Apparatus and method for measuring plurality of blood parameters by artificial intelligence

Abstract

The invention relates to a device (1, 1 ') and a method for simultaneously measuring a plurality of blood parameters by means of artificial intelligence. The invention provides for: exciting the blood with a plurality of determined wavelengths of electromagnetic radiation; receiving analog information relating to a plurality of electromagnetic and / or light responses of the blood, the plurality of electromagnetic and / or light responses comprising electromagnetic radiation retroreflected or diffused by the blood; converting the analog electromagnetic and / or light response information into digital electromagnetic and / or light response data; processing the digital electromagnetic and / or photoresponse data and actual temperature values of the blood by means of one or more neural networks (NNs); and determining a value for each parameter as a result of the processing operation. The invention provides for: determining a plurality of ratios of optical counts; providing, as input to one or more neural networks (NNs), a plurality of ratios and temperature values; processing the plurality of ratios by one or more neural networks (NNs) taking into account a plurality of data from previous measurements of the blood parameter taken during the previous training; a value for each parameter is provided as an output from one or more neural networks (NNs). The device (1, 1 ') is implemented in an embodiment in which it can be coupled to a test tube (25) or in an embodiment in which it can be directly coupled to a tube (25') of an extracorporeal blood circuit.

Description

Technical Field

[0001] The present invention relates to an apparatus and method for measuring multiple blood parameters by artificial intelligence.

[0002] The device and the method according to the invention enable the measurement of parameters in an extracorporeal blood circuit.

[0003] The invention also relates to the use of the device and to an apparatus comprising the device.

[0004] The present invention can be applied to monitoring blood parameters during treatments requiring extracorporeal blood circulation, such as hemodialysis, plasma exchange, extracorporeal respiratory assistance (ECMO), preservation of transplanted organs, and cancer treatment. Background Art

[0005] Probes are known which measure a plurality of blood parameters in an extracorporeal blood circuit.

[0006] This type of probe is known from the international patent application WO 2010136962 A1 filed in the name of the applicant in 2010. Such a probe comprises a housing for accommodating a conduit for blood flowing in an extracorporeal circuit, an emitting device and a device for detecting electromagnetic radiation, and a control unit to which the emitting device and the detection device are connected. The control unit calculates the value of the blood parameter by means of a correlation between a reference value and a ratio obtained from the light intensity value of the detected radiation; this correlation makes use of a complex mathematical formula.

[0007] The probe of WO 2010136962A1 has some disadvantages; in particular, the complex mathematical formulas used do not guarantee optimal accuracy of the measured parameters. A further disadvantage is the inaccuracy of the measurements when the blood catheter joined to the probe is changed with a blood catheter of a different formulation; such a different catheter may have a different hardness than the previous catheter (a more or less rigid catheter) and a different color (due to chemical additives used in its manufacture), but the probe does not have the necessary accuracy to take these differences into account. Therefore, the applicant points out that the measurement accuracy of the probe of WO 2010136962A1 and the speed of the measurement of the parameters can be improved and this improvement is made according to the following.

[0008] Purpose of the invention

[0009] Therefore, it is a main object of the present invention to provide a device for measuring a plurality of blood parameters which is able to overcome the above-mentioned disadvantages associated with the prior art.

[0010] It is an object of the present invention to provide a device for measuring a plurality of blood parameters which does not require calibration by the end user before using it.

[0011] A further object of the present invention is to provide a method for measuring a plurality of blood parameters such that the end user does not need to perform any calibration prior to the measurement.

[0012] An additional object of the present invention is to propose an apparatus and a method for measuring multiple blood parameters quickly, efficiently and reliably.

[0013] It is also an object of the present invention to provide a device capable of measuring a plurality of blood parameters autonomously and with high accuracy over the entire measuring range of each parameter.

[0014] These and other objects are achieved by an apparatus, a use of an apparatus, a device and one or more methods according to the following description, the appended claims and the following aspects. Summary of the invention

[0015] Various aspects of the invention are described herein. When an aspect and / or claim is reviewed by a specific correlation with one or more other aspects or claims and / or by a wording such as "the" or "said" to introduce one or more elements or steps or operations by another aspect or claim, such aspect and / or such claim may be combined with it.

[0016] The invention provides a device (probe) for measuring a plurality of blood parameters, the device being provided with a control unit enabling the device to measure the parameters based on a plurality of data of previous measurements of said parameters made during previous training sessions.

[0017] The control unit implements the intelligence of the device and is responsible for autonomously performing the measurements. To achieve this, the control unit incorporates a coded computational model (by means of a program code) derived from a plurality of data from previous measurements of blood parameters performed during previous training. In essence, the device learns from previous training and is therefore, immediately ready for use upon clinical use, without requiring prior calibration by the end user.

[0018] After a series of devices have been learned, a method is developed (e.g. a model and / or one or more algorithms encoded in the firmware) which enables subsequent devices produced to be used with a simple calibration performed, for example, in a laboratory by the manufacturer of the device. All the end user has to do is to use the device after connecting it to a tubular element (test tube) or to a tube of an extracorporeal blood circuit, without having to perform any calibration.

[0019] Each device preferably stores two types of information: a first type of information common to the devices produced after said learning and related to the "mechanism of calculation" of the values ​​of the parameters by the neural network (this type of information is preferably encoded in the firmware of the device); a second type of information, which may vary for each device, for example due to the variability of the hardware, the components and their specific geometrical position within the box of the device (this type of information is preferably stored in the device memory).

[0020] The numbered aspects of the invention are as follows.

[0021] Device for measuring multiple blood parameters

[0022] 1. A device for measuring a plurality of blood parameters by means of artificial intelligence, the device comprising:

[0023] at least one excitation member configured to excite the blood, in particular the blood flow, with electromagnetic radiation of a plurality of determined wavelengths,

[0024] - at least one electromagnetic radiation detection member, in particular at least one photodetector, configured to detect a plurality of electromagnetic responses, in particular optical responses, of the blood, said plurality of electromagnetic responses comprising electromagnetic radiation, in particular light, reflected or diffused by the blood, said electromagnetic radiation being reflected or diffused by the blood after excitation of the excitation member in an operating state of the device.

[0025] 2. A device for measuring a plurality of blood parameters by means of artificial intelligence, the device comprising:

[0026] - at least one excitation member configured to excite blood flow at a plurality of determined wavelengths,

[0027] - at least one photodetector configured to detect a plurality of electromagnetic and / or optical responses of the blood including light retroreflected or diffused by the blood in an operating state of the device after excitation by the excitation member.

[0028] 3. According to aspect 1 or 2, the apparatus further comprises a control unit configured to perform the following operations:

[0029] o commanding the at least one excitation member during an excitation step in which the excitation member excites the blood with electromagnetic radiation of a plurality of determined wavelengths,

[0030] o receiving analog information related to a plurality of electromagnetic and / or optical responses of the blood, the plurality of electromagnetic and / or optical responses comprising electromagnetic radiation, in particular light, reflected or diffused by the blood,

[0031] o converting electromagnetic and / or optical response analog information into electromagnetic and / or optical response digital data,

[0032] o processing the electromagnetic and / or optical response digital data and the actual temperature value of the blood by one or more neural networks,

[0033] o As a result of the processing operations of the one or more neural networks, a value for each of a plurality of blood parameters is determined.

[0034] 4. According to aspect 3, the control unit is configured to process the electromagnetic and / or optical response digital data and the actual temperature value of the blood by one or more neural networks and determine the value of each parameter of the plurality of blood parameters by means of the following operations:

[0035] o determining a plurality of ratios, each ratio being defined between a magnitude indicative of radiation retroreflected or diffused by the blood due to excitation at a determined wavelength and a magnitude indicative of radiation retroreflected or diffused by the blood due to excitation at another determined wavelength,

[0036] o providing the plurality of ratios and temperature values ​​as input to one or more neural networks,

[0037] o processing said plurality of ratios by means of one or more neural networks taking into account a plurality of data of previous measurements of said blood parameters performed during previous training,

[0038] o As output from one or more neural networks, values ​​for each of a plurality of blood parameters are provided.

[0039] 5. Aspect according to any of the preceding aspects, wherein the device comprises a temperature sensor configured to measure the actual temperature of the blood.

[0040] 6. An aspect according to any one of the preceding aspects, wherein the device is configured to simultaneously measure multiple blood parameters through artificial intelligence.

[0041] 7. An aspect according to any of the preceding aspects, wherein the control unit is configured to detect the value of each parameter of the plurality of blood parameters that the artificial intelligence considers to correspond to a previously determined ratio based on the plurality of data from previous measurements.

[0042] 8. An aspect according to any of the preceding aspects, wherein the device includes firmware and the control unit includes artificial intelligence information, such as one or more matrices available to the one or more neural networks, the artificial intelligence information being encoded in the firmware and enabling calculation of the multiple parameters by the one or more neural networks.

[0043] This information encoded in the firmware is "the same for all devices".

[0044] 9. An aspect according to any of the preceding aspects, wherein the device comprises a memory, the control unit being configured to provide a reference value of a first parameter to be measured and a reference value of a second parameter to be measured as input to one or more neural networks, the reference values ​​being preferably acquired and stored in the memory of the device during a calibration step prior to use of the device, the memory therefore comprising information relating to the reference values.

[0045] 10. An aspect according to any of the preceding aspects, wherein the device includes a memory comprising one or more information uniquely associated with a calibration step of the device, such as one or more matrices usable by the one or more neural networks.

[0046] 11. Aspect according to any of the preceding aspects, wherein the device calibration step is performed by the manufacturer of the device.

[0047] 12. Aspect according to any of the preceding aspects, wherein the calibration step is not performed by an end user of the device.

[0048] 13. An aspect according to any one of the preceding aspects, wherein no calibration step is performed during a state of use of the device or before a state of use of the device.

[0049] This information and / or reference parameters stored in the memory of the device are specific to each device manufactured and take into account the variability between devices (due to, for example, variability of hardware, components and their even infinitesimal geometrical positions within the housing of the device) and give a robust starting point for the measurement of oxygen saturation and hematocrit expected in the calibrated state.

[0050] 14. An aspect according to any of the preceding aspects, wherein the control unit comprises a microprocessor and the information is encoded in the firmware of the microprocessor.

[0051] 15. An aspect according to any one of the preceding aspects, wherein the device is configured to measure blood oxygen saturation, i.e. SatO2, hematocrit, i.e. Hct, and optionally also hemoglobin content, i.e. Hb, with the aid of artificial intelligence.

[0052] 16. Aspect according to any of the preceding aspects, wherein the control unit is configured to control the at least one excitation member or the two excitation members during an excitation step involving excitation of the blood flow one wavelength at a time.

[0053] 17. Aspect according to any one of the preceding aspects, wherein the device comprises a plurality of excitation members, and the control unit is configured to control the plurality of excitation members during the excitation step to activate the plurality of excitation members according to a determined time sequence.

[0054] 18. An aspect according to any one of the preceding aspects, wherein the at least one excitation component is configured to excite blood flow at least at the following wavelengths: 660nm, 805nm, 1450nm and at least one between 525nm, 940nm and 1050nm.

[0055] 19. An aspect according to any one of the preceding aspects, wherein the at least one excitation component is configured to excite blood flow at least at the following wavelengths: 660nm, 805nm, 1450nm and at least two between 525nm, 940nm and 1050nm.

[0056] 20. Aspect according to any one of the preceding aspects, wherein the at least one excitation component is configured to excite blood flow at least at the following wavelengths: 660nm, 805nm, 1450nm, 525nm, 940nm and 1050nm.

[0057] 21. An aspect according to any one of the preceding aspects, wherein the device comprises: a first excitation component, the first excitation component is configured to excite blood flow at least at a first wavelength, in particular at a first plurality of wavelengths; and a second excitation component, the second excitation component is configured to excite blood flow at least at a second wavelength, in particular at a second plurality of wavelengths.

[0058] 22. Aspect according to aspect 21, wherein the control unit is configured to activate the first excitation member and the second excitation member, preferably alternately, so as to excite the blood flow at the first wavelength or first multiple wavelengths and at the second wavelength or second multiple wavelengths, preferably alternately.

[0059] 23. Aspect according to any of the preceding aspects, wherein the device comprises a first excitation member and a second excitation member.

[0060] 24. Aspect according to aspect 23, wherein the first excitation means and the second excitation means are configured to excite blood at a respective plurality of different wavelengths.

[0061] 25. Aspect according to aspect 23 or 24, wherein the first excitation means is configured to excite blood flow at the following wavelengths: 525 nm, 940 nm, optionally 1050 nm.

