Self-adaptive system for the analysis of biomedical signals of a patient

• US 7,640,055 B2
• Filed: 10/02/2003
• Issued: 12/29/2009
• Est. Priority Date: 01/07/2002
• Status: Active Grant

• Current Assignee: Widemed Technologies Ltd.
• Original Assignee: Widemed Ltd.
• Inventors: Mergi, Arye, Pressman, Assaf, Levi, Baruch, Todros, Kobi, Geva, Amir
• Primary Examiner(s): Mallari; Patricia
• Assistant Examiner(s): Toth; Karen E
• Application Number:
  US10/677,176
• Publication Number:
  US 20040073098A1
• Time in Patent Office: 2,280 Days
• Field of Search: 600/300, 600/544
• US Class Current: 600/544
• CPC Class Codes: A61B 5/00 Measuring for diagnostic purposes radiation diagnosis A61B6/00; diagnosis by ultrasonic, sonic or infrasonic waves A61B8/00; Identification of persons
• CPC Class Codes: A61B 5/103 Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes A61B5/08 takes precedence; measuring aids for tailors A41H1/00; measuring instruments specially adapted for dentistry A61C19/04
• CPC Class Codes: A61B 5/24 Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
• CPC Class Codes: A61B 5/349 Detecting specific parameters of the electrocardiograph cycle
• CPC Class Codes: A61B 5/7203 for noise prevention, reduction or removal
• CPC Class Codes: A61B 5/7207 of noise induced by motion artifacts
• CPC Class Codes: A61B 5/726 using Wavelet transforms
• CPC Class Codes: A61B 5/7267 involving training the classification device
• CPC Class Codes: A61M 16/0051 with alarm devices
• CPC Class Codes: A61M 16/026 specially adapted for predicting, e.g. for determining an information representative of a flow limitation during a ventilation cycle by using a root square technique or a regression analysis
• CPC Class Codes: G16H 15/00 ICT specially adapted for medical reports, e.g. generation or transmission thereof
• CPC Class Codes: G16H 40/63 for local operation

Abstract

Methods and systems that (a) adaptively segment a raw data stream(s) of different type biomedical signals; (b) assign attribute values to each segment; (c) determine an attribute domain based on a point corresponding to said segment attribute value; (d) generate a cluster set(s) in said attribute domain that includes a combination of points; (e) obtain a probability of order of appearance of each cluster point, according to its property value; (f) use said probability to update each point'"'"'s property value; (g) repeat (d) through (f) while varying combinations of points in clusters according to their most updated property values and points derived from additional raw data stream adaptive segmentations; (h) associate at least one updated cluster with a normal/abnormal physiological state based on reference clusters in said domain; and (i) obtain the change probability between normal/abnormal physiological states using said probability of the order of appearance.

Claims

1. A method for sleep diagnosis, comprising:

using one or more sensors, receiving biomedical signals from a sleeping patient over a given time period;

using a processor, extracting features from the biomedical signals and grouping the features into multiple clusters;

using the processor, assigning respective vectors of membership values to a plurality of intervals within the given time period, each membership value indicating a degree of association between the biomedical signals in a corresponding interval and a respective cluster among the multiple clusters; and

with the processor, determining a condition of the patient responsively to the vectors using a state machine model which comprises machine states corresponding to physiological states of the patient.

2. The method according to claim 1, wherein evaluating the condition comprises classifying sleep stages of the patient based on the vectors.

3. The method according to claim 1, wherein receiving the biomedical signals comprises receiving electroencephalogram (EEG) signals.

4. The method according to claim 1, wherein extracting the features comprises adaptively segmenting the biomedical signals into a sequence of segments, and determining the features of each of the segments.

5. The method according to claim 1, wherein assigning the respective vectors comprises assigning fuzzy membership values to each of the intervals.

6. A method for sleep diagnosis, comprising:

using a processor, defining a state machine model comprising machine states corresponding to physiological states of a patient, and comprising respective probabilities of transitions among the states;

using one or more sensors, receiving biomedical signals from a sleeping patient over a given time period;

using the processor, segmenting the biomedical signals into a sequence of segments, and determining one or more respective features of each of the segments;

using the processor, assigning respective state classifications to the segments responsively to the respective features and to the probabilities of the transitions among the states of the state machine model; and

using the processor, evaluating a condition of the patient during the given time period responsively to the state classifications.

7. The method according to claim 6, wherein the physiological states comprise sleep stages, and wherein evaluating the condition comprises classifying sleep stages of the patient based on the state classifications.

8. The method according to claim 6, wherein defining the state machine model comprises defining a Hidden Markov Model (HMM).

9. The method according to claim 6, wherein assigning the respective state classifications comprises assigning respective fuzzy membership values to the segments, and determining the state classifications responsively to the fuzzy membership values.

10. Apparatus for sleep diagnosis, comprising:

one or more sensors, which are adapted to receive biomedical signals from a sleeping patient over a given time period; and

a processor, which is adapted to extract features from the biomedical signals and to group the features into multiple clusters, to assign respective vectors of membership values to a plurality of intervals within the given time period, each membership value indicating a degree of association between the biomedical signals in a corresponding interval and a respective cluster among the multiple clusters, and to evaluate a condition of the patient responsively to the vectors,

wherein the processor is adapted to receive a state machine model comprising machine states corresponding to physiological states of the patient, and to determine the condition using the state machine model.

11. Apparatus for sleep diagnosis, comprising:

one or more sensors, which are adapted to receive biomedical signals from a sleeping patient over a given time period; and

a processor, which is adapted to receive a definition of a state machine model comprising machine states corresponding to physiological states of a patient, the model comprising respective probabilities of transitions among the states,

wherein the processor is adapted to segment the biomedical signals into a sequence of segments and determine one or more respective features of each of the segments, to assign respective state classifications to the segments responsively to the respective features and to the probabilities of the transitions among the states of the state machine model, and evaluate a condition of the patient during the given time period responsively to the state classifications.

12. The apparatus according to claim 11, wherein the physiological states comprise sleep stages, and wherein the processor is adapted to classify sleep stages of the patient based on the state classifications.

13. The apparatus according to claim 11, wherein the state machine model comprises a Hidden Markov Model (HMM).

14. The apparatus according to claim 11, wherein the processor is adapted to assign respective fuzzy membership values to the segments, and to determine the state classifications responsively to the fuzzy membership values.