37  Wearable Technology

37.1 Learning Objectives

By the end of this section, you will be able to:

• Calculate effective detection rates for wearable health devices using the formula Effective Detection = Sensor Accuracy x User Compliance
• Evaluate sensor placement trade-offs between signal quality, user comfort, and wear compliance across wrist, chest, ear, and patch form factors
• Design adaptive sampling strategies that extend battery life 10x (0.1 Hz sleep to 25 Hz exercise) while maintaining clinically relevant accuracy thresholds
• Compare wearable market segments (consumer fitness, medical-grade, industrial safety) and map regulatory requirements (FDA 510(k), CE Mark, ATEX) to design decisions
• Analyze privacy implications of continuous physiological data collection, including data minimization, on-device processing, and GDPR/HIPAA compliance requirements

37.2 Wearable Technology

Time: ~20 min | Level: Intermediate | Unit: P03.C03.U02

Key Concepts

• Photoplethysmography (PPG): Optical heart rate sensing detecting blood volume changes under the skin using an LED and photodetector.
• Inertial Measurement Unit (IMU): Sensor combining accelerometer and gyroscope to measure motion for step counting, fall detection, and gesture recognition.
• Heart Rate Variability (HRV): Time variation between successive heartbeats; low HRV correlates with stress and reduced recovery, tracked by advanced fitness devices.
• Continuous Glucose Monitor (CGM): Subcutaneous sensor measuring interstitial glucose every 5 minutes, eliminating finger-prick tests for diabetes management.
• Activity Classification: On-device ML model distinguishing walking, running, and sleep from raw IMU data without cloud round-trips.
• Form Factor Constraint: Physical size and weight limit that directly constrains battery capacity, sensor count, and achievable battery life.
• Skin Conductance (EDA): Electrodermal activity sensor measuring sweat gland response as a proxy for stress, used in emotion-sensing wearables.

Minimum Viable Understanding

• Compliance dominates accuracy: A 99%-accurate chest strap worn 45% of the time detects fewer events than an 85%-accurate wrist sensor worn 92% of the time (Effective Detection = Accuracy x Compliance)
• Adaptive sampling extends battery 10x: Use accelerometer (0.01 mA) as wake-up sensor to scale PPG sampling from 0.1 Hz during sleep to 25 Hz during exercise, achieving 14+ day battery life on 180 mAh
• On-device data reduction of 99.8%: Raw sensor data at 5 KB/s is reduced to 10 B/s processed features before BLE transmission, which is what makes week-long battery life possible

Sensor Squad: Wearable Fitness Tracker Adventure

Wearable technology is like having a superhero sidekick that you wear on your body!

One morning, 10-year-old Maya strapped on her new fitness tracker watch. Little did she know, inside that tiny watch lived the whole Sensor Squad!

Max (motion sensor) was the first to spring into action. “She’s walking to school… 100 steps… 500 steps… I can feel every single one!” As Maya started running to catch up with her friends, Max got even more excited: “Running detected! Her feet are hitting the ground harder and faster – time to tell the others!”

Bella (bio sensor) was keeping watch on Maya’s heartbeat. She shone a tiny green light into Maya’s wrist and counted the pulses. “Heart rate is 72 beats per minute during the walk… now it’s climbing to 110 as she runs! Everything is looking healthy!” Bella knew that the green light bounces back differently each time the heart pumps blood – that is how she can count heartbeats right through the skin.

Meanwhile, Sammy (sound sensor) was listening carefully. “I can hear the environment around Maya – she’s outside near traffic now, but moving to the quieter school hallway.” And Lila (light sensor) was measuring the sunlight hitting Maya’s wrist. “Bright sunshine outside! I’ll let the screen know it needs to get brighter so Maya can read her step count.”

Deep inside the watch, a tiny battery manager carefully measured how much energy everyone was using. “We’ve got 73% battery left – plenty of power for the whole day! Max, you use almost no power. But Bella, your green light uses the most – so let’s turn it down when Maya is sitting in class and not moving.”

At the end of the day, Maya looked at her watch and smiled: 8,247 steps, 2 hours of activity, and her heart rate was perfect. The Sensor Squad had been watching over her all day long!

37.2.1 Key Words for Kids

Word	What It Means
Wearable	A smart device you wear on your body, like a watch or glasses, that can sense things about you
Fitness Tracker	A wearable that counts your steps, measures your heart rate, and helps you stay healthy
Heart Rate	How many times your heart beats in one minute – the sensor feels the tiny pulses in your wrist
Bluetooth	Invisible radio waves that let devices talk to each other over short distances
Accelerometer	A sensor that can tell when you’re moving, how fast, and in which direction

37.2.2 Try This at Home!

Make a Human Step Counter!

1. Find a friend or family member to be your “wearable”
2. Have them close their eyes and put one hand on their chest
3. Walk around the room while they count your footsteps just by feeling the floor vibrations
4. Now jump, skip, and tiptoe – can they tell the difference in movements?

This is exactly what Max does inside a fitness tracker! The accelerometer sensor feels tiny movements and figures out if you’re walking, running, or sitting still. Pretty cool, right?

For Beginners: What Makes Wearable Technology Work?

A wearable device is any small computer you wear on your body – a smartwatch, a fitness band, a ring, or even a patch stuck to your skin. These devices contain tiny sensors that measure things about your body: how fast your heart beats, how many steps you take, and even how well you sleep.

Here is the basic idea in three steps:

1. Sense: The device uses sensors to collect data from your body. For example, a green LED light shines into your wrist and a detector reads how much light bounces back. Changes in reflected light reveal your pulse because blood absorbs green light differently with each heartbeat.
2. Process: A small computer chip inside the device turns the raw sensor signals into useful numbers – your heart rate in beats per minute, your step count, or your blood oxygen level.
3. Share: The device sends those numbers to your phone using Bluetooth (a short-range wireless signal). Your phone app then shows graphs, trends, and alerts.

