ASF Standards · Home Care · Standard 10

Standard 10 — Digital Care and Artificial Intelligence

5 criteria · 2 core · 3 standard-level · Version 1.0 · Aligned to ISQua EEA Principle 7, 6th Edition

Criteria in this standard

10.1 — Digital Tools Are Genuinely Evaluated Before Adoption
10.2 — The Service Never Disadvantages a Client Who Cannot Engage With Digital Tools
10.3 — Family Members Have Genuine, Secure Access to Relevant Information
10.4 — AI-Assisted Monitoring Alerts Are Genuinely Validated, Not Trusted by Default
10.5 — Accountability for Digitally-Supported Care Decisions Is Explicitly Defined
10.6 — Clients Are Genuinely Informed When Their Care Involves AI

10.1

Digital Tools Are Genuinely Evaluated Before Adoption

Standard

Any digital tool used to support home care — scheduling systems, remote monitoring devices, caregiver check-in apps — is genuinely evaluated for reliability and appropriateness before adoption, not deployed on the assumption that any functioning app or device is automatically suitable for use in a client’s home.

In plain terms: Before a new app or monitoring device goes into clients’ homes, the provider has actually checked it works reliably there — not just assumed any functioning product is good enough.

Facility category	Crisis	Transition	Small	Standard
Applicability	Adapted	Full	Full	Full

Why this matters

A client’s home has genuinely variable conditions — internet reliability, physical layout, a client’s comfort with technology — that differ meaningfully from a controlled clinical environment. A tool that works well in a demonstration or a facility setting may not genuinely function the same way across the real diversity of client homes.

What good looks like

A tool is genuinely evaluated before introduction.
Evaluation genuinely considers the home setting specifically.
A genuine reporting pathway exists for malfunction.

Common failure modes

A tool is adopted based on vendor claims alone, with no genuine real-world testing in actual homes.

Worked example

In practice

A provider considering a new fall-detection sensor.

BeforeThe vendor’s demonstration was taken as sufficient evidence of reliability, with no genuine testing in actual client homes with varying internet connectivity.

ActionA pilot was run in five genuinely varied client homes before wider rollout, surfacing a connectivity issue in homes with weaker Wi-Fi.

AfterThe Monitor reviewed the pilot results and resulting adjustment. Verified.

If you are starting from zero — do this first

Pilot any new digital tool in a genuinely varied sample of actual client homes before full rollout.

The most common mistake: Relying on a vendor’s demonstration alone, without genuine testing in actual, varied home conditions.

Self-assessment questions

1. Is a tool genuinely evaluated before introduction? — A real, documented evaluation.
Evidence: Evaluation record

2. Does evaluation genuinely consider the home setting specifically? — A real, setting-specific consideration.
Evidence: Pilot results

3. Is there a genuine process to report a malfunctioning tool? — A real, usable feedback pathway.
Evidence: Reporting protocol

Common reasons for a PARTIAL answer

Evaluation happened but wasn’t genuinely tested across varied home conditions.

Implementation plan

When	What
Before any rollout	Pilot in a genuinely varied sample of client homes.

How the Monitor verifies this

Method	What	Detail
DOCUMENT	Pilot review	Reviews pilot results and any resulting adjustment.

Supervisor tips

Ask whether any tool was ever genuinely tested across varied home conditions before rollout.

Evidence base

World Health Organization. Global Strategy on Digital Health 2020-2025. Geneva: WHO; 2021.

ASF training courses on GMJ Academy →

Foundation courses A-00 to A-03 are live. Criterion-specific modules are being developed and will link here when published.

10.2

The Service Never Disadvantages a Client Who Cannot Engage With Digital Tools

Core

A client who cannot or does not want to engage with digital monitoring tools, apps, or portals — common among the elderly or isolated clients this service often serves — genuinely receives the same quality of care as one who does, with no digital requirement ever becoming a real barrier to appropriate home care.

In plain terms: A client who can’t or doesn’t want to use apps or digital monitoring still genuinely gets the same quality of care — digital tools support care but never become a real condition of receiving it.

Facility category	Crisis	Transition	Small	Standard
Applicability	Adapted	Full	Full	Full

Why this matters

This is marked Core because home care genuinely, disproportionately serves elderly and isolated clients — a population genuinely more likely to face real digital access or comfort barriers than the general public. A service that lets digital tools become, even unintentionally, a real condition of quality care risks excluding exactly the population it most needs to serve well.

