A configurable handheld biological analyzer for identifying biological products based on Raman spectroscopy using artificial intelligence.

Abstract

A configurable handheld biological analyzer and associated biological analysis method for identifying biological products based on Raman spectroscopy using ensemble artificial intelligence (AI) are described. A biological classification ensemble model configuration (103) is loaded into the computer memory (108) of a configurable handheld biological analyzer (102) having a processor (110) and a scanner (106). The biological ensemble classification model configuration includes a biological classification ensemble model, including unsupervised and supervised models. The processor is configured to receive a dataset of Raman-based spectra defining a biological product sample scanned by the scanner and to run a spectral preprocessing algorithm for the biological ensemble classification model configured to reduce spectral inconsistencies in the dataset of Raman-based spectra. The biological ensemble classification model identifies biological product types based on the dataset of Raman-based spectra.

Claims

1. A configurable handheld biological analyzer for identifying biological products based on Raman spectroscopy using ensemble artificial intelligence (AI), A first housing adapted for handheld operation, A first scanner held by the first housing, A first processor is communicatively connected to the first scanner, A first computer memory, which is communicably connected to the first processor, Equipped with, The first computer memory is configured to load a biological ensemble classification model configuration, the biological ensemble classification model configuration includes a biological classification ensemble model that includes an unsupervised model and a supervised model, The unsupervised model is trained using Raman-based spectral training data to configure the unsupervised model to output a first indicator of one or more biological product types. The supervised model is trained using Raman-based spectral training data to configure the supervised model to output a second indicator of one or more biological product types. The biological classification ensemble model configuration further includes one or more spectral preprocessing algorithms, and the first processor is configured to execute the one or more spectral preprocessing algorithms when the first Raman-based spectral dataset is received by the first processor to reduce spectral inconsistencies in the first Raman-based spectral dataset. A configurable handheld biological analyzer, wherein the biological classification ensemble model is configured to run on the processor, and the first processor is configured to (1) receive a dataset of first Raman-based spectra defining a first biological product sample scanned by the first scanner, and (2) use the biological classification ensemble model to identify a biological product type from among the one or more biological product types based on the dataset of first Raman-based spectra.

2. The aforementioned biological ensemble classification model configuration can be electronically transferred to a second configurable handheld biological analyzer, and the second configurable handheld biological analyzer is A second housing adapted for handheld operation, A second scanner connected to the aforementioned second housing, A second processor is communicatively connected to the second scanner, A second computer memory, which is communicably connected to the second processor, Equipped with, The second computer memory is configured to load the biological classification ensemble model configuration, the biological classification ensemble model configuration includes the biological classification ensemble model, the biological classification ensemble model is configured to run on the second processor, the second processor is configured to (1) receive a second Raman-based spectrum dataset defining a second biological product sample scanned by the second scanner, and (2) use the biological classification ensemble model to identify the biological product type based on the second Raman-based spectrum dataset. The configurable handheld biological analyzer according to claim 1, wherein the second biological product sample is a novel sample of the biological product type.

3. The spectral mismatch is an instrument-to-instrument spectral mismatch between the first Raman-based spectral dataset and one or more other Raman-based spectral datasets of one or more corresponding other handheld biological analyzers, where each of the one or more other Raman-based spectral datasets represents the biological product type. The configurable handheld biological analyzer according to claim 1, wherein the one or more spectral preprocessing algorithms are configured to reduce spectral mismatches between the analyzer and the first Raman-based spectral dataset and the one or more other Raman-based spectral datasets.

4. The one or more spectral preprocessing algorithms described above are: Applying a derivative transformation to the first Raman-based spectrum dataset generates a modified Raman-based spectrum dataset, Aligning the modified Raman-based spectrum dataset across the Raman shift axis, The modified Raman-based spectral dataset is normalized across the Raman intensity axis, A configurable handheld biological analyzer according to claim 3, including the following:

5. The modified Raman-based spectral dataset is centered in the configurable handheld biological analyzer according to claim 4.

6. The configurable handheld biological analyzer according to claim 4, wherein the derivative transformation is applied to a continuous group of 5 to 15 Raman intensity values ​​across the Raman shift axis.

