DETAILED ACTION
Response to Arguments
The amendments filed 12/30/2025 have been entered and made of record.
Applicant's amendments and arguments filed 12/30/2025 have been considered but are moot in view of the new ground(s) of rejection because the Applicant has amended at least independent claims, and Applicant's arguments in view of the amendments filed 12/30/2025 have been fully considered but they are not persuasive:
In the Applicant’s Remarks (on pages 9-12 of 15), Applicant also asserts that A) the cited references, particularly Desimone as modified by LING do not disclose new added limitation of “repetitively generates exhaled breath digital image objects at predefined time intervals based at least in part on breath continuously exhaled by a user in a breathing tube connected to the breath analyzer housing”,
However, the Examiner disagrees, because:
LING discloses repetitively generates exhaled breath digital image objects at predefined time intervals based at least in part on breath continuously exhaled by a user in a breathing tube connected to the breath analyzer housing (see LING: e.g., -- SERS spectra were measured at fixed time intervals of t=2, 5, 10, 15, 30, 60 and 90 mins. The SERS super-profiles at each time interval were analyzed using principal component analysis and compared with the blank super-profiles.—in [0107]-[0109]);
Applicant also asserts that B) there is no motivation to combine De Simone and LING,
However, the Examiner disagrees, because:
DE SIMONE and LING are combinable as they are in the same field of endeavor: machine learning model applied in assessments of breath analysis, therefore, the motivation to combine De Simone and LING is to analyze the spectra data in assessments of breath analysis both for identifying COVID-positive individuals and the diagnosis of COVID-19;
Therefore it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to further modify DE SIMONE’s apparatus using LING’s teachings by including surface-enhanced Raman scattering (SERS)-based chips images and spectra data to identify COVID-positive individuals using their breath volatile organic compounds (BVOCs) and corresponding a trained machine learning computing model to DE SIMONE’s data collection and analysis and Machine Learning model in order to predict the molecular identification and/or estimate the quantity of the one or more analytes and/or perform classification of a multi-analyte mixture (see LING: e.g. in abstract, [0107]-[0109], [0162], and [0175]);
Applicant further asserts that C) Combining De Simone and Ling would result in functional incompatibility.
However, the Examiner disagrees, because:
As discussed above addressing for the motivation and obviousness for the combination of De Simone and LING based on they are in the same field of endeavor, and provide the solutions to solve the exactly same both for identifying COVID-positive individuals and the diagnosis of COVID-19, so that Applicant’s (C) statement is baseless, and is not persuasive.
Furthermore, DE SIMONE as modified by LING further disclose the at least one spectrometric-based prediction data object provides predictions on health conditions of the user (see LING: e.g., -- providing the combined-SERS profile to a device configured with a model trained to identify the one or more analytes from the combined-SERS profile. The device may be a computer operable to process a SERS signal and capable of machine learning. The device may also be operable to carry out any chemometric analysis of a spectrum obtained via SERS, including the SERS signal and the combined-SERS profile.--, in [0162], and, -- (d) inputting the entire collective spectrum to a trained learning model and obtaining the molecular identification of and/or quantifying the one or more analytes; and (e) wherein the trained learning model having been trained through machine learning techniques to predict the molecular identification and/or estimate the quantity of the one or more analytes and/or perform classification of a multi-analyte mixture.--, in [0175]);
and transmit one or more health alerts to one or more user devices based at least in part on the at least one spectrometric-based prediction data object, wherein the one or more user devices are associated with at least one of users and healthcare providers (see DE SIMONE: e.g., --[0110] The report on the interpretation of diagnostic tests for SARS-CoV-2 (Spanish Society of Infectious Diseases and Clinical Microbiology, Instituto de Salud Carlos III) was considered for the training of the device.--, in [0110], and, -- f) processing said images by artificial intelligence to determine if said images are compatible with breath samples from sick people; g) giving a visual indication of the result.--, in claim 36, {apparently “a visual indication of the result” read on claimed “alert” transmitted to the physicians}; also see LING: e.g., --Participants with reported comorbidities are highlighted in purple, illustrating that the participants' existing health conditions do not affect their classification scores. It is to be highlighted that since the provision of this information is strictly voluntary, information for all 501 trial participants is not entirely available. Nonetheless, the accurate classification of the 70 participants with reported comorbidities indicates that the SERS sensor is able to identify specific differences in participants' breath profiles that were directly linked to whether they had COVID-19 or not.--, in [0138] {these reports are aligned with claimed “alert”}).
Therefore, claims 1-20 are still not patentably distinguishable over the prior art reference(s). Further discussions are addressed in the prior art rejection section below.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over DE SIMONE (US 20240264146 A1, Date Filed: 2022-06-03), and in view of LING (US 20240044802 A1, Date Filed: 2022-02-04).
Re Claim 1, DE SIMONE discloses an non-invasive breath analyzer apparatus (see DE SIMONE: e.g., Fig. 1, --device comprises a sample inlet, a carrier gas inlet, a homogenization sector and a gas outlet (1); at least one ionization chamber (2); optical sensor (3); and an image storage and processing system (4).--, in [0078]-[0079]; and, --[0069] The process of detecting diseases from breath using, preferably, the previously described disease detection device, comprises the following steps: [0070] a) providing a container with a breath sample; [0071] b) providing a carrier gas that mixes with the breath sample to carry said sample into an ionization chamber in a homogeneous and controlled manner; [0072] c) Ionizing said carrier gas and said breath by means of an electric arc; [0073] d) capturing and storing images of the plasma generated in said electric arc; [0074] e) evacuating the ionization chamber by circulating said pure carrier gas, in absence of a breath sample; [0075] f) processing said images by artificial intelligence to determine if said images are compatible with breath samples from sick people; [0076] g) giving a visual indication of the result.--, in [0069]-[0076]), comprising:
an image generating device that is positioned within a breath analyzer housing and generates an exhaled breath digital image data objects (see DE SIMONE: e.g.,-- provides a device that subjects said biological samples, preferably gaseous samples, more preferably breath samples, to an electrical discharge that is able to induce plasma.
[0018] Wherein said health disorders comprise diseases.
…an image sensor and means for process the images by artificial intelligence. Said electrical discharge could be a corona discharge. Therefore, this system could be named as a corona discharge induced plasma digital spectroscopy.