What good looks like

Care is genuinely equivalent regardless of digital engagement.
Caregivers are genuinely trained to identify digital limitations.
A real, documented instance shows genuine adaptation.

Common failure modes

A client without a smartphone genuinely receives a reduced or delayed version of the service.

Worked example

In practice

A client without a smartphone or reliable internet.

BeforeThe provider’s care update system relied on a client-facing app, which this client simply could not use, genuinely leaving them less informed than other clients.

ActionA parallel phone-call update process was established for this client, genuinely equivalent to the app-based updates others received.

AfterThe Monitor confirmed the client genuinely received equivalent information. Verified.

If you are starting from zero — do this first

Identify every client currently relying on a digital tool for information, and confirm a genuine non-digital equivalent exists.

The most common mistake: A digital update channel treated as the default, with no genuine equivalent for clients who can’t use it.

Self-assessment questions

1. Does a client unable to engage digitally genuinely receive equivalent care? — Real, equivalent care.
Evidence: Care records comparison

2. Are caregivers genuinely trained to identify and accommodate digital limitations? — Real, specific staff awareness.
Evidence: Training record

3. Is there a real, documented instance of care genuinely adapted for such a client? — A real, concrete example.
Evidence: Adaptation record

Common reasons for a PARTIAL answer

A non-digital equivalent exists in theory but isn’t actually offered proactively.

Implementation plan

When	What
Week 1-2	Identify affected clients and establish genuine non-digital equivalents.

How the Monitor verifies this

Method	What	Detail
DOCUMENT	Care records comparison	Compares records for a digitally-engaged and non-digitally-engaged client.

Supervisor tips

Ask about a specific client known not to use digital tools and review their actual care record.

Evidence base

World Health Organization. Global Strategy on Digital Health 2020-2025. Geneva: WHO; 2021.

ASF training courses on GMJ Academy →

Foundation courses A-00 to A-03 are live. Criterion-specific modules are being developed and will link here when published.

10.3

Family Members Have Genuine, Secure Access to Relevant Information

Standard

Where digital tools allow family members to view care updates or monitoring data, access is genuinely secure and genuinely limited to what the client has actually consented to share — not a blanket access arrangement assumed appropriate for every family member regardless of the client’s own wishes.

In plain terms: Family members only see what the client has actually agreed to share with them — not everything by default — and that access is genuinely secure.

Facility category	Crisis	Transition	Small	Standard
Applicability	Adapted	Full	Full	Full

Why this matters

Family involvement in home care is genuinely valuable and common, but a client’s own wishes about what to share — and with whom — deserve genuine respect, not an assumed blanket arrangement. Family dynamics genuinely vary, and a client may want to share different information with different family members, or none at all with some.

What good looks like

Family access is genuinely secure.
Access is genuinely limited to actual client consent.
The client can genuinely revise or withdraw access.

Common failure modes

All listed family members are given full access by default, with no genuine client-level consent step.

Worked example

In practice

A provider whose family portal access was granted by default.

BeforeAny family member listed as an emergency contact automatically received full portal access, with no genuine, separate consent step from the client.

ActionA genuine, explicit consent step was added, allowing the client to specify what each family member could see.

AfterThe Monitor reviewed the consent records. Verified.

If you are starting from zero — do this first

Add an explicit, client-level consent step for family portal access.

The most common mistake: Treating emergency-contact status as equivalent to genuine consent for full information access.

Self-assessment questions

1. Is family access genuinely secure? — Real, verified security.
Evidence: Security review

2. Is this access genuinely limited to actual client consent? — A real, client-driven consent basis.
Evidence: Consent records

3. Can the client genuinely revise or withdraw access? — A real, usable mechanism.
Evidence: Access revision process

Common reasons for a PARTIAL answer

Consent is collected once at intake but can’t actually be revised later.

Implementation plan

When	What
Week 1-2	Build explicit, revisable client consent for family access.

How the Monitor verifies this

Method	What	Detail
DOCUMENT	Consent record review	Reviews consent records for genuine, client-specific scope.

Supervisor tips

Ask whether a client has ever genuinely limited or withdrawn a family member’s access.

Evidence base

European Union. General Data Protection Regulation (GDPR), Article 7: Conditions for Consent. Brussels: European Union; 2016.

ASF training courses on GMJ Academy →

Foundation courses A-00 to A-03 are live. Criterion-specific modules are being developed and will link here when published.

