Retraining Gait in Parkinson's Disease Via a Personalised App

In the UK, falls cost the NHS >£2billion/year. Parkinson's disease (PD) is one of the most common and progressive neurological disorders, with prevalence projected to double in the next 30-years. PD increases fall risk through walking/gait disturbances, with gait variability related to an increased falls rate/risk. Approximately 60% of people with PD (PwPD) encounter many falls annually. To better understand underlying causes of falls, a gait assessment is undertaken to develop bespoke and targeted/personalised strategies to minimise fall risk.

Instrumenting fall risk Typically, a gait assessment is undertaken by visual inspection, where a physiotherapists tacit expertise/experience enables the identification of obvious and/or subtle gait disturbances. However, reliance on visual observation alone introduces inconsistency/discrepancies in designing fall reduction strategies. To overcome, digital technologies have been investigated e.g., instrumented walkways provide high-resolution gait data to inform retraining via auditory cueing (e.g., stepping to a metronome beat) but they are very costly and bulky to use in most PD services. Regardless, PwPD perceive metronome cueing as monotonous while musical alternatives lack personalisation. Accordingly, there is a need to develop and adopt affordable, pervasive and scalable approaches that are personalised and engaging.

Lab on a phone Technologies that are scalable and easily accessible could facilitate personalised fall reduction programs anywhere. Smartphones are ubiquitous to facilitate near real-time intervention. This study uses our novel app (CuePD), designed to provide cross-platform (i.e., iOS and Android) scalable and personalised auditory cueing for gait retraining to reduce fall risk. CuePD's gait and music algorithms are validated to assess and retrain gait in PwPD within a lab (87-99% accuracy). This project's aim is to deploy CuePD in the home/community, to determine its efficacy as an everyday gait retraining tool. The project is a feasibility randomized controlled trial (RCT) to inform a future grant involving a multicentre observer blind parallel group RCT.

Research questions and hypotheses

• Principal question: Is a personalised home/community gait retraining programme via CuePD feasible and more effective to reduce fall risk than advice about improving gait/walking (i.e., usual care) in people with Parkinson's disease (PwP)?

o Primary hypothesis - CuePD will be feasible and beneficial compared to advice about improving mobility (usual care) to reduce fall risk.

• Secondary questions: What is the cost-effectiveness and cost utility of CuePD compared with advice about improving mobility and usual care to reduce fall risk for PwP? What are participants' experiences of CuePD and which factors influence uptake of, and adherence to the programme? o Secondary hypothesis - CuePD will be cost-effective with positive experiences for encouragement of continued use.

Key goals

• Assess CuePD's feasibility and effectiveness to improve gait and reduce fall risk in the home.
• Understand how PwP use and value CuePD, to explore (via suggested guidance) how they incorporate it into daily life (home, community).
• Lay groundwork for larger RCT by collecting data and insights to support future scaling and seamless integration of CuePD into daily life.

Research objectives Accordingly, those objectives (overlapping with goals) will guide the trial structure, including recruitment, randomisation, protocols, and post-intervention analysis.

• The primary objective will assess CuePD's efficacy to improve gait characteristics (stride length, cadence, gait speed and stride time coefficient of variation, CoV) and thereby reduce fall risk.
• Secondary objectives are to (i) assess experience, usability factors and adherence to the CuePD intervention and (ii) determine cost-effectiveness and cost utility of CuePD.

Design This is a pragmatic, prospective, parallel observer blind, feasibility RCT to evaluate (i) gait retraining efficacy and (ii) cost-effectiveness, of CuePD versus advice (usual care) over a twelve/12-week period. It will have an allocation ratio of 1:1.

Sample size The sample size is based on a 0.01 significance level, 90% power, and an expected standardised response mean (SRM, a unitless value) of 1.0 for gait speed. Accordingly, 15-participants/group will be recruited (i.e., 30-participants total for a two-group comparison). However, to allow for a 25% attrition rate at 3-months, 40-participants will be recruited across both groups (i.e., 20/group) [1].