[0020] This system has a high reproducibility for diagnosing COVID-19. This technology could offer the possibility of diagnosing or screening health disorders in a non-invasive and rapid way with low-cost instruments.--, in [0017]-[0020];
-- [0070] a) providing a container with a breath sample; [0071] b) providing a carrier gas that mixes with the breath sample to carry said sample into an ionization chamber in a homogeneous and controlled manner; [0072] c) Ionizing said carrier gas and said breath by means of an electric arc; [0073] d) capturing and storing images of the plasma generated in said electric arc;--, in [0070]-[0073]; and, …[0098] To develop the model that generates the spectra of the images, images that represent designated “wavelengths” were used. These are the variables of the image training model. The procedure to generate the models is similar to what is done with the images to be studied: The “PIL” library is used to take the RGB composition of each pixel in the image…. [0099] The device takes images (and records a video with them) of the corona discharge, similar to the one shown in FIG. 2. These images are acquired through the “CV2” library--, in [0098]-[0099]); and
DE SIMONE however does not explicitly disclose repetitively generates exhaled breath digital image objects at predefined time intervals based at least in part on breath continuously exhaled by a user in a breathing tube connected to the breath analyzer housing;
LING discloses repetitively generates exhaled breath digital image objects at predefined time intervals based at least in part on breath continuously exhaled by a user in a breathing tube connected to the breath analyzer housing (see LING: e.g., -- SERS spectra were measured at fixed time intervals of t=2, 5, 10, 15, 30, 60 and 90 mins. The SERS super-profiles at each time interval were analyzed using principal component analysis and compared with the blank super-profiles.—in [0107]-[0109]);
DE SIMONE and LING are combinable as they are in the same field of endeavor: machine learning model applied in assessments of breath analysis, therefore, the motivation to combine De Simone and LING is to analyze the spectra data in assessments of breath analysis both for identifying COVID-positive individuals and the diagnosis of COVID-19;
Therefore it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to further modify DE SIMONE’s apparatus using LING’s teachings by including surface-enhanced Raman scattering (SERS)-based chips images and spectra data to identify COVID-positive individuals using their breath volatile organic compounds (BVOCs) and corresponding a trained machine learning computing model to DE SIMONE’s data collection and analysis and Machine Learning model in order to predict the molecular identification and/or estimate the quantity of the one or more analytes and/or perform classification of a multi-analyte mixture (see LING: e.g. in abstract, [0107]-[0109], [0162], and [0175])
DE SIMONE as modified by LING further disclose a spectrometric data analyzing device that is in electronic communication with the image generating device and comprises a processor and a memory storing a non-transitory program code, wherein the memory and the non-transitory program code are configured to, with the processor, cause the spectrometric data analyzing device to:
receive the exhaled breath digital image data objects from the image generating device (see DE SIMONE: e.g., Fig. 1, --device comprises a sample inlet, a carrier gas inlet, a homogenization sector and a gas outlet (1); at least one ionization chamber (2); optical sensor (3); and an image storage and processing system (4).--, in [0078]-[0079]; and, --[0069] The process of detecting diseases from breath using, preferably, the previously described disease detection device, comprises the following steps: [0070] a) providing a container with a breath sample; [0071] b) providing a carrier gas that mixes with the breath sample to carry said sample into an ionization chamber in a homogeneous and controlled manner; [0072] c) Ionizing said carrier gas and said breath by means of an electric arc; [0073] d) capturing and storing images of the plasma generated in said electric arc; [0074] e) evacuating the ionization chamber by circulating said pure carrier gas, in absence of a breath sample; [0075] f) processing said images by artificial intelligence to determine if said images are compatible with breath samples from sick people; [0076] g) giving a visual indication of the result.--, in [0069]-[0076];
also see: Fig. 3, and, --[0093] The image analysis process called digital spectroscopy is divided into three steps: [0094] a. Generation of the database for the image training that represents the wavelengths to generate the spectra. [0095] b. Generation of the spectrum of each training image, fitting of the data and cross validation of the model. [0096] c. Generation of spectra of the samples to be analyzed and their prediction. …[0098] To develop the model that generates the spectra of the images, images that represent designated “wavelengths” were used. These are the variables of the image training model. The procedure to generate the models is similar to what is done with the images to be studied: The “PIL” library is used to take the RGB composition of each pixel in the image…. [0099] The device takes images (and records a video with them) of the corona discharge, similar to the one shown in FIG. 2. These images are acquired through the “CV2” library…. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. This classification model will have two variables: POSITIVE (COVID-19)—NEGATIVE. The gridsearch_cv (cross validation) was used to iterate between different parameters and then obtain the optimal one. Cross validation was also used to check which was the best score obtained. In the event that the expected threshold were exceeded, it then proceeds with the last step. This model (from now on called entrenaMod) is used to predict which group the unknown measurements correspond to.
[0101] In the case of the samples to be analyzed, the procedure is similar as in the training step: the spectra table of each image is generated with the pixelMod and once the table is completed, it is predicted with the entrenaMod. This result will be exposed in a probabilistic way for each variable, and the resulting numbers are the average image probabilities of the measurement. The processing scheme is detailed in FIG. 3.--, in [0093]-[0101]);
generate a plurality of exhaled breath spectrometric data objects based at least in part on the exhaled breath digital image data objects (see DE SIMONE: e.g., -- Fig. 3, and, --[0093] The image analysis process called digital spectroscopy is divided into three steps: [0094] a. Generation of the database for the image training that represents the wavelengths to generate the spectra. [0095] b. Generation of the spectrum of each training image, fitting of the data and cross validation of the model. [0096] c. Generation of spectra of the samples to be analyzed and their prediction. …[0098] To develop the model that generates the spectra of the images, images that represent designated “wavelengths” were used. These are the variables of the image training model. The procedure to generate the models is similar to what is done with the images to be studied: The “PIL” library is used to take the RGB composition of each pixel in the image. The corresponding arbitrary length is assigned to each of these combinations of values. Once the dataframe with all the combinations has been generated, a Machine Learning (Random Forest) model is generated. This model (from now on called pixelMod) is later used to define which is the length that belongs to each pixel of the studied images.--, in [0093]-[0098]);
input the plurality of exhaled breath spectrometric data objects to at least one machine learning computing model (see DE SIMONE: e.g., -- Fig. 3, and, --[0093] The image analysis process called digital spectroscopy is divided into three steps: [0094] a. Generation of the database for the image training that represents the wavelengths to generate the spectra. [0095] b. Generation of the spectrum of each training image, fitting of the data and cross validation of the model. [0096] c. Generation of spectra of the samples to be analyzed and their prediction. …[0098] To develop the model that generates the spectra of the images, images that represent designated “wavelengths” were used. These are the variables of the image training model. The procedure to generate the models is similar to what is done with the images to be studied: The “PIL” library is used to take the RGB composition of each pixel in the image…. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. The corresponding arbitrary length is assigned to each of these combinations of values. Once the dataframe with all the combinations has been generated, a Machine Learning (Random Forest) model is generated. This model (from now on called pixelMod) is later used to define which is the length that belongs to each pixel of the studied images.--, in [0093]-[0098], and, …. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. This classification model will have two variables: POSITIVE (COVID-19)—NEGATIVE. The gridsearch_cv (cross validation) was used to iterate between different parameters and then obtain the optimal one. Cross validation was also used to check which was the best score obtained. In the event that the expected threshold were exceeded, it then proceeds with the last step. This model (from now on called entrenaMod) is used to predict which group the unknown measurements correspond to.