[0062] 26. Aspect according to aspect 23 or 24 or 25, wherein the second excitation component is configured to excite blood flow at the following wavelengths: 660nm, 805nm, 1450nm.

[0063] 27. An aspect according to any of the preceding aspects, wherein each excitation component comprises a plurality of LED elements configured to emit light radiation of said wavelength, in particular a plurality of LED elements equal in number to the wavelength at which each LED is configured to operate.

[0064] 28. Aspect according to any one of aspects 23 to 27, wherein the first actuation member and the second actuation member are arranged in the case.

[0065] 29. Aspect according to any one of aspects 23 to 28, wherein the first actuation member and the second actuation member are arranged side by side.

[0066] 30. An aspect according to any one of the preceding aspects, wherein the device comprises a first electromagnetic radiation detection component and a second electromagnetic radiation detection component, each electromagnetic radiation detection component being configured to detect multiple electromagnetic and / or optical responses of blood, in particular to detect multiple electromagnetic and / or optical responses of blood at different wavelengths.

[0067] 31. An aspect according to any one of the preceding aspects, wherein the device comprises a first photodetector and a second photodetector, each photodetector being configured to detect multiple electromagnetic and / or optical responses of the blood, in particular to detect multiple electromagnetic and / or optical responses of the blood at different wavelengths, respectively.

[0068] 32. Aspect according to aspect 31, wherein the first photodetector and the second photodetector are arranged in a box body.

[0069] 33. Aspect according to aspect 31 or 32, wherein the first photodetector and the second photodetector are arranged side by side.

[0070] 34. Aspect according to aspect 31 or 32 or 33, wherein the first photodetector and the second photodetector are configured to detect blood response at a corresponding plurality of different wavelengths.

[0071] 35. An aspect according to any one of aspects 31 to 34, wherein the first photodetector is configured to detect the light response of blood when excited at the following wavelengths: 660 nm, 805 nm, 525 nm, 1050 nm, 940 nm.

[0072] 36. Aspect according to any one of aspects 31 to 35, wherein the second photodetector is configured to detect the optical response of blood when excited at the following wavelengths: 805 nm and 1450 nm.

[0073] 37. An aspect according to any one of the preceding aspects, wherein the device comprises a housing in which the at least one or each excitation member, the at least one or each electromagnetic radiation detection member and the control unit are housed.

[0074] 37 bis. Aspect according to any of the preceding aspects, wherein the device, in particular the box body, includes a coupling part, which is configured to allow coupling of the device to a container.

[0075] 37 ter. Aspect according to aspects 37 and 37 bis, wherein the coupling portion is connected to or engaged with the box body and / or is integral with the box body.

[0076] 37-4. Aspects according to 37-2 or 37-3, wherein the coupling portion is configured to allow the box body to be coupled to the container.

[0077] 37 out of 5. An aspect according to any one of the preceding aspects, wherein the device, in particular the box body, is essentially pocket-sized.

[0078] 37 of 6. An aspect according to any one of the preceding aspects, wherein the device, in particular the box body, has one or more characteristic dimensions (e.g., width, height and length) less than 100 mm, preferably each characteristic dimension is less than 100 mm.

[0079] 37 of 6. An aspect according to any of the preceding aspects, wherein the device does not include a screen.

[0080] Stream Information

[0081] 38. Aspect according to any of the preceding aspects, wherein the control unit is configured to take into account blood flow related values, in particular the volume flow rate of the blood, when measuring the oxygen saturation and / or the hematocrit.

[0082] 39. An aspect according to any one of the preceding aspects, wherein the control unit is configured to take into account the operation of blood flow related values ​​in the measurement of oxygen saturation and / or hematocrit by correcting the measured values ​​of oxygen saturation and / or hematocrit of the blood flow.

[0083] 40. Aspect according to any of the preceding aspects, wherein the control unit is configured to make the value of each parameter of the plurality of blood parameters available, preferably at the same moment in time.

[0084] Embodiment for test tubes

[0085] 41. Aspect according to any one of the preceding aspects, wherein the device comprises a coupling portion, the coupling portion being coupled to the housing and comprising at least one coupling element, the coupling portion being configured to present:

[0086] o a coupled arrangement, wherein the coupled arrangement is coupled to a container by means of the at least one coupling element, the container being able to contain blood and / or allowing blood to flow therein,

[0087] oDecoupled configuration, wherein the decoupled configuration is not coupled to the container.

[0088] 42. Aspect according to aspect 41, wherein the at least one coupling element is magnetic and is configured to allow magnetic coupling with the container.

[0089] 43. An aspect according to any one of the preceding aspects, wherein the device comprises a first coupling element and a second coupling element, the first coupling element and the second coupling element are arranged at relative longitudinal portions of the device and are configured to couple with corresponding magnetic elements located at corresponding longitudinal portions or both ends of the container.

[0090] 44. Aspect according to any of the preceding aspects, wherein each coupling element is magnetic and is configured to allow magnetic coupling with the container.

[0091] Embodiments for tubes

[0092] 45. Aspect according to any of the preceding aspects, wherein the device is associated with a container, such as a tube, into which blood can flow in an operating state, and the device is associated with the container.

[0093] 46. ​​Aspect according to any one of the preceding aspects, wherein the control unit is configured to perform the following operations:

[0094] - detect the type of container, especially the type of tube,

[0095] - Based on the detected type of container, in particular the type of tube, a measurement is prepared.

[0096] 47. Aspect according to aspect 46, wherein the control unit is configured to prepare for measurement based on the detected type of container or tube by adapting the measurement mode of one or more parameters to operation with the detected type of container or tube.

[0097] 48. Aspect according to aspect 46 or 47, wherein the control unit is configured to detect the type of the container or tube and prepare for measurement based on the detected type of the container or tube before the operation of exciting blood flow at a plurality of determined wavelengths.

[0098] 49. Aspect according to aspect 46 or 47 or 48, wherein the control unit is configured to detect the color of the container, in particular the color of the tube.

[0099] 50. Aspect according to any one of aspects 46 to 49, wherein the control unit is configured to perform an operation of preparing for measurement based on the detected type or color of the container or tube by selecting a determined matrix from a plurality of matrices available to the neural network.

[0100] 51. Aspect according to aspect 50, wherein the control unit is configured to select the determined matrix from the plurality of matrices available to the neural network by querying a memory storing information related to the plurality of matrices.

[0101] 52. Aspect according to any of the preceding aspects, wherein the device comprises a cassette, a covering element movable relative to the cassette and a base adapted to accommodate a container, in particular a portion of a tube of an extracorporeal blood circuit, into which blood flows in an operating state of the device, the covering element being configured to operate at least between the following configurations:

[0102] - an operating configuration in which it flattens the opposing surface of a container accommodated at the base,

[0103] - Rest configuration.

[0104] 53. Aspect according to aspect 52, wherein the covering element is movable by rotation relative to the cassette body.

[0105] 54. Aspect according to aspect 52 or 53, the covering element comprises a compression element for compressing the container accommodated at the base in the operating configuration of the covering element.

[0106] 55. Aspect according to aspect 54, wherein the compression element is configured to determine a reduction of between 9% and 17% in the fluid passage cross-section of the container.

[0107] 56. Aspect according to aspect 55, wherein the reduction in the cross-section of the fluid channel occurs at a length between 15 mm and 30 mm or equal to one of these values.

[0108] 57. An aspect according to any one of aspects 52 to 56, wherein the compression element includes opposite curved ends and a flat portion defined between the ends, and the flat portion is configured to determine a flat reading area at the surface of the container, in particular, the upper surface of the container and / or the lower surface of the container.

[0109] 58. Aspect according to any one of aspects 52 to 57, wherein the device comprises at least one restraining element, the at least one restraining element being configured to maintain the covering element in the operational configuration.

[0110] 59. Aspect according to any one of aspects 52 to 58, wherein the device comprises two constraining elements, the two constraining elements being configured to maintain the covering element in the operating configuration, the two constraining elements being opposite to each other.

[0111] 60. Aspect according to any one of aspects 52 to 59, wherein the covering element is hinged to the box body at a hinge portion, and the at least one restraining element is opposite to the hinge portion.

[0112] 61. An aspect according to any one of aspects 52 to 60, wherein the device includes a sensor configured to verify the configuration (preferably the position) of the covering element, in particular to verify that the covering element is in an operating configuration (preferably in a closed position), and the control unit is configured to obtain information about the configuration (preferably the position) assumed by the covering element via the sensor.

[0113] 62. Aspect according to any of aspects 52 to 61, wherein the control unit is configured to prevent measuring the parameter when the cover element is in the rest configuration, preferably in the open position.

[0114] 63. Aspect according to any of aspects 52 to 62, wherein the control unit is configured to allow measuring the parameter only when the covering element is in the operational configuration, preferably in the closed position.

[0115] 64. Aspect according to any one of aspects 52 to 63, wherein the covering element comprises a body and a compression element configured to oscillate relative to the body.

[0116] 65. Aspect according to aspect 64, wherein the compression element is inclined relative to the body.

[0117] 66. Aspect according to any one of aspects 54 to 65, wherein the covering element drags the compression element in a rotational manner.

[0118] 67. Aspect according to aspect 64 or 65 or 66, wherein the compression element is configured to describe an angle different from the angle described by the body of the covering element.

[0119] 68. Aspect according to any one of aspects 64 to 67, wherein the compression element is configured to rotate an angle different from the rotation angle of the covering element body.

[0120] 69. Aspect according to any one of aspects 64 to 68, wherein the covering element comprises a pin, which is operatively arranged between the body and the compression element, and the compression element is configured to swing relative to the body by means of the pin.

[0121] use

[0122] 70. Use of a device according to any of the accompanying device claims and / or any of the preceding device aspects for measuring, in particular simultaneously measuring, multiple blood parameters.

[0123] 71. Aspect according to any of aspect 70, wherein the simultaneous measurement of the plurality of blood parameters is performed with the aid of artificial intelligence.

[0124] 72. Aspect according to aspect 70 or 71, wherein the use is performed in an extracorporeal blood circuit.

[0125] 73. Aspect according to aspect 70 or 71 or 72, wherein use does not involve any initial calibration of the device by an end user.

[0126] 74. Aspect according to any of aspects 70 to 73, wherein use provides measurement of multiple parameters through artificial intelligence, in particular through one or more neural networks.

[0127] 75. Aspect according to any one of aspects 70 to 74, wherein use provides for measuring the plurality of blood parameters flowing in a tube connected to an extracorporeal blood circuit.

[0128] 76. Aspect according to any one of aspects 70 to 75, wherein use provides for measuring the plurality of blood parameters at a portion of a tube of an extracorporeal blood circuit.

[0129] Components

[0130] 77. An assembly comprising:

[0131] - a device according to any of the appended device claims and / or the preceding device aspects,

[0132] - a container capable of containing blood and / or configured to allow blood to flow in the container, preferably the container is a tube or a pipe or tubing of an extracorporeal blood circuit.

[0133] equipment

[0134] 78. An apparatus comprising:

[0135] - a device according to any of the appended device claims and / or the preceding device aspects,

[0136] - medical machines, such as heart-lung machines or extracorporeal membrane oxygenation machines,

[0137] A user interface, such as a display device, is operatively connected or connectable to the apparatus and is configured to make available the measured values ​​of the plurality of blood parameters.

[0138] Method for measuring multiple blood parameters

[0139] 79. A method for measuring multiple blood parameters by artificial intelligence, the method comprising the steps of:

[0140] - exciting the blood, in particular the blood flow, with electromagnetic radiation of a plurality of determined wavelengths,

[0141] - detecting a plurality of electromagnetic responses, in particular optical responses, of the blood, the plurality of electromagnetic responses comprising electromagnetic radiation, in particular light, reflected or diffused by the blood,

[0142] - receiving analog information about a plurality of electromagnetic and / or optical responses of blood, the plurality of electromagnetic and / or optical responses including light retroreflected or diffused by the blood,

[0143] - converting electromagnetic and / or optical response analog information into electromagnetic and / or optical response digital data,

[0144] - processing said electromagnetic and / or optical response digital data and the actual temperature value of the blood by one or more neural networks,

[0145] - As a result of the processing operation of the one or more neural networks, a value for each of a plurality of blood parameters is determined.

