Digital Interventions in Neuro-Rehabilitation: Gotcha!

Background

The number of people with dementia in the UK is forecast to increase by 40% over the next 12 years and 156% over the next 38 years (Prince et al., 2014). People with all stages of Alzheimer's disease (AD) report forgetting even very familiar people's names, which patients find annoying and embarrassing (Werner, 2004). This therapy speaks to the James Lind Alliance's dementia top priority, "What are the most effective components of care that keep a person with dementia as independent as they can be at all stages of the disease in all care settings?" Several case studies and one group study have shown that patients with AD can significantly improve proper name recall with training; the effects last months after training is completed (Clare et al., 2002). Given the prevalence of mobile phone technology with inbuilt cameras and the rapid rise in usage in the 64-74 year old bracket, the investigators will develop a name-training intervention that allows people with proper name recall impairments to take pictures of and add the name of people that they wish to name. The app will include voice recognition software so that our bespoke adaptive algorithm (developed with Prof Mark Huckvale at UCL) will be able to adaptively cue users depending on their speech output.

Gotcha! will provide an app-based therapy for people with dementia who have difficulty naming people they know well.

This Digital Neuro Intervention (DNI) will provide the opportunity for the necessary increased rehabilitation to help people recover lost naming function. This will alleviate NHS Speech and Language Therapist (SALT) time and put users in control of when and where they carry out practice-based language therapy.

Identification and description of the Investigational Device

Purpose Gotcha! will provide an app-based therapy for people with dementia who have difficulty naming people they know well.

What does it do? The therapy element works via mass practice and feedback to the users on a trial-by-trial basis. The DNI is very simple in a way; it just presents a long series of pictures for users to name. This DNI will use speech recognition (SR) software in order to make a binary decision as to whether the user said the correct word or not. This affects what the next trial (object to name) is and what auditory cue (if any) is provided the next time the user has to name the same item.

*The speech will be recorded on the device but no other identifiable information will be stored on it. When the participant finishes the trial, the investigators will delete all information on the device.

What is the clinical trial design? A small, well-defined sample of patients with the potential to benefit from the DNI will be recruited. The main outcome measure is whether the DNI is effective at improving naming impairments. This comparison is within-subject and is achieved by comparing post-therapy measures to multiple baseline measures. The investigators are also comparing two different versions of the therapy (a between group comparison) to answer our secondary question of whether the therapy can be optimised. Outcome measures will be collected by the research team using a variety of standardized and non-standardized tests of language and cognition.

The DNI will be introduced to the participant when attending a testing session. One of the research team will explain how to use the DNI, provide instructions for the user, and give some time for practice items. When confident that the participant can use the DNI independently, the participant will take the device home and continue their therapy (suggested therapy time is 5-10 hours per week) which will be simultaneously monitored by the research team. The research team will also check in on the participant weekly to trouble shoot any difficulties they may have (whether motivational or technical). When the participant has completed the therapy block, the investigators will invite them in again for testing and they will return the mobile device with the DNI on it. The participants who continue to receive standard care during the therapy block will be asked to record and submit details to the research team.

Justification for the design of the clinical investigation Why is the research considered worth doing? Standard (face to face) Speech and Language Therapy (SALT) has a huge evidence-base but patients in the NHS have unmet therapy needs due to a lack of resources. The evidence from speech-therapy highlights the amount of time-on-task required to improve patients' ability to communicate. Naming problems are common and impact on the patients' wellbeing and social inclusion. There is also good evidence that current existing therapeutic approaches work.

In the NHS a patient with Alzheimer's Dementia are likely to receive little or no SALT. Our solution to the lack of available SALT therapy is to produce digital neuro-interventions that give patients the opportunity to practice scientifically validated, impairment based therapy when and where it suits them which gives them access to optimum therapy dose.