[0101] In the case of the samples to be analyzed, the procedure is similar as in the training step: the spectra table of each image is generated with the pixelMod and once the table is completed, it is predicted with the entrenaMod. This result will be exposed in a probabilistic way for each variable, and the resulting numbers are the average image probabilities of the measurement. The processing scheme is detailed in FIG. 3.--, in [0100]-[0101], [0104]);
DE SIMONE however does not explicitly disclose a trained machine learning computing model,
LING discloses a trained machine learning computing model (see LING: e.g., -- providing the combined-SERS profile to a device configured with a model trained to identify the one or more analytes from the combined-SERS profile. The device may be a computer operable to process a SERS signal and capable of machine learning. The device may also be operable to carry out any chemometric analysis of a spectrum obtained via SERS, including the SERS signal and the combined-SERS profile.--, in [0162], and, -- (d) inputting the entire collective spectrum to a trained learning model and obtaining the molecular identification of and/or quantifying the one or more analytes; and (e) wherein the trained learning model having been trained through machine learning techniques to predict the molecular identification and/or estimate the quantity of the one or more analytes and/or perform classification of a multi-analyte mixture.--, in [0175]);
DE SIMONE and LING are combinable as they are in the same field of endeavor: machine learning model applied in assessments of breath analysis. Therefore it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to further modify DE SIMONE’s apparatus using LING’s teachings by including a trained machine learning computing model to DE SIMONE’s Machine Learning model in order to predict the molecular identification and/or estimate the quantity of the one or more analytes and/or perform classification of a multi-analyte mixture (see LING: e.g. in abstract, [0162], and [0175]);
receive at least one spectrometric-based prediction data object from the at least one trained machine learning computing model (see DE SIMONE: e.g., -- Fig. 3, and, --[0093] The image analysis process called digital spectroscopy is divided into three steps: [0094] a. Generation of the database for the image training that represents the wavelengths to generate the spectra. [0095] b. Generation of the spectrum of each training image, fitting of the data and cross validation of the model. [0096] c. Generation of spectra of the samples to be analyzed and their prediction. …[0098] To develop the model that generates the spectra of the images, images that represent designated “wavelengths” were used. These are the variables of the image training model. The procedure to generate the models is similar to what is done with the images to be studied: The “PIL” library is used to take the RGB composition of each pixel in the image…. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. The corresponding arbitrary length is assigned to each of these combinations of values. Once the dataframe with all the combinations has been generated, a Machine Learning (Random Forest) model is generated. This model (from now on called pixelMod) is later used to define which is the length that belongs to each pixel of the studied images.--, in [0093]-[0098], and, …. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. This classification model will have two variables: POSITIVE (COVID-19)—NEGATIVE. The gridsearch_cv (cross validation) was used to iterate between different parameters and then obtain the optimal one. Cross validation was also used to check which was the best score obtained. In the event that the expected threshold were exceeded, it then proceeds with the last step. This model (from now on called entrenaMod) is used to predict which group the unknown measurements correspond to.
[0101] In the case of the samples to be analyzed, the procedure is similar as in the training step: the spectra table of each image is generated with the pixelMod and once the table is completed, it is predicted with the entrenaMod. This result will be exposed in a probabilistic way for each variable, and the resulting numbers are the average image probabilities of the measurement. The processing scheme is detailed in FIG. 3.--, in [0100]-[0101], [0104]; also LING: e.g., -- providing the combined-SERS profile to a device configured with a model trained to identify the one or more analytes from the combined-SERS profile. The device may be a computer operable to process a SERS signal and capable of machine learning. The device may also be operable to carry out any chemometric analysis of a spectrum obtained via SERS, including the SERS signal and the combined-SERS profile.--, in [0162], and, -- (d) inputting the entire collective spectrum to a trained learning model and obtaining the molecular identification of and/or quantifying the one or more analytes; and (e) wherein the trained learning model having been trained through machine learning techniques to predict the molecular identification and/or estimate the quantity of the one or more analytes and/or perform classification of a multi-analyte mixture.--, in [0175]); and
perform one or more prediction-based data operations based at least in part on the at least one spectrometric-based prediction data object (see DE SIMONE: e.g., -- Fig. 3, and, --[0093] The image analysis process called digital spectroscopy is divided into three steps: [0094] a. Generation of the database for the image training that represents the wavelengths to generate the spectra. [0095] b. Generation of the spectrum of each training image, fitting of the data and cross validation of the model. [0096] c. Generation of spectra of the samples to be analyzed and their prediction. …[0098] To develop the model that generates the spectra of the images, images that represent designated “wavelengths” were used. These are the variables of the image training model. The procedure to generate the models is similar to what is done with the images to be studied: The “PIL” library is used to take the RGB composition of each pixel in the image…. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. The corresponding arbitrary length is assigned to each of these combinations of values. Once the dataframe with all the combinations has been generated, a Machine Learning (Random Forest) model is generated. This model (from now on called pixelMod) is later used to define which is the length that belongs to each pixel of the studied images.--, in [0093]-[0098], and, …. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. This classification model will have two variables: POSITIVE (COVID-19)—NEGATIVE. The gridsearch_cv (cross validation) was used to iterate between different parameters and then obtain the optimal one. Cross validation was also used to check which was the best score obtained. In the event that the expected threshold were exceeded, it then proceeds with the last step. This model (from now on called entrenaMod) is used to predict which group the unknown measurements correspond to.
[0101] In the case of the samples to be analyzed, the procedure is similar as in the training step: the spectra table of each image is generated with the pixelMod and once the table is completed, it is predicted with the entrenaMod. This result will be exposed in a probabilistic way for each variable, and the resulting numbers are the average image probabilities of the measurement. The processing scheme is detailed in FIG. 3.--, in [0100]-[0101], [0104]; also see LING: e.g., -- providing the combined-SERS profile to a device configured with a model trained to identify the one or more analytes from the combined-SERS profile. The device may be a computer operable to process a SERS signal and capable of machine learning. The device may also be operable to carry out any chemometric analysis of a spectrum obtained via SERS, including the SERS signal and the combined-SERS profile.--, in [0162], and, -- (d) inputting the entire collective spectrum to a trained learning model and obtaining the molecular identification of and/or quantifying the one or more analytes; and (e) wherein the trained learning model having been trained through machine learning techniques to predict the molecular identification and/or estimate the quantity of the one or more analytes and/or perform classification of a multi-analyte mixture.--, in [0175]), wherein the at least one spectrometric-based prediction data object provides predictions on health conditions of the user (see LING: e.g., -- providing the combined-SERS profile to a device configured with a model trained to identify the one or more analytes from the combined-SERS profile. The device may be a computer operable to process a SERS signal and capable of machine learning. The device may also be operable to carry out any chemometric analysis of a spectrum obtained via SERS, including the SERS signal and the combined-SERS profile.--, in [0162], and, -- (d) inputting the entire collective spectrum to a trained learning model and obtaining the molecular identification of and/or quantifying the one or more analytes; and (e) wherein the trained learning model having been trained through machine learning techniques to predict the molecular identification and/or estimate the quantity of the one or more analytes and/or perform classification of a multi-analyte mixture.--, in [0175]); and
transmit one or more health alerts to one or more user devices based at least in part on the at least one spectrometric-based prediction data object, wherein the one or more user devices are associated with at least one of users and healthcare providers (see DE SIMONE: e.g., --[0110] The report on the interpretation of diagnostic tests for SARS-CoV-2 (Spanish Society of Infectious Diseases and Clinical Microbiology, Instituto de Salud Carlos III) was considered for the training of the device.--, in [0110], and, -- f) processing said images by artificial intelligence to determine if said images are compatible with breath samples from sick people; g) giving a visual indication of the result.--, in claim 36, {apparently “a visual indication of the result” read on claimed “alert” transmitted to the physicians}; also see LING: e.g., --Participants with reported comorbidities are highlighted in purple, illustrating that the participants' existing health conditions do not affect their classification scores. It is to be highlighted that since the provision of this information is strictly voluntary, information for all 501 trial participants is not entirely available. Nonetheless, the accurate classification of the 70 participants with reported comorbidities indicates that the SERS sensor is able to identify specific differences in participants' breath profiles that were directly linked to whether they had COVID-19 or not.--, in [0138] {these reports are aligned with claimed “alert”}).
