I85PREP
⌘ K
HKIE Professional Assessment - Biomedical Discipline study materials
HKIE Professional Assessment - Biomedical Discipline study resources

HKIE Professional Assessment - Biomedical Discipline: Health Informatics and Data

HKIE Professional Assessment - Biomedical Discipline · Work through your exam, one topic at a time.

All guide sections

Health Informatics and Data

This domain tests data interoperability, terminology and governance in clinical systems, and whether you can reason about the information risks that arise when devices and records meet. It rewards familiarity with real standards rather than generic computing language.

Interoperability standards

  • HL7 v2 remains the dominant messaging standard in hospitals, using pipe-delimited segments (MSH, PID, OBX) and event types such as ADT and ORU.
  • HL7 FHIR uses RESTful resources (Patient, Observation, DiagnosticReport, Device) with JSON or XML payloads, and is increasingly used for app-level integration and mobile access.
  • DICOM covers imaging objects and services; IHE profiles specify how these standards combine for a clinical workflow.
  • Terminology: SNOMED CT for clinical concepts, LOINC for laboratory and observation codes, ICD-10 for morbidity coding and reimbursement, and the WHO/IMDRF device nomenclature and UDI for device identification.

Governance and privacy in Hong Kong

  • The Personal Data (Privacy) Ordinance (Cap 486) sets six data protection principles: lawful and fair collection with purpose limitation, accuracy, retention, security, openness and access/correction rights.
  • The eHRSS is an opt-in, provider-participating shared record system; sharing requires patient consent within that framework.
  • Security controls: least privilege, role-based access, encryption in transit and at rest, immutable audit trails, backup and disaster recovery, plus device inventory under IEC 80001-1 when equipment joins the clinical network.

Data quality and decision support

  • Dimensions: completeness, accuracy, timeliness, consistency, provenance and fitness for purpose.
  • Clinical decision support must be validated against local data; alert burden causes fatigue and override, so sensitivity alone is a poor design target.

Worked example

A glucose result captured by a networked meter must reach the EHR as an FHIR Observation with a LOINC code, a UCUM unit (mmol/L), a timestamp, a patient identifier and provenance linking to the device. Missing units or context destroys comparability. A decision rule set to alert below 4.0 mmol/L will fire on hypoglycaemia, and the design question is whether clinicians can act on that alert before it is suppressed by fatigue.

Diagnostic performance example

With a 1% prevalence population, a 90% sensitive, 95% specific rule produces a PPV near 15%: most alerts are false positives, so the rule needs a higher-prevalence target group.

Traps that cost marks

  • Treating HL7 v2 and FHIR as interchangeable.
  • Confusing SNOMED CT (concepts) with LOINC (observations and measurements).
  • Assuming anonymisation removes all re-identification risk.
  • Ignoring audit trail and consent requirements under the PDPO.