Eleven Characteristics Selection

Eleven Characteristics Selection

Eleven Characteristics Selection

Scenario: A hospital system must choose between three remote patient monitoring (RPM) wearable devices for 5,000 chronic heart failure patients. All three meet minimum IoT requirements (Thing + Computation + Internet). Which characteristics matter most, and how do you score them?

The Three Devices:

  • Device A: $89, 48-hour battery, 15-minute data interval, FDA cleared, proprietary protocol, and no OTA updates because devices must be returned.
  • Device B: $199, 7-day battery, 5-minute data interval, FDA cleared, standard BLE plus FHIR, and automatic OTA updates.
  • Device C: $145, 72-hour battery, 1-minute data interval, FDA cleared, standard protocols, and OTA updates with manual approval.

Step 1: Prioritize Characteristics for Healthcare RPM

For a medical wearable with 5,000 deployed devices, which of the 11 characteristics are non-negotiable vs. nice-to-have?

  • Secure, critical 9/10: Patient health data needs strong protection; HIPAA compliance is required when the deployment is a covered entity or business associate.
  • Fast, high 7/10: Heart failure alerts need less than five-minute latency.
  • Low Maintenance, high 8/10: 5,000 devices with frequent battery visits would be unsustainable.
  • Upgradable, high 8/10: FDA-required firmware patches and clinical protocol changes must reach the fleet.
  • Growing, medium 6/10: The system may expand to 10,000 patients, but not 100,000.
  • Adaptable, medium 5/10: Epic EHR integration matters, but the scope is defined.
  • Smart, medium 6/10: Local anomaly detection helps, but is not the primary decision driver.
  • On Demand, low 4/10: Medical monitoring is periodic, not an instant-response consumer interaction.
  • Ubiquitous, low 3/10: The device works at home and clinic, not everywhere.
  • Agile, low 4/10: Clinical protocols change slowly.
  • Blend into Background, low 3/10: Patients expect visible medical devices.

Step 2: Score Each Device

  • Secure, 20%: Device A scores 7/10 because the proprietary design is hard to audit; Devices B and C score 9/10 with standard protocols.
  • Fast, 15%: Device A scores 5/10 because 15-minute intervals are too slow; Device B scores 9/10 at five minutes; Device C scores 10/10 at one minute.
  • Low Maintenance, 20%: Device A scores 4/10 with a two-day battery; Device B scores 10/10 with a seven-day battery; Device C scores 6/10 with a three-day battery.
  • Upgradable, 20%: Device A scores 2/10 because updates require return; Device B scores 10/10 with automatic OTA; Device C scores 7/10 with manual OTA.
  • Growing, 10%: Device A scores 4/10 due to vendor lock-in; Devices B and C score 9/10 with open standards.
  • Adaptable, 10%: Device A scores 3/10 with proprietary integration; Device B scores 10/10 with FHIR; Device C scores 9/10 with standards but a narrower integration set.
  • Smart, 5%: Device A scores 6/10 with basic alerts; Device B scores 8/10 with edge anomaly detection; Device C scores 7/10 with cloud-only processing.

Weighted Scores:

  • Device A: (7×0.20) + (5×0.15) + (4×0.20) + (2×0.20) + (4×0.10) + (3×0.10) + (6×0.05) = 4.4/10
  • Device B: (9×0.20) + (9×0.15) + (10×0.20) + (10×0.20) + (9×0.10) + (10×0.10) + (8×0.05) = 9.5/10
  • Device C: (9×0.20) + (10×0.15) + (6×0.20) + (7×0.20) + (9×0.10) + (9×0.10) + (7×0.05) = 8.1/10

Five-layer and seven-layer models expose different boundaries; compare where each places processing, service, and business decisions.

Step 3: Calculate Total Cost of Ownership (5-Year)

  • Hardware for 5,000 devices: Device A costs $445,000; Device B costs $995,000; Device C costs $725,000.
  • Battery replacements at $25 per visit: Device A costs $112,500 per year, or $562,500 over five years; Device B costs $6,250 per year, or $31,250; Device C costs $87,500 per year, or $437,500.
  • Firmware updates: Device A costs $225,000 for return shipping across three updates; Device B costs $0 with OTA; Device C costs $75,000 for two manual updates.
  • Integration: Device A costs $150,000 because of proprietary integration; Device B costs $25,000; Device C costs $40,000.
  • Five-year TCO: Device A totals $1,382,500; Device B totals $1,051,250; Device C totals $1,277,500.

Step 4: Calculate Clinical Value

State the 72%, 94% and 96% early-alert rates as hypothetical evaluation inputs, not demonstrated clinical sensitivity. - Prevented hospitalizations per 5,000 patients per year: Device A prevents 360, Device B prevents 470, and Device C prevents 480. - Cost per hospitalization: All three use the same $12,500 avoided-cost assumption. Annual savings of $4.5M, $5.875M and $6M follow only if the assumed hospitalization counts and $12,500 avoided cost hold; label this modeled value. - Five-year clinical value: Device A produces $22.5M, Device B produces $29.375M, and Device C produces $30.0M.

Decision:

Device B wins on quality score and five-year TCO; Device C has the highest modeled clinical value:

  1. Highest quality score: 9.5/10 vs. 8.1 (Device C) vs. 4.4 (Device A)
  2. Lowest TCO: $1.05M vs. $1.28M (Device C) vs. $1.38M (Device A)
  3. Clinical value: $29.4M (Device B) vs. $30M (Device C) vs. $22.5M (Device A)

Device B costs $110 more per unit than Device A, but saves $331K in 5-year TCO and delivers $6.9M more in clinical value.

Device C has slightly better clinical outcomes ($600K over 5 years) but costs $226K more than Device B and scores lower on critical characteristics (Low Maintenance, Upgradable).

Key Lessons:

  1. Not all 11 characteristics matter equally: For medical RPM, 75% of the decision weight comes from just 4 characteristics (Secure, Fast, Low Maintenance, Upgradable). Domain determines priorities.

  2. TCO ≠ purchase price: Device A’s $89 price looks attractive, but 5-year TCO is 31% higher than Device B due to battery replacement visits and firmware update logistics.

  3. Characteristics predict TCO: Low Maintenance and Upgradable directly correlate with operational costs. Devices scoring poorly on these characteristics always cost more long-term.

  4. Weighted scoring prevents bias: Without structured scoring, humans gravitate toward “favorite features.” Device C’s 1-minute data frequency is impressive but matters far less than Device B’s 7-day battery and auto-OTA for a 5,000-device deployment.

Application to Your Projects:

  1. List the 11 characteristics
  2. Prioritize 3-4 as non-negotiable for your domain
  3. Assign weights summing to 100%
  4. Score each option on the prioritized characteristics
  5. Calculate weighted total + TCO + domain-specific value
  6. Document the rationale for the decision

This framework prevents “gut feel” decisions and ensures the right trade-offs for your use case.

Back to the chapter