WO2023118947A1

WO2023118947A1 - System and method for classifying fluid substance with absorption spectroscopy

Info

Publication number: WO2023118947A1
Application number: PCT/IB2021/062310
Authority: WO
Inventors: Kai Kee Kyle WONG; Cho Yiu George SO; Ping Amanda LIM; Hong Bin TAN; Zhi Qian James FOO; Zhi Wen Steven YEO
Original assignee: Ai Innobio Limited
Priority date: 2021-12-25
Filing date: 2021-12-25
Publication date: 2023-06-29

Abstract

A system and method for identifying, quantifying, and/or classifying at least one specific substance in a biological fluid by making use of absorption spectroscopy, comprising : at least one spectrum processing unit configured to acquire a visible or optical spectrum data of the at least one specific substance in the biological fluid with absorption spectroscopy, whereby forming a spectral signature of fluid for identification, quantification, and/or classification of substance in the biological fluid; and a spectral signature processing unit operatively connected with the at least one spectrum processing unit configured to receive the spectral signature of fluid and process / analyse / learn the same with an automatic processing model built and stored inside the spectral signature processing unit for providing an immediate substance classification result preferably within seconds, wherein an accuracy of substance classification result will improve over time and past substance classification results can be retrieved, feedback, and learnt by the system for classification improvement.

Description

SYSTEM AND METHOD FOR CLASSIFYING FLUID SUBSTANCE WITH ABSORPTION SPECTROSCOPY

Technical Field

[001 ] The present disclosure relates to substance classification of biological fluid, and more particularly to systems and methods for identifying, quantifying, and/or classifying at least one specific substance, such as a pathogen substance, in a biological fluid by making use of absorption spectroscopy, and preferably a distributed system with at least one smart spectrometer utilizing absorption spectroscopy for diagnostic and healthcare purposes.

Background Art

[002] Currently there exists several body fluid substance classification systems and methods while most of them require clinical laboratories for processing and analysis. Such systems and methods use expensive tools and approach for the analysis and diagnosis, and thus might not be widely adopted. Further, the current system processing methodology classifying pathogen substances (e.g. Covid-19) of body fluids primarily utilizes Real Time Polymerase Chain Reaction (RT-PCR) and antigen test kits. RT-PCR method usually takes hours in turnaround time. Antigen test kits may take shorter time but still take around ten minutes. The results of both methods may not be connected online to match the patients and results records. In this way, it is desirable to achieve a relatively cost effective, a faster and/or a better pathogen classification and related public health management.

Brief Summary of Invention

[003] In order to at least partially address the above issues, the present disclosure has proposed a method and a system for identifying, quantifying, and/or classifying at least one specific substance, and preferably a pathogen substance, in a biological fluid by making use of absorption spectroscopy, wherein the system comprises: at least one spectrum processing unit, preferably a spectrometer and more preferably a smart spectrometer, configured to acquire a visible or optical spectrum data, and preferably a spectral signature data, of the at least one specific substance in the biological fluid with absorption spectroscopy, whereby forming a spectral signature of fluid for identification, quantification, and/or classification of substance in the biological fluid; and a spectral signature processing unit operatively connected with the at least one spectrum processing unit, preferably a server and more preferably a cloud server including an artificial intelligent (Al) server, configured to receive the spectral signature of fluid and process I analyse / learn the same with an automatic processing (Al) model built and stored inside the spectral signature processing unit for providing an immediate substance classification result preferably within seconds, wherein an accuracy of substance classification result will improve over time and past substance classification results can be retrieved, feedback, and learnt by the system for classification improvement.

[004] In some embodiments, wherein the spectral signature processing unit is a cloud server including an Al server configured to receive the spectral signature of fluid and process I analyse / learn the same with an Al model built and stored inside the Al server, wherein the Al model could be improved via a federated learning process for boosting the accuracy and generation speed of substance classification result.

[005] In several embodiments, wherein the spectral signature processing unit is a cloud server including a blockchain server configured to receive and store various data of the system including the spectral signature of fluid and the substance classification result way in a decentralized network with a plurality of nodes all across the network on which the system is run, so as to support a secure data sharing architecture from multiple spectrum processing units and secure data for a federated learning process by blockchain, such that a privacy data is well maintained by sharing data model instead of revealing an actual data thereof.

[006] In other embodiments, wherein the spectral signature processing unit is a cloud server including an API server configured to act as an intermediary allowing the spectrum processing unit to communicate with the spectral signature processing unit, and preferably the Al server and the blockchain server; and the API server enables integration with a third party system or application for public health data analysis and information access needs, and/or supports remote diagnostics needs via the third party system or application.

[007] In further embodiments, further comprising: a protocol for system processing including an Al Model Training process and a Test Running process; wherein the biological fluid is prepared with dilution using an appropriate solvent so that the spectrum processing unit, preferably the spectrometer and more preferably the smart spectrometer, can be operated in the visible light range that allows a non-destructive analysis of a target substance to be detected from the biological fluid; wherein a spectrum data collected following same protocol will be compared and analysed in the test running process for building corresponding artificial intelligence model for substance classification; and during the test running process, a similar process will be run for getting a substance classification result, which will be passed and stored in the blockchain server configured in the cloud; and/or wherein one set of protocol for a biological fluid spectrum data collection is used for both the Al Model Training process and the Testing Running process; wherein the biological fluid is diluted using appropriate solvent for the convenience of biological fluid spectrum collection with absorption spectrometry at both the Al Model Training process and the Test Running process; and following the protocol, substance classification is done by diluted biological fluid spectrum classification analysis, wherein the spectral signature data is used for verification of substance positive / negative validation; and the smart spectrometer is calibrated using environmental data to perform the substance classification, and the classification result is stored in the blockchain server for data security and validation requirements.