[0146] 80. A method for measuring a plurality of blood parameters by artificial intelligence, the method comprising the steps of:

[0147] - stimulate blood flow at multiple defined wavelengths,

[0148] - detecting a plurality of electromagnetic and / or optical responses of the blood, the plurality of electromagnetic and / or optical responses comprising light reflected or diffused by the blood,

[0149] - receiving analog information about a plurality of electromagnetic and / or optical responses of blood, the plurality of electromagnetic and / or optical responses including light reflected or diffused by the blood,

[0150] -Converting analog photoresponse information into photoresponse digital data,

[0151] - processing said light response digital data and actual blood temperature value by one or more neural networks,

[0152] - As a result of the processing operation of the one or more neural networks, a value for each of a plurality of blood parameters is determined.

[0153] 81. Aspect according to aspect 79 or 80, wherein the step of processing the electromagnetic and / or optical response digital data and the actual temperature value of the blood and determining the value of each of the plurality of blood parameters by one or more neural networks comprises:

[0154] o determining a plurality of ratios, each ratio being defined between a magnitude indicative of radiation retroreflected or diffused by the blood due to excitation at a determined wavelength and a magnitude indicative of radiation retroreflected or diffused by the blood due to excitation at another determined wavelength,

[0155] o providing the plurality of ratios as input to one or more neural networks,

[0156] o processing said plurality of ratios by means of one or more neural networks taking into account a plurality of data of previous measurements of said blood parameters performed during previous training,

[0157] o As output from one or more neural networks, values ​​for each of a plurality of blood parameters are provided.

[0158] 82. According to aspect 79 or 80 or 81, the method comprises: arranging an apparatus according to any of the accompanying apparatus claims and / or the preceding apparatus aspects, and performing the method with the aid of the apparatus.

[0159] 83. Aspect according to any of aspects 79 to 82, wherein the method is performed by means of an apparatus according to any of the accompanying apparatus claims and / or any of the preceding apparatus aspects.

[0160] 84. Aspect according to any one of aspects 79 to 83, wherein the method comprises detecting the actual temperature value of the blood, preferably by means of a temperature sensor housed in a cartridge of the device.

[0161] 85. Aspect according to any one of aspects 79 to 84, wherein the method is a method for simultaneously measuring multiple blood parameters with the aid of artificial intelligence.

[0162] 86. Aspect according to any one of aspects 79 to 85, wherein the method comprises the step of making the measured values ​​of the plurality of blood parameters available simultaneously.

[0163] 87. An aspect according to any one of aspects 79 to 86, wherein the step of processing the multiple ratios by means of one or more neural networks taking into account multiple data from previous measurements of the blood parameters performed during previous training includes: detecting the value of each parameter of the multiple blood parameters that the artificial intelligence believes corresponds to a previously determined ratio based on the multiple data from previous measurements.

[0164] 88. An aspect according to any one of aspects 79 to 87, wherein the method further comprises providing, as input to one or more neural networks, a reference value of a first parameter to be measured and a reference value of a second parameter to be measured, the reference values ​​being acquired and stored in the device during a calibration step prior to use of the device.

[0165] 89. An aspect according to aspect 87 or 88, wherein the step of detecting the value of each of the multiple blood parameters that the artificial intelligence believes corresponds to a previously determined ratio based on the multiple data from previous measurements includes: processing data and / or information with the aid of a mathematical matrix model, preferably with the aid of one or more matrices of weights and biases.

[0166] 90. Aspect according to any one of aspects 79 to 89, wherein the step of exciting blood flow at a plurality of determined wavelengths comprises activating a plurality of excitation components according to a determined time sequence.

[0167] 91. Aspect according to aspect 90, wherein the step of activating the plurality of firing members according to the determined time sequence comprises: activating the first firing member and the second firing member alternately.

[0168] 92. The aspect according to any one of aspects 79 to 91, wherein the step of detecting a plurality of blood electromagnetic responses comprises:

[0169] - detecting a plurality of electromagnetic and / or optical responses of the blood by means of an initial photodetector,

[0170] - detecting a plurality of electromagnetic and / or optical responses of the blood by means of a second photodetector.

[0171] 93. An aspect according to any one of aspects 79 to 92, wherein blood stimulation is performed in an operating state of the device, in which the blood flows into a container associated with the device.

[0172] 94. Aspect according to any of aspects 79 to 93, wherein the container associated with the device is a test tube, in particular associated with or arranged in an extracorporeal blood circuit.

[0173] 95. Aspect according to any one of aspects 79 to 94, wherein the container associated with the device is a tube or a tube part of an extracorporeal blood circuit.

[0174] 96. An aspect according to any one of aspects 79 to 95, wherein the method includes the step of making the value of each of the multiple blood parameters available on a medium that can be queried by an operator, preferably at the same time, such as on a screen of a machine associated with the device configured to perform the method for measuring multiple blood parameters by artificial intelligence.

[0175] Stream Information

[0176] 97. Aspect according to any one of aspects 79 to 96, wherein the method comprises the step of taking into account blood flow-related values, in particular the volumetric flow rate of blood, when measuring oxygen saturation and / or hematocrit.

[0177] 98. An aspect according to any one of aspects 79 to 97, wherein the step of taking into account values ​​related to blood flow when measuring oxygen saturation and / or hematocrit comprises correcting the measured values ​​of oxygen saturation and / or hematocrit according to blood flow.

[0178] Detect container type (color)

[0179] 99. Aspect according to any one of aspects 79 to 98, wherein the method further comprises the steps of:

[0180] - detect the type of container, especially the type of tube,

[0181] - adapting the mode of measurement of one or more parameters to the type of tube detected;

[0182] Preferably, the step of detecting the type of container, in particular the type of tube, comprises: detecting the color of the container, in particular the color of the tube.

[0183] 100. An aspect according to aspect 99, wherein the method comprises the step of preparing a control unit of the device for measurement based on the detected type of tube, the step comprising the step of adapting the mode of measurement of one or more parameters to the detected type of tube.

[0184] 101. Aspect according to aspect 99 or 100, wherein the step of detecting the type of container, in particular the type of tube, is performed when physiological saline or another fluid other than blood is flowing in the container.

[0185] 102. Aspect according to aspect 99 or 100 or 101, wherein the step of detecting the container type, in particular the tube type, is performed before the step of exciting the blood flow at a plurality of determined wavelengths.

[0186] 103. An aspect according to any one of aspects 99 to 102, wherein the step of adapting the pattern of measurement of one or more parameters to the type of tube detected comprises the step of selecting a determined matrix from a plurality of matrices available to the neural network.

[0187] 104. Aspect according to aspect 103, wherein the step of selecting the determined matrix from a plurality of matrices available to the neural network comprises querying a memory storing information related to the plurality of matrices.

[0188] 105. Aspect according to aspect 104, wherein the memory is a memory of an apparatus configured to implement the method.

[0189] Joint between device and container

[0190] 106. Aspect according to any of aspects 79 to 105, wherein the method comprises the step of associating the device with a container, for example with a tubular element such as a test tube or a tube such as a tube of an extracorporeal blood circuit.

[0191] 107. Aspect according to aspect 106, wherein the step of associating the device with the container includes restraining the device and the container relative to each other.

[0192] 108. Aspect according to aspect 107, wherein the step of relatively restraining the device and the container comprises:

[0193] - magnetically constraining the device and the container, for example by means of one or more magnetic coupling elements; and / or

[0194] - Mechanically constraining the device and the container, for example by means of at least one mechanical constraining element.

[0195] 109. Aspect according to aspect 108, wherein the step of mechanically constraining the device and the container comprises engaging the covering element relative to the box body of the device by means of at least one constraining element, in particular by means of two constraining elements opposite to each other.

[0196] 110. Aspect according to aspect 108 or 109, wherein the step of mechanically constraining the device and the container comprises moving the cover element proximally relative to the housing of the device, in particular rotating it.

[0197] 111. An aspect according to any one of aspects 79 to 110, wherein the method includes flattening the relative surfaces of a container associated with the device, the step being performed prior to exciting blood flow at a plurality of determined wavelengths.

[0198] 112. Aspect according to aspect 111, wherein the step of flattening the relative surface of the container associated with the device includes: compressing the tube of the extracorporeal blood circuit.

[0199] 113. Aspect according to aspect 111 or 112, wherein the step of flattening the relative surface of the container associated with the device is performed after the step of associating the device with the container.

[0200] 114. Aspect according to any one of aspects 79 to 113, wherein the method further comprises the step of compressing a portion of the container in which the blood flows or is capable of flowing.

[0201] 115. The aspect of aspect 114, wherein the step of compressing a portion of the container determines a reduction in the fluid passage cross-section of the container of between 9% and 17%.

[0202] 116. Aspect according to aspect 115, wherein the reduction in the cross-section of the fluid channel occurs at a length between 15 mm and 30 mm or equal to one of these values.

[0203] 117. An aspect according to aspect 114 or 115 or 116, wherein the step of compressing a portion of the container in which the blood flows or is capable of flowing is performed before the step of determining the value of each of the multiple blood parameters as a result of a processing operation of one or more neural networks, in particular before the step of exciting the blood.

[0204] 118. Aspect according to any one of aspects 79 to 117, wherein the method comprises the step of oscillating the compression element relative to the container in which the blood is flowing or capable of flowing.

[0205] 119. Aspect according to aspect 118, wherein the step of oscillating the compression element relative to the container is performed during the step of moving, in particular rotating, the access cover element relative to the housing of the device.

[0206] 120. Aspect according to aspect 118 or 119, wherein the step of swinging the compression element relative to the container is achieved by swinging and tilting the compression element.

[0207] 121. Aspect according to aspect 118 or 119 or 120, wherein the step of oscillating the compression element involves compressing the container stepwise and / or gently.

[0208] Other aspects

[0209] 122. An aspect according to any of the preceding aspects, wherein each ratio is a ratio of optical counts taken at a corresponding wavelength.

[0210] 123. An aspect according to any of the preceding aspects, wherein one or more neural networks take into account a plurality of data of previous measurements of said blood parameter carried out during previous training sessions with the aid of at least one matrix, in particular at least one matrix of weights and biases.

[0211] 124. An aspect according to any of the preceding aspects, wherein the container in which the blood flows is of a disposable type.

[0212] 125. An aspect according to any of the preceding aspects, wherein the device is configured to measure at least two blood parameters by artificial intelligence.

[0213] 126. The aspect according to aspect 125, wherein the at least two blood parameters are the oxygen saturation of the blood and the hematocrit value of the blood.

[0214] 127. An aspect according to any of the preceding aspects, wherein the device is configured to measure at least three blood parameters.

[0215] 128. An aspect according to aspect 125 or 126 or 127, wherein the device is configured to directly measure the at least two blood parameters by artificial intelligence, and to measure a third blood parameter as a parameter derived from the measurement of at least one of the two parameters.

[0216] 129. Aspect according to aspect 127 or 128, wherein the third parameter is hemoglobin content or concentration.

[0217] 130. An aspect according to any of the preceding aspects, wherein the device is configured to measure four blood parameters.

[0218] 131. The aspect according to aspect 130, wherein the four blood parameters are: blood oxygen saturation, hematocrit value, hemoglobin content or concentration and blood temperature.

[0219] 132. An aspect according to any of the preceding aspects, wherein the device is configured to be associated with a test tube and has no relative rotation parts to allow engagement between the test tube and the device.

[0220] 133. An aspect according to any one of aspects 31 to 132, wherein the first photodetector comprises a silicon photodiode (or Si photodiode).

[0221] 134. An aspect according to any one of aspects 31 to 133, wherein the second photodetector comprises an indium, gallium, and arsenic photodiode (or an InGaAs photodiode).

[0222] 135. An aspect according to any of the preceding aspects, wherein the artificial intelligence is capable of measuring (directly or indirectly) the following parameters: blood oxygen saturation (SatO2), hematocrit (Hct) and hemoglobin (Hb).

[0223] 136. The aspect according to any one of the preceding aspects, wherein the measurement of the plurality of parameters is performed as follows:

[0224] oDirectly measure the percentage of blood oxygen saturation, i.e. SatO2[%],

[0225] o Direct measurement of hematocrit value, i.e., Hct [%],

[0226] o Direct measurement of blood temperature, i.e., Temp. [℃],

[0227] o Indirect measurement of the hemoglobin content in the blood, ie, its concentration Hb [g / dl].

[0228] 137. The aspect according to any one of the preceding aspects, wherein the measurement of the plurality of parameters is performed as follows:

[0229] o Directly measure the percentage value of blood oxygen saturation through artificial intelligence, i.e., SatO2[%],

[0230] o Direct measurement of hematocrit value, i.e., Hct[%], by artificial intelligence,

[0231] o Use temperature sensor to directly measure blood temperature, i.e., Temp. [℃],

[0232] o Indirect measurement of the hemoglobin content of the blood, i.e., its concentration, i.e., Hb [g / dl], by artificial intelligence.