The biggest challenge in wearable design is not making the sensors more accurate. It is making the device comfortable enough that people actually wear it. A perfect medical sensor that sits in a drawer is less useful than a simple wrist band that someone wears every day. This trade-off between accuracy and comfort is the central theme of this chapter.

One number to remember: wearable devices throw away 99.8% of the raw data they collect, keeping only the important results (like your heart rate number). This extreme data reduction is what allows a tiny battery to last a full week.

37.3 How It Works: Wrist-Based Heart Rate Monitoring

How It Works: Measuring Your Pulse Through Your Skin

The big picture: Wearable fitness trackers measure heart rate without electrodes by shining green LED light into your wrist and detecting tiny changes in blood volume with each heartbeat.

Step-by-step breakdown:

1. Light emission: Green LEDs (525 nm wavelength) shine into wrist capillaries at 0.5-2.5 mA power. Real example: Fitbit uses two green LEDs pulsing 25-100 times per second to measure reflected light.

2. Blood volume detection: Photodiode measures reflected light - blood absorbs green light, so reflection decreases during heartbeat peaks when more blood flows. Real example: Each heartbeat creates 0.5-2% change in reflected light intensity; signal processing filters this from noise.

3. Adaptive sampling: Accelerometer (0.01 mA) detects activity state and adjusts PPG sampling from 0.1 Hz during sleep to 25 Hz during exercise, achieving 10x battery extension. Real example: 7-day battery life on 180 mAh by spending 95% of time in low-power sleep mode.

Why this matters: Wrist placement achieves 70-85% accuracy with 92% user compliance (worn daily), while chest straps achieve 95-99% accuracy but only 45% compliance. Effective detection = Accuracy × Compliance: wrist wins at 78% vs chest at 52%.

37.4 Wearable Sensor Placement Strategy

Understanding optimal sensor placement is critical for wearable IoT devices. Different body locations provide access to different physiological signals with varying signal quality and user comfort trade-offs.

Figure 37.1: Graph diagram showing wearable sensor placement strategy with three main body regions and trade-offs between signal quality and user comfort

Alternative View: Wearable Sensor Selection Decision Tree

This diagram helps designers choose sensor placement based on measurement goals. The key trade-off is always accuracy vs. compliance - the “best” sensor is useless if users won’t wear it.

Figure 37.2: Decision tree for wearable sensor placement starting with measurement type needed

For vital signs: chest strap gives 95-99% accuracy but only 60% compliance; wrist gives 70-85% accuracy but 95% compliance. For activity: wrist achieves 95%+ step accuracy. For biochemical: contact lenses are R&D stage, skin patches achieve 85-95% accuracy.

Wearable Sensor Placement Strategy: Different body locations offer trade-offs between signal quality, user comfort, and practical deployment. Head and chest provide the highest signal quality for physiological monitoring (ECG, EEG, glucose), while wrist and finger-worn devices prioritize user acceptance and continuous wear compliance. Smart clothing and shoes enable activity tracking without requiring users to remember to wear additional devices.

viewof sensorAccuracy = Inputs.range([50, 100], {
  value: 85,
  step: 1,
  label: "Sensor Accuracy (%)"
})

viewof userCompliance = Inputs.range([30, 100], {
  value: 92,
  step: 1,
  label: "User Compliance (%)"
})

{
  const effectiveDetection = (sensorAccuracy / 100) * (userCompliance / 100) * 100;
  
  const width = 600;
  const height = 250;
  const margin = {top: 30, right: 20, bottom: 50, left: 60};
  
  const svg = d3.create("svg")
    .attr("width", width)
    .attr("height", height)
    .attr("viewBox", [0, 0, width, height])
    .attr("style", "max-width: 100%; height: auto; font-family: Arial, sans-serif;");
  
  // Title
  svg.append("text")
    .attr("x", width / 2)
    .attr("y", 20)
    .attr("text-anchor", "middle")
    .attr("font-size", "16px")
    .attr("font-weight", "bold")
    .attr("fill", "#2C3E50")
    .text("Effective Detection Rate Calculator");
  
  // Result box
  const resultBox = svg.append("g")
    .attr("transform", `translate(${width/2}, ${height/2})`);
  
  resultBox.append("rect")
    .attr("x", -120)
    .attr("y", -40)
    .attr("width", 240)
    .attr("height", 80)
    .attr("rx", 8)
    .attr("fill", effectiveDetection >= 75 ? "#16A085" : effectiveDetection >= 50 ? "#E67E22" : "#E74C3C")
    .attr("opacity", 0.2);
  
  resultBox.append("text")
    .attr("text-anchor", "middle")
    .attr("y", -10)
    .attr("font-size", "14px")
    .attr("fill", "#2C3E50")
    .text("Effective Detection Rate");
  
  resultBox.append("text")
    .attr("text-anchor", "middle")
    .attr("y", 25)
    .attr("font-size", "32px")
    .attr("font-weight", "bold")
    .attr("fill", effectiveDetection >= 75 ? "#16A085" : effectiveDetection >= 50 ? "#E67E22" : "#E74C3C")
    .text(`${effectiveDetection.toFixed(1)}%`);
  
  // Formula display
  svg.append("text")
    .attr("x", width / 2)
    .attr("y", height - 20)
    .attr("text-anchor", "middle")
    .attr("font-size", "12px")
    .attr("fill", "#7F8C8D")
    .text(`Formula: ${sensorAccuracy}% × ${userCompliance}% = ${effectiveDetection.toFixed(1)}%`);
  
  return svg.node();
}

Interactive Calculator: Effective Detection Rate - Adjust sensor accuracy and user compliance to see how they multiply to produce the effective detection rate. This calculator demonstrates why a comfortable 85% accurate wrist sensor worn 92% of the time (effective detection: 78.2%) outperforms an uncomfortable 99% accurate chest strap worn only 45% of the time (effective detection: 44.6%).

