Medical Device Regulation, Risk & Electrical Safety

Getting a device safely to patients. Learn device classification (FDA Class I–III, EU MDR I–III, CDSCO), regulatory pathways (510(k), De Novo, PMA, CE marking), quality management (ISO 13485, design controls, DHF), risk management (ISO 14971: hazards, risk estimation, ALARP, residual risk), software (IEC 62304) and usability (IEC 62366), electrical safety (IEC 60601: leakage currents, applied parts B/BF/CF, microshock vs macroshock), clinical evaluation and post-market surveillance — with Python for risk scoring and leakage current.

Medical devices can save lives — and harm them. Regulation exists to ensure that devices are safe and effective before and after they reach the market, with requirements proportional to risk: a tongue depressor and a heart valve are treated very differently.

For engineers, regulation is not paperwork added at the end; it shapes the design process (design controls), the evidence you must generate (verification, validation, clinical data) and how you manage risk throughout the product's life.

An analogy

Regulation is the flight-safety system of medicine: like aircraft certification, it demands documented design, testing and risk analysis before 'passengers' (patients) are exposed, and continuous monitoring afterwards to catch problems early.

Classification and pathways

Devices are classified by risk: Class I (low: bandages, stethoscopes), Class II/IIa/IIb (moderate: infusion pumps, imaging), Class III (high: implants that sustain life, heart valves). In the USA, most Class II devices use 510(k) (substantial equivalence to a predicate), novel low/moderate-risk devices use De Novo, Class III uses PMA with clinical trials. In the EU, notified bodies assess conformity to the MDR for CE marking; India regulates under CDSCO's Medical Device Rules.

Quality system and design controls

ISO 13485 / FDA QMSR design controls require:

Risk, software and usability

ISO 14971: identify hazards and hazardous situations, estimate and evaluate risks, implement controls, verify effectiveness, evaluate overall residual risk against benefit, and keep the risk file alive with production and post-market data. IEC 62304 sets software lifecycle rigour by safety class (A/B/C); IEC 81001-5-1 and FDA guidance cover cybersecurity. IEC 62366 requires usability engineering: use-related risk analysis and summative testing with representative users.

Electrical safety and post-market

IEC 60601-1 addresses electrical, mechanical, thermal and radiation hazards: two means of patient protection (MOPP) — insulation, creepage/clearance distances, isolation barriers — and leakage-current limits by applied-part type (B, BF floating, CF cardiac). Microshock (current applied directly to the heart via catheters) is dangerous at microamps, so CF parts have the strictest limits. EMC is covered by IEC 60601-1-2. After launch, post-market surveillance, complaint handling, vigilance reporting (MDR/MedWatch), CAPA and field safety corrective actions (recalls) close the loop.

Formulas

Worked example: Risk score

Given: Probability level 3, severity 4 (policy: ≥ 10 unacceptable)

Find: Evaluation.

Answer: Unacceptable — add risk controls

Worked example: Leakage with a larger capacitance

Given: 230 V, 50 Hz, stray C = 1 nF

Find: Leakage current.

Answer: 72 µA — exceeds CF (10 µA), fine for B/BF normal limit (100 µA)

Worked example: Microshock vs macroshock

Given: 20 µA through a cardiac catheter vs 20 µA hand-to-hand

Find: Danger?

Answer: Catheter: potentially dangerous; skin contact: imperceptible

Risk matrix scoring and leakage-current estimate

# ISO 14971-style risk estimate and IEC 60601 leakage check
hazards = [("overheating probe", 3, 4), ("software dose error", 2, 5), ("battery swelling", 2, 3)]
for name, prob, sev in hazards:
    r = prob * sev
    level = "unacceptable" if r >= 10 else "ALARP - reduce" if r >= 5 else "acceptable"
    print(f"{name:20s}: P{prob} x S{sev} = {r:2d} -> {level}")
# Patient leakage current through a 1 kOhm body model from 2 pF stray capacitance at 230 V, 50 Hz
import math
Xc = 1 / (2 * math.pi * 50 * 2e-12)
I = 230 / (Xc + 1000)
print(f"leakage {I*1e6:.2f} uA vs CF limit 10 uA normal condition")
# overheating probe   : P3 x S4 = 12 -> unacceptable
# software dose error : P2 x S5 = 10 -> unacceptable
# battery swelling    : P2 x S3 =  6 -> ALARP - reduce
# leakage 0.14 uA vs CF limit 10 uA normal condition

Risk policies (the thresholds here are an example set by the manufacturer) drive risk-control measures in priority order: inherent safety by design, protective measures, then information for safety. Leakage via stray capacitance is small here, but long cables, filters and faults raise it — IEC 60601 tests measure it under normal and single-fault conditions.

Common mistakes

Applications

Key takeaways

Going further