Systems and methods for intravenous drug management using immittance spectroscopy

• US 8,838,395 B2
• Filed: 09/09/2011
• Issued: 09/16/2014
• Est. Priority Date: 09/09/2010
• Status: Active Grant

• Current Assignee: James W. Bennett, Leonid Matsiev
• Original Assignee: S.E.A. Medical Systems Inc.
• Inventors: Matsiev, Leonid F., Weickert, Michael J., Bennett, James W., Smith, Matthew F., Litvintseva, Svetlana, Blanke, Kit
• Primary Examiner(s): Sirmons, Kevin C
• Assistant Examiner(s): Hall, Deanna K
• Application Number:
  US13/229,597
• Publication Number:
  US 20120065617A1
• Time in Patent Office: 1,103 Days
• Field of Search: 702/1, 702/22, 702/23
• US Class Current: 702/22
• CPC Class Codes: A61M 2205/0244 Micromachined materials, e.g. made from silicon wafers, microelectromechanical systems [MEMS] or comprising nanotechnology
• CPC Class Codes: A61M 2205/3306 Optical measuring means
• CPC Class Codes: A61M 2205/3313 used specific wavelengths
• CPC Class Codes: A61M 2205/3553 remote, e.g. between patient's home and doctor's office
• CPC Class Codes: A61M 2205/3561 local, e.g. within room or hospital
• CPC Class Codes: A61M 2205/6018 providing set-up signals for the apparatus configuration
• CPC Class Codes: A61M 5/142 Pressure infusion, e.g. using pumps
• CPC Class Codes: G01N 27/026 Dielectric impedance spectroscopy electrochemical impedance spectroscopy for measuring corrosion G01N17/02
• CPC Class Codes: G01N 27/4166 measuring a particular property of an electrolyte
• CPC Class Codes: Y02A 90/10 Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Abstract

Described herein are devices, systems, and methods for determining the composition of liquids, including the identity of one or more drugs in the liquid, the concentration of the drug, and the type of diluent using immittance spectroscopy. These devices, systems and methods are particularly useful for describing the identity and, in some variations, concentration of one or more components of a medical liquid such as intravenous fluid. In particular, described herein are devices, systems and methods that may operate in low ionic strength diluents. Also described are methods of recognizing complex immittance spectrograph patterns to determine the composition of a liquid by pattern recognition.

Claims

1. A method of determining the identity of a drug or drug formulation by recognizing a pattern of complex immittance from a library of known complex immittances, the method comprising:

receiving an initial dataset comprising complex immittance spectrographic information for an unknown liquid sample, the complex immittance spectrographic information taken from a plurality of different electrode pairs each at a plurality of different frequencies wherein each pair of electrodes had different surface properties arising from a different size, different shape, or different material relative to each other, and further wherein the complex immittance spectrographic information was recorded while the electrode pairs were in contact with the unknown liquid sample;

using a processor to apply one or more pattern recognition techniques to compare the initial dataset to an identification space database comprising a plurality of identification datasets wherein the identification datasets comprise complex immittance data corresponding to known drug compositions to determine if the initial dataset matches an identification dataset from the identification space database within a threshold range; and

reporting that the initial dataset does or does not match an identification dataset, and if the initial dataset does match an identification dataset within the threshold range, reporting which drug or drugs correspond to the identification dataset matched.

2. The method of claim 1, wherein using the processor to apply one or more pattern recognition techniques comprises using a Neural Network.

3. The method of claim 1, wherein using the processor to apply one or more pattern recognition techniques comprises using a Probabilistic Neural Network.

4. The method of claim 1, wherein using the processor to apply one or more pattern recognition techniques comprises reducing the dimension of the initial dataset and performing a regression analysis.

5. The method of claim 1, wherein receiving the initial dataset comprises receiving an initial dataset having greater than 30 dimensions.

6. The method of claim 1, further comprising setting the threshold range.

7. The method of claim 1, wherein using a processor to apply one or more pattern recognition techniques comprises applying two pattern recognition techniques.

8. The method of claim 1, wherein using the processor to apply one or more pattern recognition techniques comprises initially applying a PCA method to reduce the dimension of the data and then applying another pattern recognition technique to determine if the initial dataset matches an identification dataset.