37.5 Wearable Data Architecture

A wearable IoT system involves multiple processing layers between the on-body sensor and the clinical dashboard or user app. Understanding this pipeline is critical because each layer introduces latency, power cost, and potential data loss.

Figure 37.3: Flowchart showing four-layer wearable data architecture from on-body sensors through edge processing to cloud systems

Wearable IoT data pipeline from on-body sensors through edge processing, smartphone gateway, to cloud clinical systems. Each layer reduces data volume: raw sensor data at 5 KB/s is compressed to processed features at 50 B/s for transmission.

Alternative View: Wearable Data Reduction Pipeline

This diagram shows how aggressive data reduction at each layer keeps power consumption manageable while preserving clinically relevant information.

Figure 37.4: Vertical flowchart showing data volume reduction at each processing stage in a wearable device

Data reduction at each stage: 5,000 B/s raw data is reduced 500x to 10 B/s average throughput, enabling 7+ day battery life on a 180 mAh battery.

Key architectural principle: Wearable IoT systems perform 99.8% of data reduction on the device itself. Only clinically relevant features (heart rate values, step counts, anomaly alerts) leave the wearable. This is fundamentally different from camera or audio IoT devices that must transmit raw data for cloud processing. The on-device processing model is what makes week-long battery life possible.

37.6 Wearable Sensor Types and Specifications

Understanding the sensor technologies inside wearable devices helps explain their capabilities and limitations.

Sensor	Technology	Measures	Accuracy	Power	Placement
PPG	Green LED + photodiode	Heart rate, SpO2	±3-5 BPM rest, ±10 BPM exercise	0.5-2.5 mA	Wrist, ear, finger
IMU	MEMS capacitive	Steps, activity, falls	95%+ step counting	0.01-0.05 mA	Wrist, hip, shoe
BIA	AC current injection	Body composition, hydration	±3-5% body fat	0.1-0.5 mA	Wrist multi-electrode
EDA	Skin conductance	Stress, emotional state	Relative changes only	0.01 mA	Wrist, finger
Skin temp	Thermistor / IR	Core body temp proxy	±0.1-0.5°C	0.001 mA	Wrist, ear, patch
1-lead ECG	Dry electrode contact	Heart rhythm, arrhythmia	Clinical-grade (touch)	0.3-1.0 mA	Chest patch, watch

PPG = photoplethysmography, BIA = bioimpedance analysis, and EDA = electrodermal activity.

Wearable sensor types with typical specifications. PPG is the most power-hungry continuous sensor, while the accelerometer provides the best power-to-insight ratio. {.striped .hover}

37.7 Worked Example: Fitness Tracker Battery Optimization

Scenario: A wearable manufacturer is designing a fitness tracker that must provide continuous heart rate monitoring while achieving 7-day battery life to match user expectations and reduce abandonment rates.

Given:

• Battery capacity: 180 mAh Li-Po (typical for slim fitness band form factor)
• Optical PPG sensor power: 0.8 mA at 25 Hz sampling, 2.5 mA at continuous 100 Hz
• MCU active power: 3 mA (processing PPG signal, running algorithms)
• MCU sleep power: 8 µA
• Bluetooth LE transmission: 15 mA for 3 ms per sync event
• Display (OLED): 5 mA when active, triggered by wrist raise
• Target heart rate accuracy: ±5 BPM during rest, ±10 BPM during exercise

Steps:

1. Calculate baseline continuous monitoring power budget:
   • Target: 7 days = 168 hours
   • Available energy: 180 mAh / 168 hours = 1.07 mA average current budget
   • This is extremely tight - must optimize aggressively
2. Design adaptive sampling strategy:
   • Resting state (detected via accelerometer): 1 Hz PPG sampling, 4-sample averaging
   • Walking state: 10 Hz PPG sampling with motion compensation
   • Exercise state: 25 Hz PPG sampling for accuracy during high HR
   • Night mode (no motion for 30+ min): 0.1 Hz sampling, interpolate between readings
3. Calculate power for each activity state:
   • Resting (16 hrs/day): 0.8 mA × (1/25) duty cycle + 8 µA = 0.04 mA average
   • Light activity (6 hrs/day): 0.8 mA × (10/25) + 3 mA × 0.1 = 0.62 mA average
   • Exercise (1 hr/day): 2.5 mA + 3 mA = 5.5 mA average
   • Sleep (8 hrs/day): 0.8 mA × (0.1/25) + 8 µA = 0.011 mA average
4. Calculate daily weighted average current:
   • Daily average = (16 × 0.04 + 6 × 0.62 + 1 × 5.5 + 8 × 0.011) / 24
   • Daily average = (0.64 + 3.72 + 5.5 + 0.088) / 24 = 0.41 mA
5. Add Bluetooth and display overhead:
   • BLE sync: 4 syncs/hour × 24 hours × 15 mA × 3 ms / 3600000 = 0.0012 mA
   • Display: 30 wrist raises/day × 3 seconds × 5 mA / 86400 = 0.005 mA
   • Total average: 0.41 + 0.0012 + 0.005 = 0.42 mA
6. Calculate actual battery life:
   • Battery life = 180 mAh / 0.42 mA = 428 hours = 17.8 days
   • Safety margin for battery degradation (80% at 2 years): 17.8 × 0.8 = 14.2 days
   • Exceeds 7-day target with comfortable margin

Result: Adaptive sampling strategy achieves 14+ day battery life while maintaining ±5 BPM accuracy during rest and ±10 BPM during exercise. The key insight is that 95% of battery savings come from the resting and sleep states when heart rate changes slowly.

