WO2016033186A1

WO2016033186A1 - Chemical monitoring system

Info

Publication number: WO2016033186A1
Application number: PCT/US2015/046957
Authority: WO
Inventors: Chengmeng HSIUNG; Jing Li
Original assignee: Eloret Corporation
Priority date: 2014-08-26
Filing date: 2015-08-26
Publication date: 2016-03-03

Abstract

The specification relates to a device for monitoring for a chemical event in a gas mixture. The device is capable of receiving data signals from a sensor and calculating a dynamic feature vector using these data signals. The vector is compared to an event detection model containing a plurality of chemical detection vectors. If the vector matches a chemical detection vector within the event detection model, the above steps are repeated until a threshold number of matches occur in consecutive order. When the threshold number is reached, a regression dynamic feature vector is calculated using the data signals and compared to a regression model containing a plurality of chemical concentration vectors. A concentration of the specific chemical within the gas mixture based is then determined.

Description

CHEMICAL MONITORING SYSTEM

CROSS REFERENCE TO RELATED APPLICATIONS

[0001] This application claims benefit to U.S. Provisional Application No.

62/042,124 entitled "INTEGRATED ALGORITHM FOR CONTINUOUS EVENT

DETECTION AND QUANTITATIVE REGRESSION USING AN ARRAY OF NANO CHEMICAL SENSORS," filed on August 26, 2014, hereby incorporated by reference.

STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT

[0002] This invention was made with Government support under contract number

NAS2-03144 awarded by NASA. The Government has certain rights in the invention

BACKGROUND

[0003] The subject matter described herein relates to a chemical monitoring system.

[0004] Chemical detection has drawn a lot of attention over the past few decades for early warning of hazardous events in the security and safety areas. For these systems to be beneficial, continuous monitoring of chemicals for hazardous event detection is necessary. However, the complicated background where the chemicals are present often interferes with the event detection. Imperfect sensor response signals, which inherently exist in any type of electrical measurement, can also cause inaccurate measurements and false alarms.

[0005] Many conventional algorithms try to overcome the above deficiencies by using a single sensor and only analyzing for a single chemical in a distinct background. However, these sensors still are imperfect due to cross sensitivity or interferences. In other words, the electrical signals collected from these sensors are associated with noise, drift, and fluctuation which lead to the inaccurate measurements and false alarms. In order to deal with these issues and provide a more accurate measurement, a more selective sensor array combined with a self-correcting algorithm needs to be developed.

SUMMARY

[0006] The disclosed technology relates to a chemical monitoring system. The chemical monitoring system can use Dynamic Pattern Recognition (or DPR) where an instrument continuously monitors ambient air or some other atmosphere for detection and identification of chemical gases not usually present. The chemical monitoring system can also use Static Pattern Recognition (or SPR) where a button is pressed telling the chemical monitoring system a starting time and an ending time of a sample period, e.g., an exhaled breath of a person, for detection and identification of chemicals not usually present, e.g. certain drugs or cancers.

[0007] In one implementation, the chemical monitoring system of the disclosed technology is capable of receiving raw data signals from a nanochemical sensor. These data signals are filtered for noise and outlier signals are rejected. The filtered signals are sent to a dynamic feature vector calculator and the resultant is compared to an event detection model containing a plurality of chemical detection vectors. (The event detection model was built using a machine learning algorithm and training data sets.) If it is found that the dynamic feature vector matches or closely matches a chemical detection vector within the event detection model a certain number of times, e.g., five matches in a row, an alarm can be triggered displaying a specific chemical that was identified within the gas mixture. Once the alarm is triggered, the data signals can be sent to a regression dynamic feature vector calculator where the resultant is compared to a regression model containing a plurality of chemical concentration vectors. (The regression model was also built using a machine learning algorithm and training data sets.). Based upon the regression dynamic feature vector matching or closely matching a chemical concentration vector within the regression model, a concentration of the specific chemical within the gas mixture can be determined and displayed.