[0233] 138. An aspect according to any of the preceding aspects, wherein the measurement of the parameters is performed by means of the same measurement operation or session.

[0234] 139. An aspect according to any of the preceding aspects, wherein the computational model implemented by one or more neural networks is a mathematical matrix model.

[0235] Machine Learning Methods

[0236] 140. A machine learning method based on one or more neural networks, comprising the following steps:

[0237] - providing a machine learning apparatus, the step comprising coupling a plurality of devices to corresponding containers arranged along the extracorporeal blood circuit and / or a portion of the extracorporeal blood circuit,

[0238] - performing one or more training runs in which desired blood parameters (hematocrit, oxygen saturation, temperature and optionally hemoglobin) are measured while the blood flows through the container,

[0239] - Developing a calculation method configured to allow calculating one or more of said parameters with the aid of artificial intelligence.

[0240] 141. An aspect according to aspect 140, wherein the machine learning method can include steps 1) to 7) as set out in the corresponding parts of the detailed description below.

[0241] 142. An aspect according to aspect 140 or 141, wherein the machine learning method is implemented according to a machine learning setting according to the corresponding part described in detail below.

[0242] Conventions and limitations

[0243] Note that in the following detailed description, corresponding parts / components / elements are indicated with the same reference numerals. The accompanying drawings may illustrate the subject matter of the invention by non-scaled representations; therefore, the parts / components / elements shown in the drawings and related to the subject matter of the invention may only relate to schematic representations. In the context of the present disclosure, unless otherwise specifically indicated, the use of terms such as "above", "up", "at the top", "below", "down", "at the bottom", "laterally", "laterally", "inside", "interiorly", "outside", "exteriorly", "horizontally", "horizontally", "vertically", "vertically", "front", "frontally", "back", "backward", "right", "left", similar terms and their variations refers to at least one spatial orientation that the object of the present invention can take in the state of use (see, for example, one or more of the accompanying drawings). Unless otherwise specifically stated, the terms "state" or "configuration" can be used interchangeably in the context of the present disclosure. Unless otherwise specifically stated, the expressions "upstream", "downstream" and similar or derived expressions refer to the arrangement of a part / component / element relative to a determined line or branch of a fluid flow along a fluid line or circuit or relative to the displacement of the element.

[0244] In the context of the present disclosure, one or more of the following definitions / conventions apply where appropriate, unless otherwise stated and / or unless the context excludes this (e.g. for technical reasons):

[0245] - The device is configured to measure the following parameters: oxygen saturation (ie SatO2), hematocrit (ie Hct), blood temperature and optionally also hemoglobin content (ie Hb).

[0246] - the device is preferably a probe;

[0247] - the device is preferably configured to measure said plurality of parameters simultaneously;

[0248] - "simultaneous measurement" means "measurement by means of the same measurement operation or session", in particular it means "measuring said plurality of blood parameters by means of the same measurement operation or session, by making the measured values ​​of said plurality of blood parameters available simultaneously";

[0249] - The device is configured to measure:

[0250] oThe percentage value of blood oxygen saturation, i.e., SatO2[%],

[0251] o Hematocrit value, i.e., Hct [%],

[0252] oBlood temperature, i.e., Temp. [℃],

[0253] o Optionally, the hemoglobin content or concentration in the blood, i.e., Hb [g / dl],

[0254] Among them, the oxygen saturation of the blood, the hematocrit value of the blood, and, where applicable, the hemoglobin content of the blood or its concentration are measured by artificial intelligence, and the temperature of the blood is measured using a temperature sensor;

[0255] - the plurality of blood parameters measured by the artificial intelligence may include "at least two blood parameters" (i.e., SatO2, Hct), more specifically "three blood parameters" (i.e., SatO2, Hct and Hb), even more specifically "exclusively three blood parameters" (i.e., exclusively SatO2, Hct and Hb);

[0256] - The device is configured to measure a plurality of parameters in the following manner:

[0257] o Directly measure the percentage value of blood oxygen saturation through artificial intelligence, i.e., SatO2[%],

[0258] o Direct measurement of hematocrit value, i.e., Hct[%], by artificial intelligence,

[0259] o Preferably use a temperature sensor to directly measure the blood temperature, i.e. Temp. [°C],

[0260] o Indirect measurement (derived from direct measurement by AI) of the hemoglobin content of the blood, i.e. its concentration Hb [g / dl]; note that the hemoglobin content is derived from the hematocrit

[0261] Parameters derived from direct measurement of values;

[0262] - With regard to direct measurements, the device is configured to measure with high accuracy the parameters within the corresponding ranges, namely:

[0263] o For hematocrits of 13% to 55%,

[0264] o For blood oxygen saturation 35% to 99.9%,

[0265] o For blood temperature 9°C to 42°C.

[0266] - "retroreflected electromagnetic radiation" or "retroreflected light" means electromagnetic radiation or light which, after passing through the medium (blood) and thus being partially absorbed, returns to the photodetector after exciting the blood. It corresponds to the optical counts read by the photodetector (photodiode). "Electromagnetic radiation" may also be referred to simply as "radiation";

[0267] - “light” means radiation in the visible part of the electromagnetic spectrum;

[0268] - Each "optical count" is a measurement indicating a signal (electromagnetic radiation) detected by a photodetector (photodiode) after excitation by electromagnetic radiation of a determined wavelength; optical counts are used to determine or calculate ratios and are Fig.12 It is called "count" in .

[0269] The above conventions and limitations may be used to interpret the claims when necessary. If necessary, in particular when the claims and / or aspects use one or more expressions covered by one or more conventions or limitations, one or more of the conventions and limitations may be included in one or more of the following claims and / or one or more of the aforementioned aspects. BRIEF DESCRIPTION OF THE DRAWINGS

[0270] In order to better understand the invention and appreciate its advantages, some embodiments are described below by way of example and not limitation with reference to the accompanying drawings, in which:

[0271] Figure 1 A device for measuring a plurality of blood parameters according to a first embodiment of the present invention is shown; the device is shown coupled to a test tube;

[0272] Figure 1A Shows Figure 1 a view of the device from below;

[0273] Figure 2 Shows Figure 1 an exploded view of a device and a test tube (the magnetic coupling element of the test tube is also shown in the exploded view), wherein half of the housing of the device has been removed to show the device components housed inside the housing;

[0274] Figure 3 Shows Figure 2 The device and the test tube in a coupled state to illustrate the positioning of the various internal components of the device in an operational state;

[0275] Figure 4 shows a front perspective view of a device for measuring a plurality of blood parameters according to a second embodiment of the invention, and shows the tube in an exploded view; the cover element of the device is in a rest configuration (open cover element);

[0276] Figure 5 Shows Figure 4 The device in which the tube is engaged at a suitable base;

[0277] Figure 5A Shows Figure 5 A device and a tube in which half of the housing of the device has been removed to illustrate the device components housed within the housing and the positioning of said components in an operational state;

[0278] Figure 6 Show Figure 4 A rear perspective view of the device of , with the tube engaged at the base and the cover element in the operational configuration (closed cover element);

[0279] Fig. 6A Shows Figure 6 A side view of a device showing how the covering element flattens the opposing surfaces of a tube received in a base; the flattening is achieved by compressing the tube;

[0280] Figure 7 Shows Figure 4 An exploded view of the device and tube;

[0281] Fig. 8A Shows Figure 1 The device (shown in the upper right corner of the extracorporeal blood circuit) and Figure 4 The device (shown in the lower right corner in an extracorporeal blood circuit) has possible clinical uses in the operating room, where the device can be used in conjunction with a heart-lung machine;

[0282] Figure 8B Shows Figure 1 The device (shown in the upper right corner of the extracorporeal blood circuit) and Figure 4 Another possible clinical use of the device (shown in the lower right corner of the extracorporeal blood circuit) is in the intensive care unit, where the device can be used in conjunction with an extracorporeal membrane oxygenation (ECMO) machine;

[0283] Fig. 9 shows a laboratory setup for realizing automatic learning by means of a device according to the invention;

[0284] Fig.10 shows a hematocrit calibration curve (Hct%) at fixed saturation values ​​for calibrating the device according to the invention in the laboratory; the wording on the abscissa indicates the ratio of optical counts detected by an InGaAs (Indium Gallium Arsenide) photodiode due to excitation at a wavelength of 805 nm (at the numerator) to the ratio of optical counts detected by an InGaAs photodiode due to excitation at a wavelength of 1450 nm (at the denominator);

[0285] Fig.11shows a saturation calibration curve (Sat%) at fixed hematocrit values ​​for calibrating a device according to the invention in a laboratory; the wording on the abscissa indicates the ratio of optical counts detected by a Si (silicon) photodiode due to excitation at a wavelength of 805 nm (at the numerator) to the ratio of optical counts detected by a silicon photodiode due to excitation at a wavelength of 660 nm (at the denominator);

[0286] Fig.12 The structure of a multilayer perceptron type neural network is shown by way of example as one possible structure of one or more neural networks that can be implemented in the device according to the invention;

[0287] Fig.13 It shows that it can be Fig.12 A possible computational method for implementing a neural network in order to measure the value of a blood parameter output by the neural network;

[0288] Fig.14 Shows Fig.12 The activation function of the neurons in the neural network;

[0289] Fig.15 shows the tube relative to Fig. 6A an extrusion profile of a configuration of; it shows in more detail how the cover element compresses the tube, resulting in a smooth transition between the uncompressed portion of the tube and the central portion of the tube which is compressed at the base to obtain a flat reading area;

[0290] Fig.16 Shows Figures 4 to 7 and Fig.15 An alternative covering element of the covering element shown in ; the compression element of the alternative covering element is configured to oscillate relative to the body of the covering element itself. The oscillating compression element is shown partially retracted from its base implemented in the body of the covering element;

[0291] Fig.17 shows in section a device according to a variant of the second embodiment, in which the covering element assumes an angular position in which the oscillating compression element is placed close to the tube;

[0292] Fig.18 Shown in cross section Fig.17 An apparatus in which the covering element has been removed from Fig.17 The angular position of is rotated further toward the tube, and the oscillating compression element compresses the tube. DETAILED DESCRIPTION

[0293] Device for measuring multiple blood parameters

[0294] The device according to the invention is generally indicated in the figures by the numeral 1, 1'. The device 1, 1' is configured to measure a plurality of blood parameters, in particular the following parameters: oxygen saturation (i.e. SatO2), hematocrit (i.e. Hct), blood temperature and optionally also hemoglobin content (i.e. Hb). Preferably, the device 1, 1' is configured to measure these parameters according to the following modes: measuring oxygen saturation and hematocrit by means of artificial intelligence, measuring temperature by means of a temperature sensor, and measuring hematocrit in a derived manner, in particular in a derived manner from the value of the hematocrit; details of the calculation of the parameters by means of artificial intelligence are explained below.

[0295] The device 1, 1' is configured to be associated with a container 25, 25', in particular to be coupled to a container 25, 25', in which blood can flow; the container 25, 25' is preferably a catheter. The container 25, 25' generally has a tubular shape and is therefore provided with a fluid channel cross section in which blood can flow. For example, the container can be a test tube 25 connected to the fluid of the extracorporeal blood circuit, or a tube (or a part of a tube) 25' of the extracorporeal blood circuit. The device 1, 1' is according to a first embodiment (the device indicated by the reference numeral 1, see Figures 1 to 3 ) and according to a second embodiment of a tube (or tube portion) 25' designed to be coupled to an extracorporeal blood circuit (device indicated by reference numeral 1', see Figures 4 to 7 and Fig.15 ) is described and depicted; obviously, other embodiments are contemplated. As described in detail below, the device 1, 1 'and the container 25, 25' are configured to be coupled to each other and to interact functionally. The container 25, 25' that contacts the blood during use is a disposable element, while the device 1, 1 'is reusable to perform additional parameter measurements. In particular, the container 25, 25' is the only disposable element in the assembly including the device 1, 1 'and the container 25, 25'. The common features between the first and second embodiments are described herein, and their differences are detailed below.

[0296] The device 1, 1' comprises a housing 2. The housing 2 has an internal volume in which the components described below are housed, including a control unit 3. The housing 2 has small dimensions, which provide compactness to the device 1, 1'. It should also be noted that the housing 2 may be assemblable, in particular, the housing 2 may comprise two parts 2a, 2b, which are assembled together in the operating state of the device 1, 1'. As shown in the figures, the assemblable parts may be half shells 2a, 2b; the half shells are preferably substantially symmetrical to each other.