Key Insight: Wearable battery optimization is not about reducing sampling rate uniformly - it is about matching sensor activity to physiological dynamics. Heart rate at rest changes slowly (seconds to minutes), allowing aggressive duty cycling, while exercise demands continuous monitoring. The accelerometer (0.01 mA) acts as a “wake-up” sensor that enables intelligent power management with minimal overhead.

viewof batteryCapacity = Inputs.range([100, 300], {
  value: 180,
  step: 10,
  label: "Battery Capacity (mAh)"
})

viewof restHours = Inputs.range([0, 20], {
  value: 16,
  step: 1,
  label: "Rest Hours/Day"
})

viewof activityHours = Inputs.range([0, 12], {
  value: 6,
  step: 1,
  label: "Activity Hours/Day"
})

viewof exerciseHours = Inputs.range([0, 6], {
  value: 1,
  step: 0.5,
  label: "Exercise Hours/Day"
})

viewof sleepHours = Inputs.range([0, 12], {
  value: 8,
  step: 1,
  label: "Sleep Hours/Day"
})

{
  const totalHours = restHours + activityHours + exerciseHours + sleepHours;
  
  if (totalHours !== 24) {
    const div = d3.create("div")
      .style("padding", "20px")
      .style("background", "#E74C3C")
      .style("color", "white")
      .style("border-radius", "8px")
      .style("text-align", "center")
      .style("font-family", "Arial, sans-serif");
    
    div.append("p")
      .style("margin", "0")
      .style("font-size", "16px")
      .style("font-weight", "bold")
      .text(`Error: Total hours must equal 24 (currently ${totalHours})`);
    
    return div.node();
  }
  
  // Power consumption calculations
  const restCurrent = 0.04; // mA
  const activityCurrent = 0.62; // mA
  const exerciseCurrent = 5.5; // mA
  const sleepCurrent = 0.011; // mA
  const bleOverhead = 0.0012; // mA
  const displayOverhead = 0.005; // mA
  
  const dailyAvgCurrent = (
    (restHours * restCurrent + 
     activityHours * activityCurrent + 
     exerciseHours * exerciseCurrent + 
     sleepHours * sleepCurrent) / 24
  ) + bleOverhead + displayOverhead;
  
  const batteryLifeHours = batteryCapacity / dailyAvgCurrent;
  const batteryLifeDays = batteryLifeHours / 24;
  const degradedLifeDays = batteryLifeDays * 0.8; // 80% capacity after 2 years
  
  const width = 700;
  const height = 400;
  const margin = {top: 40, right: 20, bottom: 80, left: 80};
  
  const svg = d3.create("svg")
    .attr("width", width)
    .attr("height", height)
    .attr("viewBox", [0, 0, width, height])
    .attr("style", "max-width: 100%; height: auto; font-family: Arial, sans-serif;");
  
  // Title
  svg.append("text")
    .attr("x", width / 2)
    .attr("y", 25)
    .attr("text-anchor", "middle")
    .attr("font-size", "18px")
    .attr("font-weight", "bold")
    .attr("fill", "#2C3E50")
    .text("Battery Life Optimization Calculator");
  
  // Bar chart data
  const data = [
    { state: "Rest", hours: restHours, current: restCurrent, color: "#16A085" },
    { state: "Activity", hours: activityHours, current: activityCurrent, color: "#3498DB" },
    { state: "Exercise", hours: exerciseHours, current: exerciseCurrent, color: "#E67E22" },
    { state: "Sleep", hours: sleepHours, current: sleepCurrent, color: "#9B59B6" }
  ];
  
  const chartWidth = width - margin.left - margin.right;
  const chartHeight = 180;
  
  const x = d3.scaleBand()
    .domain(data.map(d => d.state))
    .range([margin.left, width - margin.right])
    .padding(0.2);
  
  const y = d3.scaleLinear()
    .domain([0, d3.max(data, d => d.hours)])
    .nice()
    .range([margin.top + chartHeight, margin.top]);
  
  // X axis
  svg.append("g")
    .attr("transform", `translate(0,${margin.top + chartHeight})`)
    .call(d3.axisBottom(x))
    .attr("font-size", "12px");
  
  // Y axis
  svg.append("g")
    .attr("transform", `translate(${margin.left},0)`)
    .call(d3.axisLeft(y))
    .attr("font-size", "12px");
  
  svg.append("text")
    .attr("transform", "rotate(-90)")
    .attr("x", -(margin.top + chartHeight / 2))
    .attr("y", margin.left - 50)
    .attr("text-anchor", "middle")
    .attr("font-size", "12px")
    .attr("fill", "#2C3E50")
    .text("Hours per Day");
  
  // Bars
  svg.selectAll("rect.bar")
    .data(data)
    .join("rect")
    .attr("class", "bar")
    .attr("x", d => x(d.state))
    .attr("y", d => y(d.hours))
    .attr("width", x.bandwidth())
    .attr("height", d => chartHeight + margin.top - y(d.hours))
    .attr("fill", d => d.color)
    .attr("opacity", 0.8);
  
  // Hour labels on bars
  svg.selectAll("text.bar-label")
    .data(data)
    .join("text")
    .attr("class", "bar-label")
    .attr("x", d => x(d.state) + x.bandwidth() / 2)
    .attr("y", d => y(d.hours) - 5)
    .attr("text-anchor", "middle")
    .attr("font-size", "12px")
    .attr("fill", "#2C3E50")
    .text(d => `${d.hours}h`);
  
  // Results section
  const resultsY = margin.top + chartHeight + 60;
  
  // Battery life result
  const resultBox = svg.append("g")
    .attr("transform", `translate(${width/2}, ${resultsY})`);
  
  resultBox.append("rect")
    .attr("x", -150)
    .attr("y", -35)
    .attr("width", 300)
    .attr("height", 70)
    .attr("rx", 8)
    .attr("fill", degradedLifeDays >= 7 ? "#16A085" : "#E67E22")
    .attr("opacity", 0.2);
  
  resultBox.append("text")
    .attr("text-anchor", "middle")
    .attr("y", -10)
    .attr("font-size", "14px")
    .attr("fill", "#2C3E50")
    .text("Battery Life (with degradation)");
  
  resultBox.append("text")
    .attr("text-anchor", "middle")
    .attr("y", 20)
    .attr("font-size", "28px")
    .attr("font-weight", "bold")
    .attr("fill", degradedLifeDays >= 7 ? "#16A085" : "#E67E22")
    .text(`${degradedLifeDays.toFixed(1)} days`);
  
  // Details
  svg.append("text")
    .attr("x", width / 2)
    .attr("y", resultsY + 55)
    .attr("text-anchor", "middle")
    .attr("font-size", "11px")
    .attr("fill", "#7F8C8D")
    .text(`Avg current: ${dailyAvgCurrent.toFixed(2)} mA | Fresh battery: ${batteryLifeDays.toFixed(1)} days | Target: 7 days`);
  
  return svg.node();
}

Interactive Calculator: Battery Life Optimization - Adjust the hours spent in each activity state to see how adaptive sampling strategies impact battery life. This demonstrates why intelligent power management based on activity detection can extend battery life 10x compared to continuous high-rate sampling.