Randomisation + allocation concealment Participants will be allocated into control or treatment/CuePD using simple randomisation. Randomisation will be by a computer generated random number sequence and will be conducted by the researcher to support the successful blinding of the outcome observer (name1) for the intervention duration. Once allocation has occurred, the outcome observer (name1) will receive a blinded email confirming the participant number, and the outcome assessor (name2, assessing duties) will receive an un-blinded email confirming participant number and allocation from the researcher. To maintain observer blindness throughout the study period, participants will be requested not to discuss the intervention with the outcome observer.

Groups

• Control: Participants will receive usual care consisting of (co-designed by PwP) educational materials focused on maintaining good walking/gait. Materials will provide information on gait/walking factors contributing to falls, practical tips for enhancing gait, and guidance on maintaining mobility. Materials will be distributed to participants/PwP in digital and print formats to accommodate varying preferences, ensuring accessibility. Digital versions will be sent via email, enabling PwP to view them on electronic devices, while printed copies will be handed directly to participants at baseline (via an information pack). This dual-format distribution aims to enhance engagement with the content, ensuring PwP can readily access, and use, the information to manage their condition effectively and reduce fall risk. Materials will be sourced from public facing information via (i) NICE Guidelines [2], (ii) Parkinson's UK [3], (iii) Parkinson's Foundation [4], (iv) Michael J. Fox Foundation [5] and (v) NHS [6]. Participants will receive weekly calls to check adherence and safety.
• CuePD: The intervention arm (using CuePD) to deliver personalised musical cueing to retrain gait and reduce fall risk in PwP. CuePD leverages auditory stimulation via preferred music (i.e., PwP chooses their favourite songs/music) that is selected to match a new/required tempo (i.e., walking cadence) based on PwP baseline stepping cadence to improve gait outcomes such as stride length [7-9]. Cadence can be increased from +1% to +10% where a selection algorithm picks music based on its tempo (i.e., beats/minute matching equivalent cadence). Participants will be instructed how to use CuePD and given a waist attachment to wear their smartphone on the lower back, based on open and clinically validated algorithms to interpret inertial data at that body location) and wireless headphones for integration into daily routines, Figure 2. Specifically, the participant will be asked to assess their gait twice/week during the 12 weeks by placing the smartphone on their lower back - CuePD will record their gait via validated algorithms. When the participant wants to carry their smartphone as normal (e.g., pocket), they will still be able to listen to their personalised cued music. Participants will receive weekly calls to check adherence and safety but the CuePD will also be able to objectively track adherence (i.e., frequency and duration of use of the app).

Delivery At baseline, participants will be given an onboarding session in person, where they will learn how to use CuePD and discuss how they can integrate use into their daily lives. The session will tailor musical cues to each participant's baseline walking cadence/tempo, create musical playlists and adaptive adjustments (see below) to support gait retraining. Support will be available throughout the 12-week period via a dedicated helpline and email (to the researcher). CuePD integrates with Apple Music so participants will be subscribed to that music streaming service and costs will be reimbursed as well as data streaming costs.

Additional support will include weekly check-ins with a researcher or nurse via telephone, to monitor progress, address concerns/safety, and adjust the intervention as needed e.g., increasing tempo (see below). Weekly phone calls will include open-ended questions, which will be audio-recorded (with informed consent) to explore user experiences in greater depth Participants will also receive an instructional manual and video tutorial. This multimodal approach ensures that participants have the necessary guidance/support to effectively use CuePD and achieve the intervention goals.

