Home sensing · thresholds · actionability

Which sensor-derived measures actually matter at home?

Meaningful Measures began with a practical question from organisations delivering home-based care, supported living, remote monitoring and health technology: which readings from sensors and wearables are meaningful enough to support earlier intervention, safer care and better decisions for older adults living outside hospital?

Signal → threshold → evidence-weighted meaning → clinical actionability

See the finding Explore the signals
Abstract evidence visualisation showing sensor-derived physiological signals passing through a threshold into meaningful outputs

The partners asked the question

Meaningful Measures began with a practical challenge from organisations already delivering care, sensing, analytics and support in people’s homes: remote monitoring is growing quickly, but which signals are useful enough to trust?

Partner-led origin story

Not technology looking for a problem — a real care problem looking for evidence.

The industry, health and supported-living partners are the drivers of the project. They bring the settings, technologies, service models and implementation questions. The research role is to turn that need into a rigorous evidence synthesis and validation pathway.

1Care needOlder adults living outside hospital need earlier, safer support.
2Sensor capabilityPartners can capture data from homes, wearables, patches and platforms.
3Evidence questionWhich signals are meaningful enough to support action?
NSSN Grand Challenge Fund Program platform for collaborative smart-sensing research
Human Health / Ageing · university–industry collaboration · practical impact pathway
Meaningful
Measuresshared evidence question
Hospital in Your Homecare outside hospital
My Medic Watchwearables and event detection
United for Careliving environments and support contexts
InteliCarehome sensing and wellbeing analytics
Hills Health · Vlepisintegration and sensor technology
Care contextCare outside hospital
Support contextSupported living and home environments
Signal contextWearables, events and activity traces
Analytics contextHome sensing and wellbeing analytics
Technology contextSensor integration and platform capability
Evidence contextValidation, escalation and actionability
Research and program platform

Enabled by NSSN. Developed by FASTlab as research translation.

The industry, health and supported-living partners brought the practical problem. The NSW Smart Sensing Network Grand Challenge Fund provided the collaborative smart-sensing platform. FASTlab developed this public-facing synthesis, prototype evidence atlas, data-testing workflow and validation-planning pathway to help distinguish raw sensor data from measures that are interpretable, trusted and actionable in context.

Program platformNSSN Grand Challenge FundHuman Health / Ageing challenge context
Research translationFASTlabevidence method, translation design, prototype workflow and public-facing synthesis
Research governanceUniversity of Newcastleprotocol and ethics context where applicable
Clinical and industry partnersProject partnerspractical settings, device context and validation questions
The shared question

Which sensor-derived measures, or combinations of measures, are meaningful enough to help detect deterioration and support timely care for older adults living outside hospital?

01 Signal Reliable detection 02 Context Person + setting 03 Threshold Clear trigger rule 04 Validation Evidence tested 05 Action Response pathway 06 Meaningful Measure Interpretable · trusted · actionable
How to read this finding: meaningful measures are not the variables that appear most often. They are the measures that survive the translation pathway from signal, through context and threshold, into validation and action.
Evidence mapThresholdsSignal-to-action pathwayEvidence snapshot
Important interpretation: these are candidate meaningful measure families for validation and research translation. They are not universal clinical cut-offs, clinical advice or a substitute for local governance, pathway design and prospective validation.

Start with the signals

Start with the conclusion, then move through the evidence trail. Many readings can be collected at home, but only some are clear enough, validated enough and actionable enough to support care decisions.

What can we conclude?

The core finding: meaning comes from signal, context, threshold, validation and action.

Threshold explorer

See where trigger rules are explicit, partial, missing or not applicable.

Signal → action

Follow how readings become alerts, escalation and care responses.

Why did 5,509 records become 31 papers?

Systematic reviews start broad on purpose. The aim is to avoid missing relevant evidence, then narrow carefully to the studies that answer the exact registered question.

1. Search widely

The first search captures thousands of possible records across remote monitoring, sensors, wearables, older adults, community care and deterioration.