[008] The present disclosure further relates to a method for identifying, quantifying, and/or classifying at least one specific substance, and preferably a pathogen substance, in a biological fluid by making use of absorption spectroscopy, comprising: providing and configuring at least one spectrum processing unit, preferably a spectrometer and more preferably a smart spectrometer, to acquire a visible or optical spectrum data, and preferably a spectral signature data, of the at least one specific substance in the biological fluid with absorption spectroscopy, whereby forming a spectral signature of fluid for identification, quantification, and/or classification of substance in the biological fluid; and providing and configuring a spectral signature processing unit operatively connected with the at least one spectrum processing unit, preferably a server and more preferably a cloud server including an artificial intelligent (Al) server, to receive the spectral signature of fluid and process I analyse / learn the same with an automatic processing (Al) model built and stored inside the spectral signature processing unit for providing an immediate substance classification result preferably within seconds, wherein an accuracy of substance classification result will improve over time and past substance classification results can be retrieved, feedback, and learnt by the system for classification improvement.

[009] The system and method provided by the present disclosure is simple in structure, reasonable in design, high in accuracy, faster in processing time, and low in cost, such that it enables a desirable and an improved substance classification of secured biological fluid sample data through a distributed system using smart spectrometer(s) utilizing absorption spectroscopy for diagnostic and healthcare purposes.

Brief Description of Drawings

[0010] The present disclosure will be described in details below with reference to the accompanying drawings, in which:

[001 1] Figure 1 is a block diagram of a system for identifying, quantifying, and/or classifying at least one specific substance in a biological fluid sample according to a preferred embodiment of the present disclosure; [0012] Figure 2 is a block diagram of an Al model training process and a test running process according to another preferred embodiment of the present disclosure;

[0013] Figure 3 is a block diagram of a master Al model building process via federated learning according to a further preferred embodiment of the present disclosure; and

[0014] Figure 4 is a block diagram of another system for identifying, quantifying, and/or classifying at least one specific substance in a biological fluid sample according to another preferred embodiment of the present disclosure.

Detailed Description of Invention

[0015] The present disclosure will now be described in further details with reference to the accompanying drawings and embodiments, so as to make the objects, technical solutions and advantages of the present disclosure more apparent.

[0016] According to the present disclosure, absorption spectroscopy is a molecular spectroscopy method that uses the wavelength dependent absorption characteristics of materials to identify and quantify specific substances. A spectrum processing unit, such as a spectrometer, preferably a smart spectrometer (the “Smart Spectrometer”), defined herein might utilises absorption spectroscopy technology. The Smart Spectrometer utilising absorption spectroscopy and artificial intelligent (Al) algorithms may be adapted to classify substances inside body fluids, such as blood, urine, tears, cerebrospinal fluid, milk, sperm, sputum, and the like.

[0017] According to one aspect of the present disclosure, it relates to a system components and processing design (the “System”) for processing and learning data related to spectrum(s) of biological fluids (the “Biological Fluids”) comprising of at least one local party (the “Local Party”) to acquire biological samples of patients through defined protocols. The System deploys at least one Smart Spectrometer configured to acquire spectral signatures of the samples from a system end using absorption spectroscopy technology. The System includes at least one local / cloud server configured with Al server and blockchain server. The PC connected to the Smart Spectrometer and connected to the cloud is configured to receive the spectrum(s) obtained by the Smart Spectrometer(s). The cloud server(s) performs processing, analysis and learning of received data to classify substances inside biological fluids using artificial intelligence algorithms. The System utilizes federated learning process to improve the Al model. The server(s) will report the Al classification results to the users. The System supports third party devices and application integration to enhance the ecosystem of diagnostics and healthcare purposes.

[0018] Currently most body fluid substance classification methods require clinical laboratories for processing and analysis. Such laboratory methods use expensive tools and approach for the analysis and diagnosis. Body fluid substance classifications technique using Smart Spectrometers, absorption spectroscopy and artificial intelligent algorithms can save time and money. The accuracy of artificial intelligent algorithms for substance classifications will improve when more data are collected across different regions. However, the issue of bulkiness of the spectrometers currently available in the market and the concern of data privacy may prevent the multiple Local Parties from sharing the data for the improvement of the substance classifications. [0019] The System of the present disclosure can collect body fluids spectrum data from different Local Parties and can support the data sharing securely for building improved artificial intelligent algorithm models for substance classifications.

[0020] According to another aspect of the present disclosure, which provides a system and a methodology of processing biological fluids spectrum data collected by Smart Spectrometer(s) throughout multiple Local Parties. The system is configured to be fast, accurate and reliable, therefore having characteristics such as to overcome the limitations of the current system mentioned above to using biological fluids for diagnostics and public health purposes.

[0021] A further aspect of the present disclosure relates to a system, a process and a methodology of processing biological spectrum data collected by Smart Spectrometer(s) through multiple Local Parties riding on a blockchain system to secure data for federated learning to improve the artificial intelligence models used for classifying substances in body fluids. The use of federated learning may be required by multiple Local Parties who are concerned about the flow of data outside the national border.

[0022] Another further aspect of the present disclosure has proposed a system, a process and method including Smart Spectrometers(s) connected through multiple Local Parties that can support third party devices and applications, integrated through application programming interfaces (API) for diagnostic and healthcare purposes.

[0023] As stated above, the current system processing methodology classifying pathogen substances (e.g. Covid-19) of body fluids primarily utilizes RT-PCR and antigen test kits. RT-PCR method usually takes hours in turnaround time, and antigen test kits may take shorter time but still take around ten minutes; wherein the results of both methods may not be connected online to match the patients and results records. To the contrary, embodiments of the present disclosure using absorption spectroscopy can perform the classification in seconds as referring to clinical investigation. By collecting body fluids of patients through defined protocols, the system can acquire their spectral signatures and handle pathogen classification accordingly. Using artificial intelligent, the system can process and analyse the body fluid samples within seconds. The accuracy of the classification using artificial intelligent algorithms will improve over time and the past results can be retrieved and learnt for the classification improvement.