• CuePD treatment, development of intervention: At baseline, participants will be encouraged to use CuePD for up to 30mins, 3 times a week which aligns with another successful Parkinson's cueing approach (equating to a total of 36 sessions over 12-weeks) [10]. However, a co-design approach has been implemented. whereby our PwP focus group proposed that:

• Participants be asked to use CuePD as often as they can each week and during any time when they undertake community-based walks
• If during the 12-weeks the PwP would like to alter the +10% [11] from their baseline cadence (e.g., to +7%) then the researcher will enable a new tempo and new music selection. That request for dynamic functionality (faster or slower), stems from a PwP who commented during our focus group: "some days might be better than others and so I'd like the option of going faster". The dynamic function also aligns with a finding from the NIHR-INCLUDE project [12]. However, all adjustments will be discussed with the participant and wider research team to ensure PwP safety.
• CuePD adherence (e.g., wear time) and exposure time will be monitored via app functionality (like screen time usage used to monitor how often users engage with apps).

Mode of delivery Control and CuePD arms will be delivered by two providers (researcher or nurse). To attention match, control and CuePD groups will be contacted face-to-face (baseline) and subsequently by telephone once/week to check progression and safety. Both groups will keep a falls diary during the entire duration of the project i.e., 12+12-weeks/3-months. All assessments will be conducted in the home.

• Control group sessions will be undertaken independently (by participant/PwP reading educational material) with telephone support as necessary to encourage engagement with materials (maximum 11-phone calls over the 12-week period).
• CuePD group will have the same frequency of telephone calls to encourage use of the app and to make any technical adjustments e.g., changing tempo from +10% baseline cadence to a lower number.

Baseline assessment

The researcher/nurse will perform a baseline assessment following consent to participation. For characterisation and to inform treatment development, the following validated tools and data will be collected:

• Montreal Cognitive Assessment (MoCA), evaluates cognitive function (exclude those with impairment, score <21). MoCA is as a standardised neuropsychological test and is efficient for the rapid screening of global cognitive dysfunction. It examines diverse cognitive domains e.g., executive functions.
• Demographic information: Age, sex.
• Unified Parkinson's Disease Rating Scale (UPDRS), evaluates motor and non-motor symptoms, consisting of 4-sections, each targeting a specific area of evaluation (i) non-motor experiences that occur during daily living, (ii) motor experiences during daily living (iii) motor function examination and (iv) complications related to therapy.
• Hoehn and Yahr (H&Y) scale to stage disease severity.
• Falls history: Falls incidence, cause, location and consequences in the previous 12-months will be assessed by a questionnaire based on recommendations, that leads with the question "In the past year, have you had any fall including a slip or trip in which you lost your balance and landed on the floor or ground or lower level?".
• Falls Efficacy Scale-International (FES-I): Measure fear of falling during both basic and demanding activities (physical and social) using a scale from 1 (not at all concerned) to 4 (very concerned) across 16-scenarios.
• Physical activity: To concisely capture levels across a range of habitual tasks, 2-questionnaires will be used: (i) a physical activity questionnaire for the elderly with a reference period of 1-year to understand functional household and community-based activities and (ii) the International Physical Activity Questionnaire Short Form (IPAQ-SF), to assesses types of activity intensity and sitting time.
• Supervised (researcher/nurse) instrumented gait assessment via the participants own smartphone with CuePD installed. An initial gait assessment will determine baseline cadence which will define music tempo that are listened to during walks. Participants will be given (i) headphones and (ii) a belt attachment to position the smartphone on the lower back. The participant will keep the headphones and belt attachment for the duration of the intervention (12-weeks) and retain both after the study ends. CuePD will capture validated spatiotemporal gait characteristics (e.g., step time) but also the study outcomes to assess fall risk i.e., gait speed, cadence, stride length and step time coefficient of variation. Turns will also be quantified from angular velocity which can be captured by CuePD.
• To evaluate the cost-effectiveness and cost-utility of CuePD versus usual care, both groups will complete the Client Services Receipt Inventory (CSRI) and the EQ-5D:

• CSRI collects data on service use and has been used in many evaluations of health and social care interventions over the past 30-years. It records the interviewee's use of health and social care services, accommodation and living situation, income, employment and benefits. Service use data are combined with appropriate unit costs to generate total care costs/participant.
• The EQ-5D is a brief scale to measure health related quality of life (HRQoL) across 5-dimensions: mobility, self-care, usual activities, pain/discomfort and anxiety/depression. For each domain participants are asked to rate their current health using a five-point scale, describing the extent to which they can perform activities ("No problems" to "Unable to perform"). It also includes a visual analogue scale on which participants mark their current status between 0 (worst health they can imagine) and 100 (best health they can imagine). Quality Adjusted Life Years (QALYs) will be derived from the EQ-5D scores for the cost-utility analysis.
• Quality of life (QoL) + Health Realted QoL (HRQoL). In addition to EQ-5D data that will also be used to gauge QoL, HRQoL will be captured via the Centres for Disease Control and Prevention's (CDC) 14-item HRQoL "Healthy Days Measure" survey to allow participants rate their overall health status and the influence of various symptoms on daily activities.

Post-intervention assessments: 12-weeks

The following will be gathered:

• Goldsmiths Musical Sophistication Index (Gold-MSI): A self-report scale designed to assess musicality across five subscales. This tool evaluates and provides the detailed extent and nature of musical experiences and skills, but also provides a broader score on musicality, General Music Sophistication (GMS).

o The scale will be used to assess if there was a relationship between an individual's general music sophistication and their adherence to the beat of the music cue (subscale results will also be assessed).

• System Usability Scale (SUS): Evaluates perceptions of CuePD's usability, complexity, and overall satisfaction. (Higher SUS scores reflecting superior usability and used in our previous lab examination of CuePD.)
• Falls history (over the 12-week period) / falls diary.
• FES-I
• Physical activity.
• Supervised instrumented gait assessment (i.e., with researcher)
• CSRI + EQ-5D.
• QoL + HRQoL.

Outcome assessments

1-month (12+4-weeks), 3-months (12+12-weeks) Falls history (falls diary), FES-I, physical activity, supervised instrumented gait, CSRI + HQ-5D, QoL + HRQoL will be repeated (one month and three months follow-up after the intervention period), to assess retention of post-treatment effects.

Each outcome variable will be explicitly defined by type and distribution to inform the choice of statistical model. Continuous outcomes (e.g., gait speed, QoL scores) will be analysed using linear mixed-effects models. Count outcomes (falls) will be analysed using negative binomial mixed models. Ordinal outcomes (e.g., FES-I where appropriate) will be analysed using ordinal logistic regression. Highly skewed variables (e.g., healthcare costs) will be analysed using generalized linear models with appropriate distributions (e.g., gamma with log link). All models will account for repeated measures over time.

Safety and study withdrawal CuePD has been used without concerns during all lab testing and deemed safe as music is familiar and headphones allow awareness of surroundings. Bone Conducting Headphones will also be offered as an alternative option to in-ear and over-ear headphones. Moreover, participants will be contacted once/week for continuous safety oversight. Regardless, intervention safety will be evaluated by examining the occurrence of any adverse events and falls. No specific study withdrawal criteria have been set but participants can withdraw at any time, not affecting their normal care. Should a PwP withdraw, a reason for withdrawal will be sought but participants can choose to withdraw without a reason.

Data collected prior to withdrawal will be used in the study analysis unless consent for this is specifically withdrawn. If a participant does not wish to continue treatments, they will be asked if they are willing to continue to attend outcome assessments. Additionally, clinical teams, local treatment providers or investigators may withdraw PwP from the study at any time if they feel it is no longer in the participant's interest to continue e.g., because of intercurrent illness [13].

Recording, reporting adverse events This study will only report adverse events which are serious and exclude (i) pre-planned hospitalisations and (ii) scheduled treatment for pre-existing conditions. The capture of potential serious adverse events (SAEs) will take place at the study outcome assessments by including the following questions in the outcome proforma: "Are there any new medical problems since the last study assessment?" [13].