10.4

AI-Assisted Monitoring Alerts Are Genuinely Validated, Not Trusted by Default

Standard

Where an AI-assisted monitoring system (such as a fall-detection sensor) generates an alert, there is a genuine process for appropriate human validation and response — not an automated alert trusted and acted on, or dismissed, with no real human judgement genuinely involved.

In plain terms: When a sensor or AI system flags something — like a possible fall — a real person actually checks and decides what to do, rather than the alert being blindly trusted or blindly ignored.

Facility category	Crisis	Transition	Small	Standard
Applicability	Adapted	Full	Full	Full

Why this matters

AI-assisted monitoring in a client’s home — often unsupervised by clinical staff for long stretches — carries genuinely real stakes if an alert is wrongly dismissed as a false positive, or if a false positive leads to unnecessary, distressing intervention. Genuine human validation is what keeps this technology a support to good judgement rather than a replacement for it.

What good looks like

A genuine human validation process exists.
Alert accuracy is genuinely tracked.
A real instance shows the process catching an unreliable alert.

Common failure modes

Alerts trigger an automatic call-out with no genuine human review of the alert’s plausibility first.

Worked example

In practice

A provider using a fall-detection sensor with a high false-positive rate.

BeforeEvery alert triggered an automatic emergency call-out, with no genuine tracking of how many were false positives, causing real, repeated distress to the client.

ActionA brief human review step was added — a call to the client or a check of a secondary indicator — before escalation, and false-positive rates began being genuinely tracked.

AfterThe Monitor reviewed the tracked accuracy data showing a genuine reduction in unnecessary escalations. Verified.

If you are starting from zero — do this first

Add a brief human validation step before any AI-triggered alert escalates.

The most common mistake: Automatic escalation with no genuine human check, leading to repeated, distressing false alarms.

Self-assessment questions

1. Is there a genuine human validation process for an alert? — Real, genuine human involvement.
Evidence: Alert response protocol

2. Is alert accuracy genuinely tracked, including false positives and false negatives? — A real, ongoing accuracy review.
Evidence: Accuracy data

3. Is there a real, documented instance of this process genuinely catching an unreliable alert? — A real, concrete example.
Evidence: Incident record

Common reasons for a PARTIAL answer

A human validation step exists but false-positive rates aren’t actually tracked over time.

Implementation plan

When	What
Week 1-2	Build a human validation step and begin tracking alert accuracy.

How the Monitor verifies this

Method	What	Detail
DOCUMENT	Accuracy data review	Reviews tracked alert accuracy over time.

Supervisor tips

Ask what actually happens in the seconds after an alert fires, step by step.

Evidence base

US Food and Drug Administration. Artificial Intelligence and Machine Learning in Software as a Medical Device. Silver Spring: FDA; 2023.

ASF training courses on GMJ Academy →

Foundation courses A-00 to A-03 are live. Criterion-specific modules are being developed and will link here when published.

10.5

Accountability for Digitally-Supported Care Decisions Is Explicitly Defined

Core

The provider has genuinely considered and documented accountability for care decisions informed by digital monitoring or AI-assisted tools — who is responsible when a tool fails or misleads — not leaving this as an unexamined question until an actual incident forces an answer in a client’s own home, where no clinical team is physically present to intervene.

In plain terms: Everyone knows, in advance, who’s actually responsible if a monitoring tool fails or misleads — especially important since there’s no clinical team on-site in the client’s home to step in.

Facility category	Crisis	Transition	Small	Standard
Applicability	Adapted	Full	Full	Full

Why this matters

This is marked Core because the absence of on-site clinical staff in a client’s home genuinely raises the stakes of a digital tool’s failure — there is no physically present team to immediately recognise and correct a problem the way there might be in a facility. Explicit, documented accountability is what ensures a real, prompt response rather than confusion about who is actually responsible when something goes wrong.

What good looks like

Accountability is explicitly documented.
Caregivers genuinely understand their own accountability.
A defined process exists for reviewing accountability in an incident.

Common failure modes

Accountability has never been genuinely examined until an actual incident forces the question.

Worked example

In practice

A provider whose monitoring tool failed to alert on a genuine fall.

BeforeAccountability for a tool failure had never been explicitly documented, leaving real confusion about responsibility when the incident was reviewed.

ActionA clear policy was developed explicitly confirming the caregiver’s and provider’s own accountability for a tool’s failure, alongside any vendor liability.

AfterThe Monitor reviewed the policy and interviewed a caregiver who could clearly describe their own accountability. Verified.

If you are starting from zero — do this first

Document accountability for digitally-supported care decisions explicitly, before an incident forces the question.