[0297] The device 1, 1' comprises at least one excitation member 4, 5, which is configured to excite the blood flow with electromagnetic radiation of a plurality of determined wavelengths. The electromagnetic radiation is intended to induce a photoresponse of the blood; in this respect, see the wavelengths described below. In an operating state of the device 1, 1', in which at least one excitation member 4, 5 excites the blood, the blood responds to the excitation by re-reflecting and diffusing the electromagnetic radiation, including radiation of visible wavelengths (light). The wavelengths are selected so that the blood responds to these wavelengths with a significant photoresponse.

[0298] The wavelengths for blood excitation can be as follows:

[0299] -660 nm, which was chosen because it is at this wavelength that most of the absorption of oxygenated hemoglobin occurs, i.e., at this wavelength the difference in absorption between the oxygenated form and the non-oxygenated or reduced form of hemoglobin is greatest,

[0300] -805 nm, which was chosen because the isosbestic point of hemoglobin occurs (i.e., where the absorption of oxyhemoglobin and deoxyhemoglobin is equal),

[0301] -1450 nm, which was chosen because there is more absorption of water at this wavelength (for measuring hematocrit),

[0302] - 940 nm, which was chosen because of the maximum diffusion on water and the maximum absorption on blood with a high hematocrit at this wavelength (it is used in combination with 525 nm, 1450 nm and 805 nm wavelengths to detect the following states of device operation: presence of blood in the container, presence of saline, absence of container),

[0303] - 525 nm, which was chosen because there is a maximum absorption from blood at this wavelength (used to determine that the device "sees" blood),

[0304] -optionally, 1050 nm, this wavelength is chosen for possible hemoglobin changes,

[0305] - Optionally, 1550 nm, this wavelength is selected to improve the accuracy of hematocrit measurements for Hct% values ​​below 40% as there is a greater dynamic variation in the optical count at this wavelength.

[0306] Wavelengths are to be understood as falling within a defined range around the above values, for example as follows:

[0307] o 660±10nm,

[0308] o 805±5nm,

[0309] o 1450±30nm,

[0310] o 940±10nm,

[0311] o 525±10nm,

[0312] o 1050±10nm,

[0313] o 1550nm±10nm.

[0314] As shown in the drawings, the device 1, 1' comprises: a first excitation member 4, configured to excite blood flow at least at a first plurality of wavelengths; and a second excitation member 5, configured to excite blood flow at a second plurality of wavelengths. The first excitation member 4 and the second excitation member 5 are arranged in the box body 2, and in particular, the first excitation member 4 and the second excitation member 5 are arranged side by side.

[0315] Advantageously, by means of one or more excitation members 4, 5, the device 1, 1' is capable of exciting the blood with radiation from the visible region (excitation at 600nm) to the infrared, so as to "interrogate" the blood flowing in the container 25, 25' associated with the device 1, 1' at these electromagnetic wavelengths and obtain an appropriate response for calculating a plurality of parameters.

[0316] In the embodiment shown herein, the first excitation member 4 is configured to excite blood flow at the following wavelengths: 525nm, 940nm, optionally 1050nm; and the second excitation member 5 is configured to excite blood flow at the following wavelengths: 660nm, 805nm, 1450nm. In more detail, each excitation member 4, 5 comprises a plurality of individual LED elements (not shown in the drawings) configured to emit light radiation at respective wavelengths between said wavelengths. In particular, each excitation member 4, 5 comprises a plurality of individual LED elements equal in number to said wavelengths that the excitation member is configured to emit.

[0317] In the embodiments disclosed herein, the first excitation means comprises a tricolor LED means 4 configured to excite blood flow at 525 nm, 940 nm, optionally 1050 nm (a first plurality of wavelengths); and the second excitation means comprises a multi-wavelength LED means 5 configured to excite blood flow at 660 nm, 805 nm, 1450 nm (a second plurality of wavelengths). Each excitation means 4, 5 comprises a chip on which a separate LED element is mounted; the tricolor LED means 4 mounts three LED elements to emit electromagnetic radiation at wavelengths of 525 nm, 940 nm and 1050 nm, and the multi-wavelength LED means 5 mounts at least three LED elements to emit electromagnetic radiation at wavelengths of 660 nm, 805 nm and 1450 nm. Optionally, the multi-wavelength LED means 5 also mounts a fourth LED element to also emit electromagnetic radiation at a wavelength of 1550 nm; this LED element can be used to improve the measurement accuracy of the hematocrit, especially for hematocrit values ​​below 40%).

[0318] The individual LED elements can be activated (turned on) by the control unit 3 in a time-controlled manner as follows: they are turned on sequentially every 100 μs at a frequency of about 20 pulses per second and then turned off for 50 ms. The LED elements are turned on one at a time and within 1 second, each LED element (each light source) sends 20 optical count measurements. Preferably, all LED elements are turned on at the same frequency. During this time, these measurements are averaged for each individual LED element and the ratio of optical counts is calculated, which (as will be seen below) is used to calculate the parameters. The applicant notes that such timing allows a good signal-to-noise ratio to benefit the parameter measurement to be performed. Obviously, other timings that allow a good signal-to-noise ratio and / or optimal measurement of the parameters can be used.

[0319] In order to protect each excitation member 4, 5, the device 1, 1' can provide an appropriate excitation member protection window 6, 7. Each excitation member protection window 6, 7 is intended to protect the corresponding excitation member 4, 5 from electromagnetic waves having a frequency different from the frequency that the corresponding excitation member 4, 5 can emit; the device 1, 1' shown in the drawings has a first excitation member protection window 6 and a second excitation member protection window 7 (see FIG. Figure 2 and Figure 7 The protection windows 6, 7 of the excitation members are facing the respective excitation members 4, 5, so that in use, they are arranged between the respective excitation members 4, 5 and the container 25, 25' into which the blood flows (see Figure 3). Each excitation member window 6, 7 may be made of a material configured to propagate radiation of the frequency that the respective excitation member 4, 5 is configured to emit; for example, each excitation member window 6, 7 may be made of COP (Cyclic Olefin Polymer).

[0320] The device 1, 1' further comprises at least one electromagnetic radiation detection member 8, 9 configured to detect a plurality of electromagnetic blood responses comprising electromagnetic radiation or light reflected or diffused by the blood. After being excited by the one or more excitation members 4, 5, in an operating state of the device, electromagnetic radiation (including light responses) is reflected or diffused from the blood.

[0321] In more detail, as shown in the accompanying drawings, the device 1, 1 ' comprises a first electromagnetic radiation detection member 8 and a second electromagnetic radiation detection member 9. In the embodiments described herein, each electromagnetic radiation detection member is in the form of a photodetector 8, 9; therefore, the following description refers to the electromagnetic radiation detection member being a photodetector (photodiode). It should be understood that the device 1, 1 ' may include one or more electromagnetic radiation detection members different from the photodetector 8, 9.

[0322] Thus, the device 1, 1' comprises a first photodetector 8 and a second photodetector 9, each of which is configured to detect a plurality of blood photoresponses. The first photodetector 8 and the second photodetector 9 are arranged in the housing 2, in particular they are arranged side by side.

[0323] The first photodetector 8 and the second photodetector 9 are configured to detect the blood response at a respective plurality of different wavelengths. Preferably, the first photodetector 8 is configured to detect the light response of the blood upon excitation at the following wavelengths: 660 nm, 805 nm, 525 nm, 1050 nm, 940 nm. As for the second photodetector 9, the second photodetector 9 is preferably configured to detect the light response of the blood upon excitation at the following wavelengths: 805 nm and 1450 nm.

[0324] In the embodiments disclosed herein, the first photodetector is preferably a Si (silicon) type photodiode 8, i.e., it is a silicon photodiode, and the second photodetector is preferably an InGaAs (indium gallium arsenide) type photodiode 9, i.e., it is a semiconductor composed of indium, gallium and arsenic that is typically sensitive to the electromagnetic radiation band between 600nm and 2600nm.

[0325] In operation, each photodiode 8, 9 converts the electromagnetic radiation it receives into an electrical signal (current), which is converted into a voltage via a transimpedance, which is amplified by an analog-to-digital converter (ADC, part of the control unit 3) which detects a number between 0 and 4096 corresponding to the detected light counts. As will be seen in more detail below, these optical counts are used by the control unit 3 to calculate a ratio. Fig.10 and Fig.11 The ratio of optical counts is shown as "Ratio".

[0326] In order to protect each photodetector 8, 9, the device 1, 1' can provide a corresponding photodetector protection window 10, 11. Each photodetector protection window 10, 11 is intended to protect the corresponding photodetector 8, 9 from electromagnetic waves having a frequency different from the frequency that the photodetector 8, 9 can detect; the device 1, 1' shown in the figures has a first photodetector protection window 10 and a second photodetector protection window 11, which are intended to protect the first photodetector 8 and the second photodetector 9, respectively (see Figure 2 and Figure 7 The photodetector protection windows 10, 11 face the corresponding photodetectors 8, 9 so that, in use, they are arranged between the corresponding photodetectors 8, 9 and the containers 25, 25' into which the blood flows (see Figure 5A ). Each photodetector protection window 10, 11 can be configured in material to transmit radiation of a frequency that the corresponding photodetector 8, 9 is configured to detect; for example, each photodetector protection window 10, 11 can be COP (Cyclic Olefin Polymer), that is, the same material as the protection windows 6, 7 of the excitation member.

[0327] The device 1, 1' comprises a temperature sensor 12 housed in the box body 2 and configured to detect the temperature of the blood. In particular, the temperature sensor 12 is configured to detect the temperature of the blood inside the container. In order to enable easy detection of the temperature of the blood contained in the container 25, 25', the temperature sensor is preferably arranged near a coupling portion of the device 1, 1', which is configured to enable coupling with the container 25, 25'. Even more particularly, the temperature sensor 12 is arranged near the coupling portion so that the temperature sensor 12 faces the container 25, 25' in the coupled configuration of the device 1, 1'. The temperature sensor is preferably an infrared temperature sensor 12.

[0328] In order to protect the temperature sensor 12, the device 1, 1' may provide a temperature sensor protection window 13. The temperature sensor protection window 13 is intended to protect the temperature sensor 12 from electromagnetic waves having a frequency different from the frequency that the temperature sensor 12 can detect. The temperature sensor protection window 13 faces the temperature sensor 12, so that when in use, the temperature sensor protection window 13 is arranged between the temperature sensor 12 and the container 25, 25' into which the blood flows (see Figure 3 and Figure 5A ). In embodiments where the temperature sensor 12 is of an infrared type, the temperature sensor protection window 13 may be made of a material configured to transmit radiation in infrared frequencies; for example, the temperature sensor protection window 13 may be made of zinc sulfide (ZnS).

[0329] The box body 2 includes at least one hole 14, which is suitable for facing the container 25, 25' into which the blood flows when the device 1, 1' is in an operating state. The excitation component and the electromagnetic radiation detection component face the hole or corresponding holes so that they can excite the blood and receive the excitation response from the blood respectively. In addition, the temperature sensor preferably faces the container so that it can measure the temperature of the blood. The opening 14 of each component 4, 5, 8, 9 and the opening of the temperature sensor 12 can be set (see Figure 4 ). According to the embodiment shown, the electromagnetic radiation detection members 8, 9 are arranged between the excitation members 4, 5 and the temperature sensor 12. Furthermore, as shown, the excitation members 4, 5, the electromagnetic radiation detection members 8, 9 and the temperature sensor 12 can be arranged along the same longitudinal direction. Such features related to the structural arrangement of the excitation members 4, 5, the electromagnetic radiation detection members 8, 9 and the temperature sensor 12 increase the compactness of the device 1, 1 '.

[0330] In the following, the structural and functional differences between the device 1 according to the first embodiment and the device 1 ' according to the second embodiment are described, and then the operating logic of the control unit 3 of the device 1, 1 ' is described. The functional differences relate to the mode of engagement between the device 1, 1 ' and the container (test tube 25 and tube 25' respectively) to which they are configured to be coupled.

[0331] First embodiment of the device (device that can be associated with a test tube)

[0332] A first embodiment of the device 1 is envisaged to operate on a test tube 25 into which the blood circulates in the extracorporeal blood circuit during the measurement. The test tube 25 has a flat surface, in particular a flat top surface, which defines a flat reading area essential for the accuracy of the measurement; see Figure 2 .