Re Claim 2, DE SIMONE as modified by LING further disclose when generating the plurality of exhaled breath spectrometric data objects, the memory and the non-transitory program code are configured to, with the processor, cause the spectrometric data analyzing device to:
extract at least one of photographic metadata or spectrometric metadata from the exhaled breath digital image data objects (see DE SIMONE: e.g., --[0098] To develop the model that generates the spectra of the images, images that represent designated “wavelengths” were used. These are the variables of the image training model. The procedure to generate the models is similar to what is done with the images to be studied: The “PIL” library is used to take the RGB composition of each pixel in the image…. [0099] The device takes images (and records a video with them) of the corona discharge, similar to the one shown in FIG. 2. These images are acquired through the “CV2” library …[0100] To develop the training model, the same PIL library is used to extract the information of the pixels of each image to be analyzed, excluding the black pixels. Once this data has been extracted, the pixelMod is used to determine to which length each pixel belongs. After the assignment is completed, the table is generated with all the pixels of each length that each of the images has.--, in [0098]-[0100]); and
generate the plurality of exhaled breath spectrometric data objects based at least in part on the at least one of photographic metadata or spectrometric metadata (see DE SIMONE: e.g., -- Fig. 3, and, --[0093] The image analysis process called digital spectroscopy is divided into three steps: [0094] a. Generation of the database for the image training that represents the wavelengths to generate the spectra. [0095] b. Generation of the spectrum of each training image, fitting of the data and cross validation of the model. [0096] c. Generation of spectra of the samples to be analyzed and their prediction. …[0098] To develop the model that generates the spectra of the images, images that represent designated “wavelengths” were used. These are the variables of the image training model. The procedure to generate the models is similar to what is done with the images to be studied: The “PIL” library is used to take the RGB composition of each pixel in the image…. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. The corresponding arbitrary length is assigned to each of these combinations of values. Once the dataframe with all the combinations has been generated, a Machine Learning (Random Forest) model is generated. This model (from now on called pixelMod) is later used to define which is the length that belongs to each pixel of the studied images.--, in [0093]-[0098], and, …. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. This classification model will have two variables: POSITIVE (COVID-19)—NEGATIVE. The gridsearch_cv (cross validation) was used to iterate between different parameters and then obtain the optimal one. Cross validation was also used to check which was the best score obtained. In the event that the expected threshold were exceeded, it then proceeds with the last step. This model (from now on called entrenaMod) is used to predict which group the unknown measurements correspond to.
[0101] In the case of the samples to be analyzed, the procedure is similar as in the training step: the spectra table of each image is generated with the pixelMod and once the table is completed, it is predicted with the entrenaMod. This result will be exposed in a probabilistic way for each variable, and the resulting numbers are the average image probabilities of the measurement. The processing scheme is detailed in FIG. 3.--, in [0100]-[0101], [0104]).
Re Claim 3, DE SIMONE as modified by LING further disclose wherein the at least one trained machine learning computing model comprises at least one trained classification-based estimation model (see DE SIMONE: e.g., -- Fig. 3, and, --[0093] The image analysis process called digital spectroscopy is divided into three steps: [0094] a. Generation of the database for the image training that represents the wavelengths to generate the spectra. [0095] b. Generation of the spectrum of each training image, fitting of the data and cross validation of the model. [0096] c. Generation of spectra of the samples to be analyzed and their prediction. …[0098] To develop the model that generates the spectra of the images, images that represent designated “wavelengths” were used. These are the variables of the image training model. The procedure to generate the models is similar to what is done with the images to be studied: The “PIL” library is used to take the RGB composition of each pixel in the image…. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. The corresponding arbitrary length is assigned to each of these combinations of values. Once the dataframe with all the combinations has been generated, a Machine Learning (Random Forest) model is generated. This model (from now on called pixelMod) is later used to define which is the length that belongs to each pixel of the studied images.--, in [0093]-[0098], and, …. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. This classification model will have two variables: POSITIVE (COVID-19)—NEGATIVE. The gridsearch_cv (cross validation) was used to iterate between different parameters and then obtain the optimal one. Cross validation was also used to check which was the best score obtained. In the event that the expected threshold were exceeded, it then proceeds with the last step. This model (from now on called entrenaMod) is used to predict which group the unknown measurements correspond to.
[0101] In the case of the samples to be analyzed, the procedure is similar as in the training step: the spectra table of each image is generated with the pixelMod and once the table is completed, it is predicted with the entrenaMod. This result will be exposed in a probabilistic way for each variable, and the resulting numbers are the average image probabilities of the measurement. The processing scheme is detailed in FIG. 3.--, in [0100]-[0101], [0104]; also see LING: e.g., -- providing the combined-SERS profile to a device configured with a model trained to identify the one or more analytes from the combined-SERS profile. The device may be a computer operable to process a SERS signal and capable of machine learning. The device may also be operable to carry out any chemometric analysis of a spectrum obtained via SERS, including the SERS signal and the combined-SERS profile.--, in [0162], and, -- (d) inputting the entire collective spectrum to a trained learning model and obtaining the molecular identification of and/or quantifying the one or more analytes; and (e) wherein the trained learning model having been trained through machine learning techniques to predict the molecular identification and/or estimate the quantity of the one or more analytes and/or perform classification of a multi-analyte mixture.--, in [0175]).
Re Claim 4, DE SIMONE as modified by LING further disclose wherein, prior to inputting the plurality of exhaled breath spectrometric data objects to the at least one trained machine learning computing model, the memory and the non-transitory program code are configured to, with the processor, cause the spectrometric data analyzing device to: train at least one classification-based estimation model (see DE SIMONE: e.g., -- Fig. 3, and, --[0093] The image analysis process called digital spectroscopy is divided into three steps: [0094] a. Generation of the database for the image training that represents the wavelengths to generate the spectra. [0095] b. Generation of the spectrum of each training image, fitting of the data and cross validation of the model. [0096] c. Generation of spectra of the samples to be analyzed and their prediction. …[0098] To develop the model that generates the spectra of the images, images that represent designated “wavelengths” were used. These are the variables of the image training model. The procedure to generate the models is similar to what is done with the images to be studied: The “PIL” library is used to take the RGB composition of each pixel in the image…. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. The corresponding arbitrary length is assigned to each of these combinations of values. Once the dataframe with all the combinations has been generated, a Machine Learning (Random Forest) model is generated. This model (from now on called pixelMod) is later used to define which is the length that belongs to each pixel of the studied images.--, in [0093]-[0098], and, …. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. This classification model will have two variables: POSITIVE (COVID-19)—NEGATIVE. The gridsearch_cv (cross validation) was used to iterate between different parameters and then obtain the optimal one. Cross validation was also used to check which was the best score obtained. In the event that the expected threshold were exceeded, it then proceeds with the last step. This model (from now on called entrenaMod) is used to predict which group the unknown measurements correspond to.
[0101] In the case of the samples to be analyzed, the procedure is similar as in the training step: the spectra table of each image is generated with the pixelMod and once the table is completed, it is predicted with the entrenaMod. This result will be exposed in a probabilistic way for each variable, and the resulting numbers are the average image probabilities of the measurement. The processing scheme is detailed in FIG. 3.--, in [0100]-[0101], [0104]; also see LING: e.g., -- providing the combined-SERS profile to a device configured with a model trained to identify the one or more analytes from the combined-SERS profile. The device may be a computer operable to process a SERS signal and capable of machine learning. The device may also be operable to carry out any chemometric analysis of a spectrum obtained via SERS, including the SERS signal and the combined-SERS profile.--, in [0162], and, -- (d) inputting the entire collective spectrum to a trained learning model and obtaining the molecular identification of and/or quantifying the one or more analytes; and (e) wherein the trained learning model having been trained through machine learning techniques to predict the molecular identification and/or estimate the quantity of the one or more analytes and/or perform classification of a multi-analyte mixture.--, in [0175]).