Decision Framework: Wrist vs. Chest vs. Patch for Continuous Cardiac Monitoring

Factor	Wrist (Watch/Band)	Chest (Strap/Band)	Adhesive Patch	Optimal Use Case
Signal quality	70-85% (PPG, motion artifacts)	95-99% (direct ECG)	95-99% (single-lead ECG)	Chest/patch for arrhythmia detection
Wear compliance	85-95% (socially acceptable)	45-60% (uncomfortable, visible)	70-85% (skin irritation at 7+ days)	Wrist for long-term monitoring
Battery life	7-14 days (rechargeable)	12-24 hours (coin cell)	7-21 days (non-rechargeable)	Wrist for convenience
Form factor	Consumer-friendly	Athletic/clinical only	Medical-grade only	Wrist for consumer adoption
Cost per unit	$150-350 (retail)	$50-120 (wholesale)	$25-45 (disposable)	Patch for cost-sensitive programs
Regulatory path	Class II or exempt (fitness)	Class II (medical device)	Class II (medical device)	All require FDA if making medical claims
Effective detection	78% (85% accuracy × 92% compliance)	52% (99% × 53%)	76% (97% × 78%)	Wrist wins on effective detection

Decision logic:

• Consumer wellness: Wrist (compliance dominates)
• Clinical diagnostics: Chest strap or patch (accuracy requirement)
• Remote patient monitoring: Wrist (long-term compliance critical)
• Hospital recovery: Patch (7-day wear in controlled setting)
• Athletic performance: Chest strap (athletes tolerate discomfort for accuracy)

Key insight: The “best” sensor is not the most accurate sensor – it is the sensor users will actually wear. Calculate effective detection (accuracy × compliance) for every deployment scenario.

37.8 Worked Example: User Compliance Optimization

Scenario: A remote patient monitoring company must choose between a chest-strap ECG monitor (clinical-grade accuracy) and a wrist-based optical heart rate monitor for a cardiac rehabilitation program. Patient compliance determines clinical outcomes.

Given:

• Patient population: 500 post-cardiac surgery patients, average age 62
• Monitoring duration: 12 weeks of daily wear (84 days)
• Clinical threshold: Detect heart rate exceeding 120 BPM or below 50 BPM
• Chest strap accuracy: ±1 BPM (gold standard ECG)
• Wrist sensor accuracy: ±5 BPM at rest, ±12 BPM during activity
• Chest strap compliance rate: 45% wear time (uncomfortable, visible under clothing)
• Wrist sensor compliance rate: 89% wear time (comfortable, socially acceptable)
• Clinical alert threshold: Must detect 95% of true arrhythmia events

Steps:

1. Calculate effective detection capability for chest strap:
   • True detection accuracy: 99% (ECG-grade)
   • Effective detection = Accuracy × Compliance = 0.99 × 0.45 = 44.6%
   • Result: Despite perfect accuracy, low compliance means 55% of events occur when device is not worn
2. Calculate effective detection capability for wrist sensor:
   • Detection accuracy for threshold events (HR > 120 or < 50): 92% (larger deviations easier to detect despite ±12 BPM error)
   • Effective detection = 0.92 × 0.89 = 81.9%
   • Result: Higher compliance compensates for lower accuracy
3. Model clinical event detection over 12-week program:
   • Expected arrhythmia events per patient: 8 (based on cardiac rehab literature)
   • Total events across 500 patients: 4,000 events
   • Chest strap detections: 4,000 × 0.446 = 1,784 events detected
   • Wrist sensor detections: 4,000 × 0.819 = 3,276 events detected
4. Calculate false positive burden:
   • Chest strap false positives: 1% of readings during 45% wear time
   • Wrist sensor false positives: 8% of readings during 89% wear time (motion artifacts)
   • Chest strap: 500 × 84 days × 24 hrs × 0.45 wear × 0.01 FP = 4,536 false alerts
   • Wrist sensor: 500 × 84 days × 24 hrs × 0.89 wear × 0.08 FP = 71,366 false alerts
5. Design hybrid alert strategy for wrist sensor:
   • Require 3 consecutive high readings (reduces false positives by 99.5%)
   • Add accelerometer filter (reject high HR readings during detected motion)
   • Adjusted wrist sensor false positives: 71,366 × 0.005 × 0.3 = 107 alerts (manageable)
   • True positive retention: 3,276 × 0.85 (some real events filtered) = 2,785 events

Result: Wrist-based monitoring with smart filtering detects 2,785 events versus 1,784 for chest strap - a 56% improvement in clinical event detection despite lower sensor accuracy. False positive rate reduced to clinically manageable 107 alerts over 12 weeks (2.1 per patient).

Key Insight: In wearable health monitoring, the formula “Effective Detection = Accuracy × Compliance” explains why comfortable, less accurate devices often outperform clinical-grade uncomfortable devices in real-world outcomes. User compliance is the dominant variable - a 99% accurate sensor worn 45% of the time captures less data than an 85% accurate sensor worn 90% of the time. Design for compliance first, then optimize accuracy within the compliance-friendly form factor.