[0008] In one implementation, a computer-implemented method for monitoring for a chemical event in a gas mixture, the method can comprise: receiving data signals from at least one sensor on a timed basis; calculating a dynamic feature vector for event detection using the data signals; comparing the dynamic feature vector for event detection to an event detection model containing a plurality of chemical detection vectors; determining if the dynamic feature vector for event detection matches a chemical detection vector within the event detection model thereby identifying a specific chemical within the gas mixture;

repeating steps 1 to 4 until a threshold number of matches occur in consecutive order; when the threshold number is reached, calculating a regression dynamic feature vector; comparing the regression dynamic feature vector to a regression model containing a plurality of chemical concentration vectors; and determining a concentration of the specific chemical within the gas mixture based upon the regression dynamic feature vector matching a chemical concentration vector within the regression model.

[0009] In some implementations, the method can further comprise the steps of:

triggering an alarm identifying the presence of the specific chemical when the threshold number is reached. In some implementations, the method can further comprise the steps of: displaying the concentration of the specific chemical. In some implementations, the method can further comprise the steps of: filtering the data signals for noise and median readings. In some implementations, the method can further comprise the steps of: normalizing and auto scaling the dynamic feature vector for event detection. In some implementations, the data signals can be electrical resistance readings of a nanochemical sensor. In some

implementations, the gas mixture can be ambient air. In some implementations, the timed basis is in a range of once every 1 to 60 seconds on continuous basis. [00010] In some implementations, the dynamic feature vector for event detection is calculated with a formula of: DiffFVt = (Rt-R0)/R0 - (Rtl - R01)/R01 where Rt is a resistance reading at the current time, t, R0 is the resistance reading at a fixed time step, t-m, before the current time, Rtl is the resistance reading at a fixed time step, w, before the current time and R01 is the resistance reading at a fixed time step before w, w-n.

[00011] In some implementations, the regression dynamic feature vector is calculated with a formula of: FVt = (Rt-R0)/R0 where Rt is a resistance reading at the current time, t, R0 is the resistance reading at the latched reading before the chemical event happened.

[00012] In some implementations, a system for monitoring for a chemical event in a gas mixture, the system comprising: at least chemical detection unit having a chemical detection software application installed thereon; and a sensor, which sensor is coupled to the at least one mobile device, the chemical detection software application programmed to:

receive data signals from at least one sensor on a timed basis; calculate a dynamic feature vector using the data signals; compare the dynamic feature vector to an event detection model containing a plurality of chemical detection vectors; determine if the dynamic feature vector matches a chemical detection vector within the event detection model thereby identifying a specific chemical within the gas mixture; repeat steps (i) to (iv) until a threshold number of matches occur in consecutive order; when the threshold number is reached, calculate a regression dynamic feature vector; compare the regression dynamic feature vector to a regression model containing a plurality of chemical concentration vectors; and determine a concentration of the specific chemical within the gas mixture based upon the regression dynamic feature vector matching a chemical concentration vector within the regression model.

[00013] One advantage of the disclosed technology is that the dynamic feature vector corrects for baseline drift. BRIEF DESCRIPTION OF THE DRAWINGS

[00014] Figure 1 is a block diagram of an example of a system used with the disclosed technology;

[00015] Figures 2a-b are a flow chart showing an example process of the disclosed technology;

[00016] Figures 3a-b are a flow chart showing an example process of the disclosed technology;

[00017] Figure 4 is a block diagram of an example of a system used with the disclosed technology;

[00018] Figure 5 is a flow chart showing an example process of the disclosed technology; and

[00019] Figure 6 is a block diagram of an example of a system used with the disclosed technology.

DETAILED DESCRIPTION

[00020] The disclosed technology relates to a chemical monitoring system for a gas mixture, e.g., ambient air or an exhaled breath of a person.