7. The configurable handheld biological analyzer according to claim 5, wherein the corresponding derivatives of the continuous group of Raman intensity values ​​5 to 15 are specified across the Raman shift axis.

8. The configurable handheld biological analyzer according to any one of claims 1 to 7, wherein the unsupervised model is configured to detect variability associated with the identification of one or more biological product types.

9. The configurable handheld biological analyzer according to claim 8, wherein the aforementioned variation includes instrument variation or variation between sample lots.

10. The configurable handheld biological analyzer according to any one of claims 1 to 9, wherein the biological classification ensemble model identifies the biological product type when it is determined that the first indicator passes a first pass / fail threshold and the second indicator passes a second pass / fail threshold.

11. The first indicator output by the unsupervised model is based on whether one or more biological product types satisfy a threshold, as configurable handheld biological analyzer according to any one of claims 1 to 9.

12. The unsupervised model outputs a pass / fail judgment based on the threshold, as described in claim 11, a configurable handheld biological analyzer.

13. The configurable handheld biological analyzer according to claim 11 or 12, wherein the threshold is based on one or more of the following: reduced Q residual error, Hotelling's T squared value, Mahalanobis distance value, or a specific range of principal component scores.

14. A configurable handheld biological analyzer according to any one of claims 1 to 13, wherein the first biological product type and the second biological product type of the one or more biological product types have similar Raman-based spectra.

15. The configurable handheld biological analyzer according to any one of claims 1 to 14, wherein the second indicator output by the supervised model is based on whether one or more biological product types satisfy a biological product type prediction threshold.

16. The configurable handheld biological analyzer according to claim 15, wherein the supervised model outputs a pass / fail judgment based on the biological product type prediction threshold.

17. The configurable handheld biological analyzer according to any one of claims 1 to 16, wherein the computer memory is configured to load a novel biological ensemble classification model, the novel biological ensemble classification model includes an updated unsupervised model and / or an updated supervised model.

18. The configurable handheld biological analyzer according to any one of claims 1 to 17, wherein the biological classification ensemble model configuration is implemented in Extended Markup Language (XML) format.

19. A configurable handheld biological analyzer according to any one of claims 1 to 18, wherein the type of biological product is a therapeutic product.

20. The configurable handheld biological analyzer according to any one of claims 1 to 19, wherein the biological product type is identified by the biological classification ensemble model during the production of a biological product having the biological product type.

21. The configurable handheld biological analyzer according to any one of claims 1 to 20, wherein the supervised model of the biological classification ensemble model is configured to distinguish the first biological product sample having the biological product type from different biological product samples having different biological product types.

22. The configurable handheld biological analyzer according to claim 21, wherein the biological product type and the different biological product types each have distinct local characteristics within similar Raman spectral ranges.

23. The configurable handheld biological analyzer according to any one of claims 1 to 22, wherein the biological classification ensemble model is generated by a remote processor that is remote to the configurable handheld biological analyzer.

24. The configurable handheld biological analyzer according to any one of claims 1 to 23, wherein the unsupervised model is based on principal component analysis (PCA), Euclidean distance or correlation, neighborhood-based algorithm, K-means algorithm, quality threshold (QT) algorithm, centroid algorithm, Ward algorithm, or fuzzy C-means clustering algorithm.

25. The unsupervised model is a PCA model, and the PCA model includes a reduced set of principal components, according to the configurable handheld biological analyzer according to claim 24.

26. The configurable handheld biological analyzer according to any one of claims 1 to 25, wherein the supervised model is trained using a partially least squares discriminant analysis (PLSDA), linear discriminant analysis (LDA), K nearest neighbor (KNN) algorithm, soft independent modeling by class analogy (SIMCA), or logistic regression discriminant analysis (LREGDA) algorithm.

27. The supervised model is a PLSDA model, and the PLSDA model includes a reduced set of latent variables, according to the configurable handheld biological analyzer according to claim 26.

28. The configurable handheld biological analyzer according to any one of claims 1 to 27, wherein the unsupervised model is constructed based on principal component analysis (PCA), and the supervised model is constructed using partial least squares discriminant analysis (PLSDA).

