industry and society

Industries which design and manufacture digital technologies require clinical expertise in order to effectively translate products into clinical tools. Companies like Intel and Google are making huge investments into psychiatry and neuroscience, at a time when pharmaceutical advances in the area are diminishing. The skills, experience and knowledge gained through this proposal will establish a firm grounding for the next steps - working with industry partners to develop better interventions, and demonstrate effectiveness in large, randomised multisite clinical trails in diverse groups of patients. The economic benefits generated by these new approaches are likely to be substantial.