[00021] In one implementation, the chemical monitoring system is capable of receiving raw data signals from an array of nanochemical sensors. These data signals can be filtered for noise and outlier readings can be rejected. The filtered signals are sent to a dynamic feature vector calculator and the resultant is compared to an event detection model containing a plurality of chemical detection vectors. (The event detection model was built using a machine learning algorithm and training data sets.) If it is found that the dynamic feature vector matches or nearly matches a chemical detection vector within the event detection model a certain number of times, e.g., five matches in a row, an alarm can be triggered displaying a specific chemical that was identified within the gas mixture. Once the alarm is triggered, the data signals are sent to a regression dynamic feature vector calculator where the resultant is compared to a regression model containing a plurality of chemical concentration vectors. (The regression model was also built using a machine learning algorithm and training data sets.). Based upon the regression dynamic feature vector matching or nearly matching a chemical concentration vector within the regression model, a concentration of the specific chemical within the gas mixture can be determined and displayed.

[00022] As shown in Figure 1, the chemical monitoring system 10 includes one or more sensors 12 electrically coupled to a chemical monitoring unit 14 for sensing ambient air 16.

[00023] These sensors 12 can be nanosensors that are capable of detecting chemicals and volatile organic compounds within the ambient air 16 using carbon nanotubes but other sensors are contemplated.

[00024] Nanosensor technology uses nanostructures, e.g., single walled carbon nanotubes (SWNTs), combined with a silicon-based microfabrication and micromachining process. Nanotechnology based chemical sensors can provide high sensitivity, low power and low cost portable tools for in- situ chemical analysis. Each sensor in the array can consist of a nanostructure— chosen from many different categories of sensing material— and an interdigitated electrode (IDE) as a transducer. These chemical sensors can be one type of electrochemical sensor that implies the transfer of charge from one electrode to another. This means that at least two electrodes constitute an electrochemical cell to form a closed electrical circuit. Due to the interaction between nanotube devices and gas molecules, the electron configuration is changed in the nanostructured sensing device, therefore, the changes in the electronic signal such as current or voltage can be observed before and during an exposure to a gas species. By measuring the conductivity change of the nanosensor, the concentration of the chemical species, such as gas molecules, can be measured.

[00025] The chemical monitoring unit 10 can be a computing device for receiving these measurements and processing these measurements in integrated algorithms that monitor for specific chemical events and are capable of identification of chemical species and the prediction of concentration for the identified chemical species.

[00026] Specifically, the chemical monitoring unit 10 can include classification models built using machine learning algorithms, e.g., pattern recognition algorithms that aim to provide a reasonable answer for all possible inputs and to perform "most likely" matching of the inputs, taking into account their statistical variation.

[00027] The first classification model is the event detection model. The event detection models can be used as a classification algorithm that discriminates between a benign state vs a chemical detected state. That is, known benign state samples and known chemical state samples are sent to a machine learning algorithm in order to define a classification model. In some implementations, the machine learning algorithm can use ensemble learning techniques, but any type of machine learning technique can be used.

[00028] Ensemble learning is a machine learning technique where multiple algorithms are trained to solve the same problem. In contrast to ordinary machine learning approaches that try to learn one hypothesis from training data. Ensemble methods try to construct a set of hypotheses and combine them into one useable model. That is, an ensemble is a supervised learning algorithm that can be trained and then used to make predictions. The trained ensemble, therefore, represents a single hypothesis or model.

[00029] The disclosed technology will use the chemicals detected state as samples representative of data that the trained system should classify as positive, e.g. belonging to the object class of chemicals found in sample, and the benign state samples as samples representative of data that the systems should classify as negative, e.g. not belonging to the object class of chemicals not found in sample. By providing this set of known positive samples and this set of known negative samples, the computerized system can "learn" features of the positive samples that are not present in the negative samples and vice versa. The system can then look for these features in an unknown sample, and when they are present or absent, declare that the sample is a positive or negative sample as the case may be. The process of providing test samples to the system and allowing or "teaching" the system to learn prominent features is known as training the classifier. In the above implementation, two sample groups are used but in some implementations multiple sample groups may be used.