[0333] The box body 2 also has an opening 15, which is suitable for allowing connection to a cable 16; the cable 16 allows the digital data processed by the control unit 3 to be transmitted to the medical machine 90', 90" and / or the display device 91. The cable 16 also makes it possible to power the device 1. As shown in the accompanying drawings, the cable 16 can have a stress relief element 16a, which can avoid or minimize stress at the part where the cable 16 passes through the box body 2. The box body 2 has at least one gripping portion 2c, which makes it easy to grip and handle the device 1; as shown in the accompanying drawings, the gripping portion can be in the form of a pair of grooves 2c defined on opposite sides of the box body 2.

[0334] Geometrically, the box body 2 has a length L, a width T and a height H; the width T is preferably smaller than the length L and the height H. The box body 2 preferably has a configuration so that the necessary components of the device 1 'are accommodated therein in the smallest possible volume. As shown in the figures, the height H can vary along the length L; therefore, the box body 2 can have a minimum height H1 and a maximum height H2, which can be defined at the respective opposite ends as the length L of the box body 2. The length L measured along the main direction of the container can be between 40 mm and 80 mm. The width T measured orthogonally to the length L can be between 15 mm and 50 mm. The height H measured orthogonally to the length L and the width T can be between 30 mm and 80 mm. In essence, the device 1 has dimensions such that it is essentially pocket-sized. As a non-limiting example, it should be noted that in a preferred version of the first embodiment, the length L can be 65 mm, the maximum height H2 can be 58 mm, and the width L can be 35 mm. In the first embodiment, the volume of the box body 2 defines the volume of the device 1; the volume of the device 1 can be within the volume calculated according to the extreme values ​​of the above-mentioned size range. In particular, in a possible embodiment, the volume of the device 1 ′ may be between 20,000 mm 3 With 400,000mm 3 Between, preferably between 50,000mm 3 With 200,000mm 3 More specifically, between 75,000 mm 3 With 175,000mm 3 or more specifically between 90,000 mm 3 With 150,000mm 3 between.

[0335] The device 1 comprises a coupling portion 17 associated with the cassette body 2. With reference to the orientation of the device 1 shown in the drawings, the coupling portion 17 is coupled to the cassette body 2 at a lower portion of the cassette body 2. The coupling portion 17 may extend parallel to the length L of the cassette body 2 between two opposite longitudinal ends of the cassette body 2. The coupling portion 17 may be integral with the cassette body 2. The coupling portion 17 comprises a base 17a provided with a volume dimensioned to accommodate a container 25' for blood, in particular a test tube. The base may take the form of a recess 17a extending in the longitudinal direction of the extension of the coupling portion 17. As shown in the drawings, the coupling portion 17 may be in the form of a skirt coming out of the lower portion of the cassette body 2 to define the base 17a; the skirt has opposite walls 17b, 17c, between which the base 17a is defined. Since the measurement is optical, it is advantageous to shield the photosensitive area from external light components as much as possible; to this end, the walls are configured to shield the light and reduce the possibility of glare or direct light affecting the measurement. Each wall 17b, 17c can be integral with the half-shell to which it corresponds. Each wall 17b, 17c can also include one or more structural elements 17d, which can be in the form of recesses and / or ribs, configured to allow a clear coupling between the container 25 and the device 1, and additionally or alternatively to relieve (for recesses) or strengthen (for ribs) the walls 17b, 17c on which the recesses and / or ribs are defined. Preferably, the walls 17b, 17c provide the recesses / ribs 17d with the following dual functions: to relieve / reinforce the walls 17b, 17c on which they are defined; and to allow a clear coupling between the device 1 and a determined container 25, which container 25 in turn can have a corresponding structural element 25a. This allows inverting a container, such as a tube 25, between an arterial probe or a venous probe; in other words, this allows the device 1 (arterial probe) described herein to be uniquely coupled to an arterial tube due to the corresponding recesses / ribs presented by the arterial tube, thereby preventing the venous tube from coupling to the arterial probe.

[0336] The coupling portion 17 comprises at least one coupling element 17e, 17f, which allows the device 1 to assume a coupled configuration (eg Figure 1 and Figure 3 The coupling portion 17 may be configured as follows: a coupling configuration in which the housing 25 is coupled to the device 1 and a decoupling configuration in which the housing 25 is decoupled from the device 1. The blood parameter measurement is performed in the coupling configuration in which the container 25 is accommodated at the base 17a. As shown in the drawings, the coupling portion 17 may include at least a first coupling element 17e and a second coupling element 17f. The first coupling element 17e and the second coupling element 17f may be longitudinally opposite to each other (see Figure 2). The first coupling element 17e and the second coupling element 17f are configured to couple with corresponding coupling elements 25b and 25c of the container 25. Providing two coupling elements 17e, 17f arranged at opposite positions allows for greater safety and stability in the coupling between the device 1 and the container 25. The first coupling element 17e and the second coupling element 27f may be of the same type; Figures 1 to 3 Magnetic type coupling elements are shown. Each coupling element comprises a respective magnet 25b, 25c configured to couple to a container 25 (see Figure 2 ) corresponding magnets 17e, 17f; Note that the magnets 25b, 25c of the container are accommodated in the corresponding positions 25d, 25e. Figure 2 As can be seen in the figure, both the magnets 17e, 17f of the device 1 and the magnets 25b, 25c of the container 25 can extend in storage planes transverse to each other. Alternative embodiments are not excluded in which at least one of the two coupling elements 17e and 17f is not magnetic, but for example mechanical.

[0337] The device 1 may also include an additional temperature sensor, which may preferably be a thermistor; the additional temperature sensor, not shown in the drawings relating to this first embodiment, may be arranged below relative to the electromagnetic radiation detection member 9. The additional temperature sensor is configured to measure the temperature of the electromagnetic radiation detection member 9, monitoring the temperature of the electromagnetic radiation detection member 9 (photodiode InGaAs) in such a way as to be able to correct the optical responsiveness to varying temperatures. The additional temperature sensor is in contact with the electromagnetic radiation detection member and the area confined thereon.

[0338] Second embodiment of the device (a device that can be associated with a tube)

[0339] A second embodiment of the device 1 ′ is envisaged to operate directly on a portion of the extracorporeal blood circuit tubing into which the blood flows during the measurement.

[0340] The device 1' comprises a covering element 18, 18' movable relative to the box body 2 and the base 19. As previously mentioned, the box body 2 is the part that provides a housing for the excitation components 4, 5, the electromagnetic radiation detection components 8, 9 and the control unit 3. In addition to the two half-shells 2a, 2b, the box body 2 of the device 1' may also have a base part 2d, which is suitable for connecting the two half-shells 2a, 2b by means of a restraining element 2e (for example a threaded element). The base 19 is defined at the coupling part of the box body 2 and is adapted to accommodate a part of the tube 25' of the extracorporeal blood circuit. The covering elements 18, 18' are configured to be at least in an operating configuration (a closed configuration of the covering element, see Figure 6 and Fig. 6A ) with rest configuration (open configuration of overlay element, see Figure 5 and5A ), in which, in the operating configuration, the covering elements 18, 18' flatten the opposite surface of the tube 25' housed at the base 19, and the respective positions of the covering elements 18, 18' relative to the box body 2 correspond to the rest configuration. In essence, the covering element is a closing element that can be moved between a closed position and an open position. Figure 5 and Figure 6 As shown, the covering element 18, 18' is hinged to the box body 2 at the hinge part 20 of the device 1 and can be moved by rotation relative to the box body 2 in such a way as to determine the transition between the positions corresponding to the rest configuration and the operating configuration. The covering element 18, 18' can be rotated relative to the box body 2 by an angle α, α', α"; the angle α, α', α" can be defined as the angle defined by the body 18b of the covering element 18, 18' and the horizontal plane (see Figure 5 , Fig.17 and Fig.18 ). The box body 2 and the covering elements 18, 18' can present the same footprint in the plane, in particular a polygonal footprint, in particular a quadrilateral footprint; in the figures, both the box body 2 and the covering elements 18, 18' present a substantially square footprint in the plane (see Figures 4 to 7 and Fig.15 ); Such a configuration ensures the compactness of the device and minimizes its footprint. Further structural and geometric details are as follows.

[0341] In this second embodiment, it is important that, during the measurement, the upper and lower surfaces of the tube 25' are at least partially flat for the correct functioning of the device 1' and, above all, for the measurement repeatability and accuracy; to achieve this, the base and covering elements 18, 18' in which the tube 25' is assembled are designed so that a "rectangular" tube section / shape is obtained by flattening the surfaces of the tube 25' (see Fig. 6A ) (Note that such flattening is not necessary for cuvette 25, since it includes a flat surface defining a flat reading area; see Figure 2 ). Having a flat reading area allows the emission cone of electromagnetic radiation from the individual LED elements to enter the tube as far as possible during the measurement without interacting with the walls and changing their transmittance.

[0342] The cover element 18, 18' comprises a compression element 18a for compressing a portion of the tube 25' housed therein in the operating configuration. The compression element 18a is coupled to the body 18b of the cover element 18, 18'. The compression element 18a compresses the tube 25' at the measurement area so as to make the reading area flat. Essentially, when the tube portion 25' is inserted into the base 19 and the cover element 18, 18' is closed, the tube portion 25' is squeezed by the compression element 18a, thereby creating two parallel flat surfaces: one flat surface is in contact with the cover element 18, 18' and the other flat surface is in contact with the bottom of the base 19 (see Fig. 6A The compression element 18a has a geometry designed to minimize the deformation of a portion of the tube 25' so as not to significantly modify the blood flow and thus avoid the risk of triggering hemolysis of the blood caused precisely by a sudden change in the portion through which the blood flows. To this end, it is conceivable that the compression element 18a can be realized from a circular shape of the tube (defined at the uncompressed portions 25b', 25c' of the tube, see Fig.15 ) to a rectangular shape (defined at the compressed portion 25a' of the tube 25', see Fig. 6A ) in order to achieve the smallest possible deformation, with the aim of minimizing or even eliminating the effects caused by mechanical trauma (e.g. due to sudden cross-sectional contraction) to which cellular components, in particular erythrocytes, are subjected during in vitro flow. Such mechanical trauma may lead to cell lysis or various forms of sublethal damage, including changes in cell morphology and cell deformability, release of certain cellular components and shortened cell lifespan. The damage may be caused by direct contact with the solid surface, or may be due to physical forces applied to the cells. The extent of the latter damage depends on the magnitude of the shear stress to which the cells (in particular erythrocytes) are exposed, and not on the nature of the flow conditions (whether laminar or turbulent). The embodiment of the device 1 'conceived is such that the device 1 ' does not trigger damage phenomena, i.e., in the operating fluid dynamics state, it is still in the "sub-hemolytic" zone (shear stress is much lower than 100 Pa), i.e., in a region where the interaction effects of the material that is in contact with the blood surface alone dominate. To this end, the compression element 18a is provided with a defined geometry that allows a gentle squeezing of the tube 25'; essentially, the compression element 18a compresses the central portion 25a' of the tube 25' (the portion of the tube 25' that is displaced at the base 19) to obtain a flat reading area and defines a smooth transition between the uncompressed portions of the tubes 25b', 25c' defining the central compressed portion 25a'; in this regard, see Fig.15 , which shows a section of tube 25' engaged at the base 19 and partially compressed. Fig.15It is shown how compression of the tube 25' occurs at its upper surface in order to obtain a flat reading area; the surfaces that are flattened are both the upper and lower surfaces of the tube. In order to ensure optimal flattening of the tube 25' without compromising the properties of the blood (avoiding the phenomenon of hemolysis), the compression element 18a comprises a flat portion 18a' which is defined between relative curved portions (in particular the ends 18a", 18a'") which may present extensions similar or similar to each other; the flat portion 18a' is suitable for obtaining a flat reading area. The flat portion 18a' is substantially rectilinear and may have a length L18 and a depth F; the depth F may correspond to the compressed entity of the tube (or the extruded entity, i.e. the difference between the diameter D1 of the undeformed tube and the diameter of the deformed tube or the equivalent diameter D2). The depth F may be 2 mm. The length L18 is preferably between 10 mm and 50 mm, in particular between 12 mm and 45 mm. Fig.15 In the embodiment shown in , the length L18 is 30 mm; in another embodiment, the length L18 can be 15 mm. The tube 25' has a diameter D1 (undeformed diameter), a diameter or equivalent diameter D2 (compressed diameter), and a thickness T1 (wall thickness). Compression of the central portion 25a' of the tube 25' by the compression element 18a causes the cross-sectional area of ​​the blood channel to be reduced by between 9% and 17% (caused by the reduction in the diameter of the tube 25' at the length L18). For example, if the tube 25' has an undeformed diameter D1 of 14.3 mm, the reduction in the cross-sectional area of ​​the blood channel can correspond to a compression of the tube by approximately 14%, where the corresponding compressed equivalent diameter D2 of the tube 25' (D2 can be defined as the compressed equivalent diameter because the shape of the tube after compression is no longer circular) is 12.3 mm. It should be noted that in Fig.15 In an embodiment, the diameter D1 of the tube 25' is equal to 9 / 16" (dimension in inches, corresponding to approximately 14.3 mm), the thickness T1 is equal to 3 / 32" (dimension in inches, corresponding to approximately 2.38 mm), and the equivalent diameter D2 is approximately equal to 12.3 mm; therefore, the compressed solid of the tube 25' is approximately 2 mm (the difference between the undeformed diameter and the deformed diameter, as described above, is equal to approximately 14%). Fig. 6A and Fig.15 The squeezing effect visible in has a dual purpose: locking the mutual position between the device 1' and the tube 25'; and creating a substantially flat or horizontal surface on which to perform the measurements. As shown in the figures, the compression element may comprise a protrusion 18a extending in a main longitudinal direction and provided with the above-mentioned profile.

