Portable fluid sensory device with learning capabilities

Abstract

The invention discloses a device for identifying fluids or measuring their concentration. The device is configured to capture fluid sensing signals and sent to processing capabilities to be annotated, pre-processed and fed to databases of datasets and models which have learning capabilities. The device has a stick or stylus form factor which is makes it fit to be used by health care professionals or by the general public. Advantageously, the stick can be used to capture data from gas and liquid, being possibly phases of the same analyte. The device can be a package containing all processing capabilities being configured to be autonomous. It can operate in conjunction with an intermediary device of a smart phone, a PC or a POCT type. The system comprising autonomous fluid sensory devices, intermediary devices and database servers can operate in a learning mode or in a use mode. Measurements can be filtered, and normalized to statistically eliminate the differences in measurements due to bad operational conditions, differences of device configurations or differences of local parameters (temperature, hygrometry, flow rate, etc. . . . )

1 . A method to produce a model for identifying at least one of a nature of a first fluid and a concentration of the first fluid in a second fluid, the method comprising:
 receiving, at a first processing capability, a first dataset representative of the at least one of a nature of the first fluid and a concentration of the first fluid in the second fluid, the first dataset comprising a label characterizing a fluid sensing signal;   combining, at the first processing capability, the first dataset with a second dataset comprising at least one of a text, a sound and an image, representative of the at least one of a nature of the first fluid and a concentration of the first fluid in the second fluid;   transmitting from the first processing capability to a second processing capability the first dataset and the second dataset;   classifying, at the second processing capability, a plurality of pairs of first datasets and second datasets from a plurality of first processing capabilities, said classifying based on classifying of the plurality of second datasets into a plurality of classes;   selecting for each class of the plurality of classes at the output of the classifying a statistical model correlating the plurality of pairs of first datasets and second datasets, the statistical model fit for the nature of the first fluid; and   calculating parameters of the statistical model for the each class from the plurality of pairs of first datasets and second datasets.   
     
     
         2 . The method of  claim 1 , wherein one of the first dataset and the second dataset further comprises at least a value of one of a temperature, a flow rate and an hygrometry measurement at a location of the fluid sensory device. 
     
     
         3 . The method of  claim 1 , wherein the second dataset further comprises data to localize the first processing capability. 
     
     
         4 . The method of  claim 1 , wherein the receiving is performed by a fluid sensory device having an elongated form factor, with a distal end comprising a contact surface of a liquid sensor and a proximal end comprising an entry port of a gas sensor. 
     
     
         5 . A computer program product comprising computing instructions for producing a model for identifying at least one of a nature of a first fluid and a concentration of said first fluid in a second fluid, wherein the computing instructions are stored on at least a non-transitory computer storage medium and the computing instructions when executed by at least one processor, configure the at least one processor to:
 receive, at a first processing capability, a first dataset representative of the at least one of a nature of the first fluid and a concentration of the first fluid in the second fluid, the first dataset comprising a label characterizing a fluid sensing signal;   combine, at the first processing capability, the first dataset with a second dataset comprising at least one of a text, a sound and an image, representative of the at least one of a nature of the first fluid and a concentration of the first fluid in the second fluid;   transmit from the first processing capability to a second processing capability the first dataset and the second dataset;   classify, at the second processing capability, a plurality of pairs of first datasets and second datasets from a plurality of first processing capabilities, said classifying based on classifying of the plurality of second datasets into a plurality of classes;   select for each class of the plurality of classes at the output of the classifying a statistical model correlating the plurality of pairs of first datasets and second datasets, the statistical model fit for the nature of the first fluid; and   calculate parameters of the statistical model for the each class from the plurality of pairs of first datasets and second datasets.   
     
     
         6 . The computer program product of  claim 5 , wherein one of the first dataset and the second dataset further comprises at least a value of one of a temperature, a flow rate and an hygrometry measurement at a location of the fluid sensory device. 
     
     
         7 . The computer program product of  claim 6 , wherein the second dataset further comprises data to localize the first processing capability. 
     
     
         8 . The method of  claim 1  wherein the receiving, combining, transmitting classifying, selecting and calculating are repeated for the plurality of pairs of first datasets and second datasets to produce a plurality of statistical models. 
     
     
         9 . A method for identifying at least one of a nature of a first fluid and a concentration of the first fluid in a second fluid, said method comprising:
 receiving, from a fluid sensory device, at a first processing capability, said first processing capability located in the vicinity of the fluid sensory device, a first dataset representative of the at least one of a nature of the first fluid and a concentration of the first fluid in the second fluid;   obtaining at the first processing capability, a plurality of statistical models from a second processing capability;   selecting a statistical model from the plurality of statistical models;   inputting the first dataset in the statistical model at the output of the selection; and   wherein if the at least one of a nature of a first fluid and a concentration of the first fluid in the second fluid is identified, outputting a result at the first processing capability, and if the at least one of a nature of a first fluid and a concentration of the first fluid in the second fluid is not identified, redoing at least one of the obtaining and the selecting of the statistical model until either the result is obtained or a stop criteria is met.