NeuroMusic: protocol for a randomised-controlled trial of keyboard and singing music training programmes for older adults with mild cognitive impairment Authors: Zoe Menczel Schrire, Helen Frances Mitchell, Lee-Fay Low, Nicole Espinosa, Philip Eames, Joseph Toltz, Peter Walsh, Loren Mowszowski, David Espinoza, Cindy Shin-Yi Lin, Neal Peres Da Costa, Sharon L Naismith Published: 2025 ABSTRACT AbstractIntroductionMusic-based training programmes, such as learning how to play an instrument or sing in a choir, have been suggested as potential interventions for promoting healthy brain ageing in older adults at risk of cognitive decline because of their ability to enhance cognitive functions and potentially promote neuroplasticity. However, there is limited empirical evidence in older adults at risk of dementia, especially that evaluates both piano and singing interventions and their effects on cognition and neuroplasticity. In this protocol, we outline a study to assess the efficacy of keyboard and singing music training programmes on reducing cognitive decline and other outcomes in older adults with Mild Cognitive Impairment (MCI).Methods and analysisThis randomised, single-blind, controlled, parallel-group trial aims to enrol 432 individuals with MCI from the community in Sydney, Australia. Participants are randomly allocated to participate in either keyboard lessons, singing lessons or a film discussion control group once a week for 3 months. The primary objective is to assess the effectiveness of two music training programmes (keyboard and choral singing) for enhancing verbal memory after 3 months compared with control. Additionally, we will examine how these music-based interventions affect other aspects of cognition, mood, sleep, overall well-being, markers of brain plasticity and blood biomarkers of Alzheimer’s disease and neurodegeneration. Tertiary objectives are to identify factors that impact the success of the interventions, such as participation rates, engagement levels and key demographic and clinical features. Outcomes are collected at baseline and at 3 and 9 months. The primary endpoint analysis will include all randomised participants to estimate the treatment effect using intention-to-treat principles. Primary and secondary outcomes will be analysed using linear mixed models and effect size measures will be calculated.This study will be the first robust, randomised controlled trial to assess the potential and relative value of music engagement for cognitive decline in high-risk MCI individuals, as well as broader effects on other markers of mental health, well-being and neurodegeneration. Co-designed with implementation in mind, the music interventions can potentially be delivered within memory clinic or community settings.Ethics and disseminationThe Sydney University Human Research Ethics Committee (2023-026) has approved this protocol. The trial findings will be shared through conferences, publications and media.Trial registration numberAustralian and New Zealand Clinical Trials Registry (ACTRN12623000407695), Registered 21/04/2023 https://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=385552Protocol version2.02 29/11/2024. STRENGTHS AND LIMITATIONS OF THE STUDY The randomised controlled design enhances the reliability and validity of the findings by minimising selection bias. The randomised controlled design enhances the reliability and validity of the findings by minimising selection bias. The study assesses a wide range of outcomes, including cognition, neurobiological and blood biomarkers of dementia, mood and sleep, providing a holistic view of the interventions’ impacts. The study assesses a wide range of outcomes, including cognition, neurobiological and blood biomarkers of dementia, mood and sleep, providing a holistic view of the interventions’ impacts. The single-blind design may introduce bias, as participants are aware of their group allocation, potentially influencing their behaviour and self-reported outcomes; however, this is unavoidable. The single-blind design may introduce bias, as participants are aware of their group allocation, potentially influencing their behaviour and self-reported outcomes; however, this is unavoidable. The ability of the study to generalise to different settings and to diverse cultural, indigenous and socioeconomic groups may be hindered by the inner city location whereby this trial was conducted (ie, requiring travel into the city weekly for 3 months). The ability of the study to generalise to different settings and to diverse cultural, indigenous and socioeconomic groups may be hindered by the inner city location whereby this trial was conducted (ie, requiring travel into the city weekly for 3 months). During the 9 month follow-up period, we do not have the capacity to collect accurate adherence or engagement data. However, focus groups will explore key aspects of participants’ ongoing music engagement and will be reported separately. During the 9 month follow-up period, we do not have the capacity to collect accurate adherence or engagement data. However, focus groups will explore key aspects of participants’ ongoing music engagement and will be reported separately. Introduction Background and rationale Dementia is a neurodegenerative condition that affects over 50 million people globally, and Alzheimer’s disease (AD) is the most prevalent form, whereby memory loss is a key early symptom.1 Research has not yet led to a definitive cure for AD and related dementias,2 warranting efforts to focus on primary and secondary dementia prevention. Indeed, in AD, amyloid-beta accumulates in the brain 10 to 20 years before cognitive changes are evident.3 4 Once cognitive decline is apparent, up to 45% of individuals will convert to dementia within 5 years (see review by 5), presenting a critical window of opportunity for targeted secondary prevention strategies. Various terms have been used to describe this ‘at-risk period’, such as mild cognitive impairment (MCI) (see review 5) or age-associated memory impairment.6 Each term may use different thresholds, but all require evidence of cognitive decline, with the relative preservation of functional abilities. During these critical risk periods, non-pharmacological interventions have been touted as safe, efficacious and economical options to improve cognition, particularly in older adults.7 Both physical exercise and cognitive interventions (eg, memory strategy or computerised training) are recommended by the American Academy of Neurology and by Australian specialists.8 Cognitive interventions have been shown to be efficacious with associated medium effect size improvements in memory and other cognitive domains.9 10 As little as 10 cognitive training sessions have enhanced specific cognitive abilities, with effects persisting for 5 years after training.11 Cognitive interventions are cost-effective, are safe,7 have good adherence9 12 13 and promote independent living.14 They also foster neuroplasticity (the ability of the brain to reorganise or grow neural networks in response to an experience)15—enhancing metabolism, blood flow,16 brain structure and function17 18 after 12 weeks. Importantly, the effect estimate of cognitive training is modestly higher in MCI compared with healthy older adults and those with dementia, showing that this ‘at-risk’ period is the optimal time for intervention.19 However, such interventions may not appeal to all and often require specialist input, software or technology, which restricts scalability and accessibility. Although there is no definitive evidence that cognitive training can prevent dementia, further studies incorporating neuroimaging and dementia biomarkers may provide further mechanistic insights. A 3 month cognitive training programme improved neural activity, connectivity, blood flow and cognition in healthy older adults.16 Another study showed that multimodal cognitive training reduced resting-state functional connectivity between the posterior cingulate and superior frontal gyrus, correlating with improved cognition.20 A systematic review of older adults suggested that multi-domain cognitive training could enhance hippocampal functional connectivity.21 Additionally, a 6-month multicomponent cognitive training programme improved cognition and reduced glucose metabolism decline in cortical regions affected by AD in individuals with amnestic MCI.22 These findings suggest that MRI markers are likely to be important markers in assessing the efficacy of potential cognitive interventions. Building on the cognitive training literature and the general notion that brain engagement is beneficial for cognition in later life, there has also been interest in music engagement as a possible intervention. Music making is a cognitive intervention that may offer the requisite cognitive challenges23 24 while being more appealing and accessible, as some, such as choir, do not require specialist equipment, technologies or highly trained staff. Music interventions are scalable across communities and adaptable for diverse ethnic and socioeconomic groups. Music making can also be intrinsically motivating, enjoyable and potentially more rewarding than engaging in cognitive training, due to a heightened sense of achievement, motivation, socialisation, variation and activation of pleasure centres.2527 Additionally, research has shown that older adults experience positive health outcomes when they feel a sense of control or mastery28 or are in situations that provide meaningful social engagement with others.29 30 Animal studies have demonstrated that the activation of aminergic reward neurotransmitters can influence learning and plasticity.31 Music can stimulate similar pathways because of the emotions it evokes.32 Choir singing can be considered a viable intervention that combines music, language and breathing management. Singing involves left-brain hemisphere lateralisation for language perception and production, and right-hemisphere activity for