29. A configurable handheld biological analyzer according to any one of claims 1 to 28, wherein the one or more spectral preprocessing algorithms are performed to modify at least one of the training data used to train either or both of the supervised model or the unsupervised model, or (b) the product data used to generate an output from either or both of the supervised model or the unsupervised model.

30. A biological analysis method for identifying biological products based on Raman spectroscopy using ensemble artificial intelligence (AI), Loading a biological ensemble classification model configuration into the first computer memory of a first configurable handheld biological analyzer having a first processor and a first scanner, wherein the biological ensemble classification model configuration includes a biological classification ensemble model that includes an unsupervised model and a supervised model, The unsupervised model is trained using Raman-based spectral training data to configure the unsupervised model to output a first indicator of one or more biological product types. The supervised model is trained using Raman-based spectral training data to configure the supervised model to output a second indicator of one or more biological product types, and is loaded. The first processor receives a dataset of first Raman-based spectra defining a first biological product sample scanned by the first scanner, The first processor executes one or more spectral preprocessing algorithms specified by the biological ensemble classification model configuration to reduce spectral inconsistencies in the first Raman-based spectral dataset, Using the aforementioned biological classification ensemble model, the biological product type is identified based on the first Raman-based spectral dataset, Biological analytical methods, including those mentioned above.

31. Transferring the aforementioned biological ensemble classification model configuration to a second configurable handheld biological analyzer, Loading the biological classification ensemble model configuration into a second computer memory, wherein the biological classification ensemble model configuration includes the biological classification ensemble model, The second processor of the second configurable handheld biological analyzer receives a dataset of second Raman-based spectra defining a second biological product sample scanned by the second scanner, The second processor performs the biological classification ensemble model to identify the biological product type based on the second Raman-based spectral dataset, It further includes, The biological analysis method according to claim 30, wherein the second biological product sample is a novel sample of the biological product type.

32. The spectral mismatch is an instrument-to-instrument spectral mismatch between the first Raman-based spectral dataset and one or more other Raman-based spectral datasets of one or more corresponding other handheld biological analyzers, where each of the one or more other Raman-based spectral datasets represents the biological product type. The biological analysis method according to claim 30, wherein the one or more spectral preprocessing algorithms are configured to reduce spectral mismatches between the analyzer and the first Raman-based spectral dataset and the one or more other Raman-based spectral datasets.

33. The one or more spectral preprocessing algorithms described above are: Applying a derivative transformation to the first Raman-based spectrum dataset generates a modified Raman-based spectrum dataset, Aligning the modified Raman-based spectrum dataset across the Raman shift axis, The modified Raman-based spectral dataset is normalized across the Raman intensity axis, The biological analysis method according to claim 32, including the method described in claim 32.

34. The modified Raman-based spectral dataset is centered, according to the biological analysis method of claim 33.

35. The biological analysis method according to claim 33, wherein the derivative transformation is applied to a continuous group of 5 to 15 Raman intensity values ​​across the Raman shift axis.

36. The biological analysis method according to claim 34, wherein the corresponding derivatives of the continuous group of Raman intensity values ​​5 to 15 are identified across the Raman shift axis.

37. The biological analysis method according to any one of claims 30 to 36, wherein the unsupervised model is configured to detect variability associated with the identification of one or more biological product types.

38. The biological analysis method according to claim 37, wherein the aforementioned variation includes instrument variation or variation between sample lots.

39. The biological analysis method according to any one of claims 30 to 38, wherein the biological classification ensemble model identifies the biological product type when it is determined that the first indicator passes a first pass / fail threshold and the second indicator passes a second pass / fail threshold.

40. The first indicator output by the unsupervised model is based on whether one or more biological product types satisfy a threshold, according to any one of claims 30 to 38, for the biological analysis method according to any one of claims.

41. The biological analysis method according to claim 40, wherein the unsupervised model outputs a pass / fail judgment based on the threshold.

42. The biological analysis method according to claim 40 or 41, wherein the threshold is based on one or more of the following: reduced Q residual error, Hotelling's T squared value, Mahalanobis distance value, or a specific range of principal component scores.

