Connected Health and Wearables · Study deck

Wearable IoT: Privacy and Power Constraints

Body data can expose health or routine long after a wearer removes the device.

Ada is your guide for this deck.

applicationdomainswearables
iotclass.org

After studying this chapter

Learning objectives

You will be able to:

  • Explain: The review should keep raw vital signs, filtered features, decision rules, alert routing, and remote-access permissions separate so a monitoring system does not confuse a local wellness display with clinical escalation.
  • Explain: Pairing, account recovery, firmware signing, secure boot where available, encrypted storage, key rotation, and revocation must be tested against lost phones, second-hand device resale, caregiver access, and shared household devices.
  • Explain: The useful evidence is not just "breathing detected"; it is placement, sample rate, filtering, correlation window, motion artifact state, and the conditions where the respiratory-rate estimate is withheld.
iotclass.org

Major section

Body Data Needs Strong Boundaries

The architecture should keep raw sensor data, derived metrics, coaching messages, emergency alerts, and clinical claims separate.

  • A step trend, stress score, arrhythmia notification, glucose alarm, and workplace fatigue alert need different validation and governance.
  • The signal pipeline should preserve enough context to explain every metric.
  • Respiratory wearables show why validation evidence must name the reference signal.

Key terms

What privacy rule
What privacy rule is needed before launch?", "options": [ { "text": "Give managers every raw body and location record without purpose limits.", "correct": false, "feedback": "Incorrect.
iotclass.org

Major section

Body Data Needs Strong Boundaries (continued)

On-device processing protects latency and privacy for simple classification, artifact detection, and immediate alerts, but it is limited by battery, compute, memory, and thermal budget.

  • An IMU event may need sampling rate, placement, orientation, activity class, fall-detection confidence, and whether a confirmation prompt was answered.
  • A glucose event may need sensor age, calibration status, trend arrow, alarm state, and delivery path.
  • A phone can handle richer user interaction, buffering, secure sync, and local notifications.
iotclass.org

Major section

Body Data Needs Strong Boundaries (continued)

Cloud services can support longitudinal analytics, population-level model improvement, clinician portals, and research exports, but they increase privacy, security, availability, and consent obligations.

  • The strongest wearable design can trace a metric from body signal through firmware, phone app, model, uncertainty label, user action, and support or clinical workflow without pretending that a convenient sensor is automatically a clinical instrument.
  • A dance-sensing garment or body-area network node has to prove placement, motion artifact handling, short-range relay behavior, and whether the phone or tablet receives enough quality metadata to interpret the movement.
  • A fall alert needs conservative escalation and cancellation behavior.
Wearable IoT Ecosystem Architecture - Data flow from body-worn sensors through phone and cloud processing to a clinician portal
Wearable IoT Ecosystem Architecture - Data flow from body-worn sensors through phone and cloud processing to a clinician portal
iotclass.org

Major section

Body Data Needs Strong Boundaries (continued)

The review should keep raw vital signs, filtered features, decision rules, alert routing, and remote-access permissions separate so a monitoring system does not confuse a local wellness display with clinical escalation.

  • The useful evidence is not just "breathing detected"; it is placement, sample rate, filtering, correlation window, motion artifact state, and the conditions where the respiratory-rate estimate is withheld.
  • The architecture should show which layer owns each decision and what happens when BLE, cellular, or cloud access is unavailable.
  • Workplace wearables need a BYOW or BYOD policy before the pilot scales.
iotclass.org

Major section

Body Data Needs Strong Boundaries (continued)

Keeping those paths separate lets one wearable platform support multiple features without overclaiming what the body signal can prove.

  • Pairing, account recovery, firmware signing, secure boot where available, encrypted storage, key rotation, and revocation must be tested against lost phones, second-hand device resale, caregiver access, and shared household devices.
  • The engineering discipline is to keep wellness, safety, and clinical paths from blurring.
  • A wellness score can tolerate uncertainty if the app communicates trends and confidence.
Nine wearable-design principles connect first-use adoption, long-term wear, ecosystem integration, routine fit, and clear utility.
Nine wearable-design principles connect first-use adoption, long-term wear, ecosystem integration, routine fit, and clear utility.
iotclass.org

Major section

Wearable Computing History

2020s -- Medical-grade wearables: FDA-approved ECG, SpO2, and CGM features reach consumer devices.

