Apparatus and method for determining a physiological condition

• US 10,524,669 B2
• Filed: 10/16/2013
• Issued: 01/07/2020
• Est. Priority Date: 10/13/2003
• Status: Active Grant

• Current Assignee: Novo Nordisk A/S
• Original Assignee: Novo Nordisk A/S
• Inventors: Randlov, Jette, Hansen, Jon Ulrich, Poulsen, Jens Ulrik
• Primary Examiner(s): Patel, Niketa I
• Assistant Examiner(s): Baldwin, Nathan A
• Application Number:
  US14/055,508
• Publication Number:
  US 20140058218A1
• Time in Patent Office: 2,274 Days
• 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 5/14532 for measuring glucose, e.g. by tissue impedance measurement
• CPC Class Codes: G16H 40/63 for local operation
• CPC Class Codes: G16Z 99/00 Subject matter not provided for in other main groups of this subclass

Abstract

An apparatus for detecting a physiological condition in a person, comprising input structure for receiving at least two physiological parameters, processing structure for determining, from the received physiological parameters, a result signal indicative of the presence or onset of said physiological condition, and output structure for generating an output indicative of the result signal. The apparatus further comprises storage structure for storing a look-up table comprising a plurality of entries, each entry corresponding to a pre-calculated probability of the presence or onset of said physiological condition, the pre-calculated probabilities being determined by way of a mathematical process from previously monitored physiological parameters; and the processing means is adapted to process the received at least two physiological parameters to identify one of the plurality of entries of the look-up table, and to determine said result signal from the identified entry of the look-up table.

Claims

1. A method of alerting a patient to the probability of the presence of hypoglycaemia, the method comprising:

receiving, via an input of an apparatus, at least two physiological parameters corresponding to a patient state at time T_s from a group of physiological parameters comprising heart rate or pulse, heart rate variability, skin impedance, perspiration, body temperature, skin temperature, respiration frequency, respiration effort, eye movement, muscle tonus, a measurement obtained by electrooculogram (EOG), an oxygen saturation measurement, blood pressure, a carbon dioxide content of blood, and a non-invasive blood glucose measurement, and a blood glucose level;

using a processor of the apparatus, determining, from the at least two physiological parameters, a result signal indicative of the presence or onset of hypoglycaemia by processing the received physiological parameters to identify one of a plurality of entries of a look-up table, wherein each entry of the look-up up table begins as an initial set of pre-calculated probabilities of the presence or onset of hypoglycaemia, the pre-calculated probabilities being determined by an adaptive learning algorithm from previously monitored physiological parameters of the person;

wherein a loop begins with a counter τ

at the beginning of a predetermined period of time and the at least two physiological parameters are modified by a weighting parameter λ

, wherein the most recent physiological parameters are weighted more than the older physiological parameters wherein the counter τ

is provided as;

τ

=T_s−

30

wherein the predetermined period of time is 30 minutes, and if τ

<

1, τ

is set to 1 instead;

wherein the initial set of pre-calculated probabilities are adjusted by a correction term γ

that determines the relative weight of the probabilities at time τ and

at time τ

−

1;

wherein entries of the look-up table are subsequently updated during operation based on feedback received from the patient, thereby providing an adaptive detection apparatus where the probabilities in the look-up table are both adaptive with respect to the individual patient and with respect to a development of the patient over time; and

generating a notification, using the apparatus, of the result signal indicative of the presence or onset of hypoglycaemia.

2. A method according to claim 1, further comprising receiving a measured quantity indicative of the presence of hypoglycaemia.

3. A method according to claim 2, wherein the measured quantity is a measured blood glucose level.

4. A method according to claim 1, wherein the weighting parameter λ

is provided as;

τ

=1+(τ

−

T_s)/30

wherein the predetermined period of time is 30 minutes and the weighting parameter λ

is linearly increased from λ

=0 at time T_s−

30, to λ

=1 at time T_s, wherein the most recent physiological parameters are weighted more than the older physiological parameters.

5. A method according to claim 1, wherein the probabilities are adjusted by the correction term γ

modified by a learning rate a wherein the probability is adjusted according to the equation;

Δ

P_τ

=P<sub>i(τ

)</sub>+α

(r+γ

P<sub>i(τ

)</sub>−

P<sub>i(τ

−

1)</sub>)

wherein the correction term γ

determines the relative weight of the probabilities at time τ and

at time τ

−

1 and has a value between 0.97 and 0.98, and the learning rate a determines the overall strength of the correction and has a value between 0.4 and 0.6.

