POSTURE STATE CLASSIFICATION FOR A MEDICAL DEVICE

• US 20100010380A1
• Filed: 04/30/2009
• Published: 01/14/2010
• Est. Priority Date: 07/11/2008
• Status: Active Grant

• Current Assignee: Medtronic Incorporated (Medtronic Plc)
• Original Assignee: Medtronic Incorporated (Medtronic Plc)
• Inventors: Skelton, Dennis M., PANKEN, Eric J.
• Granted Patent: US 8,958,885 B2
• Time in Patent Office: Days
• US Class Current: 600/587
• CPC Class Codes: A61B 2562/0219 Inertial sensors, e.g. accelerometers, gyroscopes, tilt switches
• CPC Class Codes: A61B 5/1116 Determining posture transitions
• CPC Class Codes: A61B 5/1118 Determining activity level
• CPC Class Codes: A61B 5/686 Permanently implanted devices, e.g. pacemakers, other stimulators, biochips A61B5/6861 takes precedence
• CPC Class Codes: A61B 5/7264 Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems neural networks per se G06N3/00; expert systems per se G06N5/00
• CPC Class Codes: A61N 1/36535 controlled by body position or posture
• CPC Class Codes: A61N 1/36542 controlled by body motion, e.g. acceleration
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems

Abstract

One embodiment relates to a medical device including a sensor to provide signals indicative of a detected posture state of a patient. A memory stores one or more defined vectors, each defined vector being associated with a tolerance describing a relationship with the defined vector. A processor determines a respective similarity between the detected vector and each of one or more of the defined vectors and classifies a posture state of the patient based on whether any similarity has a relationship to the respective defined vector that is described by the associated tolerance. In one embodiment, the similarity is determined without regard to a coordinate system of the patient. Another embodiment relates to determining the similarity based on at least one of an inner product, a length of the defined posture vector and a length of the detected vector.

Claims

1. An implantable medical device, comprising:

a sensor to provide signals indicative of a detected posture vector of a patient;

a memory to store one or more defined posture vectors, each defined posture vector being associated with a tolerance describing a relationship with the defined posture vector; and

a processor to determine a respective similarity between the detected posture vector and each of one or more of the defined posture vectors and to classify a posture of the patient based on whether any similarity has a relationship to the respective defined posture vector that is described by the associated tolerance.

2. The device of claim 1, the processor further to determine each similarity based on at least one of an inner product between the detected posture vector and a defined posture vector, a squared length of the detected posture vector, and a squared length of the defined posture vector.

3. The device of claim 1, the processor to determine the defined posture vector to which the detected posture vector is closest.

4. The device of claim 1, the processor to determine each similarity as a distance between the detected vector and a defined vector.

5. The device of claim 4, the processor further to map the distance to obtain a value ranging between a selected minimum value and a selected maximum value, and to use the obtained value to classify the posture of the patient.

6. The device of claim 1, wherein multiple ones of the defined posture vectors are associated with a same posture.

7. The device of claim 6, wherein the multiple ones of the defined posture vectors associated with a same posture describe multiple regions in space interrelated by one or more logical functions.

8. The device of claim 7, wherein at least some of the multiple regions in space overlap with one another.

9. The device of claim 8, wherein the multiple regions in space include one or more toroids and one or more cones.

10. The device of claim 1, wherein the processor determines each similarity without regard to a coordinate system of the patient.

11. The device of claim 1, the processor further to classify the posture of the patient based on a first set of the one or more defined posture vectors, the processor to re-classify the posture of the patient based on a second set of the one or more defined posture vectors, each of one or more defined posture vectors in the second set being a sub-classifications of a posture in which the patient is first classified.

12. The device of claim 1, the processor further to process multiple ones of the defined posture vectors to obtain a virtual posture vector, and wherein the virtual posture vector is one of the one or more defined posture vectors used to classify the posture of the patient.

13. The device of claim 12, the processor further to determine an average of cross-products of pairs of the multiple ones of the defined posture vectors to obtain the virtual posture vector.

14. The device of claim 12, the processor further to determine a single plane that best approximates planes in which the multiple ones of the defined posture vectors reside, to select multiple vectors that describes the single plane, and to determine a vector that is normal to the single plane for use as the virtual posture vector.

