HKIE Professional Assessment - Biomedical Discipline common mistakes
HKIE Professional Assessment - Biomedical Discipline · Understand the traps and how to avoid them.
Treating IEC 60601-1 as a single checklist and quoting clause numbers without explaining the underlying principle - for example stating 'leakage current must be under 100 µA' without distinguishing patient leakage current (applied part, B/BF/CF classification) from earth leakage current and touch current, or without noting that the limits are normal-condition versus single-fault-condition dependent.
Candidates memorise a handful of numbers from summary tables and never build the classification logic (type of applied part × condition × patient/protected-earth status) that generates those numbers. Under exam pressure the memorised figure is retrieved without its qualifier, so the answer is technically wrong even though a number was produced.
What to do instead
When revising IEC 60601-1, build a matrix: rows = leakage current type (earth, touch, patient), columns = applied part classification (B, BF, CF) and condition (normal, single fault). Write the limit in each cell and one sentence on why. In the exam, always state the classification and the condition before the number. Practise deriving, not reciting.
Applying ISO 14971 risk management as a paperwork exercise: listing hazards (electric shock, infection) without linking them to reasonably foreseeable sequences of events, without a harm, and without a traceable chain from hazard → hazardous situation → harm → risk control → residual risk evaluation → benefit-risk conclusion.
Candidates learn the standard's clause headings and reproduce them as headings in an answer, but the exam asks for the reasoning that connects them. The mental model is 'document structure' rather than 'causal chain', so the answer describes a file rather than an analysis.
What to do instead
For every risk question, force the chain explicitly in writing: hazard (source of harm) → foreseeable sequence of events → hazardous situation (exposure) → harm (injury or damage to health) → severity × probability → risk control option in the ISO 14971 order (inherent safety by design, protective measures, information for safety) → verification of implementation and effectiveness → residual risk vs benefit-risk. Rehearse with three devices you know (infusion pump, electrosurgical unit, pulse oximeter).
Confusing ISO 13485 (quality management system for the organisation) with the medical device regulatory approval route, and confusing Hong Kong's Medical Device Administrative Control System (MDACS) voluntary administrative framework with a statutory registration scheme - for example asserting that MDACS listing is a legal prerequisite to supply a device in Hong Kong, or that ISO 13485 certification is itself a marketing authorisation.
The regulatory landscape is layered and candidates study standards in isolation. Hong Kong's position - a voluntary MDACS with listing of devices and conformity assessment bodies, rather than a full statutory premarket approval regime - is genuinely different from the EU MDR or US FDA pathways, and candidates default to the frameworks they know better.
What to do instead
Draw a one-page map for each jurisdiction: who the regulator is, whether participation is mandatory or voluntary, what the premarket step is (listing vs approval vs certification), what post-market obligations exist, and what the manufacturer's QMS standard is. For Hong Kong, be precise about MDACS being administered by the Medical Device Division of the Department of Health and being voluntary, and state clearly what that implies for enforcement and for a hospital's own procurement controls.
Answering clinical engineering questions about maintenance and calibration with generic 'follow the manufacturer's schedule' statements, without addressing risk-based maintenance prioritisation, calibration traceability to national or international standards, interval justification, out-of-tolerance handling (including retrospective review of results reported while the instrument was out of tolerance), and the distinction between safety testing, performance verification and calibration.
Candidates from a design or academic background treat maintenance as an operational afterthought and have not internalised the healthcare technology management lifecycle. They also conflate electrical safety testing with calibration, so the answer never reaches metrology content.
What to do instead
Structure every HTM answer around the lifecycle: needs assessment and procurement specification → acceptance testing and commissioning → inventory and criticality classification → planned preventive maintenance and calibration intervals justified by risk and history → performance verification → corrective maintenance and out-of-tolerance investigation → decommissioning and disposal. Name the traceability chain explicitly (instrument → reference standard → national metrology institute → SI) and state that out-of-tolerance findings trigger a look-back on affected patients or results.
Describing signal processing steps as a recipe ('apply a bandpass filter, then a notch filter, then wavelet denoising') without justifying the choice from the signal's spectral content and noise characteristics, and without quantifying the trade-off - for example failing to note that a narrow notch at 50 Hz can ring and distort adjacent ECG content, or that a high-order filter introduces phase distortion relevant to ST-segment analysis.
Candidates learn processing chains by example and reproduce them without the underlying analysis. The exam rewards the reasoning linking physiology to spectrum to filter specification, and the recipe answer cannot demonstrate it.
What to do instead
For each signal (ECG, EEG, EMG, SpO2, respiratory), write down the diagnostic band of interest, the dominant noise sources and their spectra, and then derive the filter specification: passband, stopband, order, phase response, and the clinical consequence of getting it wrong. Always state the trade-off and the validation method (e.g. compare against a reference annotation, check that diagnostic features survive).