2. Exclude what does not fit

Records are removed when they are the wrong setting, wrong population, inpatient-only, simulation-only, algorithm-only, or do not include a real patient population.

3. Keep the papers that answer this question

The final included papers are those that examine sensor-derived measures, thresholds, alerts, escalation or deterioration detection outside hospital.

Evidence filtering, not evidence loss

The narrow final set is the point: this project is not asking whether sensors can collect data. It is asking which sensor-derived measures are meaningful enough to inform care for older adults at home, in community care, supported living, aged care or virtual-ward contexts.

Which measures look meaningful?

A cautious, evidence-weighted view of the current extraction. Select a numbered measure on the map to see the interpretation immediately alongside it.

Measure evidence map

Position shows threshold clarity and clinical actionability. Dot size shows validation confidence. The selected interpretation appears directly beside the matrix.

Pathway-ledAction exists, but the signal is still weakly defined.
High potentialThe clearest current candidates for meaningful measures.
EmergingEarly or limited current support.
Defined but limitedA clearer signal, but weaker action pathway.
Threshold clarity →
Clinical actionability →

Evidence-weighted interpretation

Rubric used for this synthesis

  1. Threshold clarity: whether a threshold, score rule or abnormality definition is extractable.
  2. Clinical actionability: whether findings connect to review, triage, escalation or a care response.
  3. Validation confidence: whether model performance, deterioration, alert burden or decision-use evidence is reported.
  4. Evidence breadth: number and diversity of extracted studies contributing to the measure.

Threshold explorer

Thresholds are the hinge between passive sensing and clinical action. This view shows where thresholds are explicit, partial, not reported or not applicable.

Filter threshold rows

Threshold reporting profile

Missingness is not a data cleaning problem here — it is part of the finding. The site should make inconsistent threshold reporting visible.

0%

Signal-to-action pathway

Data becomes meaningful when the path from signal to threshold, alert, review and response is clear enough to trust.

Evidence snapshot

A compact, provisional view of the structured extraction currently powering the interactive atlas. It shows what has been made machine-readable so far, not the final weight of evidence across the whole review.

How to read this snapshot: the top cards describe the review funnel and the expanded structured evidence layer. The bars below show what the completed atlas can currently “see”: which measure families appear most often, and which studies report extractable validation statistics. These are evidence signals, not clinical recommendations.

Measures captured

Counts of study records in the current atlas subset that report each measure family. These are prevalence counts, not recommendations.

Extractable validation results

Model-performance values reported in the expanded structured evidence layer, shown as ROC-AUC where available. Repeated entries from one citation mean that paper reported several candidate models or signals.

Study registry

The registry keeps the full 31-paper review visible while distinguishing papers already structured for interaction from papers awaiting full extraction.

Interpretation note: the interactive layer should be read as a structured evidence view of the extracted records, not as a replacement for the full systematic review.
Show registry table

Workbook tabs as layers

The workbook tabs are mirrored as interactive layers for review and checking.

Show raw layer table

Common questions

Short answers to the questions most likely to come up.

Is this a monitoring product?

No. The microsite explains a research and validation project. The aim is to understand which measures are meaningful enough to inform future care models.

Are individuals being tracked here?

No. The public site visualises review and extraction logic. Any future participant-facing work would require appropriate consent, governance and ethics approval.

Why involve industry partners?

Because they are already delivering sensing, analytics, care-at-home, integration and living-environment capability. They bring the real-world problem the research needs to answer.

Can the site say which measure is best?

Now substantially stronger. The expanded corpus supports an evidence-weighted interpretation of which signals currently look most meaningful, while still avoiding claims of universal clinical cut-offs.

Methods & transparency

The Dictionary tab becomes the microsite’s source of definitions, tooltips and data-confidence rules. The Meta-Analysis Ready tab is treated as the evidence engine.

Evidence engine

The Meta-Analysis Ready sheet contains fields across analytic rows. It should power advanced filters and future synthesis.

Dictionary preview

Use these definitions as rollover/help text in the next build.