(Week 2) [T3-1] Biomedical Data

Question 1

Why is data so important?

Answer 1

All medical care activities involve gathering, analysing, or using data.
Data is crucial to the process of decision making and to the cycle we saw in the last class.

Question 2

Biomedical data have very high heterogeneity. Comment on that.

Answer 2

Biomedical data have very high heterogeneity. To be processed intelligently, biomedical concepts in clinical records will have to be encoded using a common terminology. Biomedical data heterogeneity could be the following.
- Numeric (temperature, pulse).
- Text (narratives).
- Images (RX).
- Signals (ECG).
- Sequences (DNA).

Question 3

What is the difference between objective data and subjective data?

Answer 3

• Objective data is collected through instruments.
• Subjective data are results from interpreting messages from the patient.

Data is collected by all sort of devices and all sorts of health professionals.

Question 4

Explain the structure of clinical data.

Answer 4

<patient, name-of-parameter, reading, date>

• Patient ID.
• Name of measured parameter (often hierarchical, elements composed by other elements). For example, ‘last name’.
• Value (reading). Could be the following.
  o Numeric: age, temperature, etc.
  i) Variable precision (3 vs 3.00).
  ii) Includes units (3g = 3000mg).
  iii) Normal range (3g on machine A = 4g on machine B.
  o Categorical: small number of discrete choices.
  i) Binary (yes/no).
  ii) Nominal: qualitative, no order (religion).
  iii) Ordinal: qualitative or quantitative, natural order (10, 11, 12, etc.).
  o Coded: taken from vocabulary. Permit relations among concepts
  (allergen == penicillin).
  o Free text or narrative values: not fully coded.
  o Null: sometimes value is not known.
  i) Explicit: value known to be not known. (It is normal to be null)
  ii) Implicit: no value for an attribute.
• Date-time of observation.
  a) Instants = points in time (1993-09-15 at 12:45:67).
  b) Points with granularity (duration of 3 months, seasons).

Question 5

What missing data tells us?

Answer 5

Missing data could be a negative sign, normal sign or unknown. The reasoning may depend on context (diagnosis).

Question 6

Which are the 3 types of data?

Answer 6

• Structured (coded). Impose structure to the collected information (lab test, exam request form, etc.).
• Non-structured (narrative). Text (written in natural language) and most clinical data in this format (admission note, nursing note, radiology, reports, etc.).
  Uses common language + technical terms (dyspnoea for ‘lack of air’, and use of abbreviations like ‘PERRLA’).
• Semi-structured (quasi-coded).

Question 7

What is the difference between narrative data and coded data?

Answer 7

Narrative data
- Universal and familiar.
- Requires little training.
- Expressive, even in new situations.
Coded data
- Encourages the introduction of all data elements (‘how can i reach you’ vs ‘explicit name, address, phone’).
- Improves the capacity of locating and reading data.
- Authors required to classify.

Question 8

What is the SOAP notes?

Answer 8

Notes for documenting medical encounters.
S: Subjective
O: Objective
A: Assessment
P: Plan