Treating medical device cybersecurity and health informatics as an IT problem separate from device safety, so answers omit that a compromised device is a hazardous situation under ISO 14971, omit the IEC 62443 and IEC 81001-5-1 framing, and omit the clinical workflow consequences of interoperability failures (wrong-patient data association, unit mismatch, alarm fatigue from a flood of low-priority alerts).
Candidates compartmentalise: safety is electrical and mechanical, cybersecurity is 'computers'. The convergence of these domains is recent relative to most study material, and candidates have not practised writing about networked devices as safety-critical systems.
What to do instead
Adopt the framing that security is a safety property: enumerate threats, map each to a hazardous situation and harm, and place controls in the ISO 14971 risk process. For informatics questions, always trace the data path from sensor to clinical decision and name the failure modes at each interface - identity, time synchronisation, units, coding terminology, and alarm prioritisation - plus the mitigation for each.
Answering tissue engineering and regenerative medicine questions with a materials-science recital (scaffold porosity, degradation rate, Young's modulus) that ignores the biological and manufacturing reality: cell sourcing and characterisation, sterility and aseptic processing for living products, potency assays, the difference between a device and an advanced therapy medicinal product, and the fact that terminal sterilisation is often unavailable for cell-bearing constructs.
Candidates with an engineering background default to the biomaterials vocabulary they are confident in, and have not engaged with the regulatory and manufacturing constraints that dominate this field. The answer sounds knowledgeable but never addresses what makes these products hard.
What to do instead
For every regenerative medicine answer, cover four axes explicitly: the biological component (cell type, source, characterisation, potency), the scaffold or carrier (material, architecture, degradation and by-products), the manufacturing and quality regime (aseptic processing, in-process controls, release testing, chain of identity), and the clinical translation path (regulatory classification, preclinical evidence, trial design, traceability of donated material).
Handling ethics and professional practice questions by naming the principle ('beneficence', 'informed consent') and stopping there, without applying it to the specific conflict in the scenario - for example a device recall where the engineer must decide whether to inform clinicians of a residual risk the manufacturer has not publicised, or a resource allocation decision where the engineer's duty to the employer and to patients diverge.
Candidates treat ethics as a vocabulary test and are reluctant to commit to a position, so they describe principles rather than resolve the case. Professional-practice questions are marked on the reasoning and the decision, not the label.
What to do instead
Use a fixed structure for every ethics scenario: identify the stakeholders and their legitimate interests, state the specific conflict, identify the relevant duties and codes (including the HKIE code of conduct and the engineer's paramount duty to public safety and health), generate at least two options, evaluate each against the duties, decide, and state how you would document and escalate. Practise committing to an answer and defending it.
Treating healthcare facility and project delivery questions as pure project management - Gantt charts, budgets, milestones - without addressing the clinical interface: disruption to live clinical services during construction, infection control during works (dust, water, ventilation), medical gas and electrical infrastructure segregation (essential, critical, and isolated power for group 1 and group 2 medical locations), shielding for imaging modalities, and commissioning and migration of equipment into a clinically validated state.
Candidates apply generic project management templates because that is what they know, and have not internalised that a hospital project's dominant risks are clinical and not schedule-based. The specific technical requirements of medical locations are also easy to overlook.
What to do instead
For any facility question, add a clinical-interface layer to the project plan: infection prevention and control review, phasing to protect live services, utility segregation and redundancy, structural and shielding requirements for heavy or radiation-emitting equipment, and a commissioning plan that ends in clinical acceptance testing rather than practical completion. Cite the applicable electrical installation requirements for medical locations and the role of the estate and clinical teams in sign-off.
Writing sustainability and emerging-technology answers as undirected enthusiasm - listing buzzwords such as AI, digital twins, 3D printing and circular economy - without an engineering evaluation: no life-cycle assessment boundary, no quantification, no discussion of regulation and validation of AI as a medical device, no consideration of clinical evidence requirements, obsolescence and end-of-life management, or the total cost of ownership.
These topics are newer to the syllabus and candidates have less structured material to draw on, so they fall back on general commentary. The exam expects the same rigour applied elsewhere: define, quantify, evaluate, and state limitations.
What to do instead
For sustainability, always define the assessment boundary and the metric (energy per procedure, kg CO2e per device, single-use versus reprocessed), then discuss trade-offs against infection risk and device performance. For AI and emerging technology, apply the medical device lifecycle: intended use and clinical claim, data provenance and bias, validation and performance monitoring, human factors and clinician trust, change control for a learning system, and post-market surveillance. Anchor every claim to a standard or a regulatory expectation where one exists.