9. The method of claim 1, wherein using the processor to apply one or more pattern recognition techniques comprises initially applying a PCA method to reduce the dimension of the dataset and then using a neural network to determine if the initial dataset matches an identification dataset.

10. The method of claim 1, wherein using the processor to apply one or more pattern recognition techniques comprises applying a linear technique selected from the group consisting of:

principal component analysis, factor analysis, projection pursuit, independent component analysis, multi-objective functions, one-unit objective functions, adaptive methods, batch-mode algorithms, and random projections methods.

11. The method of claim 1, wherein using the processor to apply one or more pattern recognition techniques comprises applying a non-linear technique selected from the group consisting of:

non-linear principle component analysis, non-linear independent component analysis, principle curves, multidimensional scaling, and topologically continuous maps.

12. The method of claim 1, further comprising interpolating to get an estimate of the concentration of the drug or drug corresponding to the matching identification dataset when the initial dataset matches the identification dataset within the threshold range.

13. The method of claim 1, wherein reporting that the initial dataset does or does not match an identification dataset comprises reporting the concentration of the drug or drugs correspond to the identification dataset when the initial dataset does match the identification dataset within the threshold range.

14. The method of claim 1, wherein using the processor to apply one or more pattern recognition techniques comprises reducing the initial dataset down to four dimensions.

15. A method of determining the identity of a drug or drug formulation by recognizing a pattern of complex immittance from a library of known complex immittances, the method comprising:

receiving an initial dataset comprising multi-dimensional, complex immittance spectrographic information for an unknown liquid sample, the complex immittance spectrographic information taken from a plurality of different electrode pairs each at a plurality of different frequencies wherein each pair of electrodes had different surface properties arising from a different size, different shape, or different material relative to each other, and further wherein the complex immittance spectrographic information was recorded while the electrode pairs were in contact with the unknown liquid sample;

reducing the dimensions of the initial dataset using a linear or non-linear technique to form a reduced dataset;

determining how closely the reduced dataset matches an identification dataset of an identification space database, wherein the identification space database comprises a plurality of identification datasets corresponding to known drug compositions; and

reporting that the known drug composition corresponding to the identification space database having the closest match to the reduced dataset if the closeness of the match is within a threshold range, or report that the unknown liquid sample does not match a known drug composition of those drugs included in the identification space database if the closeness of match is outside of the threshold range.

16. The method of claim 15, wherein reducing the dimensions of the initial dataset comprises applying a linear technique selected from the group consisting of:

principal component analysis, factor analysis, projection pursuit, independent component analysis, multi-objective functions, one-unit objective functions, adaptive methods, batch-mode algorithms, and random projections methods.

17. The method of claim 15, wherein reducing the dimensions of the initial dataset comprises applying a non-linear technique selected from the group consisting of:

non-linear principle component analysis, non-linear independent component analysis, principle curves, multidimensional scaling, and topologically continuous maps.

18. The method of claim 15, wherein reducing the dimensions of the initial dataset comprises reducing the initial dataset down to four dimensions.

19. A method of determining the identity and concentration of a drug by recognizing a pattern of complex immittance from a library of known complex immittances, the method comprising:

receiving an initial dataset comprising multi-dimensional, complex immittance spectrographic information for an unknown liquid sample, the complex immittance spectrographic information taken from a plurality of different electrode pairs each at a plurality of different frequencies wherein each pair of electrodes had different surface properties arising from a different size, different shape, or different material relative to each other, and further wherein the complex immittance spectrographic information was recorded while the electrode pairs were in contact with the unknown liquid sample;

reducing the dimensions of the initial dataset using a linear or non-linear technique to form a reduced dataset;

matching the reduced dataset to an identification space database, the identification space database comprising a plurality of identification datasets corresponding to known drug compositions;

determining the closeness of the match for the reduced dataset relative to each of the identification datasets;

determining a proposed drug composition by applying a threshold to the closeness of the match for each of the identification datasets, wherein the proposed drug composition is unknown if the closeness of match is outside of the threshold range; and

determining a concentration of drug in the unknown liquid sample by applying a regression of the proposed drug composition for the known drug composition.