[0343] The device 1' may also include a closing portion 21, which is capable of cooperating with the box body 2 to maintain the covering elements 18, 18' in an operating configuration (closed position). The closing portion 21 may be integral with the covering elements 18, 18' and / or may have opposing operating ends 21a, 21b. Each operating end 21a, 21b is provided with a base, which is preferably defined on the outside of the operating end 21a, 21b; the base may be a groove, for example a groove with a curved profile. The closing portion 21 may extend longitudinally parallel to the compression element 18a. In the illustrated embodiment (see Figure 5 and Figure 5A ), the closing part 21 is integral and has opposite operating ends 21a, 21b and is connected by a middle part 21c. The closing part 21 is easy to clean because it has no recesses that are difficult to access and / or difficult to clean.

[0344] The device 1' preferably includes at least one constraint element 22, 23, which is configured to keep the covering element 18, 18' in the operating configuration; in essence, the constraint element 22, 23 allows the covering element 18, 18' to remain closed when it is in the operating configuration (during parameter measurement). As shown in the accompanying drawings, the device 1' preferably includes two constraint elements 22, 23, which are configured to keep the covering element 18, 18' in the operating configuration. The constraint elements 22, 23 are engaged to the box body 2 near the upper surface of the box body 2 and at corresponding positions 2f defined on the box body 2. The two constraint elements 22, 23 are opposite to each other to ensure a stable and symmetrical closure; in addition, the symmetry of the closure helps to ensure that the defined measurement surface at the reading area is flat. In order to achieve a stable closure of the covering elements 18, 18', the constraint elements 22, 23 are preferably opposite to the hinged part 20. As shown in the figures, each constraining element 22, 23 can provide, for example, a movable element 22a, 23a with a curved head (for example, at least partially spherical); when closed, each movable element 22a, 23a can be moved by the relative operating end 21a, 21b of the closing part 21; such movement determines the engagement of the constraining elements 22, 23 with their respective bases. The engagement is preferably determined by accommodating the curved head 22a, 23a of each constraining element 22, 23 in the curved groove of the corresponding operating end 21a, 21b of the closing part. The movement of the constraining elements 22, 23 preferably occurs with a mutual movement away from each other. Each movable element 22a, 23a can be moved at least in the direction determining the constraint with the closing part 21 (external direction, moving away from each other), in particular, each movable element 22a, 23a can be moved in multiple directions, preferably substantially in each direction. The movable elements 22a, 23a are substantially sealed; therefore, providing closure of the covering elements 18, 18' by means of the substantially sealed movable elements 22a, 23a and by means of the easily cleanable closure part 21, in addition to a stable closure, also ensures a high degree of cleanability of the device 1'; this is obviously advantageous from a hygiene point of view.

[0345] The box body 2 also has a device 24 for connection (preferably a connector) to achieve connection with the cable 16; the cable 16 enables the digital data processed by the control unit 3 to be transmitted to the medical machine 90', 90" and / or the display device 91. The cable 16 also realizes the power supply of the device 1'. As shown in the accompanying drawings, the device 24 for connection can be arranged in the vicinity of the hinged part 20, in particular at the side of the box body 2 and below the hinged part 20.

[0346] The device 1' may also include an additional temperature sensor 12', which may preferably be a thermistor (see Figure 7). The additional temperature sensor 12' is configured to measure the temperature of the electromagnetic radiation detection member 9 to monitor the temperature of the electromagnetic radiation detection member 9 (InGaAs photodiode) so as to be able to correct the optical responsiveness to varying temperatures. The additional temperature sensor 12' is in contact with the electromagnetic radiation detection member 9 and with the area confined thereto.

[0347] Geometrically, the device 1' has a length L, a width T and at least one height, in particular a minimum height H1 and a maximum height H2; in the illustrated embodiment, the width T is substantially equivalent to the length L (essentially a square profile). The width T and the length L are dimensions with respect to the box body 2, while the heights H1, H2 are dimensions defined by the fit between the box body 2 and the covering elements 18, 18'. The box body 2 preferably has the following configuration so that the necessary components of the device 1' can be accommodated therein with as small a volume as possible. As shown in the accompanying drawings, the heights H1, H2 of the device 1' are a function of the configuration presented by the covering elements 18, 18'; therefore, the box body 2 can present a minimum height H1 corresponding to the use state of the device 1' (the height of the covering element in the operating configuration) and a maximum height H2 corresponding to the rest state of the device 1' (the height of the covering element in the rest configuration). The length L can be between 40 mm and 80 mm. The width T measured orthogonally to the length L can be between 40 mm and 85 mm. The minimum height H1 measured orthogonally to the length L and the width T may be between 30 mm and 60 mm. In essence, the device 1 'has the following dimensions, making it substantially pocket-sized. As a non-limiting example, it should be noted that in a preferred version of the second embodiment, the length L may be 64 mm, the minimum height H1 may be 44 mm, and the width L may be 67 mm; the length L and the width T being similar values, the box body 3 may be substantially square in shape (see Figures 4 to 7 ). In a second embodiment, the volume of the device 1' is defined by the box 2 and the covering elements 18, 18'; in the volumes described below, reference will be made to the volume of the device 1' calculated using the minimum height value H1 as the height value. The volume of the device 1' may be included in the volume calculated according to the extreme values ​​of the above-mentioned size interval. In particular, in a possible embodiment, the volume of the device 1' may be between 48,000 mm 3 With 408,000mm 3 Between, preferably between 75,000mm 3 With 330,000mm 3 More specifically, between 120,000 mm 3 With 260,000mm 3 or more specifically between 150,000 mm 3 With 220,000mm 3The above-mentioned quadrilateral configuration, in particular the substantially square configuration, is the minimum configuration that encompasses the physical dimensions of the components (including the restraining elements 22, 23 and the closing portion 21).

[0348] The advantage of this embodiment is that it avoids the use of components that increase the cost of using the device 1 ′, namely the test tube 25; moreover, by coupling the device 1 ′ directly to the tube 25 ′ of the extracorporeal blood circuit, the device 1 ′ allows monitoring of blood parameters “on the go”, i.e. when a medical procedure has begun.

[0349] Fig.16 , Fig.17 and Fig.18 Shown according to Figures 4 to 7 and Fig.15 18' of an alternative variant of the covering element 18 shown in . According to this variant, the covering element 18' provides a compression element 18o, which is configured to oscillate relative to the body 18b of the covering element 18' (to tilt the compression element). Apart from the ability to oscillate, the compression element 18o can have the same properties as the previously described compression element 18a, in particular the same configuration, and can therefore have parts 18a', 18a", 18a'" (see Fig.16 ).

[0350] The development of the oscillating compression element 180 supports the need to achieve a blood passage cross section that is as smooth as possible. By providing the possibility of adapting the compression of the tube 25', the oscillating compression element 180 makes it possible to increase the repeatability of the measurements that are essentially optical, which should be completely unaffected by the non-fixed and repeatable positioning of the tube 25' in front of the optical component protection window previously described.

[0351] In order to allow oscillation of the compression element 18o, the cover element 18' comprises a pin 18c housed in a corresponding seat 18d defined on the body 18b. Fig.16 , Fig.17 and Fig.18 A base 18d is shown which is inversely shaped to the circular profile of the pin 18c. As shown in these figures, the pin 18c may be made integrally with the oscillating compression element 18o; in an alternative embodiment, they may form two separate but coupled components. Fig.16 , Fig.17 and Fig.18 As shown, in order to retain the pin 18c, the base 18d can substantially define an undercut; with respect to the assembly of the compression element 18o to the body 18b, the pin 18c can be inserted into the base 18d (see Fig.16). Obviously, instead of the pin 18c, a device such as a kinematic device (for example, a cam or a parallelogram kinematic device) can be provided to allow the oscillation of the compression element 18o relative to the body 18b or to allow the oscillation of the entire covering element 19' relative to the housing 25'. The oscillating compression element 18o is rotationally pulled by the body 18b of the covering element 18'. Although being rotationally dragged by the covering element 18', the compression element 18o can assume an angular position according to angles β', β" different from the angles α, α', α" of the covering element 18' due to the pin 18c. In particular, the two rotation configurations of the covering element 18' are Fig.17 and Fig.18 As can be seen in the figure, the body of the covering element 18' is positioned according to corresponding angles α', α", while the compression element 18' presents corresponding and different angles β', β". The angles β', β" described by the compression element 18o, although different from the angles α, α', α" described by the body 18b of the covering element 18', are still related thereto, because certain angular positions of the covering element 18' correspond to corresponding angular positions of the compression element 18o. The device 1' may provide that: the body 18b of the covering element 18' rotates relative to the hinged part 20 according to a first rotation direction (for example, clockwise), and the compression element 18o rotates according to the first rotation direction and a second rotation direction opposite to the first rotation direction (for example, counterclockwise); this is possible because the compression element 18o is inclined and is therefore configured to adapt its rotation direction to the relative position between the covering element 18' and the tube 25' by oscillating around the pin 18c.

[0352] Continuing to close the cover element 18' in the direction of the cassette 2, the compression element 18o does not compress the tube 25' (see Fig.17 ) is transformed into a configuration in which the compression element 18o contacts the tube 25' and oscillates and then gradually rotates as the angles α, α', α" decrease. Therefore, according to this variant, it can be advantageously envisioned that the compression element 18o gradually adapts to the configuration of the tube 25' and gently compresses the central portion 25a' of the tube 25' in a manner that does not significantly change the blood flow within the tube 25'. Fig.18 The configuration of the cover element 18' before complete closure on the box body 2 is shown; in this figure, it can be seen that the central portion 25a' of the tube is not yet completely compressed and therefore the reading area is not yet flat. Thus, despite the fact that the cover element 18' is simply hinged to the box body 2, the access of the compression element 18o can be obtained in an almost parallel manner.

[0353] Control unit of the device

[0354] The device 1 , 1 ′ further comprises a control unit 3 . The logic and characteristics of the control unit 3 described below are common to both embodiments of the device 1 , 1 ′ described above. The control unit 3 is accommodated in the box 2 .

[0355] Due to the training described below, the control unit 3 preferably enables the device 1, 1' to perform the measurement of the parameter without the need for an initial calibration. In addition, the control unit 3 preferably enables the device 1, 1' to perform the measurement without any interaction with a processing or computing unit external to the device 1, 1'; therefore, the processing of information and data for measuring a plurality of blood parameters is performed autonomously by the device 1, 1'. In this way, no database external to the device 1, 1' is required; the device 1, 1' provides all the components for performing the measurement of the parameter of interest within the box 2. With the help of the control unit 3, the device 1, 1' is able to perform the measurement of the parameter based on a plurality of data of previous measurements of the blood parameter performed during a previous training. This is possible because the control unit 3 can provide an encoded computational model. Preferably, the training is performed in a laboratory (e.g., by the supplier of the device 1, 1' upon request of the end user) on a series of devices 1L, 1L' to be trained before supplying the device 1, 1' in question. As will be discussed further below, the previous training gives rise to learning, in particular machine learning, which the control unit 3 takes into account when measuring the parameter during clinical use. Each device 1, 1' produced after training a certain number of devices 1L, 1L' (this aspect will be described in detail below) is calibrated in the laboratory so that the end user of the device 1, 1' has an instrument that is immediately ready for use. Since there is no initial calibration, the device 1, 1' is particularly suitable for use in emergency situations. Examples of emergency situations are those situations that require treatment with an extracorporeal membrane oxygenation machine (ECMO or extracorporeal membrane oxygenation machine), i.e. cardiocirculatory support, such as in patients suffering from cardiac arrest or lung trauma, in the most acute and otherwise untreatable Covid-19 disease stages (Coronavirus disease 19 caused by the SARS-CoV-2 virus, also referred to as "Covid" hereinafter).