the perception and production of singing tasks.3335 Studies have indicated that singing can improve response times to musical nuances36 and improve vocal function.37 Importantly, singing has been shown to enhance overall quality of life38 and psychological well-being.3942 In a self-selected sample, singing has been demonstrated to improve self-reported physical health, decrease doctor visits, lower medication use, decrease instances of falls and reduce the occurrence of other health problems.43 Community choirs have also been shown to reduce loneliness and increase interest in life among diverse older adults.44 Furthermore, singing has been shown to improve speech perception in noise,45 46 which is particularly beneficial as hearing loss is a risk factor for dementia.47 Qualitative research indicates that singing may enhance cognitive function by providing increased stimulation, opportunities for reminiscing and learning experiences.42 48 One pre-post quasi-experiment showed improvements in memory, language, speech information processing and executive function after 12 weeks of group singing.49 Another cross-sectional study showed that older adults engaged in choir singing performed better in verbal flexibility, the domain of cognitive function that focuses on one’s ability to adapt their communication style.50 In summary, choir singing emerges as a multifaceted intervention that may confer significant benefits for cognitive, mental and physical health, particularly among older adults. Playing a keyboard has also been suggested to enhance cognitive skills. For example, the interplay between motor movements and cognitive tasks can improve cognitive flexibility.51 Additionally, learning to read sheet music requires repetitive use of cognitive resources, which can also improve cognition.52 53 More recently, there has been some research investigating the effects of piano training in individuals with MCI; however, the range of neuropsychological outcomes explored was limited.54 Based on a cross-sectional study, it has been suggested that playing a musical instrument may have a greater impact on learning, working memory and processing speed than singing.55 Older adults with prior music experience have been shown to have better executive functioning compared with those without.56 Interestingly, preliminary evidence suggests that learning how to play the piano later in life may promote structural and functional changes in the brain.57 Previous research on non-musicians aged 24–38 years old has shown that learning a musical sequence on the piano through sensorimotor-auditory training can induce neuroplastic reorganisation in the auditory cortex.58 In healthy older adults, four studies have been conducted to investigate the effects of music-making interventions on neurobiological measures, but none have focused on individuals with MCI.59 Among these, only two studies showed a positive effect of piano training intervention on brain markers, showing increased frontal theta power60 and grey matter volume in the dorsolateral prefrontal cortex and cerebellum61 in the piano group compared with the control group after the intervention. However, since the study by West et al61 had a small sample size (n=20) and high attrition rates, the findings should be interpreted with caution. Finally, two studies of 155 older adults from the same participant pool found a stabilisation effect in response to piano training. The control group showed a decline in fibre density in the fornix62 and grey matter volume in a section of the primary auditory cortex.63 Overall, while these findings suggest that while there is promising evidence for the cognitive and neurobiological benefits of piano training, more research is needed to fully understand its impact on the ageing brain, specifically within people with MCI. Convergent, yet limited quality data are supportive of music interventions for the prevention of cognitive decline. While many studies have examined music listening for behavioural symptoms in dementia, few have researched music making or learning in MCI. When examining the efficacy of active music interventions combined with physical activity, one randomised controlled trial in MCI found that both playing musical instruments and dance improved general cognitive function but only dance significantly improved memory recall compared with a health education control group.64 Another ongoing trial in MCI is comparing music practice and psychomotor training, but results are not yet available (NCT04546451).65 A recent systematic review of 19 studies involving 1024 older adults with MCI or dementia was conducted to investigate the effects of music-based interventions including singing, percussion and music listening on cognitive subdomains.66 The study found significant improvements in global cognition (as assessed by the Mini Mental State Examination), executive function (as assessed by the Frontal Assessment Battery) and episodic memory (assessed by the Rey Auditory Verbal Learning Test) for those who participated in music-based interventions. Notably, however, of the three studies that included participants with MCI, two involved music listening or music with movement, while only one small study (n=26) investigated karaoke training.67 These existing studies identified beneficial effects on memory, an important outcome that is intricately linked to early hippocampal compromise68 and changes in functional brain connectivity.69 Additionally, in older adults at risk of dementia, currently playing a musical instrument is associated with improved markers of brain health, in comparison to those who have ceased or never played an instrument.70 In summary, while music-making and choral interventions may benefit mood, quality of life, socialisation and physical health in healthy older adults,64 71 72 only a few studies with limited sample sizes and less common music interventions have focused on participants with MCI. Targeting memory decline in people with MCI addresses a key early, significant feature of MCI that prompts health seeking within primary care, as well as more detailed assessment and intervention within the memory clinic or specialist setting. Moreover, the effects of music interventions on biomarkers of AD pathology or neurodegeneration are unknown. This trial was designed to address these extant gaps. Objectives Here, we report the protocol of the NeuroMusic trial, which will boost the evidence base for the efficacy of music engagement interventions for people at-risk for dementia. The NeuroMusic trial will specifically examine the efficacy of keyboard and choral singing interventions developed via co-design and informed by an implementation science framework. This led us to examine two music interventions that feature frequently used, practical media.73 Specifically, in older people with MCI, we aim to evaluate whether, compared with an active control condition, music interventions (keyboard and choral singing) are associated with: Less decline in memory (primary outcome) at 3 months and 9 months. Less decline in memory (primary outcome) at 3 months and 9 months. Improved aspects of broader health and well-being including mood, sleep and global cognition at 3 months and 9 months. Improved aspects of broader health and well-being including mood, sleep and global cognition at 3 months and 9 months. Less accumulation of neurodegenerative pathology in blood and imaging biomarkers at 9 months. Less accumulation of neurodegenerative pathology in blood and imaging biomarkers at 9 months. We hypothesise that in comparison to the control condition, both music-based training programmes will be associated with less decline in verbal episodic memory in older adults with MCI. Further, we hypothesise that in comparison to the control condition, both music-based training programmes will be associated with less accumulation of neurodegenerative pathologyin blood and imaging biomarkers and associated with improved mood, sleep, quality of life, social connectedness and global cognition. Trial design The current Randomised Controlled Trial (RCT) began recruitment in May 2023 and will continue until the end of 2027. This study is a parallel, 3 month, single-blind, three-armed RCT in which 432 community-dwelling, musically naive older Australians (60–90 years), including permanent residents and those living temporarily in Australia, with objective memory difficulties (>1 SD below normative scores), are randomised (1:1:1) to either the keyboard intervention, choral singing intervention, or an active control condition (ie, film-viewing and discussion sessions). The trial design is summarised in table 1. Primary and secondary outcomes are measured at 3 months. There is also a 9 month long-term follow-up, captured 6 months after intervention completion. Individuals are encouraged not to participate in music-based learning outside their assigned medium before the final follow-up. For example, participants in the choral singing intervention may join a choir after completing the study intervention but should not start taking keyboard lessons. This protocol is written following the SPIRIT Recommendations for Interventional Trials statement74 (online supplemental file 2) and the WHO Trial Registration Dataset (online supplemental file 3). Table 1Neuropsychological batteryNeuropsychological testCognitive domain measuredCognitive compositeMontreal Cognitive Assessment audiovisual (MoCA audiovisual, DeYoung110)Note: The letter ‘P’ is used in the place of ‘F’ for fluency task to mitigate interference with the administration of COWAT, see belowCognitive function and declineThis test is only administered at baseline to characterise the cognitive functioning of the sampleWechsler Test of Adult Reading (WTAR, Wechsler111)Estimated premorbid IQThis test is only administered at baseline to determine premorbid functioning.Rey Auditory Verbal Learning Test (RAVLT, Rey75)*Learning and memoryPACC, verbal learning composite (trials 1–5 