[00030] Specifically, training an object classification or object detection system involves extracting features in either a supervised or unsupervised manner to learn the differences and similarities between the positive and negative samples. Once determined, these indicators can be applied to new samples to determine whether they should be classified as belonging to the object class or not belonging to the object class.

[00031] In one example, the event detection model can be built using support vector machine (SVM). SVM is a pattern classifier that selects a hyper-plane based on maximizing a separation margin between classes. Its solution can depend on a small subset of training examples, e.g., dynamic or static support vectors. And it can be easily extended to deal with datasets with nonlinear separation through the kernel mapping scheme. Confidence in the form of decision value or probability can be calculated together with the winner class for a predicted chemical compound. Typically, normalization followed by autoscaling

pretreatments is performed on the support vectors before they are sent for SVM modeling.

[00032] The second classification model is the regression model. Once the "chemicals detected" state has been confirmed, the algorithm will trigger the regression calculation step where a regression model based on detected chemical will be used to calculate that chemical's concentration. The regression model can be built using multivariate Partial Least Square (PLS). For example, a regression model can be built by first receiving a set of known samples from a sensor. The sample data can be filtered for noise and autoscaled or mean centered. A dynamic training model is then created using machine learning vectors for the sample data and a non-agent can be added to the training model for robustness.

[00033] The chemical monitoring unit 10 can also include an event detection vector calculator and a regression vector calculator. The event detection vector can use a conventional feature vector scheme (see equation (1), below) or a differencing feature vector scheme (see equation (2) or (3) below) where the differencing feature vector (diffFV) compensates for a signal having significant baseline drift.

[00034] Conventional feature vectors can be calculated using the formula:

FVt = (Rt-R0)/R0 (1)

where Rt is a resistance reading at the current time, t, and R0 is the resistance reading at a fixed time step, t-w called a moving window width, before the current time.

[00035] The differencing feature vectors for event detection are calculated using the formula:

DiffFVt = FVt - FV0 (2)

where FVt is the current feature vector and FV0 is the feature vector at a fixed time step, w, before the current time translating into a formula of:

DiffFVt = (Rt-R0)/R0 - (Rtl - R01)/R01 -— (3) where Rt is a resistance reading at the current time, t, R0 is the resistance reading at a fixed time step, w, before the current time, Rtl is the resistance reading at a fixed time step, w, before the current time and R01 is the resistance reading at a fixed time step 2w before current time. [00036] In the case of regression calculation to find a chemical's concentration, since a chemical's concentration is proportional to the absolute magnitude of sensor's response (Beer's law), the R0 reading (reference resistance) should be latched at the sensor's response before event detection just happened. The dynamic feature vector for regression is calculated by the same formula as for conventional event detection (or equation (1) ). The only difference is the calculation of R0 which will be the latched reading at the time before event detection was just triggered (or one time step before sensors' response fall inside the target chemical's decision boundary) translating into same formula like equation (1) above:

FVt = (Rt-R0)/R0

where Rt is a resistance reading at the current time, t, R0 is the resistance reading at the latched reading before the start of the event.

[00037] The disclosed technology found that an approach with a differencing feature vectors has the advantage of enhancing pattern discrimination between real patterns collected with a sloping baseline vs the pattern of that background.

[00038] In use, the algorithm of the chemical detection unit 10 integrates signal preprocessing and postprocessing with a dynamic training model built from median virtual sensors for continuous monitoring of a chemical or a few chemicals present in an ambient environment.

[00039] Figures 2a-2b is a flowchart of an embodiment of the algorithm of the present invention. In use, a detection software unit can be uploaded to a computing device, e.g., a cell phone, laptop or any other portable device. (Stepl). The detection unit can include an event detection model and a regression model. Since model files are loaded as structure data type, they can be unwrapped into each parameter's variable data type and stored in the computing device. These model parameters can be used in both the classification and regression calculations' models parameters settings. [00040] A nanosensor can be coupled to the computing device. (Step 2). The sensor can read raw responses based on electric resistance readings from the sensor array. These responses can be received by the detection software unit. (Step 3). The responses can be, e.g., multivariable time series data. The responses can be applied to a filter (Step 4), e.g., Savitzky-Golay filtering, to remove noisy data and a median reading can be taken from a subset of redundant sensors to remove the influence of any outlier sensor(s).

