Patient condition display

• US 20040148140A1
• Filed: 12/01/2003
• Published: 07/29/2004
• Est. Priority Date: 05/31/2001
• Status: Active Grant

• Current Assignee: Oxford University Innovation Limited (University of Oxford)
• Original Assignee: Isis Innovation Limited
• Inventors: Tarassenko, Lionel, Townsend, Neil William
• Granted Patent: US 7,031,857 B2
• Time in Patent Office: Days
• US Class Current: 702/189
• CPC Class Codes: A61B 5/0205 Simultaneously evaluating both cardiovascular conditions and different types of body conditions, e.g. heart and respiratory condition
• CPC Class Codes: A61B 5/7267 involving training the classification device
• CPC Class Codes: A61B 5/742 using visual displays A61B5/339 takes precedence
• CPC Class Codes: A61B 5/7445 Display arrangements, e.g. multiple display units
• CPC Class Codes: G16H 30/40 for processing medical images, e.g. editing
• CPC Class Codes: G16H 40/63 for local operation
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: G16H 50/70 for mining of medical data, e.g. analysing previous cases of other patients

Abstract

Data from a plurality of sensors representing a patient'"'"'s condition, including the measurement signals and also secondary parameters derived from the measurement signals, are displayed in a simple way by mapping them from the multi-dimensional measurement space to a two-dimensional visualisation space. This can be achieved using a mapping which preserves the topography of the data points, for instance by ensuring that the inter-point distances in the visualisation space match as closely as possible the corresponding inter-point distances in the measurement space. Such a mapping, for instance Sammon'"'"'s mapping is achieved by a suitably trained artificial neural network. The parameters are normalised before the mapping process and the normalisation and mapping are such that mapped points from a patient whose condition is normal appear in the centre of the visualisation space, whereas points from a patient whose condition is abnormal appear at the edge of the visualisation space. The artificial neural network may be trained using data points from a single patient or from a group of patients, and data may be thinned out using a pre-clustering algorithm.

Claims

1. Apparatus for displaying a graphical representation of a patient'"'"'s condition as measured by n parameters, where n>

3, comprising a processor which maps data points represented by said n parameters from an n-dimensional measurement space into an m-dimensional visualisation space, where m<

n, using a dimensionality reduction mapping, and a display which displays the visualisation space and the data points mapped into it, and which is adapted to the display of dynamically changing values of said parameters by means of the mapping being carried out by a trained artificial neural network.

2. Apparatus according to claim 1 wherein the n parameters are selected from:

a respiration measurement, an oxygen saturation measurement, blood pressure, skin temperature, S-T segment elevation/depression, heart rate, heart rate variability and respiration rate.

3. Apparatus according to claim 2 wherein the respiration measurement is an impedance pneumography measurement.

4. Apparatus according to claim 2 or 3 wherein the oxygen saturation measurement is a pulse oximetry measurement.

5. Apparatus according to any one of the preceding claims wherein said processor normalises the parameters prior to said mapping.

6. Apparatus according to claim 5 wherein the processor normalises the parameters using a zero mean, unit variance transformation.

7. Apparatus according to claim 5 wherein the processor normalises the parameters using an empirical transformation.

8. Apparatus according to any one of the preceding claims wherein m=3.

9. Apparatus according to any one of claims 1 to 7 wherein m=2.

10. Apparatus according to any one of the preceding claims wherein the dimensionality reduction is a distance-preserving mapping.

11. Apparatus according to claim 10 wherein the distance preserving mapping minimises the difference in the distance between pairs of points in the measurement space and the corresponding pairs of points in the visualisation space.

12. Apparatus according to claim 11 wherein the distance-preserving mapping is Sammon'"'"'s mapping.

13. Apparatus according to any one of the preceding claims wherein the artificial neural network is a Radial Basis Function (RBF) neural network.

14. Apparatus according to any one of the preceding claims wherein the artificial neural network is trained with data comprising a plurality of sets of said parameters from the patient.