43. The biological analysis method according to any one of claims 30 to 42, wherein the first biological product type and the second biological product type of the one or more biological product types have similar Raman-based spectra.

44. The biological analysis method according to any one of claims 30 to 43, wherein the second indicator output by the supervised model is based on whether or not one or more biological product types satisfy a biological product type prediction threshold.

45. The biological analysis method according to claim 44, wherein the supervised model outputs a pass / fail judgment based on the biological product type prediction threshold.

46. The method of biological analysis according to any one of claims 30 to 45, wherein the computer memory is configured to load a new biological classification ensemble model, and the new biological classification ensemble model includes an updated unsupervised model and / or an updated supervised model.

47. The biological analysis method according to any one of claims 30 to 46, wherein the biological classification ensemble model configuration is implemented in Extended Markup Language (XML) format.

48. The biological analysis method according to any one of claims 30 to 47, wherein the type of biological product is a therapeutic product.

49. The biological analysis method according to any one of claims 30 to 48, wherein the biological product type is identified by the biological classification ensemble model during the production of a biological product having the biological product type.

50. The biological analysis method according to any one of claims 30 to 49, wherein the supervised model of the biological classification ensemble model is configured to distinguish the first biological product sample having the biological product type from different biological product samples having different biological product types.

51. The biological analysis method according to claim 50, wherein the biological product type and the different biological product types each have distinct local characteristics within similar Raman spectral ranges.

52. The biological analysis method according to any one of claims 30 to 51, wherein the biological classification ensemble model is generated by a remote processor that is remote to the configurable handheld biological analyzer.

53. The biological analysis method according to any one of claims 30 to 52, wherein the unsupervised model is based on principal component analysis (PCA), Euclidean distance or correlation, neighborhood-based algorithm, K-means algorithm, quality threshold (QT) algorithm, centroid algorithm, Ward algorithm, or fuzzy C-means clustering algorithm.

54. The unsupervised model is a PCA model, and the PCA model includes a reduced set of principal components, according to claim 53, the method of biological analysis.

55. The method of biological analysis according to any one of claims 30 to 54, wherein the supervised model is trained using a partially least squares discriminant analysis (PLSDA), linear discriminant analysis (LDA), K nearest neighbor (KNN) algorithm, soft independent modeling by class analogy (SIMCA), or logistic regression discriminant analysis (LREGDA) algorithm.

56. The method of biological analysis according to claim 55, wherein the supervised model is a PLSDA model, and the PLSDA model includes a reduced set of latent variables.

57. The biological analysis method according to any one of claims 30 to 56, wherein the unsupervised model is constructed based on principal component analysis (PCA), and the supervised model is constructed using partial least squares discriminant analysis (PLSDA).

58. The biological analysis method according to any one of claims 30 to 57, wherein the one or more spectral preprocessing algorithms are performed to modify at least one of the training data used to train either or both of the supervised model or the unsupervised model, or (b) the product data used to generate an output from either or both of the supervised model or the unsupervised model.

59. A tangible, non-transient, computer-readable medium storing instructions for identifying biological products based on Raman spectroscopy using ensemble artificial intelligence (AI), wherein the instructions, when executed by one or more processors of a configurable handheld biological analyzer, are transmitted to one or more processors of the configurable handheld biological analyzer. Loading a biological ensemble classification model configuration into the first computer memory of a first configurable handheld biological analyzer having a first processor and a first scanner, wherein the biological ensemble classification model configuration includes a biological classification ensemble model that includes an unsupervised model and a supervised model, The unsupervised model is trained using Raman-based spectral training data to configure the unsupervised model to output a first indicator of one or more biological product types. The supervised model is trained using Raman-based spectral training data to configure the supervised model to output a second indicator of one or more biological product types, and is loaded. The first processor receives a dataset of first Raman-based spectra defining a first biological product sample scanned by the first scanner, The first processor executes one or more spectral preprocessing algorithms specified by the biological ensemble classification model configuration to reduce spectral inconsistencies in the first Raman-based spectral dataset, Using the aforementioned biological classification ensemble model, the biological product type is identified based on the first Raman-based spectral dataset, A tangible, non-temporary, computer-readable medium that enables the operation of a computer.

Images