The most common mistake: Leaving accountability genuinely unexamined until an actual incident forces an answer under pressure.

Self-assessment questions

1. Is accountability for a digitally-supported care decision explicitly documented? — A real, clear answer.
Evidence: Documented accountability policy

2. Do caregivers genuinely understand their own accountability? — Real, demonstrated understanding.
Evidence: Staff interview

3. Is there a defined process for reviewing accountability in a safety incident? — A real, usable process.
Evidence: Incident review protocol

Common reasons for a PARTIAL answer

A policy exists but caregivers genuinely haven’t been told about it.

Implementation plan

When	What
Week 1-2	Document accountability explicitly and communicate it to all caregivers.

How the Monitor verifies this

Method	What	Detail
ASK	Staff interview	Asks a caregiver to describe their accountability for a digital tool’s output.

Supervisor tips

Ask a caregiver directly: “If the sensor missed a real fall, who’s responsible?”

Evidence base

European Commission. Ethics Guidelines for Trustworthy Artificial Intelligence. Brussels: European Commission; 2019.

ASF training courses on GMJ Academy →

Foundation courses A-00 to A-03 are live. Criterion-specific modules are being developed and will link here when published.

10.6

Clients Are Genuinely Informed When Their Care Involves AI

Standard

A client whose care involves an aspect delivered with AI-assisted support is genuinely informed of this — not left to assume every decision was made by their caregiver alone, with disclosure treated as optional rather than a genuine, standard part of informed home care.

In plain terms: If AI is actually involved in part of a client’s care — a fall-detection sensor, for instance — the client or their family genuinely knows this, not left to assume everything is purely caregiver-monitored.

Facility category	Crisis	Transition	Small	Standard
Applicability	Adapted	Full	Full	Full

Why this matters

A client’s or family’s genuine right to understand care includes knowing, in plain terms, when an AI-assisted monitoring tool is actually part of the care arrangement rather than purely human observation. This is especially relevant in home care, where AI-assisted monitoring often runs continuously and unseen in the background — easy for a client or family to never genuinely realise is there at all unless actively disclosed.

What good looks like

Clients and families are genuinely informed when AI is involved in care.
Disclosure is genuinely understandable, in plain language.
A client or family member can genuinely confirm they were told.

Common failure modes

AI-assisted monitoring runs in the background with no genuine disclosure ever made to the client or family.

Worked example

In practice

A provider using an AI-assisted fall-detection sensor in a client’s home.

BeforeThe sensor’s role was mentioned only in a lengthy general service agreement, which the family signed without genuine awareness that an AI tool was actually monitoring their relative.

ActionA brief, plain-language conversation at the start of care now explains: “A sensor helps alert your caregiver to a possible fall; your caregiver always checks before responding.”

AfterThe Monitor interviewed a family member who could genuinely confirm they understood AI was involved. Verified.

If you are starting from zero — do this first

Add a brief, plain-language disclosure point at the start of care for any AI-assisted monitoring in use.

The most common mistake: AI-assisted monitoring that runs continuously with no genuine disclosure ever actually reaching the client or family.

Self-assessment questions

1. Is a client (or family, where the client cannot meaningfully participate) genuinely informed when AI-assisted monitoring is part of their care? — Real, standard disclosure.
Evidence: Disclosure protocol

2. Is this disclosure genuinely understandable, not buried in a general service agreement? — A real, plain-language explanation.
Evidence: Service agreement wording

3. Can a client or family member asked directly confirm they were genuinely told? — A real, concrete confirmation.
Evidence: Family interview

Common reasons for a PARTIAL answer

Disclosure exists in the service agreement but families genuinely cannot recall or explain it when asked directly.

Implementation plan

When	What
Week 1-2	Draft plain-language disclosure wording for the care start conversation and train caregivers to deliver it.

How the Monitor verifies this

Method	What	Detail
ASK	Family interview	Asks a family member whether they were genuinely told AI was involved in their relative’s care.

Supervisor tips

Ask a family member directly rather than relying on the service agreement’s existence alone.

Evidence base

World Health Organization. Ethics and Governance of Artificial Intelligence for Health: Guidance on Large Multi-Modal Models. Geneva: WHO; 2024.

ASF training courses on GMJ Academy →

Foundation courses A-00 to A-03 are live. Criterion-specific modules are being developed and will link here when published.

← Home Care Standards — all standards
Self-assess against these criteria →
← Standard 9 — Sustainable Care
Standard 11 — Supporting the Care Workforce →