6. A method for detecting a probability of hypoglycaemia in a person, the method comprising:

providing a storage device adapted to store a look-up table comprising a plurality of entries, wherein each entry is a pre-calculated probability of the presence or onset of hypoglycaemia and each entry begins as an initial set of pre-calculated probabilities, the pre-calculated probabilities being determined by an adaptive learning algorithm from previously monitored physiological parameters of the person corresponding to a patient state at time T_s;

providing a processor adapted to receive at least two physiological parameters and identify one of the plurality of entries of the look-up table, wherein the identification comprises comparison of at least two physiological parameters against stored values of the look-up table, and wherein the processor is further adapted to determine, from the received physiological parameters, a result signal indicative of the presence or onset of hypoglycaemia wherein the entries of the look-up table are subsequently updated during operation based on feedback received from the patient;

wherein a loop begins with a counter τ

at the beginning of a predetermined period of time and the at least two physiological parameters are modified by a weighting parameter λ

, wherein the most recent physiological parameters are weighted more than the older physiological parameters wherein the counter τ

is provided as;

τ

=T_s−

30

wherein the predetermined period of time is 30 minutes, and if τ

<

1, τ

is set to 1 instead;

wherein the initial set of pre-calculated probabilities are adjusted by a correction term γ

that determines the relative weight of the probabilities at time τ and

at time τ

−

1;

wherein entries of the look-up table are subsequently updated during operation based on feedback received from the patient, thereby providing an adaptive detection apparatus where the probabilities in the look-up table are both adaptive with respect to the individual patient and with respect to a development of the patient over time;

providing an output device adapted to generate an output indicative of the result signal corresponding to a pre-calculated probability of the presence or onset of hypoglycaemia;

receiving a user input indicative of the presence or absence of hypoglycaemia wherein the processor is further adapted to update at least the identified entry of the look-up table according to the adaptive learning algorithm process and in response to a user input indicative of the presence or absence of hypoglycaemia and/or a received measured quantity indicative of the presence of hypoglycaemia.

7. A method according to claim 6, wherein the adaptive learning algorithm comprises a reinforcement learning process.

8. A method according to claim 6, wherein the processor is further adapted to discretize the received physiological parameters to obtain one of a plurality of possible combinations of discrete physiological parameters, each of the plurality of combinations being associated to a corresponding one of the plurality of entries of the look-up table.

9. A method according to claim 8, wherein the processor is adapted to discretize the received physiological parameters by identifying, for each of the received physiological parameters, one of a corresponding number of intervals in which the received physiological parameter is determined to fall;

resulting in a combination of identified intervals, the combination identifying a corresponding entry in the look-up table.

10. A method according to claim 9, wherein each of the physiological parameters are discretized with respect to 2-10 respective intervals.

11. A method according claim 6, further comprising an input for receiving the at least two physiological parameters.

12. A method according to claim 6, wherein the at least two physiological parameters consist of 2-10 parameters.

13. A method according to claim 12, where the at least two physiological parameters are selected from the set of parameters consisting of pulse, heart rate variability, skin temperature, skin impedance, respiration frequency, eye movements, muscle tonus, CO₂ content of the blood, a blood glucose measurement.

14. A method according to claim 13, wherein one of the at least two physiological parameters is a non-invasive blood glucose measurement.

15. A method according to claim 6, wherein the at least two physiological parameters further include the rate of change of at least one of the physiological parameters.

16. A method according to claim 6, wherein the processor is further adapted to compare the identified entry of the look-up table with a predetermined threshold and where the result signal is indicative of the result of said comparison.

17. A method according to claim 6, wherein the output device comprises a structure for generating an audible alarm.

18. A method according to claim 6, wherein the weighting parameter λ

is provided as;

τ

=1+(τ

−

T_s)/30

wherein the predetermined period of time is 30 minutes and the weighting parameter λ

is linearly increased from λ

=0 at time T_s−

30, to λ

=1 at time T_s, wherein the most recent physiological parameters are weighted more than the older physiological parameters.

19. A method according to claim 6, wherein the probabilities are adjusted by the correction term γ

modified by a learning rate a wherein the probability is adjusted according to the equation;

Δ

P_τ

=P<sub>i(τ

)</sub>+α

(r+γ

P<sub>i(τ

)</sub>−

P<sub>i(τ

−

1)</sub>)

wherein the correction term γ

determines the relative weight of the probabilities at time τ and

at time τ

−

1 and has a value between 0.97 and 0.98, and the learning rate a determines the overall strength of the correction and has a value between 0.4 and 0.6.