total score), verbal memory composite (trial 7)Wechsler Memory Scale, third edition, Logical Memory subtest (Logical Memory, Wechsler76)MemoryPACC, verbal learning composite (subtest I); verbal memory composite (subtest II)Controlled Oral Word Association Test (COWAT, Benton, Hamsher112)Phonemic and semantic fluencyPACC, executive functioning composite (letters F, A, S), language composite (animal naming)Wechsler Adult Intelligence Scale Third Edition Digit Span (Digit Span, Wechsler113)Working memoryPACCSymbol Digit Modalities Test, oral version (SDMT, Smith114)Processing speedPACCOral Trail-Making Test (OTMT, Ricker and Axelrod115)Processing speed, executive functioningPACC, processing speed composite (part A), executive functioning (part B)Delis-Kaplan Executive Function System (DKEFS) Colour Word Interference Test (DKEFS CWIT, Delis, Kaplan116)Processing speed, executive functioningProcessing speed composite (conditions 1 and 2), executive functioning composite (conditions 3 and 4)15-item Boston Naming Test (BNT, Kaplan, Goodglass117)LanguageLanguage composite*Primary OutcomePACC, Preclinical Alzheimer’s Disease Global Cognitive Composite. Neuropsychological battery Primary Outcome PACC, Preclinical Alzheimer’s Disease Global Cognitive Composite. Methods: participants, interventions and outcomes Patient and public involvement Consumer focus groups informed the design of the music interventions. The focus groups canvassed ideas regarding preferred types of music and music making, desirability and feasibility of a music making intervention, preferred session type and length and probability of long-term engagement with our target population. The final ideas were tested with older people with cognitive impairment in a further focus group and refined as necessary, resulting in finalised interventions for the RCT. The RCT was then presented to the Sydney Dementia Network Lived Experience Expert Advisory Panel, consisting of people with MCI or dementia or caregivers, which guided recruitment strategies and evaluated impact from a real-world perspective. Study setting Participants are being recruited in ‘waves’ comprising a maximum of up to 36 participants. Each wave consists of a maximum of 12 participants per group (depending on the randomisation sequence and speed of recruitment), and all participants in each wave begin and end the intervention on the same dates. Eligibility criteria Inclusion criteria Adults aged 60-90 years. Adults aged 60-90 years. MCI defined by >1 SD below the age-normed average score on any memory test performed at screening, including:The Rey Auditory Verbal Learning Test (List A: sum of trials 1-5, trial 7 (after 20-minute delay), trial 7/trial 5 or trial 7/trial 6 (List B))75Logical Memory I or II subtests of the Wechsler Memory Scale (3rd Edition).76A standardised z-score is computed for each test score using published normative data.76 77 We define MCI as >1SD below predicted levels relative to the mean, unless the participant was considered to have higher than average premorbid functioning. Two neuropsychologists judged premorbid functioning by consensus for the first 50 cases; after this, one neuropsychologist has been deemed sufficient to make the decision, and consensus will only be sought as necessary. Note, this cut-score is not as stringent as the −1.5 SD threshold often applied in MCI, to balance feasible recruitment against a clinically enriched population. MCI defined by >1 SD below the age-normed average score on any memory test performed at screening, including: The Rey Auditory Verbal Learning Test (List A: sum of trials 1-5, trial 7 (after 20-minute delay), trial 7/trial 5 or trial 7/trial 6 (List B))75 The Rey Auditory Verbal Learning Test (List A: sum of trials 1-5, trial 7 (after 20-minute delay), trial 7/trial 5 or trial 7/trial 6 (List B))75 Logical Memory I or II subtests of the Wechsler Memory Scale (3rd Edition).76 Logical Memory I or II subtests of the Wechsler Memory Scale (3rd Edition).76 A standardised z-score is computed for each test score using published normative data.76 77 We define MCI as >1SD below predicted levels relative to the mean, unless the participant was considered to have higher than average premorbid functioning. Two neuropsychologists judged premorbid functioning by consensus for the first 50 cases; after this, one neuropsychologist has been deemed sufficient to make the decision, and consensus will only be sought as necessary. Note, this cut-score is not as stringent as the −1.5 SD threshold often applied in MCI, to balance feasible recruitment against a clinically enriched population. Able to provide informed written consent. Able to provide informed written consent. Fluent English. Fluent English. Appropriateness to participate and engage in a group setting, based on clinical judgement during the screening process. Appropriateness to participate and engage in a group setting, based on clinical judgement during the screening process. Willingness and ability to travel to study locations for interventions. Willingness and ability to travel to study locations for interventions. Exclusion criteria Participants are excluded if they meet the following criteria: A previous diagnosis of dementia. Participants who are later diagnosed with dementia during the trial by external sources remain eligible. A previous diagnosis of dementia. Participants who are later diagnosed with dementia during the trial by external sources remain eligible. More significant cognitive decline, as indicated with a score on the Montreal Cognitive Assessment-Audiovisual78 of <18, a score on the Modified Telephone Interview for Cognitive Status (TICS-M) of <22 79 or as assessed by a clinician during the screening process. More significant cognitive decline, as indicated with a score on the Montreal Cognitive Assessment-Audiovisual78 of <18, a score on the Modified Telephone Interview for Cognitive Status (TICS-M) of <22 79 or as assessed by a clinician during the screening process. Played a musical instrument or sung in an organised group in the last 6 months and/or had a history of employment as a professional musician. Played a musical instrument or sung in an organised group in the last 6 months and/or had a history of employment as a professional musician. Attended formal music lessons for more than 3 years. Attended formal music lessons for more than 3 years. Problems affecting hand dexterity that could affect their participation in the trial. Problems affecting hand dexterity that could affect their participation in the trial. Impaired and uncorrected visual or auditory accuracy that could affect trial participation. Impaired and uncorrected visual or auditory accuracy that could affect trial participation. History of major head injury (with loss of consciousness for ≥30 min) or cerebrovascular event (stroke, transient ischaemic attack). History of major head injury (with loss of consciousness for ≥30 min) or cerebrovascular event (stroke, transient ischaemic attack). History of neurological disorder (eg, Parkinson’s disease, multiple sclerosis, epilepsy). History of neurological disorder (eg, Parkinson’s disease, multiple sclerosis, epilepsy). Current illicit substance use or harmful alcohol intake (Alcohol Use Disorders Identification Test Consumption 80score >8). Current illicit substance use or harmful alcohol intake (Alcohol Use Disorders Identification Test Consumption 80score >8). Current severe major depression as defined by a score >20 on the Patient Health Questionnaire (PHQ-981) and/or suicidal ideation (score of >1 on Q9 of the PHQ-9) or history of severe psychiatric or developmental disorders (eg, schizophrenia, bipolar disorder, autism). Current severe major depression as defined by a score >20 on the Patient Health Questionnaire (PHQ-981) and/or suicidal ideation (score of >1 on Q9 of the PHQ-9) or history of severe psychiatric or developmental disorders (eg, schizophrenia, bipolar disorder, autism). Previously participated in any music training programme or currently enrolled in another interventional study. Previously participated in any music training programme or currently enrolled in another interventional study. Participated in a cognitive training intervention within 4 weeks of enrolment. Participants on self-assigned brain training (eg, online training) were allowed to continue if they did not engage in any new levels and maintained their levels of activity consistent with those before they began the trial. Participated in a cognitive training intervention within 4 weeks of enrolment. Participants on self-assigned brain training (eg, online training) were allowed to continue if they did not engage in any new levels and maintained their levels of activity consistent with those before they began the trial. Commenced antidepressant medication within the last 4 weeks or if they were experiencing side effects related to this medication. Commenced antidepressant medication within the last 4 weeks or if they were experiencing side effects related to this medication. Memory decline caused by external factors, such as new medications or surgeries. Memory decline caused by external factors, such as new medications or surgeries. Participants on chemotherapy are asked to wait for 6 months after cessation before screening. Recruitment Participants with memory complaints are recruited from the community using Facebook advertising, media (ie, news, podcasts, websites and radio), community events, flyers posted in community centres, letterbox drops and word of mouth. The trial is also being promoted to Memory and Cognition services in Sydney, Australia, and via the StepUp for Dementia Research online recruitment tool.82 Participants are directed to the study website where they can download the Participant Information Statement and provide self-consent for the online pre-screen. This determines if they meet the eligibility criteria and if so, prompts them to enter their contact information so that a member of the