Skip to content
This is NOT an official site of the Government of Canada. Click here for the official AI registry.

Helpful Places

Government of Canada – DTPR AI Register

AI-Assisted Physiological Stress Detection from Facial and Radar Data

Healthcare · Employment & Work

What it collects that can identify you

Biometric
Identifiable data
  • Facial imagery and millimetre-wave (mmWave) radar signals captured from individuals without physical contact. Facial data is directly identifiable.

Also collects sensitive personal information, which is anonymized data.

Run by
Canadian Space Agency (CSA)
Where
No fixed location
Kept
Not stated by the Helpful Places.
Shared with
Accountable organization, Vendor
Your copy
You cannot see the data it holds about you. What you can do

What it is for

HealthcareEmployment & Work

This system uses machine learning to estimate physiological stress in individuals by analysing facial video and millimetre-wave (mmWave) radar signals — all without physical contact. It is being developed for use with Government of Canada employees by a vendor (Thales) on behalf of the Canadian Space Agency. Because it processes facial imagery and health-related signals, it involves personal information, and the use of AI is disclosed to users.

What it collects and what happens to it

Data taken in

Biometric
Identifiable data
  • Facial imagery and millimetre-wave (mmWave) radar signals captured from individuals without physical contact. Facial data is directly identifiable.
Sensitive personal information
Anonymized data
  • Health database and de-identified datasets used for model training. The register notes these training sources are de-identified.

Processing

Computer Vision
  • Computer vision processes facial video frames to detect physiological signals associated with stress.
Classification & Prediction
  • Deep learning and machine learning models classify or score physiological stress from fused facial and mmWave inputs.

What it does

Sensing (Perceptive AI)
Human decides
  • The system senses physiological signals from facial video and mmWave radar, converting raw sensor data into structured stress indicators. The system is still in development; human review of outputs is assumed but not explicitly documented.
Deciding (Analytical AI)
Human decides
  • Machine learning and deep learning models score or classify stress levels from the sensed data. Results are advisory — a human is expected to decide on any follow-up action.

Outputs

A recommendation or prediction
Anonymized data
  • The system produces stress scores or classifications as advisory outputs. No binding decision about an individual is described; outputs are intended to inform human reviewers.

Run by

Canadian Space Agency (CSA)
  • The Canadian Space Agency is the federal department accountable for deploying this AI system for use with Government of Canada employees.

Government of Canada AI Register — CARPS (2526-CSA-ASC-001)

Built by

Thales
  • Thales is the vendor that developed the CARPS system under contract to the Canadian Space Agency.

Government of Canada AI Register — CARPS (2526-CSA-ASC-001)

Kept for

Not stated by the Helpful Places.

Shared with

Available to the accountable organization
  • Stress monitoring outputs are expected to be available to the Canadian Space Agency as the deploying organization. Specific access control details are not documented in the register.
Available to vendor
  • Thales, as the vendor, has access to system outputs and data during the development phase. Post-deployment access is not documented.
Not available to me
  • Individual employees do not appear to have direct access to their own stress monitoring records based on available register information. No self-access mechanism is described.

Stored

Not stated by the Helpful Places.

How to read the colours

Can it identify you?

Anonymized data
Data about people with the link to who is broken. Stripped of identifiers, blurred, aggregated, or noised so this system can’t reasonably tie a record back to an individual.
Pseudonymous data
Each person’s data is tied to a token (hash, ID, template) that lets this system recognise the same person across events, but the token itself doesn’t reveal a name. Reidentification is possible with extra information.
Identifiable data
The data either contains a direct identifier (name, address, account name, recognisable face or voice, plate number) or carries a token this system uses to look up legal identity during processing.

Who completes the loop?

Human decides
This mode suggests; a person decides what to do next. The AI is always advisory — a human is in the loop on every decision. Example: a triage tool ranks cases for a clinician who chooses which to see first.
Human executes
This mode decides; a person carries out the result. Example: an optimizer plans the day’s trash-collection routes, and drivers run them.
Autonomous
This mode decides and acts on its own. No person reviews each decision or carries out the resulting action.

Definitions from the DTPR standard. Amber is about your data, violet about who decides. The fuller the shape and the deeper the colour, the more identifying the data or the less a person is involved.

What you can do

Ask about this system

Questions go to the Helpful Places, not the vendor.

Your rights

  • Right to Be Informed of AI UseThe register records that AI use is disclosed to users (Government of Canada employees). Employees are informed that an AI system is in operation. Further details on how disclosure is made are not available in the register entry.
  • Right to a Human ReviewGiven that the system's outputs are framed as advisory stress scores rather than binding decisions, and given the workplace health context, employees should have the right to have any stress assessment reviewed by a human before consequential action is taken. This right is not explicitly documented in the register.
  • Right to Non-discriminationPhysiological stress detection models trained on health databases may perform unevenly across demographic groups. Employees have the right not to be discriminated against on the basis of race, age, disability, or other protected characteristics as a result of differential model accuracy. No bias audit or fairness testing is documented in the register entry.

Risks and safeguards

  • Psychological harmContinuous physiological stress monitoring in a workplace setting carries risks of employee anxiety, stigma, and chilling effects on behaviour. The system processes sensitive health-related signals from employees who may feel unable to opt out in a work context. Mitigation details are not documented in the register entry; this risk warrants explicit consent mechanisms, opt-out pathways, and well-being safeguards.
  • Civil liberties harmContactless biometric and physiological surveillance of employees in the workplace raises civil liberties concerns, including the right to privacy and freedom from covert monitoring. Facial data is directly identifiable. The register discloses AI use to employees, which is a baseline mitigation, but no further safeguards such as purpose limitation, data minimisation, or independent oversight are described.
  • Reputational harmMisclassification of stress levels could stigmatise employees — for example, incorrectly flagging a person as highly stressed, with potential downstream effects on performance reviews or work assignments. No specific accuracy thresholds, correction mechanisms, or human review requirements are documented in the register entry.