[0024] Compared to alternative or prior art solutions, the technical solution proposed by the present disclosure is faster to achieve pathogen classification and related public health management. After the artificial intelligence algorithms have been established, the time needed for the server(s) to report the results to the users is less than a minute. The time taken to acquire the biological samples of patients through defined protocols varies. After the samples have been prepared, the time taken for the Smart Spectrometer to classify substances inside the biological fluids is less than a minute.

[0025] In some embodiments, the present disclosure supports machine learning for continuing improvement. The Smart Spectrometer is combined with artificial intelligence for substance classifications. As a result, the system can adapt to various mutations, as the mutated samples are being fed to the system to learn and for improvement.

[0026] Relative to alternative solutions, the technical solution of the present disclosure is more scalable for the deployment. In some embodiments, the Smart Spectrometer component for the system is extremely portable (less than 500g) and is of the size of a palm. It can be deployed at point-of-care sites or seaports/terminals/airports with ease. Given its portability, it is ideal as a screen device at multiple sites. Applying absorption spectroscopy to get spectral data of body fluids from multiple Local Parties for federated learning to improve artificial intelligent algorithm models for diagnostic and healthcare purpose is more convenient owing to the portability of the system.

[0027] According to an embodiment of the present disclosure, the System might comprises: at least one Smart Spectrometer device configured to acquire spectral signatures of biological fluids using absorption spectroscopy technology and to be processed, analysed, or learnt through the System; and/or at least one protocol used for preparing biological fluids for using classification needs; and/or at least one artificial intelligent server configured for classifying substances inside biological fluids; and/or at least a blockchain server configured to support the secure data sharing architecture from multiple Local Parties, wherein the blockchain system is designed to support the secure data sharing architecture from multiple Local Parties; and/or an application programming interface (API) centre adapted for enabling different diagnostic or healthcare applications to be integrated to the System.

[0028] According to the present disclosure, it appears to the applicant that a novel and inventive technical solution with new System components and process using newly invented spectrometer system components is proposed. The applicant notes that there is no any exact similar prior art in the present technical field can be located or referred. However, some prior arts in other similar or related technical field can be referred for facilitating the easy understanding of individual or overall system components and processing designs of the present disclosure

[0029] Referring to US 2019/0252045 A1 , which discloses a system for acquisition, transmission and processing data related to biological fluids comprising at least one support configured to acquire at least one biological sample of a patient, at least one infrared spectrometer configured to acquire the infrared spectrum of the sample, at least one server connected to the infrared spectrometer by internet and configured to receive the spectrum obtained by the infrared spectrometer, performing a step of post-processing and analysis of received data. To the contrary, the present disclosure differs in that, among others, the system of the present disclosure uses spectrometer working in the visible light range. It provides fast and high resolution spectral signatures acquisition. The system processing methodology of using smart spectrometer(s) based on absorption spectroscopy on classifying pathogen substances of body fluids using artificial intelligent algorithm models learnt from federated learning through a blockchain system with multiple Local Parties securely. The system processing methodology of using smart spectrometer(s) based on absorption spectroscopy for getting spectrum data of body fluids from multiple Local Parties for federated learning to improve artificial intelligent algorithm models for diagnostic and healthcare purposes. The system processing methodology of using smart spectrometer(s) based on absorption spectroscopy for forming multiple Local Parties blockchain system to connect third party application or devices through application programming interfaces (API) for diagnostic and healthcare purposes.

[0030] Regarding US 10,504,020 B2, which discloses an application provisioning system and method. A server provides an application provisioning service. A user of a client provides a schema defining an application. The application interacts with peripherals coupled to the client and receives input from sensors coupled to the peripherals. The sensor data is provided to the server for processing, including by neural networks. The application includes a workflow defining a finite state machine that traverses states at least partially based on the response to sensor data. The server may provide dynamic reallocation of compute resources to resolve demand for classifier training job requests; use of jurisdictional certificates to define data usage and sharing; and data fusion. Applications include manufacturing verification, medical diagnosis and treatment, genomics and viral detection. However, the present disclosure differs in that, among others, it is specific to the Smart Spectrometer and detects the spectral signature of body fluid. An artificial intelligent algorithm is adopted in our system that can provide federated learning to improve model performance as more systems are integrated to the network for public health purpose. The system processing methodology of using smart spectrometer(s) based on absorption spectroscopy for getting spectrum data of body fluids from multiple Local Parties for federated learning to improve artificial intelligent algorithm models for diagnostic and healthcare purposes. Our invention is based on absorption spectroscopy for forming multiple Local Parties blockchain system to connect third party application or devices through application programming interfaces (API) for diagnostic and healthcare purposes.

[0031] In addition, US 2017/0089761 A1 discloses an approach to noninvasively and remotely detect the presence, location, and/or quantity of a target substance in a scene via a spectral imaging system comprising a spectral filter array and image capture array. For a chosen target substance, a spectral filter array is provided that is sensitive to selected wavelengths characterizing the electromagnetic spectrum of the target substance. Elements of the image capture array are optically aligned with elements of the spectral filter array to simultaneously capture spectrally filtered images. These filtered images identify the spectrum of the target substance. Program instructions analyze the acquired images to compute information about the target substance throughout the scene. A color-coded output image may be displayed on a smartphone or computing device to indicate spatial and quantitative information about the detected target substance. The system desirably includes a library of interchangeable spectral filter arrays, each sensitive to one or more target substances. However, the present disclosure differs in that, among others, it utilizes absorption spectrometry in the visible light spectrum. This allows the classification of pathogen substance in the point of interest in a non-destructive manner. The system processing methodology of using smart spectrometer(s) utilizing absorption spectroscopy for getting spectrum data of body fluids from multiple Local Parties for federated learning to improve artificial intelligent algorithm models for diagnostic and healthcare purposes. The system processing methodology of using smart spectrometer(s) based on absorption spectroscopy for forming multiple Local Parties blockchain system to connect third party application or devices through application programming interfaces (API) for diagnostic and healthcare purposes.

