Calculations

How to Calculate Wearable Medical Sensor Production Metrics: 5 Core Formulas with Worked Examples

Work the five formulas that size a wearable sensor line, calibration time, patch yield, battery runtime, test capacity, and flashing throughput, with real numbers and units.

Five calculations govern almost every wearable medical sensor line: calibration time per unit, adhesive patch yield, battery runtime, wireless test capacity, and firmware flashing throughput. Get any one wrong and the error compounds. A 10 second miss on calibration cycle time across 500,000 units per year is 1,389 hours of hidden chamber demand. This guide works each formula with real units, shows where every input comes from, and points to the calculator that automates it. Keep a consistent unit convention throughout: seconds for cycle times, milliamp hours for battery capacity, and units per hour, UPH, for station throughput. Mixing minutes and seconds is the single most common source of a 60x sizing error.

Start with calibration because it is usually the bottleneck. Per unit calibration time equals thermal soak plus the number of reference points multiplied by dwell time per point, plus handling. A skin temperature sensor calibrated at 3 setpoints with 45 second dwells and a 120 second soak needs 120 + (3 × 45) = 255 seconds, or 4.25 minutes. If your chamber holds 32 sensors per fixture, batch cycle time is 255 seconds plus 90 seconds of load and unload, giving 32 units per 345 seconds, about 334 UPH per chamber. The Sensor Calibration Time calculator handles the batch math and lets you test whether 2 point calibration with software linearization buys back capacity.

Adhesive patch yield is a rolled throughput calculation, not a single number. Multiply the yield of every converting step: lamination × die cut × sensor bond × release liner. Example: 0.985 × 0.992 × 0.978 × 0.995 = 0.951, so you keep 95.1 percent of starts. Then compute theoretical patches per roll: usable roll length divided by repeat pitch. A 100 meter roll with 2 meters of splice and edge loss leaves 98 usable meters; 98 / 0.055 at a 55 millimeter pitch gives 1,781 blanks, and at 95.1 percent rolled yield, 1,694 good patches. The Adhesive Patch Yield calculator chains the step yields for you. Track each step separately, because a blended number hides which laminator is bleeding material.

Battery runtime in hours equals usable capacity in milliamp hours divided by average current draw in milliamps. Average current for a duty cycled device is the time weighted sum of each operating state. Example: a patch samples at 2 mA for 5 ms each second, transmits BLE at 8 mA for 3 ms, and sleeps at 4 µA the rest of the time. Average draw = (2 × 0.005) + (8 × 0.003) + (0.004 × 0.992) = 0.038 mA. With a 40 mAh cell derated to 85 percent for temperature and end of life, runtime = 34 / 0.038 = 895 hours, about 37 days. The Battery Runtime calculator applies the derate automatically. Pull sleep current from the datasheet at body temperature, 35 °C, not the 25 °C headline figure.

Wireless test capacity per fixture is 3600 divided by test cycle time in seconds, multiplied by utilization, divided by 1 plus the retest fraction. A 75 second BLE test at 85 percent fixture utilization with a 6 percent retest rate delivers 3600 / 75 × 0.85 / 1.06 = 38.5 UPH per fixture. To hit 300 UPH of demand you need 300 / 38.5 = 7.8, so 8 fixtures. The Bluetooth Test Capacity calculator runs this sizing and shows the sensitivity: cutting the RF sweep from 75 to 55 seconds saves 2 fixtures at that volume. Measure cycle time with a stopwatch across 30 real cycles, including operator handling, never from the test script alone.

Firmware flashing throughput follows the same station logic with a different cycle time build up: flash time = image size / programming speed + verify time + socket handling. A 512 KB image at 150 KB/s programs in 3.4 seconds; add 2 seconds of verify and 4 seconds of handling for 9.4 seconds per device. One socket gives 3600 / 9.4 = 383 UPH; an 8 socket gang programmer gives roughly 3,060 UPH, since handling overlaps across sockets. The Firmware Flashing Throughput calculator compares single versus gang setups. Watch the verify step: switching from full readback verify to a CRC check typically cuts 1 to 3 seconds per unit, which matters more than a faster programmer.

Tie it together with the Final Functional Test Load calculator, which converts demand into required stations: stations = (annual demand × test minutes per unit) / (available minutes per station per year × utilization). At 500,000 units, a 3 minute functional test, and 2 shifts of 105,000 available minutes each per year at 80 percent utilization, you need 1,500,000 / 168,000 = 8.9, so 9 stations. Run every formula above at launch yield and again at mature yield, because retest and scrap inflate effective demand by 5 to 12 percent in year one. When the numbers disagree with the floor, audit inputs in this order: cycle time first, then utilization, then yield.

Published 2026-07-02.