15. The device of claim 12, the processor further to classify the posture of the patient as being Lying Down based on whether the detected posture vector lies outside of a cone surrounding the virtual posture vector and further based on whether the detected posture vector lies within any of one or more cones disposed about one or more of the defined posture vectors associated with the Lying Down posture.

16. The device of claim 1, the sensor further to provide signals indicative of an activity state of the patient, and the processor further to classify a posture state of the patient based on whether any similarity has a relationship to the respective defined posture vector that is described by the associated tolerance and further based on the indicated activity state of the patient.

17. The device of claim 1, further including a therapy module to deliver electrical stimulation therapy to the patient based on the posture classification.

18. A method for use with a medical system having a sensor, comprising:

obtaining a defined vector indicative of a defined posture state;

obtaining from the sensor a detected vector that is indicative of a posture state of a patient;

determining a similarity between the defined vector and the detected vector, the similarity being determined without deriving an angle;

classifying the posture state of the patient based on the similarity; and

initiating via the medical system an action related to care provided to the patient, the action being based on the classification of the posture state of the patient.

19. The method of claim 18, wherein classifying the posture state includes classifying the patient as being in at least one of a posture and an activity state.

20. The method of claim 18, wherein the similarity is a cosine of an angle between the defined vector and the detected vector.

21. The method of claim 20, further comprising:

deriving the cosine of the angle between the defined vector and the detected vector;

obtaining a tolerance that is associated with the defined vector; and

if the cosine satisfies a relationship described by the tolerance, classifying the patient as being in the posture state.

22. The method of claim 18, further comprising:

associating a tolerance with the defined vector; and

wherein classifying the posture state comprises determining, based on the similarity, whether the detected vector has a relationship to the defined vector described by the tolerance.

23. The method of claim 22, wherein the tolerance describes a relationship involving at least one of a cone and a toroid disposed about the defined vector.

24. The method of claim 22, wherein the tolerance describes multiple regions in space and wherein classifying the posture state comprises determining, based on the similarity, whether the detected vector lies within one or more of the multiple regions.

25. The method of claim 22, wherein the tolerance interrelates multiple regions in space using one or more logical functions.

26. The method of claim 22, further comprising:

associating the defined vector with a condition; and

wherein classifying the posture state comprises;

evaluating the condition;

selecting a relationship based on the evaluation of the condition; and

determining based on the similarity whether the detected vector has the relationship to the defined vector.

27. The method of claim 18, further comprising:

obtaining multiple defined vectors;

deriving a respective similarity between the detected vector and each of the multiple defined vectors;

determining, based on the similarities, which of the defined vectors is closest to the detected vector; and

classifying the posture state based on the determination.

28. The method of claim 27, wherein determining which of the defined vectors is closest comprises determining with which of the multiple defined vectors the detected vector has a largest inner product.

29. The method of claim 27, wherein deriving the respective similarity comprises obtaining at least one constant value that is based on at least one of a squared length of the defined vector and a cosine of an angle describing a region in space with respect to the defined vector.

30. The method of claim 18, wherein obtaining the defined vector includes obtaining the defined vector from a sensor implanted in the patient while the patient assumes the posture state.

31. The method of claim 30, wherein the defined vector and the detected vector are described in a coordinate system of the sensor and without regard to an orientation of the sensor in relation to the patient.

32. The method of claim 18, further comprising:

obtaining multiple tolerances, each describing a relationship with the defined vector;

evaluating a condition associated with the patient;

selecting one of the multiple tolerances based on the evaluation; and

classifying the posture state based on whether the similarity has a relationship to the defined vector described by the selected tolerance.

33. The method of claim 18, further comprising:

obtaining multiple defined vectors; and

processing the multiple defined vectors to obtain a virtual vector; and

wherein determining the similarity between the defined vector and the detected vector further comprises determining the similarity between the virtual vector and the detected vector.

34. The method of claim 33, wherein processing the multiple defined vectors includes deriving cross-products between pairs of the multiple defined vectors and averaging the cross-products.

35. The method of claim 33, further comprising:

associating multiple regions in space with at least one of the virtual vector and one or more of the multiple defined vectors; and

wherein classifying the posture state of the patient further comprises determining whether the detected vector lies within any of the multiple regions in space.

36. The method of claim 18, wherein the multiple regions in space include at least one toroid and at least one cone.

37. The method of claim 18, wherein the multiple regions in space include a toroid disposed about the virtual vector, and multiple cones, each disposed about a respective one of the multiple defined vectors.

