Observation Series for V3

1
Observation Series for V3

• A way to represent table-like, multi-dimensional
  Observations in HL7s Version 3.

Presented by Barry D. Brown Mortara Instrument,
Inc. September 30, 2002
2
Background

• FDA wishes to receive annotated ECGs in support
  of New Drug Applications.
• Work is underway to create clinical trials
  information models in HL7 V3 (RCRIM TC, CDISC).
• Need a way to cast ECGs and similar types of
  information in HL7 V3 information model.

3
Single-valued Observations

• The value attribute of Observation has the data
  type ANY.
• ANY can hold a wide range of values.
• It is clear how to communicate single-valued
  observations in V3, but not so clear when
  multiple values are observed.

4
Multi-valued Observations

• An Observation instance has one cd attribute, so
  all values must represent the same type of
  observation.
• Generic collections like BAGltgt, LISTltgt and SETltgt
  can be used for the value attribute.
• When LISTltgt is used for a multi-valued
  observation, call it an Observation Sequence.

5
Example

• Cardiac stress exam.
• Technician asks patient perceived level of
  exertion at different points in the exam.
• Technician records those observations in a list.
• This is an Observation Sequence with a cd for
  perceived exertion.

Perceived Exertion (Borg)
7
10
10
13
15
15
17
19
6
Correlated Sequences

• A single observation sequence communicates an
  ordinal relationship between the values, but we
  dont know if its temporal or something else.
• Correlating each value in the list with some
  other type of observation gives more information.

7
Example Continued

• Technician observes the clock in the room and
  records the time in a parallel list.
• There are 2 observation sequences, and the values
  in each row are correlated.
• Call this Correlated Observation Sequences.

Perceived Exertion (Borg) Time Of Day
7 1002
10 1010
10 1012
13 1013
15 1015
15 1020
17 1022
19 1027
8
Table-like Observations

• A set of observations that can be represented in
  a table can be Correlated Observation Sequences.
• Each column is an Observation Sequence.
• Each row is a correlation between the values of
  each sequence.

9
Example Continued

• Observe HR and BP.

Perceived Exertion (Borg) Time Of Day Heart Rate (BPM) Systolic BP (mmHg) Diastolic BP (mmHg)
7 1002 75 120 72
10 1010 94 130 77
10 1012 108 148 84
13 1013 116 156 93
15 1015 124 163 101
15 1020 154 175 108
17 1022 163 189 115
19 1027 168 202 123
- 1035 145 176 118
- 1050 108 164 107
- 1058 89 157 92
- 1104 83 154 85
- 1110 78120 142 78
10
Example Continued

• Observe ECG Leads.

Test time (ms) Lead I (µV) Lead II (µV) Lead V5 (µV) Other leads
0 33 54 35
2 34 67 37
4 33 93 39
6 68 115 72
8 102 124 106
10 155 143 157
12 145 165 147
14 104 147 106
16 57 107 59
18 12 78 15
20 -43 46 -42
22 -78 23 -75

11
Correlated Sequences Model
12
Observation Series

• Multiple tables (correlated sequences) observed
  in the same frame of reference can be grouped
  into an Observation Series.
• Sequences of the same type (cd) appearing in
  different tables can be compared.
• E.g. relative time, ECG electrode placement,
  patient position

13
Example Continued

• Record relative time so observations can be
  compared with ECG.

Perceived Exertion (Borg) Test time (ms) Heart Rate (BPM) Systolic BP (mmHg) Diastolic BP (mmHg)
7 120000 75 120 72
10 600000 94 130 77
10 720000 108 148 84
13 780000 116 156 93
15 900000 124 163 101
15 1200000 154 175 108
17 1320000 163 189 115
19 1620000 168 202 123
- 2100000 145 176 118
- 3000000 108 164 107
- 3480000 89 157 92
- 3840000 83 154 85
- 4200000 78120 142 78
14
Observation Series CMET
15
Region

• Secondary observations can be made from other
  observations.
• Series observations contain many observation
  values.
• Need a way to identify subsets of the Series from
  which the secondary observation is made.
• Call it a Region.

16
Region Boundary

• Each Region is defined by a set of boundaries,
  one for each observed Sequence type.
• The Region Boundary cd is the same as the
  Observation Sequence cd.
• The Region Boundary value defines the interval
  inside the Region.

17
Region CMET
18
Observations on Regions

• Secondary observations can be made on the Regions
  within the Series.
• No special Observation class is required.
• Use the supported by relationship.

19
Observations on Regions