[0032] Further, US 2020/0211692 A1 discloses techniques that facilitate integrating artificial intelligence (Al) informatics in healthcare systems using a distributed learning platform. In one embodiment, a computer-implemented is provided that comprises interfacing, by a system operatively coupled to a processor, with a medical imaging application that provides for viewing medical image data. The method further comprises, facilitating, by the system, generation of structured diagnostic data according to a defined ontology in association with usage of the imaging application to perform a clinical evaluation of the medical image data. The method further comprises providing, by the system, the structured diagnostic data to one or more machine learning systems, wherein based on the providing, the one or more machine learning systems employ the structured diagnostic data as training data to generate or train one or more diagnostic models configured to provide artificial intelligence-based diagnostic evaluations of new medical image data.. However, the present disclosure differs in that, among others, it uses artificial intelligence to classify spectral data of any scene of interest. Using a distributed learning platform from that fetch data across multiple systems, this technique can provide better evaluations of new spectral signature data. The system processing methodology of using smart spectrometer(s) riding on absorption spectroscopy for getting spectrum data of body fluids from multiple Local Parties for federated learning to improve artificial intelligent algorithm models for diagnostic and healthcare purposes. The system processing methodology of using smart spectrometer(s) riding on absorption spectroscopy for forming multiple Local Parties blockchain system to connect third party application or devices through application programming interfaces (API) for diagnostic and healthcare purposes.

[0033] According to an embodiment of the present disclosure, an example system for identifying, quantifying, and/or classifying at least one specific substance in a biological fluid by making use of absorption spectroscopy might comprises one or more of the following parts:

[0034] Part 1 , wherein at least one smart spectrometer configured for acquiring the spectral signatures of biological samples is used to provide a spectral signature for classification, as described in Figure 1 and named as “Local Party System”.

[0035] Part 2, wherein at least one protocol is required for the preparation of samples at the scene of interest to prepare and train the artificial intelligence model, as described by Figure 2 and named as “Training and Test Running Process”.

[0036] Part 3, wherein at least one artificial intelligent server configured with an Al model is employed for classifying substances inside biological fluids; and/or at least one blockchain server is configured to support the secure data sharing from multiple Local Parties. Preferably, federated learning for machine learning is incorporated for the cross regional master Al Model building. More preferably, federated learning by blockchain is incorporated so that privacy data is well maintained by sharing data model instead of revealing the actual data as described by Figure 3 and named as “Master Al Model Building by Federated Learning Process”.

[0037] Part 4, wherein at least one application programming interface (API) server is configured for enabling one or more different diagnostic or healthcare applications to be integrated to the System. Preferably, at least one blockchain server is configured to store data security and ensure security for integrating with third (3^rd) party enquires into the system, as depicted by Figure 4 and named as “The system components for API integration”.

[0038] According to the present disclosure, it generally relates to system components and processing methodology of using smart spectrometer(s) utilizing absorption spectroscopy to classify pathogen substances inside body fluids using artificial intelligent algorithm models. The classification model learnt from federated learning through a blockchain system with multiple Local Parties connected securely. The Local Party System components are described with reference to the Local Party system invention. The system process invention is described with reference to the Al Model Training process and the Test Running process. More details will be described hereinunder with respect to Figure 1 and/or Figure 2.

[0039] The system processing methodology of using smart spectrometer(s) utilizing absorption spectroscopy for getting spectrum data of body fluids from multiple Local Parties for federated learning to improve artificial intelligent algorithm models for diagnostic and healthcare purposes. The master Al Model building, blockchain and federated learning system components and process design will be described with reference to Part 3 of the system of the present disclosure. The relevant system components and flow are described below with respect to Figure 3.

[0040] The system processing methodology of using smart spectrometer(s) riding on absorption spectroscopy for forming multiple Local Parties blockchain system to connect third party application or devices through application programming interfaces (API) for diagnostic and healthcare purposes. The API system design for integrating with third party system for diagnostic and healthcare purposes are described with reference to Part 4 of the system of the present disclosure. The relevant system components and flow will be described under with respect to Figure 4.

[0041] Now referring to Figure 1 , which depicts a block diagram of a classification system for identifying, quantifying, and/or classifying at least one specific substance in a biological fluid sample according to a preferred embodiment of the present disclosure. [0042] In some embodiments, the Part 1 or the Local party system of the above classification system comprises: at least one smart spectrometer for acquiring the spectral signatures of biological samples is used to provide a spectral signature for classification.

[0043] The system processing methodology of using smart spectrometer(s) is riding on absorption spectroscopy on classifying substances of body fluids, wherein at least one Smart Spectrometer device is configured to acquire spectral signatures of biological fluids using absorption spectroscopy technology, such that spectral signatures might be processed, analysed, or learnt through the system. The rapid acquisition of spectral signatures in the visible light range of the target substances of body fluids can provide a spectral signature for classification.

[0044] The classification system including the Part 1 and the work flow is illustrated under with reference to Figure 1 , wherein, at a Local Party, at least one smart spectrometer [B] connected with a PC/Laptop (C) is set up for generating text file (ii) spectrum for a body fluid Sample in a new cuvette [A] with token-controlled cuvette ID inserted to the smart spectrometer [B] for substance classification. The smart spectrometer generates text file spectrum as spectral signature using absorption spectroscopy technology riding on the specimen of body fluid stored inside a cuvette and inserted to the smart spectrometer. The text file spectrum data is uploaded to a Cloud server (D) through the internet connected PC [C]. In some embodiments, the Cloud server is configured to include an Al server, a Blockchain Server, and/or an API server. Preferably, Al model built inside the Al server is run for substances classification. The Blockchain server is built for managing the spectrum text files transmission and storage. Blockchain technology is adopted to overcome threats of sensitive data transmissions and allows to decentralize sensitive local operations while preserving a high level of security. The spectrum data (II) uploaded to the cloud and classification results (I) received through the Al algorithm analysis will be stored in the configured blockchain server securely. Passing through the API server, the results are returned through the internet o the local PC (C).