[00041] These treated responses can be sent to a calculator for calculating a dynamic feature vector for event detection, as described above. (Step 5). The calculated feature vector is applied to the event detection model to predict class assignment and confidence. (Step 6). If a match is found (Step 7), to ensure detection, alarm logic can be applied. That is, usually when we normalize sensors' response to ambient air, because it has the effect of amplifying responses, it will generate random noisy dynamic feature vectors. In order to avoid a random noisy dynamic feature vector from triggering event detect alarm(s), alarm logic can be used that sets a certain alarming threshold in the form of a consecutive triggering alarm (this means the current response falls inside the decision boundary of a certain chemical class) for say 5 time steps before the alarm is triggered and the chemical class assignment is predicted. (Step 8).

[00042] If an alarm threshold is reached, trigger an alarm and display the detected chemical. (Step 9). Simultaneously, resistance readings can be retrieved for calculating a dynamic feature vector for regression, as described above. (Step 10). The calculated regression vector is applied to the regression model to calculate concentration reading. (Step 11). The concentration reading or indicator can then be displayed. (Step 12).

[00043] Figures 3a-3b is a flowchart of an embodiment of the algorithm of the present invention. In this implementation, a device can receive data signals from at least one sensor on a timed basis. (Step Al). A dynamic feature vector is calculated using the data signals. (Step A2). The vector is compared to an event detection model containing a plurality of chemical detection vectors to determine if the dynamic feature vector matches a chemical detection vector within the event detection model thereby identifying a specific chemical within the gas mixture. (Step A3). If a match is found (Step A4), these steps are repeated until a threshold number of matches occur in consecutive order. (Step A5). When the threshold number is reached, a regression dynamic feature vector is calculated and compared to a regression model containing a plurality of chemical concentration vectors. (Step A6). A concentration of the specific chemical within the gas mixture is then determined based upon the regression dynamic feature vector matching a chemical concentration vector within the regression model.

[00044] In another implementation, as shown in Figure 4, the chemical monitoring system 20 includes one or more sensors 22 electrically coupled to a chemical monitoring unit 24 for sensing exhaled air 26.

[00045] The chemical monitoring system 20 can be used to sample a single event, e.g., a single breath from a person. In this implementation, the unit 24 can have a button that instructs the device to begin to analyze an incoming samples for a number of cycles, e.g., once a second for ten seconds. These samples can be applied to static vector calculations that are similar to the dynamic vector calculations, above.

[00046] The differencing feature vectors for event detection are calculated using the formula:

DiffFVt = FVt - FV0

where FVt is the end feature vector and FV0 is the start feature vector translating into a formula of:.

DiffFVt = (Rt-R0)/R0 - (Rtl - R01)/R01 where Rt is a resistance reading at an end time, t, RO is the resistance reading at a fixed time step, w, before the current time, R01 is the resistance reading at 2*w time steps before current time, and Rtl is the resistance reading at a fixed time step w time steps before current time.

[00047] The feature vectors for regression are calculated using the formula:

FVt = (Rt-R0)/R0

where Rt is a resistance reading at current time, t, R0 is the resistance reading at a fixed time step, w, before the current time.

[00048] Figure 5 is a flowchart of an embodiment of the algorithm of the present invention. A device for monitoring for a chemical event in a sample of exhaled air using a sensor and a mobile computing device. A chemical detection unit is loaded onto the mobile unit device. The device receives data signals from at least one sensor. A dynamic feature vector is calculated using the data signals (Step Bl). The vector is compared to an event detection model containing a plurality of chemical detection vectors. (Step B2). If the dynamic feature vector matches a chemical detection vector within the event detection model thereby identifying a specific chemical within the gas mixture. (Step B3). If a match is found (step B4), a regression dynamic feature vector is calculated (Step B5) and compared to a regression model containing a plurality of chemical concentration vectors. (Step B6). A concentration of the specific chemical within the gas mixture is determined based upon the regression dynamic feature vector matching a chemical concentration vector within the regression model. (Step B7).