Re Claim 5, DE SIMONE as modified by LING further disclose wherein, when training the at least one classification-based estimation model, the memory and the non-transitory program code are configured to, with the processor, cause the spectrometric data analyzing device to:
retrieve a plurality of training exhaled breath spectrometric data objects associated with a training spectrometric-based prediction data object (see DE SIMONE: e.g., -- Fig. 3, and, --[0093] The image analysis process called digital spectroscopy is divided into three steps: [0094] a. Generation of the database for the image training that represents the wavelengths to generate the spectra. [0095] b. Generation of the spectrum of each training image, fitting of the data and cross validation of the model. [0096] c. Generation of spectra of the samples to be analyzed and their prediction. …[0098] To develop the model that generates the spectra of the images, images that represent designated “wavelengths” were used. These are the variables of the image training model. The procedure to generate the models is similar to what is done with the images to be studied: The “PIL” library is used to take the RGB composition of each pixel in the image…. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. The corresponding arbitrary length is assigned to each of these combinations of values. Once the dataframe with all the combinations has been generated, a Machine Learning (Random Forest) model is generated. This model (from now on called pixelMod) is later used to define which is the length that belongs to each pixel of the studied images.--, in [0093]-[0098], and, …. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. This classification model will have two variables: POSITIVE (COVID-19)—NEGATIVE. The gridsearch_cv (cross validation) was used to iterate between different parameters and then obtain the optimal one. Cross validation was also used to check which was the best score obtained. In the event that the expected threshold were exceeded, it then proceeds with the last step. This model (from now on called entrenaMod) is used to predict which group the unknown measurements correspond to.
[0101] In the case of the samples to be analyzed, the procedure is similar as in the training step: the spectra table of each image is generated with the pixelMod and once the table is completed, it is predicted with the entrenaMod. This result will be exposed in a probabilistic way for each variable, and the resulting numbers are the average image probabilities of the measurement. The processing scheme is detailed in FIG. 3.--, in [0100]-[0101], [0104]; also see LING: e.g., -- providing the combined-SERS profile to a device configured with a model trained to identify the one or more analytes from the combined-SERS profile. The device may be a computer operable to process a SERS signal and capable of machine learning. The device may also be operable to carry out any chemometric analysis of a spectrum obtained via SERS, including the SERS signal and the combined-SERS profile.--, in [0162], and, -- (d) inputting the entire collective spectrum to a trained learning model and obtaining the molecular identification of and/or quantifying the one or more analytes; and (e) wherein the trained learning model having been trained through machine learning techniques to predict the molecular identification and/or estimate the quantity of the one or more analytes and/or perform classification of a multi-analyte mixture.--, in [0175]);
input the plurality of training exhaled breath spectrometric data objects to the at least one classification-based estimation model (see DE SIMONE: e.g., -- Fig. 3, and, --[0093] The image analysis process called digital spectroscopy is divided into three steps: [0094] a. Generation of the database for the image training that represents the wavelengths to generate the spectra. [0095] b. Generation of the spectrum of each training image, fitting of the data and cross validation of the model. [0096] c. Generation of spectra of the samples to be analyzed and their prediction. …[0098] To develop the model that generates the spectra of the images, images that represent designated “wavelengths” were used. These are the variables of the image training model. The procedure to generate the models is similar to what is done with the images to be studied: The “PIL” library is used to take the RGB composition of each pixel in the image…. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. The corresponding arbitrary length is assigned to each of these combinations of values. Once the dataframe with all the combinations has been generated, a Machine Learning (Random Forest) model is generated. This model (from now on called pixelMod) is later used to define which is the length that belongs to each pixel of the studied images.--, in [0093]-[0098], and, …. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. This classification model will have two variables: POSITIVE (COVID-19)—NEGATIVE. The gridsearch_cv (cross validation) was used to iterate between different parameters and then obtain the optimal one. Cross validation was also used to check which was the best score obtained. In the event that the expected threshold were exceeded, it then proceeds with the last step. This model (from now on called entrenaMod) is used to predict which group the unknown measurements correspond to.
[0101] In the case of the samples to be analyzed, the procedure is similar as in the training step: the spectra table of each image is generated with the pixelMod and once the table is completed, it is predicted with the entrenaMod. This result will be exposed in a probabilistic way for each variable, and the resulting numbers are the average image probabilities of the measurement. The processing scheme is detailed in FIG. 3.--, in [0100]-[0101], [0104]; also see LING: e.g., -- providing the combined-SERS profile to a device configured with a model trained to identify the one or more analytes from the combined-SERS profile. The device may be a computer operable to process a SERS signal and capable of machine learning. The device may also be operable to carry out any chemometric analysis of a spectrum obtained via SERS, including the SERS signal and the combined-SERS profile.--, in [0162], and, -- (d) inputting the entire collective spectrum to a trained learning model and obtaining the molecular identification of and/or quantifying the one or more analytes; and (e) wherein the trained learning model having been trained through machine learning techniques to predict the molecular identification and/or estimate the quantity of the one or more analytes and/or perform classification of a multi-analyte mixture.--, in [0175]);
receive a testing spectrometric-based prediction data object from the at least one classification-based estimation model (see DE SIMONE: e.g., -- Fig. 3, and, --[0093] The image analysis process called digital spectroscopy is divided into three steps: [0094] a. Generation of the database for the image training that represents the wavelengths to generate the spectra. [0095] b. Generation of the spectrum of each training image, fitting of the data and cross validation of the model. [0096] c. Generation of spectra of the samples to be analyzed and their prediction. …[0098] To develop the model that generates the spectra of the images, images that represent designated “wavelengths” were used. These are the variables of the image training model. The procedure to generate the models is similar to what is done with the images to be studied: The “PIL” library is used to take the RGB composition of each pixel in the image…. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. The corresponding arbitrary length is assigned to each of these combinations of values. Once the dataframe with all the combinations has been generated, a Machine Learning (Random Forest) model is generated. This model (from now on called pixelMod) is later used to define which is the length that belongs to each pixel of the studied images.--, in [0093]-[0098], and, …. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. This classification model will have two variables: POSITIVE (COVID-19)—NEGATIVE. The gridsearch_cv (cross validation) was used to iterate between different parameters and then obtain the optimal one. Cross validation was also used to check which was the best score obtained. In the event that the expected threshold were exceeded, it then proceeds with the last step. This model (from now on called entrenaMod) is used to predict which group the unknown measurements correspond to.
