Layer 2 — Intelligence Synthesis
Layer 3 — Outreach Targeting
Layer 4 — Precision Screening
Layer 5 — Triage & Action
Layer 1 — Population Signal Ingestion
Public Data Sources & Digital Exhaust
ONS Census 2021
Over-75s living alone. LSOA granularity (~1,500 households)
Public · Free
NHS Fingertips API
Frailty prevalence, unplanned admissions, falls rates by PCN
API · Free
IMD 2019
Deprivation index. Strong frailty proxy. Ward + LSOA level
Public · Free
DWP Benefits Data
Attendance Allowance & PIP claims by postcode district
Public · Free
GP Registration Data
Patient counts by age band per practice. NHS published quarterly
API · Free
CQC Register
Care home locations, capacity, ratings. Service gap indicator
API · Free
Pharmacy Dispensing
Polypharmacy rates (5+ meds) by practice. Strong frailty signal
Public · Free
ACORN / Mosaic
Geodemographic household classification. Vulnerability proxies
Commercial · Licensed
⚠ Linkage challenge: datasets use different geographic units (LSOA, ward, PCN, postcode sector). The normalisation step is non-trivial — all must be mapped to a common spatial index before scoring.
↓
Layer 2 — Composite Frailty Risk Scoring
Geographic Intelligence Agent
Agent Logic — Scoring Pipeline
Step 01
Normalise Geography
Map all inputs to LSOA. Use ONS lookup tables for PCN → ward → LSOA crosswalk.
Step 02
Weight & Score
Apply domain-informed weights: age alone (high), polypharmacy (high), IMD (medium), DWP (medium), A&E rate (high).
Step 03
Composite Index
Generate Frailty Emergence Probability (FEP) score per LSOA on 0–100 scale.
Step 04
Service Gap Overlay
Cross-reference CQC capacity data. High FEP + low care provision = priority zones.
Step 05
Change Detection
Agent reruns quarterly. Flags LSOAs with rising FEP score — early trajectory signal.
Step 06
Uncertainty Flag
Marks zones where data is sparse or proxies are weak. Prevents false precision in commissioner reports.
Output: ranked heat map of LSOAs by Frailty Emergence Probability. Exportable as GeoJSON for mapping, CSV for commissioners, or direct API feed to Layer 3.
↓
Layer 3 — Community Touchpoint Identification
Outreach Targeting Agent
GP Surgeries
Anchor point. PCN-funded proactive outreach duty for over-75s
PCN Contract
Pharmacies
High-frequency contact for older adults. Polypharmacy patients already identified
Community
Libraries
Trusted civic space. Digital access points. Often already run wellbeing sessions
LA Partnership
Faith Communities
Older demographic concentration. High trust. Lunch clubs, coffee mornings
Voluntary Sector
Food Banks / VCSE
Reaches under-served populations. IMD-correlated. High-need intersection
VCSE
Domiciliary Care
Carers already in homes. Natural facilitated screening opportunity
Assistiv Network
Agent maps touchpoints within each priority LSOA using Google Places API + LA open data. Outputs a ranked outreach plan: which venues, in which order, with estimated population reach per venue.
↓
Layer 4 — Precision Screening Deployment
resiliencetools.xyz — Deployed at Touchpoint
Voice-First Screener
PRISMA-7 + FRAIL Scale. Self-administered or facilitated. Gives before it asks.
Live
Carer Observational
Completed by domiciliary carer, family member, or pharmacist. Parallel signal.
Live
Triple Tap Intelligence
Synthesises self-report + carer observation + population signal into unified frailty profile
Core IP
Frailty Profile Radar
Visual SVG output. Shareable with GP, family, or care coordinator.
Live
Three-Layer Consent
Data minimisation. Individual controls what is shared and with whom.
Clinical Design
Pattern Safety Interrupts
Detects minimisation. Routes to safety pathway if crisis indicators present.
Clinical Design
Context injection: each screening session receives the LSOA FEP score as background context. Agent calibrates sensitivity thresholds accordingly — higher-risk zones trigger lower referral thresholds.
↓
Layer 5 — Triage, Referral & Population Intelligence
Action & Feedback Loop
🏥
Clinical Referral
Single or dual-person referral to GP / geriatric service with Frailty Profile Radar attached
📊
Population Dashboard
PCN / ICB commissioner view. Aggregate trends, screening uptake, referral rates by LSOA
🔁
FEP Score Feedback
Anonymised screening outcomes fed back to Layer 2. Validates and recalibrates composite scores over time
🏠
Assistiv Onboarding
High-need individuals routed into Design Your Life — graduated sensor installation pathway
📍
Outreach Efficacy Report
Which touchpoints yielded highest screening uptake and referral conversion. Informs next cycle.
⚠️
Safety Escalation
Immediate routing to social care / crisis team where pattern interrupts fire during screening
⚙ Strategic Positioning
PCN Statutory Duty
PCNs must proactively identify and support their over-75 population. This architecture gives them the geographic intelligence layer they currently lack, plus the screening tool to act on it. Funded pathway.
Health Innovation KSS
Layer 2 composite scoring + Layer 3 outreach map is a standalone commissioner-facing product. Demonstrable to Matt Peach before full platform is built.
IP Boundary
Data sources are public. The value — and the IP — is in the FEP scoring algorithm, the Layer 2→3 routing logic, and the Triple Tap synthesis. Defensible.
Feedback Loop = Moat
Every screening outcome recalibrates the FEP model. Over time, Assistiv's scoring outperforms generic deprivation proxies. The dataset becomes the competitive advantage.