[00049] FIG. 6 is a schematic diagram of an example of a chemical detection system

100. The chemical detection system 100 can include a sensor array 110, chemical detection unit 130, one or more processors 121, application programming interface (API) 125, one or more display devices 127, e.g., CRT, LCD, one or more interfaces 123, input devices 126, e.g., touchscreen, keyboard, mouse, scanner, activation button, etc., and one or more computer-readable mediums 124. These components exchange communications and data using one or more buses, e.g., EISA, PCI, PCI Express, etc. The term "computer-readable medium" refers to any non-transitory medium that participates in providing instructions to processor 121 for execution. The computer-readable mediums further include operating system 122.

[00050] The operating system 122 can be multi-user, multiprocessing, multitasking, multithreading, real-time, near real-time and the like. The operating system 122 can perform basic tasks, including but not limited to: recognizing input from input device 126; sending output to display devices 127; keeping track of files and directories on computer-readable mediums 124, e.g., memory or a storage device; controlling peripheral devices, e.g., disk drives, printers, etc.; and managing traffic on the one or more buses. The operating system 122 can also run algorithms (e.g. detection vector calculator 131 and regression vector calculator 134) associated with the system 100, accessing the detection model 132 and regression model 135, and running the detection comparator 133 and regression comparator 136.

[00051] Implementations of the subject matter and the operations described in this specification can be done in electronic circuitry, or in computer software, firmware, or hardware, including the structures disclosed in this specification and their structural equivalents, or in combinations of one or more of them. Implementations of the subject matter described in this specification can be done as one or more computer programs, e.g., one or more modules of computer program instructions, encoded on a computer storage media for execution by, or to control the operation of, data processing apparatus.

Alternatively or in addition, the program instructions can be encoded on an artificially- generated propagated signal, e.g., a machine-generated electrical, optical, or electromagnetic signal that is generated to encode information for transmission to suitable receiver apparatus for execution by a data processing apparatus. The computer storage medium can be, or can be included in, a computer-readable storage device, a computer-readable storage substrate, a random or serial access memory array or device, or a combination of one or more of them.

[00052] The operations described in this specification can be implemented as operations performed by a data processing apparatus on data stored on one or more computer- readable storage devices or received from other sources. The term "data processing apparatus" encompasses all kinds of apparatus, devices, and machines for processing data, including by way of example a programmable processor, a computer, a system on a chip, or combinations of them. The apparatus can include special purpose logic circuitry, e.g., an FPGA (field programmable gate array) or an ASIC (application-specific integrated circuit). The apparatus can also include, in addition to hardware, code that creates an execution environment for the computer program in question, e.g., code that constitutes processor firmware, a protocol stack, a repository management system, an operating system, a cross- platform runtime environment, e.g., a virtual machine, or a combination of one or more of them. The apparatus and execution environment can realize various different computing model infrastructures, e.g., web services, distributed computing and grid computing infrastructures.

[00053] A computer program (also known as a program, software, software

application, script, or code) can be written in any form of programming language, including compiled or interpreted languages, declarative or procedural languages, and it can be deployed in any form, including as a stand-alone program or as a module, component, subroutine, object, or other unit suitable for use in a computing environment. A computer program can, but need not, correspond to a file in a file system. A program can be stored in a portion of a file that holds other programs or data, e.g., one or more scripts stored in a markup language document, in a single file dedicated to the program in question, or in multiple coordinated files, e.g., files that store one or more modules, sub-programs, or portions of code. A computer program can be deployed to be executed on one computer or on multiple computers that are located at one site or distributed across multiple sites and interconnected by a communication network.

[00054] The processes and logic flows described in this specification can be performed by one or more programmable processors executing one or more computer programs to perform functions by operating on input data and generating output. The processes and logic flows can also be performed by, and apparatus can also be implemented as, special purpose logic circuitry, e.g., an FPGA (field programmable gate array) or an ASIC (application- specific integrated circuit).