38. The method of claim 18, wherein the similarity is a distance between the detected vector and the defined vector, and wherein classifying the posture state includes mapping the distance to a value that is directionally proportional to how similar the defined vector is to the detected vector.

39. The method of claim 18, wherein classifying the posture state of the patient further comprises:

classifying the patient in a first posture state;

obtaining multiple defined vectors associated with the first posture state;

determining a similarity between the defined vector and each of the multiple defined vectors associated with the first posture state; and

re-classifying the posture state of the patient based on the similarities between the defined vector and each of the multiple defined vectors associated with the first posture state.

40. The method of claim 39, wherein the first posture state is a Lying Down posture state and wherein the multiple defined vectors associated with the first posture state are associated with posture states that are sub-classifications of the Lying Down posture state.

41. The method of claim 39, wherein classifying the patient in the first posture state further comprises determining whether the detected vector is in one or more regions in space associated with the defined posture vector, and wherein determining a similarity between the defined vector and each of the multiple defined vectors associated with the first posture state comprises determining to which of the multiple defined vectors the detected vector is closest.

42. The method of claim 18, wherein classifying the posture state of the patient classifies the patient as being in at least one of a posture and an activity state.

43. A medical system, comprising:

a sensor to provide a detected vector indicative of a posture state of a patient;

a storage device to store posture state definitions, one or more of which are associated with defined vectors;

a processor to derive a similarity between the detected vector and each of one or more of the defined vectors, wherein deriving the similarity does not require derivation of angles, and to classify the posture state of the patient based on the derived similarities; and

a response module to generate a response based on the posture state classification.

44. The system of claim 43, wherein the response module is a therapy module to deliver therapy to the patient based on the posture state classification.

45. The system of claim 43, wherein at least one of the sensor and the response module is implanted within the patient.

46. The system of claim 43, wherein a programmer comprises the processor, the programmer to allow a user to create one or more of the posture state definitions.

47. The system of claim 43, the sensor to provide each of the defined vectors while the patient is assuming a respectively-associated posture state.

48. The system of claim 47, wherein the defined vectors and the detected vector are each expressed in a coordinate system of the sensor without reference to the coordinate system of the patient, and the processor derives each of the similarities without reference to the coordinate system of the patient.

49. The system of claim 43, the processor further to employ ones of the defined vectors to derive at least one additional defined vector, the processor to derive a similarity between the detected vector and the at least one additional defined vector, and to classify the posture state of the patient based on the similarity between the detected vector and the at least one additional defined vector.

50. The system of claim 49, wherein the at least one additional defined vector includes a virtual upright vector, the processor further to derive a similarity between the detected vector and one of the defined vectors to determine whether to classify the patient as being in an Upright posture state, the processor further to derive a similarity between the detected vector and the virtual upright vector to determine whether to classify the patient as being in a Lying Down posture state.

51. The system of claim 50, wherein if the patient is classified in the Lying Down posture state, the processor further to derive a similarity between the detected vector and each of ones of the defined vectors associated with the Lying Down posture state to classify the patient in a posture state that is a sub-classification of the Lying Down posture state.

52. The system of claim 43, wherein each similarity between the detected vector and a defined vector is derived based on at least one of an inner product between the detected vector and the defined vector, a length of the detected vector, and a length of the defined vector.

53. The system of claim 52, wherein each of the defined vectors are normalized to have a same selected length.

54. The system of claim 43, the processor further to utilize the detected vector to evaluate a condition and to classifying the posture state of the patient based on the evaluation.

55. The system of claim 54, wherein the evaluation determines at least one of a size and shape of a region in space, the processor further to classify the posture state of the patient based on a relationship of the detected vector to the region in space as indicated by the derived similarities.

56. The system of claim 43, wherein the detected vector is indicative of acceleration or velocity.

57. The system of claim 47, wherein the posture state is which the patient is classified is indicative of posture and activity state.

58. A storage medium for storing instructions to cause a digital processor to:

obtain a defined vector indicative of a defined posture state;

obtain from the sensor a detected vector that is indicative of a posture state of a patient;

determine a similarity between the defined vector and the detected vector, the similarity being determined without deriving an angle;

classify the posture state of the patient based on the similarity; and

initiate an action related to at least one of caring for, collecting data describing, and diagnosing, the patient.