Additionally, the investigators will specifically enquire about falls during weekly telephone consultations. As members of the research team become aware of SAEs at times other than at outcome assessment appointments, the SAE form will also be used to directly capture these events. All SAEs regardless of causality or expectedness will be reported to the chief investigator (name) and co-investigator (name2) at the trial sponsor (Northumbria University) in line with local policies with onward discussion to clinical-based partners (name3, name4).

4.11.1 Data monitoring and ethics committee All SAEs (related and unexpected) will be collated and reported to a Data Management and Ethics Committee (DMEC) within 24-48 hours of the chief investigator (name1) becoming aware of the event. The DMEC comprises experts from related fields at independent institutions. The DMEC will review SAE's and liaise with the chief investigator as necessary (up to every 6-months).

The DMEC can recommend premature closure of the trial and can unblind the data if required. The DMEC, throughout the trial, will take responsibility for:

• Monitoring data and making recommendations to the TSC as to whether there are any ethical or safety reasons why the trial should not continue.
• Monitoring patient safety data (SAEs).
• Providing advice to the chief investigator, PPIE Advisory Group, funder, or sponsor, as appropriate.

Post intervention focus group Participants (control and intervention) will be invited to focus groups to expand findings. Consenting views of al CuePD intervention and control participants will be sought in a series of focus groups. Focus groups will be separate to the main study and undertaken after the 12-week outcome assessments. Questions (e.g., did a carer encourage use?) will explore participant experiences and factors influencing adherence. Focus groups will be voice recorded and transcripts will be analysed to identify themes and connections.

Methods of analysis

Efficacy of intervention/CuePD:

Intention to treat analysis will be our primary model for efficacy analysis. The investigators will conduct a linear regression of change from baseline to 12-week follow-up (see SAP, Table 1), adding control and CuePD arms. A secondary analysis via SRM and FES-I/diaries will examine changes between baseline and end of the 12-week intervention as well as 1- and 3-month follow-ups. If there are some missing data at follow-up the investigators can attempt to extrapolate based on the data from participants with complete datasets if there are less than 25% missing data. Additionally, focus groups will gather insights from the lived experiences of CuePD.

Cost-effectiveness + cost-utility analyses:

CSRI generated costs will be added to costs of the intervention and will be derived from information on staff time involved in delivering the intervention and other non-staff costs. The endpoint for the economic analysis will be 12-weeks when, if costs are lower in the CuePD group and outcome better, then the CuePD intervention will have been cost-effective. In contrast, if costs are higher and outcomes better then the investigators will construct cost-effectiveness acceptability curves to show the probability that CuePD is cost-effective for different values placed on a change in outcome. Cost-utility analyses will be conducted in a similar way (using QALYs as the outcome measure). Uncertainty around the results will be explored using cost-effectiveness planes and acceptability curves.

Weekly phone calls and focus group(s) A thematic analysis will analyse phone calls and the focus group data as it is a flexible method that allows themes to emerge from the data and has previously been used by the team to analyse data following discussions relating to the technology to assess gait and reduce fall risk within PD. The thematic approach can produce insights via the participants voice to help answer the research questions. Moreover, a thematic analysis is one of the most used and routine methods of analysing qualitative data by being less time-consuming and with a flexible approach.

Closing study The study will be brought to a close once all participants have completed the 24 week follow up of the study or at a date no later than 31-12/Dec-2027.

Inclusion criteria

• Able to walk unaided.
• Diagnosis of idiopathic Parkinson's.
• Score ≥21/30 on Montreal Cognitive Assessment (MoCA) which is used to classify non-demented Parkinson's (Parkinson's dementia is <21/30).
• Uses a smartphone.

Exclusion criteria

• Non-English speakers
• Use of any mobility aids e.g., walking stick
• History of stroke, traumatic brain injury or other neurological disorders (other than Parkinson's)
• Unable to comply with the testing protocol or currently participating in another interfering research project.
• Does not use a smartphone.
• Body mass index ≥35 (i.e., severe to morbid obesity)