study team can call them. The research staff then contact provisionally eligible participants to explain the study’s details, including its objectives, benefits, risks and requirements, and verbally confirm that they do not meet any exclusion criteria. Participants are also administered the TICS-M as part of the pre-screening process. If a participant’s score falls below 22, they do not proceed to the screening stage. In this circumstance, the researcher provides additional resources to participants who express concern about their memory, including links and locations of memory clinics for further diagnostic investigations and management, outside the trial. Before enrolment, participants are encouraged to contact the study team if they have any questions or concerns. Study procedures are summarised in figure 1. Participants who have satisfied the aforementioned requirements are booked for a 90 minute online screening assessment. At this stage, participants are sequentially assigned a unique, three-digit participant screening number in ascending order. The results of screening tests are provided to ineligible participants on request. Study flow diagram. TICS-M, Modified Telephone Interview for Cognitive Status. Who will take informed consent? During screening with the neuropsychologist, participants are given ample time to discuss the study before electronically signing the consent form via REDCap. Data collection and management Plans for assessment and collection of outcomes After obtaining informed consent, the neuropsychologist administers standardised neuropsychological tests, which are then scored and used to inform eligibility. If participants are eligible, their screening assessment is used as their baseline assessment. Participants are excluded if they score too highly on memory testing (ie, they do not have memory impairment) or, in rare circumstances, are found to score in a range that might signify dementia, as determined by the clinician. If this occurs, the participants are provided with feedback and with recommendations for further investigation and are provided with options as appropriate. Note that if participants have been referred from a memory and cognition clinic and if neuropsychological assessment was performed within the prior 3 months, consent is sought to access their scores, and these tests are not repeated. Eligible participants proceed to complete baseline measures. The same measures are completed for a 3 month follow-up. Questionnaires, cognitive tests and actigraphy are administered at the 9 month follow-up. Blood collection occurs at baseline and at 9 month follow-up (figure 1). Magnetic Resonance Imaging A subset of participants (the first four waves enrolled) undergoes a 60 min 3T MRI brain scan at the Brain and Mind Centre (GE scanner, 32 channel coil). The MRI protocol includes the following sequences: (i) Sag 3D T1 BRAVO, (ii) Sag 3D T2 FLAIR, (iii) two T2 high-resolution hippocampus, (iv) Axe SWI, (v) Diffusion weighted imaging (DWI) and (vi) Resting state functional MRI (rsfMRI). The follow-up MRI is conducted at a 3 month follow-up without the Sag 3D T2 FLAIR and Axe SWI to last 45 min to reduce participant burden. These MRI sequences are collected to allow for the assessment of key brain measures such as hippocampal volume, prefrontal cortical thickness and white matter lesion volume, as well as the functional integrity of resting-state brain networks (rsfMRI) and the structural integrity of white matter tracts (DWI). Blood collection A fasting 30 mL blood sample is collected by a trained phlebotomist. Blood samples are processed within 30 min after collection, centrifuged and aliquoted into 500 µL (plasma and serum) and 1 mL (buffy coat) screw cap tubes and stored at −80°C for future batch analysis. Questionnaires Geriatric Depression Scale (GDS-15): The GDS is a 15-item self-report questionnaire developed to assess depressive symptoms and screen for depression among older people.83 Geriatric Depression Scale (GDS-15): The GDS is a 15-item self-report questionnaire developed to assess depressive symptoms and screen for depression among older people.83 Geriatric Anxiety Scale (GAS): The GAS consists of 30 items on a 4-point Likert scale (0=not at all, 1=sometimes, 2=most of the time, 3=all of the time) designed to assess typical common anxiety symptoms in older adults.84 Geriatric Anxiety Scale (GAS): The GAS consists of 30 items on a 4-point Likert scale (0=not at all, 1=sometimes, 2=most of the time, 3=all of the time) designed to assess typical common anxiety symptoms in older adults.84 British Columbia Cognitive Complaints Inventory: The British Columbia Cognitive Complaints Inventory is a 6-item screening tool that assesses perceived cognitive difficulties in the past 7 days. Each item uses a 4-point Likert scale to capture a rating (0=not at all; 3=very much) which adds up to broadly normal (0–4); mild cognitive complaints (5-8); moderate cognitive complaints (9-14); and severe cognitive complaints (15+).131685 British Columbia Cognitive Complaints Inventory: The British Columbia Cognitive Complaints Inventory is a 6-item screening tool that assesses perceived cognitive difficulties in the past 7 days. Each item uses a 4-point Likert scale to capture a rating (0=not at all; 3=very much) which adds up to broadly normal (0–4); mild cognitive complaints (5-8); moderate cognitive complaints (9-14); and severe cognitive complaints (15+).131685 EuroQol Five Dimensions Questionnaire (EQ-5D): The EQ5D is a brief questionnaire about health-related quality of life that queries five dimensions of everyday experience (mobility, self-care, usual activities, pain/discomfort, anxiety/depression) that are rated using a 5-point scale (no problems/slight problems/moderate problems/severe problems/extreme problems). The EQ-5D is a well-validated instrument that provides a single index of overall health status.86 EuroQol Five Dimensions Questionnaire (EQ-5D): The EQ5D is a brief questionnaire about health-related quality of life that queries five dimensions of everyday experience (mobility, self-care, usual activities, pain/discomfort, anxiety/depression) that are rated using a 5-point scale (no problems/slight problems/moderate problems/severe problems/extreme problems). The EQ-5D is a well-validated instrument that provides a single index of overall health status.86 Pittsburgh Sleep Quality Index (PSQI): The PSQI is a self-report questionnaire that measures sleep quality over a 1 month time interval. The measure consists of 19 individual items, creating seven components that produce one global score and takes 5–10 minutes to complete. The global PSQI, sleep efficiency index and sleep quality will be examined.87 Pittsburgh Sleep Quality Index (PSQI): The PSQI is a self-report questionnaire that measures sleep quality over a 1 month time interval. The measure consists of 19 individual items, creating seven components that produce one global score and takes 5–10 minutes to complete. The global PSQI, sleep efficiency index and sleep quality will be examined.87 Healthy Brain Ageing - Functional Assessment Questionnaire (HBA-FAQ): The HBA-FAQ is a self-report questionnaire that assesses subjective perception of functional change in daily life over the last 12 months. This measure consists of 44 items across 10 subdomains. Participants are asked to record their perception of functional changes using a 4-point Likert scale ranging from 0 (‘no change’) to 3 (‘unable/no longer completes independently’). The total HBA-FAQ score is calculated as the sum of averaged scores for each of the 10 subdomains (maximum total score=30), with higher scores reflecting more pronounced functional change.88 The questionnaire will be administered to both participants and a carer/informant of their choice. Healthy Brain Ageing - Functional Assessment Questionnaire (HBA-FAQ): The HBA-FAQ is a self-report questionnaire that assesses subjective perception of functional change in daily life over the last 12 months. This measure consists of 44 items across 10 subdomains. Participants are asked to record their perception of functional changes using a 4-point Likert scale ranging from 0 (‘no change’) to 3 (‘unable/no longer completes independently’). The total HBA-FAQ score is calculated as the sum of averaged scores for each of the 10 subdomains (maximum total score=30), with higher scores reflecting more pronounced functional change.88 The questionnaire will be administered to both participants and a carer/informant of their choice. Multifactorial Memory Questionnaire (MMQ): The MMQ consists of three scales (Satisfaction, Ability and Strategy) assessing various aspects of metamemory. Items are rated on a 5-point Likert scale (strongly agree, agree, undecided, disagree, strongly disagree). Each of the three MMQ scales is scored separately, and higher scores indicate better ratings.89 Multifactorial Memory Questionnaire (MMQ): The MMQ consists of three scales (Satisfaction, Ability and Strategy) assessing various aspects of metamemory. Items are rated on a 5-point Likert scale (strongly agree, agree, undecided, disagree, strongly disagree). Each of the three MMQ scales is scored separately, and higher scores indicate better ratings.89 Cognitive Stimulating Activity Questionnaire (CSA-Q): The CSA-Q is a brief questionnaire that measures everyday cognitive activities. Participants are asked to rate how frequently they participated in 13 common cognitively engaging activities during the previous month. Each item uses a 5-point Likert scale (never (1), once or twice per month (2), once or twice per week (3), several times a week (4) and every day or about every day (5)). Responses are averaged across all items to provide the mean level of activity participation (ie, CSA-Q mean) from 0 to 5. Higher scores reflect a higher participation frequency.90 Cognitive Stimulating Activity Questionnaire (CSA-Q): The CSA-Q is a brief questionnaire that measures everyday cognitive activities. Participants