[0101] In the case of the samples to be analyzed, the procedure is similar as in the training step: the spectra table of each image is generated with the pixelMod and once the table is completed, it is predicted with the entrenaMod. This result will be exposed in a probabilistic way for each variable, and the resulting numbers are the average image probabilities of the measurement. The processing scheme is detailed in FIG. 3.--, in [0100]-[0101], [0104]; also see LING: e.g., -- providing the combined-SERS profile to a device configured with a model trained to identify the one or more analytes from the combined-SERS profile. The device may be a computer operable to process a SERS signal and capable of machine learning. The device may also be operable to carry out any chemometric analysis of a spectrum obtained via SERS, including the SERS signal and the combined-SERS profile.--, in [0162], and, -- (d) inputting the entire collective spectrum to a trained learning model and obtaining the molecular identification of and/or quantifying the one or more analytes; and (e) wherein the trained learning model having been trained through machine learning techniques to predict the molecular identification and/or estimate the quantity of the one or more analytes and/or perform classification of a multi-analyte mixture.--, in [0175]); and
adjust the at least one classification-based estimation model based at least in part on the testing spectrometric-based prediction data object and the training spectrometric-based prediction data object (see DE SIMONE: e.g., -- Fig. 3, and, --[0093] The image analysis process called digital spectroscopy is divided into three steps: [0094] a. Generation of the database for the image training that represents the wavelengths to generate the spectra. [0095] b. Generation of the spectrum of each training image, fitting of the data and cross validation of the model. [0096] c. Generation of spectra of the samples to be analyzed and their prediction. …[0098] To develop the model that generates the spectra of the images, images that represent designated “wavelengths” were used. These are the variables of the image training model. The procedure to generate the models is similar to what is done with the images to be studied: The “PIL” library is used to take the RGB composition of each pixel in the image…. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. The corresponding arbitrary length is assigned to each of these combinations of values. Once the dataframe with all the combinations has been generated, a Machine Learning (Random Forest) model is generated. This model (from now on called pixelMod) is later used to define which is the length that belongs to each pixel of the studied images.--, in [0093]-[0098], and, …. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. This classification model will have two variables: POSITIVE (COVID-19)—NEGATIVE. The gridsearch_cv (cross validation) was used to iterate between different parameters and then obtain the optimal one. Cross validation was also used to check which was the best score obtained. In the event that the expected threshold were exceeded, it then proceeds with the last step. This model (from now on called entrenaMod) is used to predict which group the unknown measurements correspond to.
[0101] In the case of the samples to be analyzed, the procedure is similar as in the training step: the spectra table of each image is generated with the pixelMod and once the table is completed, it is predicted with the entrenaMod. This result will be exposed in a probabilistic way for each variable, and the resulting numbers are the average image probabilities of the measurement. The processing scheme is detailed in FIG. 3.--, in [0100]-[0101], [0104]; also see LING: e.g., -- providing the combined-SERS profile to a device configured with a model trained to identify the one or more analytes from the combined-SERS profile. The device may be a computer operable to process a SERS signal and capable of machine learning. The device may also be operable to carry out any chemometric analysis of a spectrum obtained via SERS, including the SERS signal and the combined-SERS profile.--, in [0162], and, -- (d) inputting the entire collective spectrum to a trained learning model and obtaining the molecular identification of and/or quantifying the one or more analytes; and (e) wherein the trained learning model having been trained through machine learning techniques to predict the molecular identification and/or estimate the quantity of the one or more analytes and/or perform classification of a multi-analyte mixture.--, in [0175]).
Re Claim 6, DE SIMONE as modified by LING further disclose when performing the one or more prediction-based data operations based at least in part on the at least one spectrometric-based prediction data object, the memory and the non-transitory program code are configured to, with the processor, cause the spectrometric data analyzing device to:
determine whether the at least one spectrometric-based prediction data object satisfies a health condition threshold (see DE SIMONE: e.g., -- Fig. 3, and, --[0093] The image analysis process called digital spectroscopy is divided into three steps: [0094] a. Generation of the database for the image training that represents the wavelengths to generate the spectra. [0095] b. Generation of the spectrum of each training image, fitting of the data and cross validation of the model. [0096] c. Generation of spectra of the samples to be analyzed and their prediction. …[0098] To develop the model that generates the spectra of the images, images that represent designated “wavelengths” were used. These are the variables of the image training model. The procedure to generate the models is similar to what is done with the images to be studied: The “PIL” library is used to take the RGB composition of each pixel in the image…. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. The corresponding arbitrary length is assigned to each of these combinations of values. Once the dataframe with all the combinations has been generated, a Machine Learning (Random Forest) model is generated. This model (from now on called pixelMod) is later used to define which is the length that belongs to each pixel of the studied images.--, in [0093]-[0098], and, …. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. This classification model will have two variables: POSITIVE (COVID-19)—NEGATIVE. The gridsearch_cv (cross validation) was used to iterate between different parameters and then obtain the optimal one. Cross validation was also used to check which was the best score obtained. In the event that the expected threshold were exceeded, it then proceeds with the last step. This model (from now on called entrenaMod) is used to predict which group the unknown measurements correspond to.
[0101] In the case of the samples to be analyzed, the procedure is similar as in the training step: the spectra table of each image is generated with the pixelMod and once the table is completed, it is predicted with the entrenaMod. This result will be exposed in a probabilistic way for each variable, and the resulting numbers are the average image probabilities of the measurement. The processing scheme is detailed in FIG. 3.--, in [0100]-[0101], [0104]; also see LING: e.g., -- providing the combined-SERS profile to a device configured with a model trained to identify the one or more analytes from the combined-SERS profile. The device may be a computer operable to process a SERS signal and capable of machine learning. The device may also be operable to carry out any chemometric analysis of a spectrum obtained via SERS, including the SERS signal and the combined-SERS profile.--, in [0162], and, -- (d) inputting the entire collective spectrum to a trained learning model and obtaining the molecular identification of and/or quantifying the one or more analytes; and (e) wherein the trained learning model having been trained through machine learning techniques to predict the molecular identification and/or estimate the quantity of the one or more analytes and/or perform classification of a multi-analyte mixture.--, in [0175], and, --[0280] Participant statistics for categorical variables such as age and gender were presented as number (%). Continuous variables such as intensity ratios were presented as mean±standard deviation. The statistical significance of each variable between blanks and COVID-positive, blanks and COVID-negative, and COVID-positive and COVID-negative were assessed with the Mann-Whitney rank sum test. All tests were two-tailed with p<0.05 as the significance threshold. Calculations were performed using the OriginPro 9.0 software. The statistical significance of each confounding factor on the classification was assessed using either a t test (for continuous variable) or a χ.sup.2 test (categorical variable). The choice of statistical test depends on several parameters including the variable type (categorical/continuous) and distributions (normal/non-normal).--, in [0280]); and
in response to determining that the at least one spectrometric-based prediction data object satisfies the health condition threshold, transmit a predicted health condition indication to a non- invasive breath analyzer server (see DE SIMONE: e.g., -- Fig. 3, and, --[0093] The image analysis process called digital spectroscopy is divided into three steps: [0094] a. Generation of the database for the image training that represents the wavelengths to generate the spectra. [0095] b. Generation of the spectrum of each training image, fitting of the data and cross validation of the model. [0096] c. Generation of spectra of the samples to be analyzed and their prediction. …[0098] To develop the model that generates the spectra of the images, images that represent designated “wavelengths” were used. These are the variables of the image training model. The procedure to generate the models is similar to what is done with the images to be studied: The “PIL” library is used to take the RGB composition of each pixel in the image…. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. The corresponding arbitrary length is assigned to each of these combinations of values. Once the dataframe with all the combinations has been generated, a Machine Learning (Random Forest) model is generated. This model (from now on called pixelMod) is later used to define which is the length that belongs to each pixel of the studied images.--, in [0093]-[0098], and, …. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. This classification model will have two variables: POSITIVE (COVID-19)—NEGATIVE. The gridsearch_cv (cross validation) was used to iterate between different parameters and then obtain the optimal one. Cross validation was also used to check which was the best score obtained. In the event that the expected threshold were exceeded, it then proceeds with the last step. This model (from now on called entrenaMod) is used to predict which group the unknown measurements correspond to.