[0045] Referring to Figure 2, which is a block diagram of an Al model training process and a test running process according to another preferred embodiment of the present disclosure, wherein, at least one protocol is required for the preparation of samples at the point of interest to prepare and train the artificial intelligence model.

[0046] According to the embodiment, a protocol for the system processing should include both an Al Model Training process and a Test Running process. The Biological Fluids should be prepared with dilution using appropriate solvents so that the smarter spectrometer can be operated in the visible light range that allows a non-destructive analysis of the target substances to be detected from the samples. The preparation of samples depends on the point of interest and the system is versatile to be customized to the processing methodology at the local party. Spectrum data collected following the same protocol will be compared and analysed in the Test Running for building the corresponding artificial intelligence model for substances classifications. During the test running process, similar process will be run for getting the substances classification results. The results will be passed and stored in the blockchain server configured in the cloud.

[0047] As illustrated in Figure 2, one set of protocol for a Biological Fluid spectrum data collection might be used for both the Al Model Training (A) process and, also the Testing Running (B) process. The Biological Fluid is diluted using appropriate solvent for the convenience of Biological Fluid spectrum collection using a Smart Spectrometer noted PC riding on absorption spectrometry at both the Al Model Training stage and, also the Test Running stage, by using the corresponding artificial intelligence model trained, substance classification or Positive / Negative Validation can be achieved. Following the protocol, substance classification is done by diluted Body Fluids spectrum classification analysis. The spectral signature is used for verification of substance positive I negative validation. The token controlled smart spectrometer is calibrated using environmental data to perform the classification test. The classification results are stored in the blockchain server configured at the cloud for data security and validation requirements.

[0048] For the Al Model Training process according to the embodiment, there are following eight steps as shown in Figure 2.

[0049] A1 . Collect a body fluid sample with known classification result and diluted using a standard set of protocol (the “Sample”).

[0050] A2. Put the Sample in to a cuvette certified to be used in the Smart Spectrometer.

[0051] A3. Put the cuvette in the Smart Spectrometer that is connected to a PC.

[0052] A4. Run the Spectrometer for getting the spectral signature data of the Sample (the “Spectrum”).

[0053] A5. Send the Spectrum to the cloud with an Al server configured.

[0054] A6. Analyse the Spectrum using the Al model built and stored inside the Al server.

[0055] A7. Input the known classification and compare the Al model result

[0056] A8. Process the machine learning algorithms for finetuning the Al model.

[0057]

[0058] For the Test Running process according to the embodiment, there might be eight steps as follows.

[0059] B1 . Collect a body fluid sample and diluted using a standard set of protocol (the “Sample”).

[0060] B2. Put the Sample in to a cuvette certified to be used in the Smart Spectrometer.

[0061] B3. Put the cuvette in the Smart Spectrometer that is connected to a PC.

[0062] B4. Run the Spectrometer for getting the spectral signature data of the Sample (the “Spectrum”). [0063] B5. Send the Spectrum to the cloud with an Al server configured.

[0064] B6. Analyse the Spectrum using the Al model built and stored inside the Al server.

[0065] B7. Return the classification result to the PC connected to the Smart Spectrometer.

[0066] B8. Transfer the Store the Cuvette ID and result into the blockchain server configured in the cloud system.

[0067] It appears to the inventor of the present disclosure that the existing or prior art methods may not support getting the classification results immediately. Existing methods may use methods like PCR that require testing at a lab. Lab tests may take long time and no immediate results can be received. Besides, the specimen cannot be reused for other kinds of tests. Spectral signatures of the body fluid substances are not used for the substance classifications.

[0068] According to the present disclosure, riding on a smart spectrometer installed at the Local Party for collecting the spectral signature of the body fluid following a predefined protocol, immediate substances classifications can be done. The Smart Spectrometer, through absorption spectroscopy technology enable spectral of biological fluids to be acquire and processed through the System. The spectral signature acquired can be classified after they have been passed through the System. By having at least one Smart Spectrometer, each of them can contribute to acquiring spectral signatures to be learnt through the System.

[0069] Instead of waiting long for the classification result at the local party, the classification can be done in seconds by having the smart spectrometer be connected to the desktop and the local party system. Having more than one Local Party systems connected, it allows parallel processing of samples that can acquire a lot of spectral signatures for the System at the same time. The Al model can be improved over time as more data are collected for Al Model Training. The Test Running using blockchain token technology can handle the data protection and security concern.

[0070] In this regard, the value of the above described components and processes lies in that the employment of the protocol might support the faster and simplified way of classification. The token controlled smart spectrometer, with the protocol that is used for the testing service launch can prepare the system to be deployed at any scene of interest. The classification accuracy improvement overtime is handled through the connection with the cloud with a shared Al model configured.

[0071] According to the present disclosure, the classification result can be obtained in seconds and the classification model can be improved over time in a comparatively cost-effective way. Using more than one Smart Spectrometer to acquire spectral signatures for the System allows multiple parties to contribute to collecting the require data for the System. It allows a quick way to acquire sufficient data for a better model. With more than one Smart Spectrometer being employed in the system, it can be connected to one another to form a better System by having more data for processing and analysing, to better classify substances of body fluids.

[0072] Now referring to Figure 3, which is a block diagram of a master Al model building process via federated learning according to a further preferred embodiment of the present disclosure Part 3 Invention: Master Al Model Building by Federated Learning Process, wherein, at least one artificial intelligent server and with an Al model is employed for classifying substances inside biological fluids. Al models improve as more data are collected at a Local Party and multiple local parties. At least a blockchain server is configured to support the secure data sharing from multiple Local Parties. Tokens are used for creating trust in the blockchain. Tokens are used for managing the access rights of the classification tests. Blockchain technology is used to hold the testing results. Blockchain can store the results in a way that is accessible to the authenticated parties on the network and is completely immutable. Federated learning by blockchain is incorporated so that privacy data is well maintained by sharing data model instead of revealing the actual data. [0073] According to the present disclosure, Federated learning might be used as the machine learning technique that trains the Al model algorithm across multiple decentralized local clouds holding local data samples, without exchanging them. Federated learning enables multiple local parties to build a common, robust machine learning model without sharing data, thus allowing to address cross regional data privacy, data security, data access rights kinds of issues. The federated learning data training approach brings solutions to train a global Al model for substance classification while respecting security constraints. Blockchain technology authenticates the data and federated learning trains the Al model among different local regions while preserving the privacy of local regional data. The relevant process is illustrated in Figure 3, including the following local process steps run at each local region (C):

[0074] C1 . Body fluid Sample collection using standardised protocol and hold inside a cuvette.