15. Apparatus according to any one of claims 1 to 13 wherein the artificial neural network is trained with data comprising a plurality of sets of said parameters from a group of patients.

16. Apparatus according to any one of the preceding claims wherein the display is adapted to display at least one of:

a region of normality, and normal trajectory, in the visualisation space.

17. Apparatus according to any one of the preceding claims further comprising an alarm responsive to a departure of mapped data points from at least one of:

a region of normality, and normal trajectory, in the visualisation space.

18. Apparatus according to any one of the preceding claims wherein the points are displayed in the visualisation space in a visually distinguishable manner depending on the distance in the measurement space between each point and predetermined point or set of points representing a normal condition for the patient.

37. A method of training an artificial neural network for use in the apparatus of any one of claims 1 to 18 or the method of any one of claims 19 to 36, further comprising the step of pre-clustering in the measurement space data points each comprising a set of said parameters, deriving the centres of the clusters, and using the centres of the clusters as training data for the artificial neural network.

38. A method according to claim 37 wherein the data points are pre-clustered using the K-means method.

39. A patient condition monitor comprising apparatus according to any one of claims 1 to 18, or operating in accordance with the method of any one of claims 19 to 36.

19. A method of displaying a graphical representation of a patient'"'"'s condition as measured by n parameters, where n>

3, comprising mapping data points represented by said n parameters from an n-dimensional measurement space into an m-dimensional visualisation space, where m<

n, using a dimensionality reduction mapping, and displaying the visualisation space and the data points mapped into it, and which is adapted to the display of dynamically changing values of said parameters by means of the mapping being carried out by a trained artificial neural network.

20. A method according to claim 19 wherein the n parameters are selected from:

a respiration measurement, an oxygen saturation measurement, blood pressure, skin temperature, S-T segment elevation/depression, heart rate, heart rate variability and respiration rate.

21. A method according to claim 20 wherein the respiration measurement is an impedance pneumography measurement.

22. A method according to claim 20 or 21 wherein the oxygen saturation measurement is a pulse oximetry measurement.

23. A method according to any one of claims 19 to 22 further comprising normalising the parameters prior to said mapping.

24. A method according to claim 23 wherein the normalisation is by a zero mean, unit variance transformation.

25. A method according to claim 23 wherein the normalisation is by an empirical transformation.

26. A method according to any one of claims 19 to 25 wherein m=3.

27. A method according to any one of claims 19 to 25 wherein m=2.

28. A method according to any one of claims 19 to 25 wherein the dimensionality-reduction is a distance-preserving mapping.

29. A method according to claim 28 wherein the distance preserving mapping minimises the difference in the distance between pairs of points in the measurement space and the corresponding pairs of points in the visualisation space.

30. A method according to claim 29 wherein the distance-preserving mapping is Sammon'"'"'s mapping.

31. A method according to any one of claims 19 to 30 wherein the artificial neural network is a Radial Basis Function (RBF) neural network.

32. A method according to any one of claims 19 to 31 wherein the artificial neural network is trained with data comprising a set of said parameters from the patient.

33. A method according to any one of claims 19 to 32 wherein the artificial neural network is trained with data comprising a plurality of sets of said parameters from a group of patients.

34. A method according to any one of claims 19 to 33 further comprising displaying at least one of:

a region of normality, and normal trajectory, in the visualisation space.

35. A method according to any one of claims 19 to 34 further comprising generating an alarm responsive to a departure of mapped data points from at least one of:

a region of normality, and normal trajectory, in the visualisation space.

36. A method according to any one of claims 19 to 35 wherein the points are displayed in the visualisation space in a visually distinguishable manner depending on the distance in the measurement space between each point and predetermined point or set of points representing a normal condition for the patient.

40. A computer program comprising program code means for executing on a programmed computer system the method of any one of claims 19 to 38.

41. A computer readable storage medium storing a computer program according to claim 40.