[0101] In the case of the samples to be analyzed, the procedure is similar as in the training step: the spectra table of each image is generated with the pixelMod and once the table is completed, it is predicted with the entrenaMod. This result will be exposed in a probabilistic way for each variable, and the resulting numbers are the average image probabilities of the measurement. The processing scheme is detailed in FIG. 3.--, in [0100]-[0101], [0104]; also see LING: e.g., -- providing the combined-SERS profile to a device configured with a model trained to identify the one or more analytes from the combined-SERS profile. The device may be a computer operable to process a SERS signal and capable of machine learning. The device may also be operable to carry out any chemometric analysis of a spectrum obtained via SERS, including the SERS signal and the combined-SERS profile.--, in [0162], and, -- (d) inputting the entire collective spectrum to a trained learning model and obtaining the molecular identification of and/or quantifying the one or more analytes; and (e) wherein the trained learning model having been trained through machine learning techniques to predict the molecular identification and/or estimate the quantity of the one or more analytes and/or perform classification of a multi-analyte mixture.--, in [0175], and, --[0280] Participant statistics for categorical variables such as age and gender were presented as number (%). Continuous variables such as intensity ratios were presented as mean±standard deviation. The statistical significance of each variable between blanks and COVID-positive, blanks and COVID-negative, and COVID-positive and COVID-negative were assessed with the Mann-Whitney rank sum test. All tests were two-tailed with p<0.05 as the significance threshold. Calculations were performed using the OriginPro 9.0 software. The statistical significance of each confounding factor on the classification was assessed using either a t test (for continuous variable) or a χ.sup.2 test (categorical variable). The choice of statistical test depends on several parameters including the variable type (categorical/continuous) and distributions (normal/non-normal).--, in [0280]).
Re Claim 7, DE SIMONE as modified by LING further disclose when performing the one or more prediction-based data operations based at least in part on the at least one spectrometric-based prediction data object, the memory and the non-transitory program code are configured to, with the processor, cause the spectrometric data analyzing device to:
retrieve a previous spectrometric-based prediction data object associated with a previous time point (see DE SIMONE: e.g., --[0092] To perform the identification of the samples and determine the discrimination capability of the device, the following procedure was carried out to train and identify the analyzed samples. These samples correspond to volunteers previously diagnosed with or without COVID-19 disease.
[0093] The image analysis process called digital spectroscopy is divided into three steps: [0094] a. Generation of the database for the image training that represents the wavelengths to generate the spectra. [0095] b. Generation of the spectrum of each training image, fitting of the data and cross validation of the model. [0096] c. Generation of spectra of the samples to be analyzed and their prediction.--, in [0092]-[0096]; also see LING: e.g., --[0297] The PLSDA score and loadings plot are further used to highlight how different receptor spectral regions influence the classification outcome, so as to establish a robust relationship between the classification results and previously identified regions which showed distinct differences (FIGS. 42B and 42C). The first two LVs of are important in describing regions that contribute to the largest variances between the two classes. From the score plot, it can be observed that COVID-positive breath samples show more positive LV 2 scores while COVID-negative breath samples show more negative LV 2 scores (FIG. 42B).--, in [0297]);
retrieve a subsequent spectrometric-based prediction data object associated with a subsequent time point (see DE SIMONE: e.g., -- Fig. 3, and, --[0093] The image analysis process called digital spectroscopy is divided into three steps: [0094] a. Generation of the database for the image training that represents the wavelengths to generate the spectra. [0095] b. Generation of the spectrum of each training image, fitting of the data and cross validation of the model. [0096] c. Generation of spectra of the samples to be analyzed and their prediction. …[0098] To develop the model that generates the spectra of the images, images that represent designated “wavelengths” were used. These are the variables of the image training model. The procedure to generate the models is similar to what is done with the images to be studied: The “PIL” library is used to take the RGB composition of each pixel in the image…. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. The corresponding arbitrary length is assigned to each of these combinations of values. Once the dataframe with all the combinations has been generated, a Machine Learning (Random Forest) model is generated. This model (from now on called pixelMod) is later used to define which is the length that belongs to each pixel of the studied images.--, in [0093]-[0098], and, …. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. This classification model will have two variables: POSITIVE (COVID-19)—NEGATIVE. The gridsearch_cv (cross validation) was used to iterate between different parameters and then obtain the optimal one. Cross validation was also used to check which was the best score obtained. In the event that the expected threshold were exceeded, it then proceeds with the last step. This model (from now on called entrenaMod) is used to predict which group the unknown measurements correspond to.
[0101] In the case of the samples to be analyzed, the procedure is similar as in the training step: the spectra table of each image is generated with the pixelMod and once the table is completed, it is predicted with the entrenaMod. This result will be exposed in a probabilistic way for each variable, and the resulting numbers are the average image probabilities of the measurement. The processing scheme is detailed in FIG. 3.--, in [0100]-[0101], [0104]; also see LING: e.g., -- providing the combined-SERS profile to a device configured with a model trained to identify the one or more analytes from the combined-SERS profile. The device may be a computer operable to process a SERS signal and capable of machine learning. The device may also be operable to carry out any chemometric analysis of a spectrum obtained via SERS, including the SERS signal and the combined-SERS profile.--, in [0162], and, -- (d) inputting the entire collective spectrum to a trained learning model and obtaining the molecular identification of and/or quantifying the one or more analytes; and (e) wherein the trained learning model having been trained through machine learning techniques to predict the molecular identification and/or estimate the quantity of the one or more analytes and/or perform classification of a multi-analyte mixture.--, in [0175], and, --[0280] Participant statistics for categorical variables such as age and gender were presented as number (%). Continuous variables such as intensity ratios were presented as mean±standard deviation. The statistical significance of each variable between blanks and COVID-positive, blanks and COVID-negative, and COVID-positive and COVID-negative were assessed with the Mann-Whitney rank sum test. All tests were two-tailed with p<0.05 as the significance threshold. Calculations were performed using the OriginPro 9.0 software. The statistical significance of each confounding factor on the classification was assessed using either a t test (for continuous variable) or a χ.sup.2 test (categorical variable). The choice of statistical test depends on several parameters including the variable type (categorical/continuous) and distributions (normal/non-normal).--, in [0280]); and
generate a predicted condition progression indication based at least in part on comparing the subsequent spectrometric-based prediction data object with the previous spectrometric-based prediction data object (see DE SIMONE: e.g., -- Fig. 3, and, --[0093] The image analysis process called digital spectroscopy is divided into three steps: [0094] a. Generation of the database for the image training that represents the wavelengths to generate the spectra. [0095] b. Generation of the spectrum of each training image, fitting of the data and cross validation of the model. [0096] c. Generation of spectra of the samples to be analyzed and their prediction. …[0098] To develop the model that generates the spectra of the images, images that represent designated “wavelengths” were used. These are the variables of the image training model. The procedure to generate the models is similar to what is done with the images to be studied: The “PIL” library is used to take the RGB composition of each pixel in the image…. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. The corresponding arbitrary length is assigned to each of these combinations of values. Once the dataframe with all the combinations has been generated, a Machine Learning (Random Forest) model is generated. This model (from now on called pixelMod) is later used to define which is the length that belongs to each pixel of the studied images.--, in [0093]-[0098], and, …. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. This classification model will have two variables: POSITIVE (COVID-19)—NEGATIVE. The gridsearch_cv (cross validation) was used to iterate between different parameters and then obtain the optimal one. Cross validation was also used to check which was the best score obtained. In the event that the expected threshold were exceeded, it then proceeds with the last step. This model (from now on called entrenaMod) is used to predict which group the unknown measurements correspond to.