[0075] C2. The cuvette is inserted to a Smart Spectrometer connected to a PC linked to the cloud.

Absorption spectrometry technology is used for generating the spectral signature of the Sample.

[0076] C3. The spectral signature of the Sample is generated as a set of unique Spectrum data by the PC with the Smart Spectrometer driver installed.

[0077] C4. The Spectrum data is send to the cloud with local Al and blockchain servers configured. The data is stored in the blockchain server and handled by the Al server. Al algorithms in the Al server is used to classify substances in the Sample as positive or negative.

[0078] C5. For an unknown Sample result, the spectrum data will be used as input data for substance classification by the Al Model. The results recorded in the blockchain will be returned to the PC through the internet and cloud as shown in step C2 and described in Figure 2 (B) Test Running Steps using the system components as described in Figure 1 . For a known Sample result, the spectrum data will be used as input data for machine learning and step f will be followed.

[0079] In some embodiments, which comprise following cross regional process steps for federated learning (D):

[0080] D1 . Federated learning as a machine learning technique is used to train the Al algorithm across multiple local servers holding local Spectrum data samples.

[0081] D2. The master Al model is trained and fine-tuned for improving the local Al models as shown in step C5.

[0082] It appears to the inventor of the present disclosure that the existing or prior art method may not support gathering and enhancing the classification results immediately. Existing methods may use methods like PCR that require testing at a lab. Cross regional data sharing may have data security concern. Fast spectral signatures collection of the body fluid substances is not commonly used for the substance classifications. The related analyses results are not commonly cumulated for the model training and improvement.

[0083] According to the present disclosure, instead of waiting long for the classification result from the local labs, the classification can be done in seconds by having a spectrometer be connected to the internet linked PC. Having more than one local region cloud connected, it allows parallel processing of samples that can acquire a lot of spectral signatures for the System at the same time. The method may support gathering and fine tuning the classification results through the model trained and improved. Cross regional data sharing security concern is handled through the blockchain technology. Fast spectral signatures collection of the body fluid substances is used for the substance classifications. The related spectral signatures analyses results are cumulated for the model training and improvement.

[0084] In this regard, the advantageous effect or value of the above described process lies in that:

[0085] as tokens are used for creating trust in the blockchain, access rights can be managed. As blockchain technology is used to store the spectrum data and testing results, the data is immutable. The Al model can be improved through machine learning with data collected among different regions. The balance between data privacy and transparency for model improvement purpose can be handled by the blockchain and federated learning technologies. Improved master Al models-built riding on the federated learning can improve the accuracy of substances classification.

[0086] Referring to Figure 4, which is a block diagram of another system for identifying, quantifying, and/or classifying at least one specific substance in a biological fluid sample according to another preferred embodiment of the present disclosure, wherein, at least one application programming interface (API) server is configured for one or more different diagnostic or healthcare application to be integrated to the System. In some embodiments, at least one blockchain server is configured to store data security and ensure security for integrating with 3rd party enquires into the system.

[0087] According to the present disclosure, the API server acts as a software intermediary that allows the Smart Spectrometer application to talk to the Al system and Blockchain storage layer. The API server application sends message to the PC upon success of connection and data uploading. The local party with a local party application connected as a user interface for a user sends request for body fluid classification. The text file spectrum data stored in the blockchain sever connected through the API Server is analysed and classified riding on an Al analysis done by the Al server connected through the API server also. The record is received and marked through the blockchain server under smart contracts. Also, the API server manage other third-party healthcare related applications connecting to the system to support more healthcare needs. The system of the present embodiment is illustrated in Figure 4, which comprises one or more of the following features.

[0088] The front-end layer with an application program and includes at least one PC connected Smart Spectrometer.

[0089] Third party healthcare related application

[0090] The API server connects the messing among the applications in the System. It also connects the third-party applications for connecting to the system.

[0091] The Al system sever which manages the machine learning for building the Al model and the Al model for the classification support.

[0092] The blockchain server layer with Node manager to manage the spectrum data and classification results transactions.

[0093] The front-end layer (A) with an application program and includes at least one PC connected Smart Spectrometer collects body fluid spectral signatures in form of spectrum data files. The spectrum data files are sent to the blockchain system through the API layer (C). The data files handled by the blockchain layer immutable and be transmitted securely.

[0094] Third party healthcare related application (B) can be linked to the System through API for supporting public health needs if any. The API server connects the messaging among the applications in the System. It also connects the third-party applications for connecting to the system. The Al system layer (D) with the Al system sever which manages the machine learning for building the Al model and the Al model for the classification support. The spectrum data files stored under the blockchain server are sent to the Al system layer through the API for substances classification using the trained Al model. The results are sent back the blockchain layer through the API. The blockchain server layer with Node manager to manage and store the spectrum data and classification results transactions.

[0095] The inventor notes that the existing or prior art method may not support integration with 3rd party health system. Also, data may not be shared securely among different platforms for public health data analysis and information access needs.

[0096] However, the current system components support integration with 3rd party health system. Also, blockchain system managed by nodes can store data securely as API server support integration with 3rd Party systems for public health data analysis and information access needs. Also, different 3rd party healthcare applications can integrate to support the remote diagnostics needs.

[0097] In this regard, the advantageous effect or value of the present embodiment lies in that:

[0098] APIs process transactions that respond to client requests to access the blockchain data libraries.