Therapy can be effortful for patients who can sometimes be too fatigued when the therapist comes to do their session. This digital neuro-intervention will provide optimum availability of therapy and therefore place the patient in control of how much they do. However, the therapy will be engaging through adaptive and exciting software. The adaptiveness will serve to reward the patients for their efforts without ever making them feel like they are failing. This will hopefully boost how long they spend in therapy and therefore make greater gains.

There are therapy apps available for patients but the scientific basis for them is lacking and none are targeted for proper name anomia. By using speech recognition software our intervention will be uniquely designed to be used by patients with word retrieval impairments.

What new information will the research provide? The concept of the therapy component is based on standard SALT practice and thus has a strong proof of concept basis, as do our proposed outcome measures. The novel component is packaging it in a user- friendly web-app that is designed for and by people with word-retrieval problems. The investigators will test the efficacy in a randomized-controlled trial (Phase 2). These digital neuro-interventions will be in contrast to the plethora of apps available online claiming cognitive training without having scientific proof.

The Gotcha! research study is a small-scale, randomised clinical trial for participants with mild to moderate dementia who have difficulty naming the people they know well (proper name anomia). The main research question is whether Gotcha! therapy improves participants' ability to freely name the people they know well. will test this hypothesis by assessing participants' naming abilities on two matched lists of people, half are trained and half remain untrained.

A secondary question relates to maintenance of therapy gains, to test this hypothesis two different versions of Gotcha! are compared, 'standard Gotcha!' and 'Gotcha! Maintenance'. These two versions are identical for the first block of therapy but differ thereafter in that the maintenance version continues to top up participants with brief training sessions over the next block while those in the standard group do not have any further interactions with the therapy. Participants will be randomised (using minimisation) to one of the two versions of the therapy. The main outcome measure will be a clinically relevant improvement on the naming ability of the trained items (compared to untrained control items). Secondary outcomes will include improvements in social activity and participation and participant and/or carer reported outcome measures (PROMS).

The investigators also plan to capture brain-based data (functional and structural MRI and MEG) in order to test a series of secondary hypotheses relating to both how and why the therapy may work in some participants but not others. All brain imaging will be outside of standard care for the participants and take place at UCL by trained staff.

Hypotheses and Objectives

H1-Primary:

Does Gotcha! improve naming performance on trained items?

This is a within-subject comparison. The criteria for success is a 10% (raw, or unstandardized) relative improvement in performance. This effect size was used in the recent Big Cactus study.

H2-Secondary:

Does Gotcha! maintenance keep participants at their maximum achieved level compared to standard Gotcha?

This is two factor analysis with both a within-subject factor (trained vs. untrained items) and a between-group factor (Gotcha! maintenance vs. standard Gotcha). The criteria for success is a 10% (raw, or unstandardized) relative difference in performance. This effect size was used in the recent Big Cactus study.

Primary Objective Does Gotcha! improve naming performance on trained items?

Secondary Objective(s)

The investigators aim to show that using this DNI will improve word retrieval on trained items by providing a known therapy used by SLT's through a digital app and by enabling an increased dose of therapy. However, the investigators also believe that by using this DNI and practicing every day, there will be changes in participants' brain structure and in the health of the participants and their carers.

• Does Gotcha! maintenance keep participants at their maximum achieved level compared to standard Gotcha?
• Are the therapy effects item-specific or do is there generalization to untrained items?
• Are the effects of the DNI therapy limited to the language domain alone or do they improve in other cognitive domains (e.g. sustained attention)?
• Does the DNI improve participant's social activity and participation?
• Is the DNI acceptable to participants (can they use it easily)?
• Is the DNI acceptable to participants' carers?
• Does the DNI improve carer health/well-being?
• Does the DNI have any economic benefits?
• Can baseline structural brain imaging explain participants' responses to the DNI?
• Can baseline structural brain imaging be used to predict incoming participants' response to the DNI?
• Can repeated-measures structural or functional brain imaging identify participant's brain areas that change in response to the DNI?