[00055] Processors suitable for the execution of a computer program include, by way of example, both general and special purpose microprocessors, and any one or more processors of any kind of digital computer. Generally, a processor can receive instructions and data from a read-only memory or a random access memory or both. The elements of a computer comprise a processor for performing or executing instructions and one or more memory devices for storing instructions and data. Generally, a computer can also include, or be operatively coupled to receive data from or transfer data to, or both, one or more mass storage devices for storing data, e.g., magnetic, magneto-optical disks, or optical disks.

However, a computer need not have such devices. Moreover, a computer can be embedded in another device, e.g., a mobile telephone, a personal digital assistant (PDA), a mobile audio or video player, a game console, a Global Positioning System (GPS) receiver, or a portable storage device, e.g., a universal serial bus (USB) flash drive, to name just a few. Devices suitable for storing computer program instructions and data include all forms of non- volatile memory, media and memory devices, including by way of example semiconductor memory devices, e.g., EPROM, EEPROM, and flash memory devices; magnetic disks, e.g., internal hard disks or removable disks; magneto-optical disks; and CD-ROM and DVD-ROM disks. The processor and the memory can be supplemented by, or incorporated in, special purpose logic circuitry.

[00056] To provide for interaction with a user, implementations of the subject matter described in this specification can be implemented on a computer having a display device, e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor, for displaying information to the user and a keyboard and a pointing device, e.g., a mouse or a trackball, by which the user can provide input to the computer. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback, e.g., visual feedback, auditory feedback, or tactile feedback; and input from the user can be received in any form, including acoustic, speech, thought or tactile input. In addition, a computer can interact with a user by sending documents to and receiving documents from a device that is used by the user.

[00057] While this specification contains many specific implementation details, these should not be construed as limitations on the scope of the disclosed technology or of what can be claimed, but rather as descriptions of features specific to particular implementations of the disclosed technology. Certain features that are described in this specification in the context of separate implementations can also be implemented in combination in a single

implementation. Conversely, various features that are described in the context of a single implementation can also be implemented in multiple implementations separately or in any suitable subcombination. Moreover, although features can be described above as acting in certain combinations and even initially claimed as such, one or more features from a claimed combination can in some cases be excised from the combination, and the claimed

combination can be directed to a subcombination or variation of a subcombination. [00058] Similarly, while operations are depicted in the drawings in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order, or that all illustrated operations be performed, to achieve desirable results. In certain circumstances, multitasking and parallel processing can be advantageous. In some cases, the actions recited in the claims can be performed in a different order and still achieve desirable results. Moreover, the separation of various system components in the implementations described above should not be understood as requiring such separation in all implementations, and it should be understood that the described program components and systems can generally be integrated together in a single software product or packaged into multiple software products.

[00059] The foregoing Detailed Description is to be understood as being in every respect illustrative, but not restrictive, and the scope of the disclosed technology disclosed herein is not to be determined from the Detailed Description, but rather from the claims as interpreted according to the full breadth permitted by the patent laws. It is to be understood that the implementations shown and described herein are only illustrative of the principles of the disclosed technology and that various modifications can be implemented without departing from the scope and spirit of the disclosed technology.

Claims

IN THE CLAIMS

1. A computer-implemented method for monitoring for a chemical event in a gas mixture, the method comprising: receiving data signals from at least one sensor on a timed basis; calculating a dynamic feature vector for event detection using the data signals; comparing the dynamic feature vector for event detection to an event detection model containing a plurality of chemical detection vectors; determining if the dynamic feature vector for event detection matches a chemical detection vector within the event detection model thereby identifying a specific chemical within the gas mixture; repeating steps 1 to 4 until a threshold number of matches occur in consecutive order; when the threshold number is reached, calculating a regression dynamic feature vector; comparing the regression dynamic feature vector to a regression model containing a plurality of chemical concentration vectors; and determining a concentration of the specific chemical within the gas mixture based upon the regression dynamic feature vector matching a chemical concentration vector within the regression model.