are asked to rate how frequently they participated in 13 common cognitively engaging activities during the previous month. Each item uses a 5-point Likert scale (never (1), once or twice per month (2), once or twice per week (3), several times a week (4) and every day or about every day (5)). Responses are averaged across all items to provide the mean level of activity participation (ie, CSA-Q mean) from 0 to 5. Higher scores reflect a higher participation frequency.90 The HBA Exercise Questionnaire is an adaptation of the Active Australia Survey, popular 8-item measure of activity that asks participants to report the frequency and duration of physical activity in the past week, including walking, gardening and moderate and vigorous activity.91 This questionnaire has been piloted in a memory clinic sample. The HBA Exercise Questionnaire is an adaptation of the Active Australia Survey, popular 8-item measure of activity that asks participants to report the frequency and duration of physical activity in the past week, including walking, gardening and moderate and vigorous activity.91 This questionnaire has been piloted in a memory clinic sample. Apathy Evaluation Scale (Self-Reported) (AES-S): The AES-S is a brief questionnaire that is a global measure of apathy in adults and older adults. The scale has 18 items rated on a 4-point Likert scale, which captures three themes: overt goal-directed behaviour, goal-related cognitions and goal-related emotional responses. Higher scores indicate higher levels of apathy.92 This scale will only be administered at baseline. Apathy Evaluation Scale (Self-Reported) (AES-S): The AES-S is a brief questionnaire that is a global measure of apathy in adults and older adults. The scale has 18 items rated on a 4-point Likert scale, which captures three themes: overt goal-directed behaviour, goal-related cognitions and goal-related emotional responses. Higher scores indicate higher levels of apathy.92 This scale will only be administered at baseline. Social Bonding Questionnaire: Developed by our research team, this questionnaire has two sets of questions. The first set, asked at all three time points, assesses if the intervention increases general feelings of social bonding, belonging and coalition signalling over time. The second set, asked only at follow-ups, evaluates if choir or piano lessons result in higher experiences of these feelings with the rest of the participants in the trial. Social Bonding Questionnaire: Developed by our research team, this questionnaire has two sets of questions. The first set, asked at all three time points, assesses if the intervention increases general feelings of social bonding, belonging and coalition signalling over time. The second set, asked only at follow-ups, evaluates if choir or piano lessons result in higher experiences of these feelings with the rest of the participants in the trial. Neuropsychological Tests Neuropsychological assessments are conducted by a trained neuropsychologist. To strengthen our cognitive conclusions, we will consolidate these test results into composite scores that reflect performance in five cognitive domain composites: executive functioning, verbal memory, verbal learning, processing speed and language (see table 1 for details). To create these cognitive composites, age-corrected standard z-scores will be calculated for each test. Then, for each participant, the mean z-score for all tests within a composite will be calculated, giving equal weight to each individual test used in a composite (eg, Executive functioning: ((Delis-Kaplan Executive Function System Colour Word Interference Test) (DKEFS CWIT) condition 3 z-scores+DKEFS CWIT condition 4 z-scores/2) + OTMT part B z-scores+COWAT z-scores+Digit span z-scores)/4). Actigraphy monitoring and sleep diary To objectively measure at-home sleep-wake patterns (such as sleep efficiency, nocturnal awakenings, sleep timing and total sleep time), all participants, except those who join within 1 week of the intervention starting, are offered an actigraphy watch (GeneActiv, Activinsights Ltd., Cambridgeshire, UK) to wear for 1 week before the intervention, 1 week after the intervention and during the 9 month follow-up. This is a non-invasive method of monitoring human rest/activity cycles and uses an accelerometer to record movement over a given threshold. The device is set to a measurement frequency of 30 Hz. During the screening process, the researcher explains the actigraphy watch and offers it to participants as an optional part of the study. Participants are given the watch when they attend in person for their blood collection and are given a paid return parcel to mail back. A 7 day sleep diary to record sleep quality and habits is given to participants with the actigraphy device. Additional consent provisions for collection and use of participant data and biological specimens Participants provide online consent for participation and have the option to withdraw it at any time. Any information kept on paper will be stored in a locked cabin within a locked office at the Brain and Mind Centre at the University of Sydney, or at the Sydney Conservatorium of Music. Any electronic data is stored in REDCap (Research Electronic Data Capture, hosted at University of Sydney), where access is limited to specified users. Participants also sign on to the Participant Information Statement that they understand that their deidentified data may be used in collaborations with other research programmes, subject to approval by the relevant Human Research Ethics Committee. Assignment of interventions: allocation and implementation Sequence generation and randomisation Randomisation occurs after completing all baseline assessments and is on a 1:1:1 basis to keyboard, choir or control group, stratified by age (60–65 or older than 65 years). The blinded study statistician created a computer-generated randomisation sequence and uploaded it to REDCap for the research team to follow. For each participant, the study coordinator opens their online RedCap record and clicks ‘randomise’ which allocates them the next group assignment in the pre-uploaded stratified sequence. Concealment mechanism Following baseline assessments, the unblinded study coordinator randomises participants. Each participant is assigned a unique randomisation number sequentially assigned in ascending order. To ensure the participants’ anonymity, all documents use the participants’ unique screening and randomisation numbers. Assignment of interventions: blinding Who will be blinded To maintain allocation concealment, only the necessary study team members are aware of group allocation (eg, the study coordinator conducting randomisation and informing participants of their group, relevant research assistants, and the music intervention facilitators). All study members that conduct assessments (eg, phlebotomy or neuropsychology) and analyses are blinded. Procedure for unblinding if needed As participants are not blinded to the intervention, there is no procedure for unblinding. Interventions Explanation for the choice of comparators Group description Duration All groups aim to complete 24 hours of group training (or active control) over 12 weeks. All participants are requested to attend a 2 hour session once a week (minimum request of 80% participation). However, once randomised, if participants miss more than 80% of sessions, they are still invited to attend, pending approval from the group facilitator as missing too many sessions in a row may disrupt the rest of the class. Location All three interventions occur at the Sydney Conservatorium of Music and are facilitated by professional musicians. Music interventions Approach Each session is structured with an initial warm-up followed by a brief review of material, and the remaining portion of the class focuses on learning and practising new skills and concepts and practising repertoire. Straight after class, there is a 25 minute socialisation session with light refreshments in a dedicated room. The control group also attend a socialisation session. Participants cover new and previously covered repertoire each week. There is no pressure on perfection but encouragement to attempt activities and participate. Facilitators focus on both gradual levels of short-term achievement and long-term goals. Their inclusion increases the size of the ensemble to reduce vocal exposure and provide more blending of voices. At 3 months, participants are invited to be part of a group performance in front of the unblinded study team and the other music intervention group and selected family or friends. Progress and practice Progress is recorded for each participant during each class session. Participants are recommended to practise for 30 minutes outside of the formal session, six times a week. If participants decide to practise at home, they receive weekly guidance and access to support resources. A series of weekly practice guides are provided to review course materials and recordings of each weekly class. This will also allow participants who were absent to catch up. Participants are asked to record their practice (eg, time spent, type of activity, level of engagement) in a practice diary provided by the instructor. Keyboard Keyboard intervention: Participants allocated to this arm receive keyboard lessons from a professional piano instructor. These lessons consist of basic piano technique and literature, music theory, dexterity exercises, conventional music notation reading and rhythmic training. At each training session, participants are expected to practise, and at times, perform technical exercises, sight reading examples and piano pieces. Lessons include solo and ensemble playing out loud as well as individual practice. To encourage practice, participants in the keyboard intervention are provided with small keyboards to take home until the end of the 9 month follow-up. They are also provided with headphones to wear (to avoid causing disturbance