Putting Numbers to It: Compliance-Adjusted Detection

Given: 12-week cardiac monitoring, 500 patients, 8 events/patient expected

\[\text{Chest ECG effective} = 0.99\,\text{(accuracy)} \times 0.45\,\text{(compliance)} = 0.446\] \[\text{Wrist PPG effective} = 0.92\,\text{(accuracy)} \times 0.89\,\text{(compliance)} = 0.819\] \[\text{Detection ratio} = \frac{0.819}{0.446} = 1.84\times\,\text{more events detected}\]

Real impact: At 4,000 total events, wrist detects 3,276 vs chest 1,784 - capturing 1,492 additional cardiac events that would prevent strokes and complications, despite 7% lower sensor accuracy.

viewof device1Accuracy = Inputs.range([50, 100], {
  value: 99,
  step: 1,
  label: "Device 1 Accuracy (%) - Chest Strap"
})

viewof device1Compliance = Inputs.range([20, 100], {
  value: 45,
  step: 1,
  label: "Device 1 Compliance (%) - Chest Strap"
})

viewof device2Accuracy = Inputs.range([50, 100], {
  value: 85,
  step: 1,
  label: "Device 2 Accuracy (%) - Wrist"
})

viewof device2Compliance = Inputs.range([20, 100], {
  value: 92,
  step: 1,
  label: "Device 2 Compliance (%) - Wrist"
})

viewof totalEvents = Inputs.range([1000, 10000], {
  value: 4000,
  step: 500,
  label: "Total Expected Events"
})

{
  const device1Effective = (device1Accuracy / 100) * (device1Compliance / 100);
  const device2Effective = (device2Accuracy / 100) * (device2Compliance / 100);
  
  const device1Detected = Math.round(totalEvents * device1Effective);
  const device2Detected = Math.round(totalEvents * device2Effective);
  
  const width = 700;
  const height = 400;
  const margin = {top: 40, right: 20, bottom: 60, left: 80};
  
  const svg = d3.create("svg")
    .attr("width", width)
    .attr("height", height)
    .attr("viewBox", [0, 0, width, height])
    .attr("style", "max-width: 100%; height: auto; font-family: Arial, sans-serif;");
  
  // Title
  svg.append("text")
    .attr("x", width / 2)
    .attr("y", 25)
    .attr("text-anchor", "middle")
    .attr("font-size", "18px")
    .attr("font-weight", "bold")
    .attr("fill", "#2C3E50")
    .text("Compliance vs Accuracy Trade-off Comparison");
  
  // Comparison bars
  const data = [
    { 
      device: "Device 1 (Chest)", 
      accuracy: device1Accuracy,
      compliance: device1Compliance,
      effective: device1Effective * 100, 
      detected: device1Detected,
      color: "#E67E22" 
    },
    { 
      device: "Device 2 (Wrist)", 
      accuracy: device2Accuracy,
      compliance: device2Compliance,
      effective: device2Effective * 100, 
      detected: device2Detected,
      color: "#16A085" 
    }
  ];
  
  const chartHeight = 200;
  
  const x = d3.scaleBand()
    .domain(data.map(d => d.device))
    .range([margin.left, width - margin.right])
    .padding(0.3);
  
  const y = d3.scaleLinear()
    .domain([0, 100])
    .range([margin.top + chartHeight, margin.top]);
  
  // Y axis
  svg.append("g")
    .attr("transform", `translate(${margin.left},0)`)
    .call(d3.axisLeft(y).ticks(5))
    .attr("font-size", "11px");
  
  svg.append("text")
    .attr("transform", "rotate(-90)")
    .attr("x", -(margin.top + chartHeight / 2))
    .attr("y", margin.left - 50)
    .attr("text-anchor", "middle")
    .attr("font-size", "12px")
    .attr("fill", "#2C3E50")
    .text("Effective Detection Rate (%)");
  
  // Bars for effective detection
  svg.selectAll("rect.effective-bar")
    .data(data)
    .join("rect")
    .attr("class", "effective-bar")
    .attr("x", d => x(d.device))
    .attr("y", d => y(d.effective))
    .attr("width", x.bandwidth())
    .attr("height", d => chartHeight + margin.top - y(d.effective))
    .attr("fill", d => d.color)
    .attr("opacity", 0.7);
  
  // Effective % labels
  svg.selectAll("text.effective-label")
    .data(data)
    .join("text")
    .attr("class", "effective-label")
    .attr("x", d => x(d.device) + x.bandwidth() / 2)
    .attr("y", d => y(d.effective) - 5)
    .attr("text-anchor", "middle")
    .attr("font-size", "14px")
    .attr("font-weight", "bold")
    .attr("fill", d => d.color)
    .text(d => `${d.effective.toFixed(1)}%`);
  
  // Device labels
  svg.selectAll("text.device-label")
    .data(data)
    .join("text")
    .attr("class", "device-label")
    .attr("x", d => x(d.device) + x.bandwidth() / 2)
    .attr("y", margin.top + chartHeight + 20)
    .attr("text-anchor", "middle")
    .attr("font-size", "12px")
    .attr("fill", "#2C3E50")
    .text(d => d.device);
  
  // Detail labels
  svg.selectAll("text.detail-label")
    .data(data)
    .join("text")
    .attr("class", "detail-label")
    .attr("x", d => x(d.device) + x.bandwidth() / 2)
    .attr("y", margin.top + chartHeight + 35)
    .attr("text-anchor", "middle")
    .attr("font-size", "10px")
    .attr("fill", "#7F8C8D")
    .text(d => `${d.accuracy}% acc × ${d.compliance}% comp`);
  
  // Results section
  const resultsY = margin.top + chartHeight + 70;
  
  const winner = device2Detected > device1Detected ? data[1] : data[0];
  const difference = Math.abs(device2Detected - device1Detected);
  const percentDiff = ((difference / Math.min(device1Detected, device2Detected)) * 100).toFixed(1);
  