The blockchain system validate an integrated client through the API server to ensure they have the permissions to access the data in the blockchain. The blockchain server is configured for a database stored can benefit the APIs for getting stored data more securely. Blockchains are immutable. This means that once the healthcare data is stored on the blockchain storage system, it can't be deleted or changed. The immutable nature of blockchain database makes them tamper proof while processing large volumes of transactions. As APIs are used with blockchain technology the core functionality of APIs stays the same but a database on a blockchain addresses the key security issue of potential hacks.

[0099] In other embodiments according to the present disclosure, which proposed a system for identifying, quantifying, and/or classifying at least one specific substance, and preferably a pathogen substance, in a biological fluid by making use of absorption spectroscopy, wherein the system comprises: at least one spectrum processing unit, preferably a spectrometer and more preferably a smart spectrometer, configured to acquire a visible or optical spectrum data, and preferably a spectral signature data, of the at least one specific substance in the biological fluid with absorption spectroscopy, whereby forming a spectral signature of fluid for identification, quantification, and/or classification of substance in the biological fluid; and a spectral signature processing unit operatively connected with the at least one spectrum processing unit, preferably a server and more preferably a cloud server including an artificial intelligent (Al) server, configured to receive the spectral signature of fluid and process / analyse / learn the same with an automatic processing (Al) model built and stored inside the spectral signature processing unit for providing an immediate substance classification result preferably within seconds, wherein an accuracy of substance classification result will improve over time and past substance classification results can be retrieved, feedback, and learnt by the system for classification improvement.

[00100] The present disclosure further relates to a method for identifying, quantifying, and/or classifying at least one specific substance, and preferably a pathogen substance, in a biological fluid by making use of absorption spectroscopy, comprising: providing and configuring at least one spectrum processing unit, preferably a spectrometer and more preferably a smart spectrometer, to acquire a visible or optical spectrum data, and preferably a spectral signature data, of the at least one specific substance in the biological fluid with absorption spectroscopy, whereby forming a spectral signature of fluid for identification, quantification, and/or classification of substance in the biological fluid; and providing and configuring a spectral signature processing unit operatively connected with the at least one spectrum processing unit, preferably a server and more preferably a cloud server including an artificial intelligent (Al) server, to receive the spectral signature of fluid and process I analyse I learn the same with an automatic processing (Al) model built and stored inside the spectral signature processing unit for providing an immediate substance classification result preferably within seconds, wherein an accuracy of substance classification result will improve over time and past substance classification results can be retrieved, feedback, and learnt by the system for classification improvement. In some embodiments, wherein the spectral signature processing unit is a cloud server including an Al server configured to receive the spectral signature of fluid and process / analyse / learn the same with an Al model built and stored inside the Al server, wherein the Al model could be improved via a federated learning process for boosting the accuracy and generation speed of substance classification result.

[00101] In several embodiments, wherein the spectral signature processing unit is a cloud server including a blockchain server configured to receive and store various data of the system including the spectral signature of fluid and the substance classification result way in a decentralized network with a plurality of nodes all across the network on which the system is run, so as to support a secure data sharing architecture from multiple spectrum processing units and secure data for a federated learning process by blockchain, such that a privacy data is well maintained by sharing data model instead of revealing an actual data thereof.

[00102] In other embodiments, wherein the spectral signature processing unit is a cloud server including an API server configured to act as an intermediary allowing the spectrum processing unit to communicate with the spectral signature processing unit, and preferably the Al server and the blockchain server; and the API server enables integration with a third party system or application for public health data analysis and information access needs, and/or supports remote diagnostics needs via the third party system or application.

[00103] In further embodiments, further comprising: a protocol for system processing including an Al Model Training process and a Test Running process; wherein the biological fluid is prepared with dilution using an appropriate solvent so that the spectrum processing unit, preferably the spectrometer and more preferably the smart spectrometer, can be operated in the visible light range that allows a non-destructive analysis of a target substance to be detected from the biological fluid; wherein a spectrum data collected following same protocol will be compared and analysed in the test running process for building corresponding artificial intelligence model for substance classification; and during the test running process, a similar process will be run for getting a substance classification result, which will be passed and stored in the blockchain server configured in the cloud; and/or wherein one set of protocol for a biological fluid spectrum data collection is used for both the Al Model Training process and the Testing Running process; wherein the biological fluid is diluted using appropriate solvent for the convenience of biological fluid spectrum collection with absorption spectrometry at both the Al Model Training process and the Test Running process; and following the protocol, substance classification is done by diluted biological fluid spectrum classification analysis, wherein the spectral signature data is used for verification of substance positive I negative validation; and the smart spectrometer is calibrated using environmental data to perform the substance classification, and the classification result is stored in the blockchain server for data security and validation requirements.

[00104] The present disclosure is described according to specific embodiments, but those skilled in the art will appreciate that various changes and equivalents might be made without departing from the scope of the present disclosure. In addition, many modifications might be made to the present disclosure without departing from the scope of the invention in order to adapt to specific circumstances or components of the present disclosure. Accordingly, the present disclosure is not limited to the specific embodiments disclosed herein, and shall include all embodiments falling within the scope of the claims.

Claims

1. A system for identifying, quantifying, and/or classifying at least one specific substance, and preferably a pathogen substance, in a biological fluid by making use of absorption spectroscopy, comprising: at least one spectrum processing unit, preferably a spectrometer and more preferably a smart spectrometer, configured to acquire a visible or optical spectrum data, and preferably a spectral signature data, of the at least one specific substance in the biological fluid with absorption spectroscopy, whereby forming a spectral signature of fluid for identification, quantification, and/or classification of substance in the biological fluid; and a spectral signature processing unit operatively connected with the at least one spectrum processing unit, preferably a server and more preferably a cloud server including an artificial intelligent (Al) server, configured to receive the spectral signature of fluid and process I analyse I learn the same with an automatic processing (Al) model built and stored inside the spectral signature processing unit for providing an immediate substance classification result preferably within seconds, wherein an accuracy of substance classification result will improve over time and past substance classification results can be retrieved, feedback, and learnt by the system for classification improvement.

