ACT — Anticipatory Care Tool

What ACT is

ACT — the Anticipatory Care Tool — is the engine of the MCDC programme. It is the analytics and observational layer that turns individual moments of musical engagement into longitudinal trajectories, and it is simultaneously being piloted as a standalone observational app for anyone in regular contact with a person living with dementia.

The two framings are deliberate. Within MCDC, ACT is Component 3 of the four-component framework: it ingests data from M-CST sessions (via the M-CST-ob observation layer) and from daily Memory Tracks song-task associations, and it surfaces trends, red flags, and domain-specific trajectories that feed back into care planning. As a standalone tool, ACT delivers the same observational discipline to family members, informal carers, and non-clinical staff — people who see small changes first but rarely have a structured way to record or act on them.

Both framings serve the same anticipatory principle: identify small but significant changes in physical and mental wellbeing early, so that care can adapt before crisis.

The problem it addresses

Existing dementia assessment is episodic, clinic-bound, and dependent on verbal self-report — all of which work against people living with dementia. Standard cognitive tests miss decline between testing points, lose sensitivity at floor and ceiling, and require the person being assessed to articulate symptoms they may no longer recognise. Meanwhile, the people in daily contact — family, personal carers, activity coordinators, care-home staff — observe meaningful change continuously but lack a shared, structured instrument to capture it.

The consequence is visible at a population scale. In the UK, roughly one in four hospital beds is occupied by a person living with dementia, average stays run to several weeks, and a significant proportion of admissions are avoidable if change is spotted early. Anticipatory care appears in national policy (the NHS Long Term Plan; the Scottish Government's framework descending from Julian Tudor Hart's 1974 Milroy Lecture), but its implementation remains largely informal.

ACT exists to make anticipation routine — built into everyday care contact rather than gated behind specialist assessment.

The approach

ACT's core instrument is sixteen observational questions, organised into four domains:

• Physical health — day-to-day physical wellbeing, mobility, appetite, sleep
• Wellbeing — mood, affect, distress cues, engagement
• Behavioural — changes in routine behaviour, responsiveness, initiation
• Cognitive — memory, orientation, communication

Each question is answered on a 1–5 scale (1 = lowest concern, 5 = highest). The sixteen questions were derived from a structured analysis of eight established clinical assessment instruments, per the under-review ACT manuscript (Brill & Whalley): the Six-item Cognitive Impairment Test (6CIT), Addenbrooke's Cognitive Examination-III (ACE-III), the Comprehensive Geriatric Assessment (CGA), the Montreal Cognitive Assessment (MoCA), the Mini-Mental State Examination (MMSE), the Neuropsychiatric Inventory (NPI), the Barthel Index of Activities of Daily Living, and the Electronic Frailty Index (eFI). Each was mapped for domain coverage, assessment methodology, and practicality for non-clinical use; ACT's 16-item framework preserves their core domains while translating the questions into observational language lay caregivers can use without clinical training.

The ACT algorithm applies weighted values per question and per section, and runs a red-flag detection system built on a moving average — so a sudden dip against a resident's own baseline is surfaced even when any single score is within "normal" range. Within the MCDC framework, ACT's inputs extend beyond the sixteen questions to include the session-level data from M-CST-ob and the daily engagement data from Memory Tracks.

ACT is deliberately non-diagnostic. It does not generate clinical decisions. It surfaces patterns for human judgement and facilitates appropriate referral when changes are detected.

Current state

• A live observational web app is running as an AWS-hosted pilot at anticipatory.care, with multi-device support (mobile, tablet, desktop) so the tool is usable at the point of care.
• A working analysis demo demonstrates the algorithm's current red-flag and trend-detection behaviour against sample data.
• The MCDC white paper (Whalley, James, Brill & Cunningham, 2026) sets out ACT's role as Component 3 of the framework, including its data inputs, phase-1 and phase-2 capabilities, and its place in the closed-loop care cycle.
• The first live-setting test is the planned Pendine Park pilot in North Wales — 12–20 participants, 7 weeks of Music CST with baseline and follow-up, described in Section 9 of the white paper.

What's next

The roadmap for ACT is explicit, staged, and tied to evidence from the pilot:

1. API integration with care-management systems — a licence-model integration so ACT's outputs reach the daily care record. ACT-API entered beta in April 2026; Person Centred Software is named in the MCDC white paper as the priority integration target post-pilot. Further consumer integrations will be confirmed as they sign.
2. Machine-learning time-series prediction — moving from rule-based red flags to pattern prediction once longitudinal data accumulates.
3. Computer vision — AI-assisted visual analysis to supplement structured observation.
4. Semantic analysis — natural-language processing of free-text care notes and verbal observations.

Each stage is contingent on the prior: no ML prediction without pilot data; no NLP without a shared vocabulary; no clinical use without validation.

Collaborators and funding

ACT began in 2023 under Dr J. Harry Whalley and Mark Brill at the University for the Creative Arts, funded by a UKRI Zinc Catalyst Award of £60,000 (2023–2024). Co-design partners are Memory Matters (Plymouth) and Lifecare (Edinburgh), whose workshops and interviews with professional care staff, family caregivers, healthcare professionals, and twelve people living with dementia shaped the 16-item framework. A 24-week trial at the Lifecare dementia club in 2023 (n = 16) was the first live test of the algorithm; it generated two GP referrals on the strength of ACT alerts. The MCDC-integrated pilot takes place at Pendine Park in North Wales.

More at anticipatory.care. The full development and technical-validation paper is at Papers: Brill & Whalley, Development of a Digital Anticipatory Care Tool for People Living with Dementia.

Related

• Paper: Music-Centred Dementia Care: A Dual-Purpose Framework — the framework within which ACT sits (Component 3).
• Research strand: [PLACEHOLDER LINK: Research → M-CST-ob] — the observational data layer that feeds ACT within MCDC.
• Research strand: [PLACEHOLDER LINK: Research → ACT-API] — the in-development integration layer exposing ACT outputs to care-record systems.

Open questions for Harry

• Is there a concise, public one-liner for Innovate UK / Zinc Catalyst Healthy Ageing funding we should attribute more specifically (grant reference, period)?
• Does "16 observational questions" include the domain breakdown I've listed (4 × 4), or does the distribution across sections differ? Confirm from the live app if so.
• [PLACEHOLDER: exact Innovate UK grant reference if you want it cited]
• [PLACEHOLDER: a date for the Pendine Park pilot — planned month/year of start]