to co-inhabitants). Keyboard intervention: Participants allocated to this arm receive keyboard lessons from a professional piano instructor. These lessons consist of basic piano technique and literature, music theory, dexterity exercises, conventional music notation reading and rhythmic training. At each training session, participants are expected to practise, and at times, perform technical exercises, sight reading examples and piano pieces. Lessons include solo and ensemble playing out loud as well as individual practice. To encourage practice, participants in the keyboard intervention are provided with small keyboards to take home until the end of the 9 month follow-up. They are also provided with headphones to wear (to avoid causing disturbance to co-inhabitants). Choral singing intervention: The sessions are conducted by an experienced choir director and involve different strategies to learn new songs, practise listening to other participants, develop a unified sound, gain awareness of singing technique and musical terminology, and review previously learnt songs. The repertoire is selected based on a mix of familiar and unfamiliar, a wide range of common and uncommon sung languages, and to address specific musical aspects. The conductor accompanies the choir when required on piano, but the repertoire is often attempted as a cappella. Choral singing intervention: The sessions are conducted by an experienced choir director and involve different strategies to learn new songs, practise listening to other participants, develop a unified sound, gain awareness of singing technique and musical terminology, and review previously learnt songs. The repertoire is selected based on a mix of familiar and unfamiliar, a wide range of common and uncommon sung languages, and to address specific musical aspects. The conductor accompanies the choir when required on piano, but the repertoire is often attempted as a cappella. Documentation Keyboard and choral singing intervention facilitators are asked to maintain a weekly journal on their experiences leading the musical interventions. They reflect on the groups’ progress and note any specific challenges to tackle in future sessions. Facilitators are also interviewed regarding their experience after the intervention as outlined below. Active control group The active control group in this trial involves attending weekly film-viewing events and taking part in a semi-structured discussion. The amount of time exposed to the control group is matched to the two active interventions. This discussion focuses on personal questions of taste and reaction to the film (from the participants). Importantly, the facilitator does not bring musical expertise or instruction into the discussion—the emphasis is on the participants’ perspectives. The facilitator might introduce discussion questions that address the film’s sociological, historical, political and cultural aspects to prompt discussions within the group. Thus, control participants experience social elements similar to those of the intervention groups and hence, the duration and the potential positive impacts of socialisation in the intervention groups are closely aligned. Participants watch the film for 45–60 minutes. Then, depending on the length of the film screening, the facilitator initiates a 20–30 minute semi-structured discussion. Copyright permission to use the films for educational purposes has been obtained. After completion of the trial at 9 months, participants in the active control group are offered the opportunity to engage with keyboard or choral singing interventions, depending on availability in those classes. Strategies to improve adherence to interventions and plans to promote participant retention and complete follow-up If a participant does not attend a weekly session of any of the groups, a member of the research team attempts to contact them over 2 weeks via telephone or email to establish the reason for the non-contact, checks the participant’s condition and encourages attendance. If these phone calls fail to reach the participant, the researcher emails them. However, the study team still attempts to contact the participant near the 3 and 9 month time points to collect outcome intention to treat data if the participant is willing. Participants are offered a $50 AUD gift voucher at the conclusion of their participation in the trial. Participants who withdraw but still are willing to complete follow-up are also offered this gift card. Criteria for discontinuing or modifying allocated interventions If all reasonable attempts to contact over a reasonable period fail, the investigator will declare the participant ‘lost to follow-up’. Relevant concomitant care permitted or prohibited during the trial During the trial, participants have no restrictions apart from engaging in any of the exclusion criteria. They are asked not to engage in any music learning, apart from that which pertains to the group in which they are allocated. Provisions for post-trial care The Trial Steering Committee (TSC) reviews any significant on-study incidents such as injury or death, dropout rates and participants lost to follow-up. The TSC regularly considers participant safety, confidentiality and data integrity issues and recommends study continuation. We do not anticipate the need for post-trial care due to the nature of the trial. Outcomes Primary outcome The primary outcome is change in verbal memory performance measured by the Rey Auditory Verbal Learning Test (RAVLT)75 delayed recall (ie, trial 7) score (number of correct answers/15 after 20 min delay) between baseline and 3 months. This outcome was chosen as it (a) is meaningful to patients who present to health services with memory concerns (b) is an early feature of dementia (c) has been shown to be sensitive to changes after 10–12 week cognitive training,13 93 94 sleep94 and other139396 interventions. While verbal memory tests such as the RAVLT have been shown to have practice effects when repeated, these would be present in all randomised conditions. Hence, only the standard version of the RAVLT was utilised. Secondary and mediating outcomes The secondary objective is to evaluate the effects of keyboard and choral singing interventions on brain plasticity, cognition, psychosocial outcomes and blood-based biomarkers compared with a control condition. The schedule of study procedures is shown in figure 1. Effect sizes and variability (means, SDs, CIs) of each group on each of the following measurements are recorded: Biomarkers of neurodegeneration and AD sensitive to changes that occur in MCI (due to AD and other neuropathologies) MRI scanning (baseline and 3 months—for the first four waves) is conducted on a 3-Tesla General Electric (GE) Discovery MR750 scanner (GE Medical Systems, USA) with a 32-channel phased-array head coil at the Brain and Mind Centre. Cortical thickness in the prefrontal, parietal and temporal regions is captured using a 3D-T1-weighted BRAVO Spoiled Gradient-Recalled (repetition time=7.4 ms; echo time=3.0 ms; flip angle=11; matrix 256×256; 1.0 mm isotropic voxels). Total hippocampal and subfield volumes are captured using high-resolution T2 images using a 2D-T2-weighted FSE-XL (repetition time=3000.0 ms; echo time=51.48 ms; flip angle=111; matrix 512×512; 0.25, 0.25, 2.0 mm isotropic voxels). Brain connectivity within the brain’s default network is measured using resting state fMRI using a single-echo multiband EPI resting state fMRI (repetition time=1000 ms; echo time=29 ms; 60 flip angle; FOV=128×128 mm2; voxel size=1.875×1.875 x2.5mm; number of slices=51; 2.50 mm slices). Increases in white matter fibre connectivity in the hippocampal and cortical-subcortical regions are captured using a multi-shell DWI (repetition time=4000 ms; echo time=95.2 ms; 90 flip angle; FOV=256×256mm2; voxel size=1.0×1.0 x2.0mm; 66×2 mm slices). The MRI data are pre-processed using FastSurfer97 for cortical thickness and the automated segmentation of hippocampal subfields (ASHS)98 for hippocampal subfield volumes. Functional connectivity and structural white matter integrity are preprocessed using fMRI prep99 and MRItrix3,100 respectively. These MRI software tools will ensure high-quality data processing and accurate measurement of the brain structures and functions of interest. MRI scanning (baseline and 3 months—for the first four waves) is conducted on a 3-Tesla General Electric (GE) Discovery MR750 scanner (GE Medical Systems, USA) with a 32-channel phased-array head coil at the Brain and Mind Centre. Cortical thickness in the prefrontal, parietal and temporal regions is captured using a 3D-T1-weighted BRAVO Spoiled Gradient-Recalled (repetition time=7.4 ms; echo time=3.0 ms; flip angle=11; matrix 256×256; 1.0 mm isotropic voxels). Total hippocampal and subfield volumes are captured using high-resolution T2 images using a 2D-T2-weighted FSE-XL (repetition time=3000.0 ms; echo time=51.48 ms; flip angle=111; matrix 512×512; 0.25, 0.25, 2.0 mm isotropic voxels). Brain connectivity within the brain’s default network is measured using resting state fMRI using a single-echo multiband EPI resting state fMRI (repetition time=1000 ms; echo time=29 ms; 60 flip angle; FOV=128×128 mm2; voxel size=1.875×1.875 x2.5mm; number of slices=51; 2.50 mm slices). Increases in white matter fibre connectivity in the hippocampal and cortical-subcortical regions are captured using a multi-shell DWI (repetition time=4000 ms; echo time=95.2 ms; 90 flip angle; FOV=256×256mm2; voxel size=1.0×1.0 x2.0mm; 66×2 mm slices). The MRI data are pre-processed using FastSurfer97 for cortical thickness and the automated segmentation of hippocampal subfields (ASHS)98 for hippocampal subfield volumes. Functional connectivity and structural white matter integrity are preprocessed using fMRI prep99 and MRItrix3,100 respectively. These MRI software tools will ensure high-quality data processing and accurate measurement of the brain structures and functions of interest. Blood