[0101] In the case of the samples to be analyzed, the procedure is similar as in the training step: the spectra table of each image is generated with the pixelMod and once the table is completed, it is predicted with the entrenaMod. This result will be exposed in a probabilistic way for each variable, and the resulting numbers are the average image probabilities of the measurement. The processing scheme is detailed in FIG. 3.--, in [0100]-[0101], [0104]; also see LING: e.g., -- providing the combined-SERS profile to a device configured with a model trained to identify the one or more analytes from the combined-SERS profile. The device may be a computer operable to process a SERS signal and capable of machine learning. The device may also be operable to carry out any chemometric analysis of a spectrum obtained via SERS, including the SERS signal and the combined-SERS profile.--, in [0162], and, -- (d) inputting the entire collective spectrum to a trained learning model and obtaining the molecular identification of and/or quantifying the one or more analytes; and (e) wherein the trained learning model having been trained through machine learning techniques to predict the molecular identification and/or estimate the quantity of the one or more analytes and/or perform classification of a multi-analyte mixture.--, in [0175], and, --[0280] Participant statistics for categorical variables such as age and gender were presented as number (%). Continuous variables such as intensity ratios were presented as mean±standard deviation. The statistical significance of each variable between blanks and COVID-positive, blanks and COVID-negative, and COVID-positive and COVID-negative were assessed with the Mann-Whitney rank sum test. All tests were two-tailed with p<0.05 as the significance threshold. Calculations were performed using the OriginPro 9.0 software. The statistical significance of each confounding factor on the classification was assessed using either a t test (for continuous variable) or a χ.sup.2 test (categorical variable). The choice of statistical test depends on several parameters including the variable type (categorical/continuous) and distributions (normal/non-normal).--, in [0280]).
Re Claims 8-14, claims 8-14 are the corresponding product claim to claims 1-7, respectively. Claims 8-14 thus are rejected for the similar reasons for claims 1-7. See above discussions with regard to claims 1-7 respectively. Further, DE SIMONE as modified by LING further disclose a computer program product for non-invasive breath analysis, the computer program product comprising at least one non-transitory computer-readable storage medium having computer-readable program code portions stored therein, the computer-readable program code portions comprising an executable portion configured to perform the steps (see DE SIMONE: e.g., -- Fig. 3, and, --[0093] The image analysis process called digital spectroscopy is divided into three steps: [0094] a. Generation of the database for the image training that represents the wavelengths to generate the spectra. [0095] b. Generation of the spectrum of each training image, fitting of the data and cross validation of the model. [0096] c. Generation of spectra of the samples to be analyzed and their prediction. …[0098] To develop the model that generates the spectra of the images, images that represent designated “wavelengths” were used. These are the variables of the image training model. The procedure to generate the models is similar to what is done with the images to be studied: The “PIL” library is used to take the RGB composition of each pixel in the image…. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. The corresponding arbitrary length is assigned to each of these combinations of values. Once the dataframe with all the combinations has been generated, a Machine Learning (Random Forest) model is generated. This model (from now on called pixelMod) is later used to define which is the length that belongs to each pixel of the studied images.--, in [0093]-[0098], and, …. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. This classification model will have two variables: POSITIVE (COVID-19)—NEGATIVE. The gridsearch_cv (cross validation) was used to iterate between different parameters and then obtain the optimal one. Cross validation was also used to check which was the best score obtained. In the event that the expected threshold were exceeded, it then proceeds with the last step. This model (from now on called entrenaMod) is used to predict which group the unknown measurements correspond to.
[0101] In the case of the samples to be analyzed, the procedure is similar as in the training step: the spectra table of each image is generated with the pixelMod and once the table is completed, it is predicted with the entrenaMod. This result will be exposed in a probabilistic way for each variable, and the resulting numbers are the average image probabilities of the measurement. The processing scheme is detailed in FIG. 3.--, in [0100]-[0101], [0104]; also see LING: e.g., -- the method may include providing the combined-SERS profile to a device configured with a model trained to identify the one or more analytes from the combined-SERS profile. The device may be a computer operable to process a SERS signal and capable of machine learning. The device may also be operable to carry out any chemometric analysis of a spectrum obtained via SERS, including the SERS signal and the combined-SERS profile.--, in [0162]).
Re Claims 15-20, claims 15-20 are the corresponding method claim to claims 1-6, respectively. Claims 15-20 thus are rejected for the similar reasons for claims 1-6. See above discussions with regard to claims 1-6 respectively. Further, DE SIMONE as modified by LING further disclose computer-implemented method of performing the steps (see DE SIMONE: e.g., -- Fig. 3, and, --[0093] The image analysis process called digital spectroscopy is divided into three steps: [0094] a. Generation of the database for the image training that represents the wavelengths to generate the spectra. [0095] b. Generation of the spectrum of each training image, fitting of the data and cross validation of the model. [0096] c. Generation of spectra of the samples to be analyzed and their prediction. …[0098] To develop the model that generates the spectra of the images, images that represent designated “wavelengths” were used. These are the variables of the image training model. The procedure to generate the models is similar to what is done with the images to be studied: The “PIL” library is used to take the RGB composition of each pixel in the image…. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. The corresponding arbitrary length is assigned to each of these combinations of values. Once the dataframe with all the combinations has been generated, a Machine Learning (Random Forest) model is generated. This model (from now on called pixelMod) is later used to define which is the length that belongs to each pixel of the studied images.--, in [0093]-[0098], and, …. the table is generated with all the pixels of each length that each of the images has. This table is the new basis for generating a ML (Random Forest) training model. This classification model will have two variables: POSITIVE (COVID-19)—NEGATIVE. The gridsearch_cv (cross validation) was used to iterate between different parameters and then obtain the optimal one. Cross validation was also used to check which was the best score obtained. In the event that the expected threshold were exceeded, it then proceeds with the last step. This model (from now on called entrenaMod) is used to predict which group the unknown measurements correspond to.
[0101] In the case of the samples to be analyzed, the procedure is similar as in the training step: the spectra table of each image is generated with the pixelMod and once the table is completed, it is predicted with the entrenaMod. This result will be exposed in a probabilistic way for each variable, and the resulting numbers are the average image probabilities of the measurement. The processing scheme is detailed in FIG. 3.--, in [0100]-[0101], [0104]; also see LING: e.g., -- the method may include providing the combined-SERS profile to a device configured with a model trained to identify the one or more analytes from the combined-SERS profile. The device may be a computer operable to process a SERS signal and capable of machine learning. The device may also be operable to carry out any chemometric analysis of a spectrum obtained via SERS, including the SERS signal and the combined-SERS profile.--, in [0162]).
Conclusion
Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any extension fee pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to WEI WEN YANG whose telephone number is (571)270-5670. The examiner can normally be reached on 8:00 - 5:00 pm.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Amandeep Saini can be reached on 571-272-3382. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/WEI WEN YANG/Primary Examiner, Art Unit 2667