  // Winner announcement
  const resultBox = svg.append("g")
    .attr("transform", `translate(${width/2}, ${resultsY})`);
  
  resultBox.append("rect")
    .attr("x", -200)
    .attr("y", -30)
    .attr("width", 400)
    .attr("height", 60)
    .attr("rx", 8)
    .attr("fill", winner.color)
    .attr("opacity", 0.2);
  
  resultBox.append("text")
    .attr("text-anchor", "middle")
    .attr("y", -5)
    .attr("font-size", "14px")
    .attr("fill", "#2C3E50")
    .text(`${winner.device} detects ${difference.toLocaleString()} more events (+${percentDiff}%)`);
  
  resultBox.append("text")
    .attr("text-anchor", "middle")
    .attr("y", 18)
    .attr("font-size", "16px")
    .attr("font-weight", "bold")
    .attr("fill", winner.color)
    .text(`${device1Detected.toLocaleString()} vs ${device2Detected.toLocaleString()} total events detected`);
  
  return svg.node();
}

Interactive Calculator: Compliance vs Accuracy Trade-off - Compare two wearable devices to see which detects more total clinical events. Adjust accuracy and compliance values to explore the trade-off space. This calculator shows why the formula “Effective Detection = Accuracy × Compliance” is the key metric for real-world wearable deployment success.

37.9 Knowledge Check

37.10 Wearable Market Segments and Requirements

Different wearable market segments impose fundamentally different design constraints. A fitness tracker, a clinical-grade medical device, and a workplace safety wearable share sensor technology but diverge in accuracy requirements, regulatory burden, and user expectations.

Figure 37.5: Graph diagram showing three wearable market segments with distinct requirements

Three wearable market segments with distinct requirements. Consumer fitness prioritizes battery life and comfort; medical grade demands regulatory approval and clinical accuracy; industrial safety requires ruggedness and real-time alerting.

Alternative View: Wearable Design Decision Matrix

This comparison helps product teams select the right design trade-offs for their target market segment.

Wearable design parameters vary significantly across market segments. Medical and industrial segments face regulatory constraints that consumer devices avoid.

Design Parameter	Consumer Fitness	Medical Grade	Industrial Safety
Form factor priority	Aesthetics, slim	Signal quality, skin contact	Ruggedness, glove compatibility
IP rating	IP67 (swim-proof)	IPX4 (splash-proof)	IP68+ (dust/water/chemical)
Operating temp	0 to 45°C	15 to 40°C (body temp)	-20 to 60°C
Data privacy	User consent, app privacy	HIPAA / GDPR health data	Employer data, union agreements
Update mechanism	OTA via phone app	FDA-controlled firmware	IT-managed fleet updates
Failure mode	Graceful (show last known)	Alert clinician immediately	Alarm + lockout if safety-critical
Typical lifecycle	2-3 years	5-7 years	3-5 years

37.11 Common Pitfalls in Wearable IoT Design

Common Pitfalls and Misconceptions

• Optimizing for accuracy instead of compliance: Choosing a chest-strap ECG (±1 BPM) over a wrist PPG (±5 BPM) seems logical, but chest straps achieve only 45% average wear time while wrist devices achieve 89%. Calculate Effective Detection = Accuracy × Compliance for every design decision – a 10% accuracy gain that causes a 20% compliance drop reduces net detection capability.

• Ignoring motion artifacts in PPG signals: Lab-measured heart rate accuracy on stationary subjects does not transfer to exercise. During movement, the sensor shifts on the wrist, ambient light leaks in, and blood flow patterns change, creating 30+ BPM errors. A “99% accurate” lab sensor may be only 70% accurate during a jog. Always specify accuracy per activity state and implement motion compensation using concurrent accelerometer data.

• Underestimating skin diversity impact on optical sensors: PPG accuracy varies significantly with skin tone (melanin absorbs green light), tattoo ink (blocks transmission entirely), wrist hair density, and skin thickness. Studies show SpO2 overestimation of 2-4% in darker skin tones (Fitzpatrick V-VI). Validate across all six Fitzpatrick skin types and use multiple LED wavelengths (green + infrared).

• Designing battery life for average use, not worst case: Advertising “7-day battery life” based on typical profiles (16 hrs rest, 1 hr exercise) misleads active users. A marathon runner training 3+ hours daily with continuous GPS will drain the same 180 mAh battery in 18-36 hours. Publish battery life for light, moderate, and heavy usage profiles and implement power reserve mode for core health monitoring.

• Treating regulatory classification as a hardware decision: The boundary between “wellness device” (no FDA approval) and “medical device” (510(k) required) is determined by software claims, not sensor capabilities. The same PPG sensor that displays heart rate (wellness) becomes FDA-regulated when its algorithm alerts to atrial fibrillation. Plan regulatory strategy before writing detection algorithms, not after.

37.12 Regulatory Landscape for Wearable Health Devices

Wearable IoT devices that make health claims face an increasingly complex regulatory environment. The boundary between “wellness” and “medical” determines whether a device needs months or years of regulatory approval.

FDA regulatory pathways for wearable health devices. The distinction between “wellness” and “medical” claims determines regulatory burden.

Category	Examples	Regulatory Path	Timeline	Data Requirements
General Wellness	Step counters, calorie trackers, sleep quality	No FDA approval needed	Immediate	Self-certification
Low-Risk Medical	HR monitors with arrhythmia detection (e.g., Apple Watch ECG)	FDA 510(k) or De Novo	6-12 months	Clinical validation study
Moderate-Risk Medical	Continuous glucose monitors, SpO2 for clinical use	FDA 510(k) with clinical data	12-24 months	Multi-site clinical trial
High-Risk Medical	Implantable cardiac monitors, closed-loop insulin	FDA PMA (Pre-Market Approval)	2-5 years	Large-scale randomized trial

Regulatory boundary: A fitness tracker that displays heart rate is a “wellness” device. The same tracker that alerts users to irregular heart rhythm suggesting atrial fibrillation becomes a medical device requiring FDA clearance. The software algorithm – not the hardware sensor – often determines the regulatory classification. This is why many consumer wearable manufacturers carefully word their features as “wellness insights” rather than “medical diagnoses.”