biomarkers (baseline and 9 months): Recent breakthroughs have enabled the detection in blood plasma of neurotoxic proteins implicated in AD and neurodegenerative conditions (well-validated against gold-standard Positron Emission Tomography (PET) imaging), demonstrating overall accuracy and showing sensitivity to change longitudinally.101 The field has now moved to using such markers as primary trial endpoints, in a range of AD (lecanemab, donanemab102 and neurodegeneration studies to evaluate the effect of pharmacological and non-pharmacological interventions in MCI and early AD from as early as 8 weeks (eg, clinicaltrials.gov IDs: NCT05877196; NCT05015803; NCT05172128). Such trial endpoints are more feasible, accessible and cost-effective than the use of PET scans or invasive lumbar puncture for CSF samples. A fasting 30 mL blood sample is collected, processed and stored at −80°C using established protocols aligned with the Australian Dementia Network. Analyses occur at trial completion for markers of neurodegeneration (neurofilament light chain, NfL; Glial fibrillar acidic protein, GFAP), and AD (pTau217, Amyloid-β 42/40). ApoE genotype and kidney function are obtained, as potential mediating factors. Blood biomarkers (baseline and 9 months): Recent breakthroughs have enabled the detection in blood plasma of neurotoxic proteins implicated in AD and neurodegenerative conditions (well-validated against gold-standard Positron Emission Tomography (PET) imaging), demonstrating overall accuracy and showing sensitivity to change longitudinally.101 The field has now moved to using such markers as primary trial endpoints, in a range of AD (lecanemab, donanemab102 and neurodegeneration studies to evaluate the effect of pharmacological and non-pharmacological interventions in MCI and early AD from as early as 8 weeks (eg, clinicaltrials.gov IDs: NCT05877196; NCT05015803; NCT05172128). Such trial endpoints are more feasible, accessible and cost-effective than the use of PET scans or invasive lumbar puncture for CSF samples. A fasting 30 mL blood sample is collected, processed and stored at −80°C using established protocols aligned with the Australian Dementia Network. Analyses occur at trial completion for markers of neurodegeneration (neurofilament light chain, NfL; Glial fibrillar acidic protein, GFAP), and AD (pTau217, Amyloid-β 42/40). ApoE genotype and kidney function are obtained, as potential mediating factors. Markers of functioning, well-being and global cognition (baseline, 3 months and 9 months) Mood: Depressive symptoms as assessed by the GDS-1583; anxiety symptoms as assessed by the GAS.84 Mood: Depressive symptoms as assessed by the GDS-1583; anxiety symptoms as assessed by the GAS.84 Global cognition: As per table 1, domains of processing speed and executive functioning will be computed as well as a Preclinical Alzheimer’s Disease Global Cognitive Composite (PACC).103 Global cognition: As per table 1, domains of processing speed and executive functioning will be computed as well as a Preclinical Alzheimer’s Disease Global Cognitive Composite (PACC).103 Sleep: 7 days of actigraphy according to established protocols for sleep efficiency and duration104 and self-reported sleep quality (Pittsburgh Sleep Quality Index).87 Sleep: 7 days of actigraphy according to established protocols for sleep efficiency and duration104 and self-reported sleep quality (Pittsburgh Sleep Quality Index).87 Daily functioning is determined by the HBA-FAQ88 88(participants and informants). Daily functioning is determined by the HBA-FAQ88 88(participants and informants). Subjective cognitive complaints: BCCI.85 Subjective cognitive complaints: BCCI.85 Overall health status is assessed by the EuroQol 5-dimension scale.86 Overall health status is assessed by the EuroQol 5-dimension scale.86 Metamemory (satisfaction, ability and strategy) is assessed by the MMQ.105 Metamemory (satisfaction, ability and strategy) is assessed by the MMQ.105 Participation in cognitively stimulating tasks is determined by the CSA-Q.90 106 Participation in cognitively stimulating tasks is determined by the CSA-Q.90 106 Exploratory outcomes Treatment adherence is determined by a pre-specified attendance value of 80% of sessions at 3 months. Treatment adherence is determined by a pre-specified attendance value of 80% of sessions at 3 months. Overall engagement is assessed by self-reported practice time outside intervention attendance and self-reported goal completion scores at 3 months. This outcome will be examined as a mediator of the relationship between music intervention and MRI and cognitive outcomes. Overall engagement is assessed by self-reported practice time outside intervention attendance and self-reported goal completion scores at 3 months. This outcome will be examined as a mediator of the relationship between music intervention and MRI and cognitive outcomes. Apathy will be assessed using the AES-S 92 at baseline. This outcome will be examined as a mediator of the relationship between music intervention and MRI and cognitive outcomes. Apathy will be assessed using the AES-S 92 at baseline. This outcome will be examined as a mediator of the relationship between music intervention and MRI and cognitive outcomes. Social bonding is determined by the Social Bonding Questionnaire (developed in house). Social bonding is determined by the Social Bonding Questionnaire (developed in house). Semi-structured interviews and focus groups assess the participants’ experience of the music-based training programme at weeks 4, 8, 12 and at 9 months. Each interview lasts approximately 30 minutes and focus groups last up to 45 minutes due to the nature of the group interaction. Topic areas and prompt questions address participant enjoyment, the challenge of the musical tasks and the broader impacts of the intervention. The interviews and focus groups are audio-recorded and transcribed using Zoom, and a thematic framework will be used for qualitative analysis. Sample size Statistical power analysis for the primary aim has been completed by our trial statistician (DE) using SAS (V.9.4) GLMPOWER. Our existing wave 1 and 2 data show an attrition rate of<10% for our target sample. Controlling for a type I error of 0.05 (two-tailed) with 80% statistical power and assuming a correlation of 0.5 between baseline and 3 month primary outcome RAVLT-7 scores (estimated from blinded analysis of wave 1 and 2 data), a sample size of 232 would be required to detect an effect size of 0.34 between the experimental groups (keyboard+choral) and control group. To accommodate comparisons of each intervention group with control, a final sample size of 308 would provide 0.80 power at alpha 0.05 to detect an effect size of 0.34 (as per meta-analysis).66 Accounting for attrition (estimated at 10%), and the clustering effect of the waves introduced by the delivery method of the intervention (intraclass correlation, ICC, of 0.02), the target sample size is 432 participants (n=144 per randomised condition). All statistical analyses will be described in a predefined analysis plan (see ‘Statistical Methods’ section). Data management The datasets generated during the current study will not be available for data sharing until the final analysis and publication of results. Other investigators can formally request access to the anonymised dataset if a formal request describing their plans is approved by the Steering Committee, ethics committee and trial sponsor, and appropriate data-sharing agreements are in place. Confidentiality Participant data in this study are assigned a code number only after the participant has consented to participate. This code will be used to identify the participants throughout the study. The key that links the participant’s identity to their assigned code is stored in a password-encrypted file. To ensure confidentiality, no information that could potentially identify a participant is used in any publication or presentation without their explicit permission. All data collected during this research is treated as confidential and securely stored. Plans for collection, laboratory evaluation and storage of biological specimens for genetic or molecular analysis in this trial/future use All blood samples will be stored in a safe and secure storage facility at the University of Sydney until analysis and will only be used for the purposes of this study. All test samples taken will be labelled with a unique code. The chemistry and haematology samples will be destroyed when the final analysis is completed. Statistical methods Statistical methods for primary and secondary outcomes All randomised participants will be included in the primary endpoint analysis to estimate the treatment effect using an intention-to-treat principle. Descriptive analyses will estimate means and proportions, CIs and measures of variability of all measures. Linear mixed models (LMMs) will be implemented to investigate changes in verbal memory performance measured by the RAVLT trial 7 between and within each group (keyboard, choral singing and control) across different time points (baseline and 3 months). The primary and secondary outcomes will be analysed via LMMs using maximum likelihood estimation. The model will include random intercepts, fixed effects of time (baseline and 3 months), intervention (keyboard, choral singing and control), and their interactions. Assessment of the treatment effect will be via the treatment-by-time interaction. Two contrasts will be performed to assess differences in changes in the primary and secondary outcomes between: The Choral Music Group and the control. The Choral Music Group and the control. The Keyboard Music Group and the control. The Keyboard Music Group and the control. Appropriate effect size measures will be calculated (eg, ICC for random effects and f2 for fixed effects). The blinded study statistician will prepare a statistical analysis plan, including the planned interim analysis after the conclusion, and finalise it