2. The computer-implemented method of Claim 1 further comprising the step of: triggering an alarm identifying the presence of the specific chemical when the threshold number is reached.

3. The computer-implemented method of Claim 1 further comprising the step of: displaying the concentration of the specific chemical.

4. The computer-implemented method of Claim 1 further comprising the step of: filtering the data signals for noise and median readings.

5. The computer-implemented method of Claim 1 further comprising the step of: normalizing and auto scaling the dynamic feature vector for event detection.

6. The computer-implemented method of Claim 1 wherein the data signals are electrical resistance readings of a nanochemical sensor.

7. The computer-implemented method of Claim 1 wherein the gas mixture is ambient air.

8. The computer-implemented method of Claim 1 wherein the timed basis is in a range of once every 1 to 60 seconds on continuous basis.

9. The computer-implemented method of Claim 1 wherein the dynamic feature vector for event detection is calculated with a formula of: DiffFVt = (Rt-R0)/R0 - (Rtl - R01)/R01 where Rt is a resistance reading at the current time, t, RO is the resistance reading at a fixed time step, t-m, before the current time, Rtl is the resistance reading at a fixed time step, w, before the current time and R01 is the resistance reading at a fixed time step before w, w-n.

10. The computer-implemented method of Claim 1 wherein the regression dynamic feature vector is calculated with a formula of: FVt = (Rt-R0)/R0 where Rt is a resistance reading at the current time, t, RO is the resistance reading at the latched reading just before the chemical event happened.

11. A system for monitoring for a chemical event in a gas mixture, the system

comprising:

(a) at least chemical detection unit having a chemical detection software application installed thereon; and (b) a sensor, which sensor is coupled to the at least one mobile device, the chemical detection software application programmed to:

(i) receive data signals from at least one sensor on a timed basis;

calculate a dynamic feature vector using the data signals

(iii) compare the dynamic feature vector to an event detection model containing a plurality of chemical detection vectors;

(iv) determine if the dynamic feature vector matches a chemical detection vector within the event detection model thereby identifying a specific chemical within the gas mixture;

(v) repeat steps (i) to (iv) until a threshold number of matches occur in consecutive order;

(vi) when the threshold number is reached, calculate a regression dynamic feature vector;

(vii) compare the regression dynamic feature vector to a regression model containing a plurality of chemical concentration vectors; and

(viii) determine a concentration of the specific chemical within the gas mixture based upon the regression dynamic feature vector matching a chemical concentration vector within the regression model.

12. The system of Claim 11 wherein the chemical detection software application is programmed to: trigger an alarm identifying the presence of the specific chemical when the threshold number is reached.

13. The system of Claim 11 wherein the chemical detection software application is programmed to: display the concentration of the specific chemical.

14. The system of Claim 11 wherein the chemical detection software application is programmed to: filter the data signals for noise and median readings.

15. The system of Claim 11 wherein the chemical detection software application is programmed to: normalize and auto scale the dynamic feature vector for event detection.

16. The system of Claim 11 wherein the data signals are electrical resistance readings of a nanochemical sensor.

17. The system of Claim 11 wherein the gas mixture is ambient air.

18. The system of Claim 11 wherein the timed basis is in a range of once every 1 to 60 seconds on continuous basis.

19. The system of Claim 11 wherein the dynamic feature vector for event detection is calculated with a formula of: DiffFVt = (Rt-R0)/R0 - (Rtl - R01)/R01 where Rt is a resistance reading at the current time, t, R0 is the resistance reading at a fixed time step, t-m, before the current time, Rtl is the resistance reading at a fixed time step, w, before the current time and R01 is the resistance reading at a fixed time step before w, w-n.

20. The system of Claim 11 wherein the regression dynamic feature vector is calculated with a formula of: FVt = (Rt-R0)/R0 where Rt is a resistance reading at the current time, t, R0 is the resistance reading at the latched reading just before the chemical event happened.