Biomedical¶
Medical devices and human factors¶
Rules found in this domain: P25, P59.
Illustrative, not evidence.
Scenario. A team designs a portable infusion pump for nurses on a busy night ward. The pump is safety-critical and regulated. The team writes each requirement so it can be tested. One example is that the dose entry screen must be readable in dim light. Each requirement is linked to a user need, to the design element that meets it and to a test. Nurses use a prototype in a dim, noisy ward while interrupted, and the team watches how they enter doses.
- P25: A person operates the pump, so the team designs the interface for real nurses in the real ward, with stress and interruptions. Critical tasks, such as entering a dose, are tested with real users.
- P59: The pump is safety-critical and regulated, so every requirement is linked in both directions to design and test. A traceability report shows no requirement without a test and no test without a requirement.
Surgery¶
Rules found in this domain: P71.
Illustrative, not evidence.
Scenario. A surgical team in an operating theatre prepares to start a planned operation. The list is running late, so the start is done under time pressure. Before the first incision, the team pauses while a nurse reads a short checklist that covers patient identity, the procedure and the marked site. Each item is confirmed by pointing at it, such as the wristband or the site mark, and saying its status aloud. After a near miss, the team adds one item to the list.
- P71: The procedure is critical and done under time pressure, so the team uses a short checklist and confirms each item by pointing and saying its status aloud. The list is updated after the near miss, as the check requires.
Clinical engineering¶
Rules found in this domain: P75.
Illustrative, not evidence.
Scenario. A hospital's clinical engineering team monitors patient monitors and infusion pumps. The alarm system raises thousands of alerts a week. Staff stop reacting to the most frequent ones. The team gives each alert type an owner and a defined response. It measures the share of alerts that lead to no action, and it removes or changes the alerts that nobody acts on.
- P75: The system raises alerts to people, so each alert gets an owner and a defined response. The team measures the share of alerts with no action, keeps it low and removes or changes alerts that nobody acts on.
Bioinformatics¶
Rules found in this domain: P13.
Illustrative, not evidence.
Scenario. A lab reports which gene variants appear in a group of samples. The report must be reproducible. The analysis runs as a recorded pipeline. The pipeline records the version of the reference genome, the input data, the code and each tool. A reviewer re-runs it with one command and gets the same variant list.
- P13: The result must be reproducible, so it is produced only by automated, recorded steps with versions of data, code and tools. It can be regenerated from recorded inputs and versions with one command.
Biomaterials¶
Rules found in this domain: P29.
Illustrative, not evidence.
Scenario. A team designs a hip implant with a metal stem and head and a polymer cup liner. The parts touch each other and the body's fluids. The team does not assume the materials are compatible. It tests each pair of materials in a fluid that imitates body conditions, to check for wear and corrosion. A pair that fails the test is separated by a coating or replaced. Passing pairs go on a compatibility list.
- P29: Parts of different kinds touch, so the team verifies each pair in the real environment, here simulated body fluid. Pairs that are not compatible are separated, and each tested pair is on the compatibility list.
Biomechanics¶
Rules found in this domain: P25.
Illustrative, not evidence.
Scenario. A company designs a prosthetic knee. A person walks on it, climbs stairs and stumbles on uneven ground. The company does not design for an average user. It recruits users who differ in weight, gait and age. They test the knee in the places where it will really be used, such as stairs, wet floors and a hurried walk to a bus.
- P25: A person uses the system, so the design fits real users in real environments, including stress, and the full range of users, not an average user. Critical tasks, such as a step on stairs, are tested with real users.
Medical imaging (radiation protection)¶
Rules found in this domain: Master.
Illustrative, not evidence.
Scenario. A hospital plans a new CT scanner room. A scan with a higher dose can give a clearer image, but it exposes the patient to more radiation. The consequence of an error is high, and the staff are new to the scanner. For each scan type, the team lowers the dose step by step while the image still answers the clinical question. It applies stricter checks to the high-dose protocols than to low-dose ones, and extra checks to all of them while the staff are new.
- Master: The master rule scales the rigor of every rule to the consequence of failure times the uncertainty. High-dose protocols carry the larger consequence, and a scanner new to the staff adds uncertainty, so both raise the checks. The team keeps reducing dose until the cost of the next reduction, an image that could miss the diagnosis, is grossly disproportionate to the small dose it saves (ALARP).
Neural engineering¶
Rules found in this domain: P72.
Illustrative, not evidence.
Scenario. A team makes a deep brain stimulator, which is implanted and delivers pulses. The process repeated for each unit is manufacture and sterilization. The team names the few steps where control is essential, such as the seal of the casing and the sterilization cycle. It ranks them by consequence. Each has a limit, monitoring, a corrective action, a record and a recent check that the control works.
- P72: The process is repeated and a few steps need control. The team defines those critical control points, ranks them by consequence and gives each one a limit, monitoring, a predefined corrective action, a record and regular verification in practice.