before unblinding the data. It will contain additional details on the methods described. Data will be analysed by the trial statistician (DE) from the NHMRC Clinical Trials Centre. Local or international collaborators may use de-identified data for future research purposes. Interim analyses An interim analysis took place one quarter of the way through recruitment (April 2025), based on the blinded data, with the purpose to confirm or update the sample size based on the findings. Based on the interim analysis, the sample size did not change. Methods for additional analyses (eg, subgroup analyses) A planned sub-analysis of the primary outcome in the participants that screen positive for AD based on pTau217 results is planned. Further exploratory analyses will evaluate whether changes in the blood outcomes mediate changes in the imaging and cognitive outcomes. There will also be an analysis to investigate if physical activity change, as captured by the HBA Exercise Questionnaire, affects the cognitive outcomes. Methods in analysis to handle protocol non-adherence and any statistical methods to handle missing data Multiple imputation (eg, JM, FCS) will be implemented where necessary. Plans to give access to the full protocol, participant-level data and statistical code There are no current plans for granting public access to the full protocol, participant-level dataset or statistical code. Oversight and monitoring Composition of the coordinating centre and TSC The randomised controlled trial will comply with the protocol published in the Australian and New Zealand Clinical Trials Registry (ACTRN12623000407695), the International Conference for Harmonisation on Good Clinical Practice, and all other applicable regulatory requirements. This clinical trial protocol has been drafted according to the SPIRIT 2013 Statement74 (online supplemental file 2), and the final trial paper will be reported according to the CONSORT 2010 Statement.107 The composition of the coordinating centre and TSC is detailed in online supplemental file 1. Composition of the data monitoring committee, its role and reporting structure Due to the nature of the trial, we do not anticipate adverse events or serious adverse events. As such, a Data and Safety Monitoring Board was considered to be unnecessary, given minimal risk to participants. The Steering Committee takes charge of any risks to data validity or integrity. Adverse event reporting and harms All personnel collecting data, including phlebotomists, neuropsychologists and study doctors, have the necessary qualifications and training, including ICH-GCP certification. The study doctor reviews the MRI reports and contacts participants promptly if unexpected (eg, incidental) findings arise, advising them to consult their GP. Frequency and plans for auditing trial conduct Data are audited yearly according to a data monitoring plan by a researcher not otherwise involved in the trial. To date, there has been one audit conducted in January 2024, and another currently in progress that began October 2025. At the completion of each wave, a summary including number prescreened, screened and eligible, confirmation that all consents have been checked and saved appropriately, and any other noteworthy information is presented to the study PI (SLN) and an independent researcher with expertise in clinical trials who signs off. Plans for communicating important protocol amendments to relevant parties (eg, trial participants, ethical committees) Important protocol modifications will be communicated to relevant parties (investigators, trial participants and trial registries) as soon as they come into effect. If any important/significant protocol modifications or amendments are made, standard steps will be taken to gain approval from ethics for relevant documents and to re-consent affected participants with the new protocol as applicable. Ethics and dissemination This research protocol has been given approval by the University Human Research Ethics Committee (2023/026). All protocol amendments are submitted to the ethics committee and the Clinical Trials Governance Team at the University of Sydney and are only implemented when approved. Where applicable, the study results will be communicated through conferences, publications and media. At the end of the trial, participants will be fully informed of the outcomes. Eligible authors will include investigators involved in the study’s conception and design, the trial’s conduct, the analysis of the results, and the reporting and presentation of the study findings. Discussion The results of the current study have the potential to provide valuable evidence regarding the effectiveness of music interventions in preventing cognitive decline and fostering neuroplasticity in older adults with MCI. Despite the promising theoretical foundation for music interventions, there remains a lack of robust evidence from randomised controlled trials specifically targeting this population. If the findings from this Phase 3 RCT support our hypothesis, the next step will be to focus on the practical translation of these interventions into sustainable, community-based health practices. If the interventions were scaled, other allied health facilitators could potentially deliver the interventions; however, this would require sufficient training, further testing and validation. Translation will involve co-designing programmes with the community to ensure they are culturally relevant, accessible and scalable, ultimately aiming to integrate them into public health services tailored to older adults with memory difficulties, such as memory and cognition clinics. The existing body of literature examining music interventions for individuals with MCI is diverse, with an integrative review identifying eight studies that utilised 25 cognitive outcome measures. These studies also employed a variety of music interventions, including passive music listening, musical stimuli delivered via digital devices, multitask movement music therapy, recorder lessons and percussion instruction.108 While the review highlights some positive results, the studies showed significant variability in scope, cognitive profiles and quality, which made it difficult to draw consistent conclusions. Importantly, the review did not include studies focused on keyboard learning or singing—an area our protocol uniquely addresses. This gap underscores the contribution of our study, which will examine in a definitive trial, a previously under-researched aspect of music intervention that could provide new insights into cognitive enhancement for older adults with MCI. A critical challenge for scaling music-based interventions is ensuring equitable access, particularly among older adults, socioeconomically or cognitively disadvantaged populations, diverse ethnic groups, or those living in rural or regional communities. It is essential that these interventions are not only effective but also feasible, cost-efficient and appealing to the target populations. Co-designing the programme with input from the community and key stakeholders has been a vital step to help ensure that the interventions are tailored to the specific needs and preferences of older adults with cognitive impairment. By actively involving the community in the design and implementation phases, we can ensure that the intervention is culturally relevant and resonates with participants, thereby enhancing engagement and adherence. Additionally, for such programmes to be adopted widely in public health practice, they must demonstrate broader benefits beyond cognitive outcomes, including improvements in quality of life, emotional well-being and daily functioning.109 This study has several notable strengths. The randomised controlled design enhances the internal validity by minimising selection bias, while the comprehensive assessment of outcomes, including cognition, neurobiological and blood biomarkers of dementia, mood and sleep, offers a holistic understanding of the interventions’ effects. Nonetheless, some limitations should be acknowledged. First, as a single-blind trial, participants were aware of their group allocation, which may have influenced their behaviour or self-reported outcomes. This limitation is largely unavoidable in behavioural intervention studies. Second, the study’s ability to generalise to diverse cultural, Indigenous and socioeconomic groups may be constrained by its inner-city setting, which required participants to travel weekly for 3 months. Third, the 3 month intervention duration, although chosen for feasibility, may not be sufficient to detect meaningful changes in some neurodegenerative markers. Finally, during the 9 month follow-up period, it was not possible to collect accurate adherence or engagement data; however, focus groups qualitatively explored participants’ ongoing music engagement, with findings to be reported separately. Despite these limitations, the rigorous design and breadth of outcome measures enhance the study’s contribution to understanding how structured music interventions may influence cognitive and emotional health in older adults with MCI. In conclusion, this study seeks to build a robust evidence base for the use of keyboard or choral singing interventions for people with MCI. If the Phase 3 RCT results are promising, the next phase will focus on translating the intervention into a scalable model that is accessible, cost-effective and beneficial to older adults at risk for dementia. The involvement of the community in the co-design process will be crucial to ensuring that the intervention meets the needs of participants and can be effectively implemented in real-world settings. The goal is to incorporate these music-based interventions into public health practices, providing older adults with a sustainable, engaging and effective tool for cognitive health and overall well-being. Trial status Recruitment began in May 2023 and is estimated to last for 4 years. Supplementary material