2. The system of claim 1 , wherein the spectral signature processing unit is a cloud server including an Al server configured to receive the spectral signature of fluid and process / analyse I learn the same with an Al model built and stored inside the Al server, wherein the Al model could be improved via a federated learning process for boosting the accuracy and generation speed of substance classification result.

3. The system of claim 1 or 2, wherein the spectral signature processing unit is a cloud server including a blockchain server configured to receive and store various data of the system including the spectral signature of fluid and the substance classification result way in a decentralized network with a plurality of nodes all across the network on which the system is run, so as to support a secure data sharing architecture from multiple spectrum processing units and secure data for a federated learning process by blockchain, such that a privacy data is well maintained by sharing data model instead of revealing an actual data thereof.

4. The system of any one of claims 1 - 3, wherein the spectral signature processing unit is a cloud server including an API server configured to act as an intermediary allowing the spectrum processing unit to communicate with the spectral signature processing unit, and preferably the Al server and the blockchain server; and the API server enables integration with a third party system or application for public health data analysis and information access needs, and/or supports remote diagnostics needs via the third party system or application.

5. The system of any one of claims 1 - 4, further comprising a protocol for system processing including an Al Model Training process and a Test Running process; wherein the biological fluid is prepared with dilution using an appropriate solvent so that the spectrum processing unit, preferably the spectrometer and more preferably the smart spectrometer, can be operated in the visible light range that allows a non-destructive analysis of a target substance to be detected from the

E:\TT\PIP21\P21171325IB - Smart Spectrometer\P21171325IB - Spec (24 Dec 2021).docx 1 biological fluid; wherein a spectrum data collected following same protocol will be compared and analysed in the test running process for building corresponding artificial intelligence model for substance classification; and during the test running process, a similar process will be run for getting a substance classification result, which will be passed and stored in the blockchain server configured in the cloud; and/or wherein one set of protocol for a biological fluid spectrum data collection is used for both the Al Model Training process and the Testing Running process; wherein the biological fluid is diluted using appropriate solvent for the convenience of biological fluid spectrum collection with absorption spectrometry at both the Al Model Training process and the Test Running process; and following the protocol, substance classification is done by diluted biological fluid spectrum classification analysis, wherein the spectral signature data is used for verification of substance positive / negative validation; and the smart spectrometer is calibrated using environmental data to perform the substance classification, and the classification result is stored in the blockchain server for data security and validation requirements.

6. A method for identifying, quantifying, and/or classifying at least one specific substance, and preferably a pathogen substance, in a biological fluid by making use of absorption spectroscopy, comprising: providing and configuring at least one spectrum processing unit, preferably a spectrometer and more preferably a smart spectrometer, to acquire a visible or optical spectrum data, and preferably a spectral signature data, of the at least one specific substance in the biological fluid with absorption spectroscopy, whereby forming a spectral signature of fluid for identification, quantification, and/or classification of substance in the biological fluid; and providing and configuring a spectral signature processing unit operatively connected with the at least one spectrum processing unit, preferably a server and more preferably a cloud server including an artificial intelligent (Al) server, to receive the spectral signature of fluid and process / analyse / learn the same with an automatic processing (Al) model built and stored inside the spectral signature processing unit for providing an immediate substance classification result preferably within seconds, wherein an accuracy of substance classification result will improve over time and past substance classification results can be retrieved, feedback, and learnt by the system for classification improvement.

7. The method of claim 6, wherein the spectral signature processing unit is a cloud server including an Al server configured to receive the spectral signature of fluid and process / analyse / learn the same with an Al model built and stored inside the Al server, wherein the Al model could be improved via a federated learning process for boosting the accuracy and generation speed of substance classification result.

8. The method of claim 6 or 7, wherein the spectral signature processing unit is a cloud server including a blockchain server configured to receive and store various data of the system including the spectral signature of fluid and the substance classification result way in a decentralized network with a plurality of nodes all across the network on which the system is run, so as to support a secure data sharing architecture from multiple spectrum processing units and secure data for a federated learning process by blockchain, such that a privacy data is well maintained by sharing data model instead of revealing an actual data thereof.

E:\TT\PIP21\P21171325IB - Smart Spectrometer\P21171325IB - Spec (24 Dec 2021).docx 2 The method of any one of claims 6 - 8, wherein the spectral signature processing unit is a cloud server including an API server configured to act as an intermediary allowing the spectrum processing unit to communicate with the spectral signature processing unit, and preferably the Al server and the blockchain server; and the API server enables integration with a third party system or application for public health data analysis and information access needs, and/or supports remote diagnostics needs via the third party system or application. . The method of any one of claims 6 - 9, further comprising: providing a protocol for system processing including an Al Model Training process and a Test Running process; wherein the biological fluid is prepared with dilution using an appropriate solvent so that the spectrum processing unit, preferably the spectrometer and more preferably the smart spectrometer, can be operated in the visible light range that allows a non-destructive analysis of a target substance to be detected from the biological fluid; wherein a spectrum data collected following same protocol will be compared and analysed in the test running process for building corresponding artificial intelligence model for substance classification; and during the test running process, a similar process will be run for getting a substance classification result, which will be passed and stored in the blockchain server configured in the cloud; and/or wherein one set of protocol for a biological fluid spectrum data collection is used for both the Al Model Training process and the Testing Running process; wherein the biological fluid is diluted using appropriate solvent for the convenience of biological fluid spectrum collection with absorption spectrometry at both the Al Model Training process and the Test Running process; and following the protocol, substance classification is done by diluted biological fluid spectrum classification analysis, wherein the spectral signature data is used for verification of substance positive I negative validation; and the smart spectrometer is calibrated using environmental data to perform the substance classification, and the classification result is stored in the blockchain server for data security and validation requirements.

E:\TT\PIP21\P21171325IB - Smart Spectrometer\P21171325IB - Spec (24 Dec 2021).docx 3