37.13 Wearable Deployment Tradeoffs

Tradeoff: Continuous Monitoring vs. Periodic Sampling

Option A: Continuous 24/7 monitoring - Captures all events including rare occurrences, enables real-time intervention, and provides complete longitudinal data. Drawbacks: shorter battery life, higher data costs, potential alarm fatigue. Option B: Periodic sampling (every 15 min, hourly, or on-demand) - Extends battery life to years instead of days, reduces data volume and costs, and may be sufficient for slowly-changing conditions. Risk: may miss acute events between samples. Decision factors: Event dynamics (arrhythmias are brief and require continuous ECG; blood pressure changes slowly and can be sampled hourly), battery constraints (implants need 5+ year battery life), and clinical value of captured data (is every heartbeat necessary, or just daily averages?).

37.14 Wearable IoT Privacy and Data Ownership

Wearable devices generate intimate physiological data 24/7. Unlike smartphone data, wearable health data reveals medical conditions, emotional states, sleep patterns, and physical capabilities – information with significant implications for insurance, employment, and personal relationships.

Key privacy considerations for wearable IoT:

• Data minimization: Transmit only derived features (heart rate values), not raw sensor data (PPG waveforms), unless clinically necessary. Raw waveforms can reveal biometric identity, drug use, and conditions the user did not consent to share.
• On-device processing: Perform health analysis on the wearable or phone, not in the cloud. This reduces exposure surface and latency while improving user trust.
• Granular consent: Allow users to share step counts without sharing heart rate, or share with their doctor but not with their employer’s wellness program.
• Data retention limits: Wearable data accumulates at 250 KB/day per user. A 1-million-user platform stores 250 GB/day. Define retention policies (raw data for 30 days, aggregated data for 5 years, anonymized data indefinitely).
• Right to delete: GDPR and CCPA require the ability to delete all user data on request, including derived models trained on that data.

Cross-Reference: Privacy and Security

For detailed coverage of IoT privacy regulations and compliance frameworks, see:

• IoT Security Fundamentals – encryption and secure data transmission
• Introduction to Privacy – GDPR and privacy-by-design principles
• IoT Security Common Mistakes – avoiding data breach vulnerabilities

37.15 Concept Relationships

How wearable concepts connect across IoT architecture and design:

This Chapter Concept	Related Chapter	How They Connect
Adaptive sampling strategy	Energy and Power Management	Accelerometer wake-up sensor enables 10x battery extension
BLE data transmission	Bluetooth BLE	3 KB sync every 5 minutes reduces battery drain vs continuous streaming
PPG sensor accuracy	Sensor Types	Green LED wavelength absorption by hemoglobin enables pulse detection
On-device processing	Edge Computing	99.8% data reduction (5 KB/s → 10 B/s) happens on-device before transmission
FDA classification	Privacy and Compliance	Software claims determine whether device is “wellness” or “medical”

37.16 See Also

Related chapters for wearable implementation details:

• Healthcare IoT - Clinical deployment of wearable monitoring systems
• Elderly Care IoT - Fall detection with multi-sensor fusion
• BLE Power Optimization - Connection intervals and advertising for week-long battery life
• Sensor Calibration - Correcting PPG signal drift over time
• Introduction to Privacy - GDPR/HIPAA compliance for health data

In 60 Seconds

Wearable IoT devices capture continuous biometric data—heart rate, activity, sleep—enabling personalised health insights while balancing battery life constraints, sensor accuracy across diverse body types, and user comfort expectations.

Interactive Quiz: Match Wearable IoT Concepts

Interactive Quiz: Sequence the Steps

Label the Diagram

💻 Code Challenge

37.17 Summary

This chapter covered the engineering principles behind wearable IoT technology:

Core Formula: Effective Detection = Sensor Accuracy × User Compliance

Chapter summary: key topics and takeaways for wearable IoT design.

Topic	Key Takeaway
Sensor placement	Different body locations trade signal quality for comfort; wrist achieves 95% compliance versus 60% for chest
Data architecture	On-device processing reduces raw sensor data by 99.8% (5 KB/s to 10 B/s average), enabling week-long battery life
Battery optimization	Adaptive sampling (0.1 Hz sleep to 25 Hz exercise) extends battery 10x versus continuous high-rate sampling
Compliance vs. accuracy	Wrist PPG at 89% compliance detects 56% more cardiac events than chest ECG at 45% compliance
Market segments	Consumer, medical, and industrial wearables share sensors but diverge in accuracy, regulatory, and durability requirements
Common pitfalls	Motion artifacts, skin tone bias in PPG, accuracy-over-compliance optimization, and average-case battery claims
Regulatory landscape	Software algorithms (not hardware) often determine whether a device is “wellness” or “medical” for FDA classification
Privacy	Wearable data reveals medical conditions and biometric identity; design for data minimization and granular consent

Design principles to remember:

1. Compliance dominates accuracy in real-world detection effectiveness
2. The accelerometer is the cheapest, lowest-power sensor and enables intelligent power management for all other sensors
3. On-device data reduction is what makes wearable IoT architecturally distinct from other IoT categories
4. Regulatory classification depends on health claims made by software, not sensor hardware capabilities

37.18 What’s Next

Next Topic	Description
Smart Contact Lenses and Advanced Wearables	AR contact lenses, battery-free implantables, and flexible bio-integrated sensors
Healthcare IoT Impact	Clinical deployment of wearable monitoring systems and latency tier architecture
Elderly Care IoT	Fall detection, behavioral analytics, and remote patient monitoring
Medication Monitoring	Smart pill bottles, adherence tracking, and EHR integration

Continue to Smart Contact Lenses and Advanced Wearables ->