  • Figure: Wearable IoT Ecosystem Architecture - Data flow from shows what the word wearable now covers, which is far more than the thing on the wrist.
  • The phone tier merges, compresses, and encrypts before anything leaves the user's hands, and it is also where sharing is configured.
  • The cloud tier adds models and a store with clinical obligations attached.
Wearable IoT Ecosystem Architecture - Data flow from body-worn sensors through phone and cloud processing to a clinician portal
Wearable IoT Ecosystem Architecture - Data flow from body-worn sensors through phone and cloud processing to a clinician portal
iotclass.org

Deck summary

Key takeaways

The architecture should keep raw sensor data, derived metrics, coaching messages, emergency alerts, and clinical claims separate.

  • On-device processing protects latency and privacy for simple classification, artifact detection, and immediate alerts, but it is limited by battery, compute, memory, and thermal budget.
  • Cloud services can support longitudinal analytics, population-level model improvement, clinician portals, and research exports, but they increase privacy, security, availability, and consent obligations.
  • The review should keep raw vital signs, filtered features, decision rules, alert routing, and remote-access permissions separate so a monitoring system does not confuse a local wellness display with clinical escalation.
iotclass.org

Retrieval practice

Recall check 1 of 4

Ada says: answer from memory, then check your reasoning.

Q1An employer offers a fitness wearable to staff. What privacy rule is needed before launch?

AGive managers every raw body and location record without purpose limits.
BKeep employee records forever because digital storage is inexpensive.
CLimit access by purpose and consent, with deletion and opt-out rules.
DTreat silence as consent to share wearable records with insurers.
Show answer

Answer: C The rule sets who may see data and how a worker can refuse or erase it.

iotclass.org

Retrieval practice

Recall check 2 of 4

Ada says: answer from memory, then check your reasoning.

Q2A wearable uploads a heart-rate metric after exercise. What context helps explain whether it is trustworthy?

AContact quality, motion, window, and algorithm context
BThe numeric value without its sampling window
CThe account’s plan name instead of sensor quality
DThe latest cloud upload time as the activity state
Show answer

Answer: A The chapter preserves these details along with timestamp and activity state.

iotclass.org

Retrieval practice

Recall check 3 of 4

Ada says: answer from memory, then check your reasoning.

Q3A team chooses a wrist PPG device for intense activity. Which issue should it test before strengthening its claim?

AChest-strap comfort as the wrist signal’s accuracy proof
BThe sensor part number without on-body evidence
CAn app’s trend graph as proof of clinical fitness
DMotion and contact artifacts at the chosen placement
Show answer

Answer: D The section describes wrist PPG as vulnerable to motion and fit effects.

iotclass.org

Retrieval practice

Recall check 4 of 4

Ada says: answer from memory, then check your reasoning.

Q4A wearable has 600 mAh usable capacity. It draws 2 mA for 20 hours/day and 30 mA for 4 hours/day. A radio burst also pulls its rail below the minimum voltage. Which estimate and decision fit?

A40 mAh/day, 15 days, counting the 2 mA mode but omitting 30 mA hours.
B160 mAh/day, 3.75 ideal days, and a separate fix for radio rail sag.
C160 mAh/day, 3.75 days, because that runtime also establishes the rail is safe.
D600 mAh/day, one day, mistaking capacity for daily consumption.
Show answer

Answer: B 20 × 2 + 4 × 30 = 160 mAh/day and 600/160 = 3.75 ideal days; capacity math does not waive the failed peak-load rail test.

iotclass.org

Print reference

Answers

Answer key.

  1. C · The rule sets who may see data and how a worker can refuse or erase it.
  2. A · The chapter preserves these details along with timestamp and activity state.
  3. D · The section describes wrist PPG as vulnerable to motion and fit effects.
  4. B · 20 × 2 + 4 × 30 = 160 mAh/day and 600/160 = 3.75 ideal days; capacity math does not waive the failed peak-load rail test.
iotclass.org