[0356] In operation, the control unit 3 is configured to implement at least the following operations:

[0357] o controlling the excitation means 4, 5 during an excitation step in which the excitation means 4, 5 excite the blood flow with electromagnetic radiation of a plurality of determined wavelengths,

[0358] o receiving analog information (read by electromagnetic radiation detection means 8, 9) relating to a plurality of electromagnetic and / or optical responses of the blood, the plurality of electromagnetic and / or optical responses comprising electromagnetic radiation, in particular light, reflected or diffused by the blood,

[0359] o converting electromagnetic and / or optical response analog information into electromagnetic and / or optical response digital data,

[0360] o processing the digital electromagnetic and / or optical response data and the actual blood temperature value via one or more neural networks NN,

[0361] o As a result of the processing operations performed using one or more neural networks NN, a value for each of a plurality of blood parameters is determined.

[0362] In order to perform these operations, the control unit 3 is operatively connected to the excitation means 4 , 5 and to the electromagnetic radiation detection means 8 , 9 .

[0363] Continuing to a higher level of detail regarding the processing and measurement of parameters by the apparatus 1 , 1 ′, it is indicated that the control unit 3 is configured to:

[0364] oDetermination of multiple ratios (optical count ratios),

[0365] o providing a plurality of ratios and temperature values ​​as input to one or more neural networks NN,

[0366] o processing of a plurality of ratios by means of one or more neural networks NN taking into account a plurality of data from previous measurements of blood parameters carried out during previous training (laboratory training),

[0367] o As an output from one or more neural networks NN, values ​​for each of a plurality of blood parameters are provided.

[0368] Each ratio is defined between an amount indicating radiation reflected or diffused by the blood due to excitation of a determined wavelength (counts calculated from radiation received by a photodiode) and an amount indicating radiation reflected or diffused by the blood due to excitation of another determined wavelength (counts calculated from radiation received by the same photodiode). For example, Fig.10 The x-axis of the graph in shows the "Ratio 805InGaAs / 1450", which is the ratio of optical counts to the radiation read from the InGaAs photodiode 9 in accordance with excitation radiation at 805 nm (back-reflected radiation) (the numerator of the ratio) and the radiation read from the same InGaAs photodiode 9 in accordance with excitation radiation at 1450 nm (back-reflected radiation). To provide a further example, Fig.11 Indicated on the horizontal axis of the graph in is the “Ratio 805 / 660”, which is the ratio of the optical counts relative to the radiation (back-reflected radiation) read from the silicon photodiode 8 for excitation radiation at 805 nm (the numerator of the ratio) and the radiation (back-reflected radiation) read from the same silicon photodiode 8 for excitation radiation at 660 nm.

[0369] Basically, the control unit 3 receives as input the actual temperature value of the blood and the simulated electromagnetic and / or optical response information from the photodetectors 8, 9, processes it based on previous training, and measures the parameter value. The actual blood temperature value is measured by the temperature sensor 12, which detects the temperature of the blood flowing into the container 25, 25'. Regarding the input of the temperature value, it is useful to indicate that it is at least because as the temperature changes, the excitation member has a different emission state of electromagnetic radiation, and the response of the blood to the excitation electromagnetic radiation changes as a function of temperature.

[0370] The control unit 3 is configured to control the plurality of excitation members 4, 5 during an excitation step, which comprises: activating the plurality of excitation members 4, 5 according to a specified time sequence. For example, the control unit 3 may activate the plurality of excitation members 4, 5 by activating the LED elements one at a time according to the following time sequence, so that the blood is excited: activating the LED element causing excitation at 660 nm, then activating the LED element causing excitation at 805 nm; then activating the LED element causing excitation at 1450 nm; and finally activating the LED element causing excitation at 1550 nm; and also activating the other LED elements of 525 nm, 940 nm and 1050 nm. In essence, it should be understood that following a specific time sequence as described above is not necessary for the measurement of multiple parameters and is therefore not limiting in any way; therefore, the control unit 3 may work by exciting the blood according to other excitation sequences.

[0371] Preferably, the control unit 3 is configured to activate the first excitation member 4 and the second excitation member 5, in particular alternately, to stimulate blood flow in particular alternately at a first wavelength or a first plurality of wavelengths selected from a first plurality of wavelengths (525nm, 940nm, optionally 1050nm) and in particular alternately at a second wavelength or a second plurality of wavelengths selected from a second plurality of wavelengths (660nm, 805nm, 1450nm). Preferably, the control unit 3 is configured to activate the first excitation member 4 and the second excitation member 5, in particular alternately, to stimulate blood flow in particular alternately at the first plurality of wavelengths and the second plurality of wavelengths.

[0372] The control unit 3 is configured to control at least one excitation member or both excitation members 4, 5 during an excitation step in which the blood flow is excited one wavelength at a time.

[0373] In more detail, the control unit 3 is configured to activate (turn on) each LED element in a timed manner so that within a given reference time unit (e.g. 1 second), there are tens of measurements of the optical count of transmission for each LED element (each light source). In particular, the control unit 3 is configured to activate each LED element in a timed manner according to the timing described previously so that within 1 second there are 20 measurements of the optical count of transmission for each LED element. As the number of measurements within a given time unit increases, the influence of noise on the measurement has a smaller weight (the acquired measurements can be averaged), and therefore the accuracy of the measurement is improved because the data used is more statistically reliable.

[0374] Essentially, the control unit 3 constitutes a member of the device 1 , 1 ′ equipped with artificial intelligence (based on machine learning described below) and configured to detect the value of each parameter that the artificial intelligence considers to correspond to a previously determined ratio based on a plurality of data from previous measurements.

[0375] In particular, the control unit 3 comprises artificial intelligence information, such as one or more matrices usable by said one or more neural networks, which is encoded in its firmware and enables the calculation of a plurality of parameters with the aid of the one or more neural networks NN and with the aid of a given calculation model.

[0376] In addition to the information encoded in the firmware, the control unit 3 is also configured to provide, as input to one or more neural networks, reference values ​​of a first parameter to be measured (oxygen saturation) and a reference value of a second parameter to be measured (hematocrit). These reference values ​​are acquired during a calibration step prior to use of the device 1, 1' and stored in a memory of the device (e.g., an EEPROM memory); the memory may be part of the control unit 3. Such calibration prior to use of the device 1, 1' may be performed by the manufacturer of the device 1, 1', for example in a laboratory; in any case, the calibration is not performed by the end user.

[0377] The information encoded in the firmware, in particular the artificial intelligence information, is identical for each device 1, 1' produced, whereas the aforementioned information stored in the memory may vary between devices 1, 1' since it takes into account the variability of each individual device 1, 1' by means of a calibration performed in the laboratory, which may be associated with the variability of the hardware, components and their specific, even infinitesimal, geometrical position within the housing of the device.

[0378] When measuring oxygen saturation and / or hematocrit, the control unit 3 may also be configured to take into account values ​​related to blood flow, in particular volumetric blood flow rate. As described below, the control unit 3 may perform this operation by correcting the measured values ​​of oxygen saturation and / or hematocrit according to blood flow.

[0379] In terms of components, the control unit 3 may include at least a microprocessor MP and an analog-to-digital converter (ADC). The microprocessor MP may determine and supervise the execution of the above operations, and the ADC performs the operation of converting analog information into digital data. The artificial intelligence information is encoded in the firmware of the microprocessor MP. The control unit 3 may include the previously mentioned memory, such as an EEPROM memory.

[0380] The control unit 3 may also comprise an analog "AFE" front end (AFE chip), i.e. a system integrating the analog technologies required to achieve an optimal interface with an analog / digital converter. This interface fundamentally involves the adaptation of the analog signal captured by the sensor (photodetector) to the functional specifications of the analog / digital converter regarding amplitude dynamics (amplification) and frequency bandwidth (anti-aliasing filtering), i.e. regarding the principle of sampling analog signals. In fact, the analog front end performs complex analog signal processing generally called "conditioning", most of which strictly depends on the application and the nature of the sensor (photodetector).

[0381] The control unit 3 may also include at least one printed circuit board assembly PCB1. In essence, the printed circuit board assembly PCB1 is a board filled with, i.e., having arranged thereon, certain electronic components that enable the assembly to perform its intended function or functions. As shown in the accompanying drawings, the control unit 3 may include a first printed circuit board PCB1 for the microprocessor MP and a second printed circuit board PCB2 for the analog front end. Figure 7 As shown in FIG. 1 , a restraining element 2 e , in particular a screw element, may be provided to fix the printed circuit board assembly PCB1 , the exciting components 4 , 5 and the electromagnetic radiation detecting components 8 , 9 .

[0382] The values ​​of oxygen saturation and hematocrit are calculated essentially at the same moment by the artificial intelligence of the device 1, 1'; therefore, the measurement of these parameters is essentially instantaneous. In addition, the output parameters (hematocrit, saturation, hemoglobin) are provided by the device 1, 1' via the control unit 3 at the same moment; the temperature data measured without the assistance of artificial intelligence are also synchronized with the other parameters and output together. Therefore, the end user of the device 1, 1' has all the parameters measured by the device 1, 1' available at the same time and, in particular, these parameters can be displayed on a user interface such as a display device 91 that is operatively connected or connectable to the device 1, 1'. The parameters can be made available at a certain rhythm (for example, every second, every 5 seconds, every 10 seconds) or continuously and additionally or alternatively on demand (for example, at the request of the medical machine 90', 90").

[0383] In a second embodiment of the device 1 ′, according to which the device can be associated with a tube 25 ′, the control unit 1 is configured to perform the following operations:

[0384] - the type of detection tube 25',

[0385] - Preparation for measurement based on the detected type of pipe 25'.

[0386] Among the properties of the tubes 25' that may differ, it is important to note their color or hue, as it may modify the response of the blood to the electromagnetic excitation. The tubes 25' are typically plasticized polyvinyl chloride (PVC), or at least based on plasticized polyvinyl chloride, and are not completely transparent. Furthermore, the sterilization to which the tubes 25' may be subjected before their use may also result in changes in the color or hue of the tubes 25'. It should be remembered that if the actual color or hue of the tubes 25' on which the measurements are made is not properly taken into account, the blood response received by the electromagnetic radiation detection member will be incorrect (for example, because the device 1, 1' may expect a transparent container, while in fact the tube is not a transparent container, thus modifying the electromagnetic response "recorded" by the electromagnetic radiation detection member), and therefore the measurement of the desired parameter will not be reliable.

[0387] The detection of the tube 25' type then involves the detection of the tube 25's color or hue; the control unit 3 then prepares itself for measurement based on the detected color or hue of the tube 25', for example by adapting the measurement mode of one or more parameters to the detected container or tube 25' type.

[0388] The detection of the color or hue of the tube 25' is performed before blood stimulation, in particular before blood flows into the tube 25'. When the color or hue of the tube 25' is detected, fluid other than saline or blood preferably flows into the tube 25'.

[0389] The control unit 3 is configured to perform an operation of preparing for measurement, based on the color or hue detected for the tube 25', by selecting a given matrix from a plurality of matrices available to the neural network NN; each matrix corresponding to a given color or hue of the tube 25'. This selection operation can be performed by consulting a memory of the device 1, 1' in which information relating to the plurality of matrices is stored.

[0390] The technical features disclosed herein relating to the functionality of the device 1, 1' or its parts / components / elements, in particular to the operation of the control unit 3, are applicable in the context of the corresponding use of the device or method steps described below and can therefore be used to describe such use and methods in the attached claims.

[0391] Machine learning setup based on one or more neural networks

[0392] The studies were conducted by extensive ex vivo laboratory tests, i.e. using extracorporeal blood prepared and circulated in a setup 50, wherein the conditions of the blood and the environment were altered to simulate essentially all possible conditions to which the device 1 according to the invention might be subjected in clinical use (see Fig. 9 ). The simulated conditions include blood temperature, blood flow rate, ambient temperature, blood oxygen saturation, and hematocrit. The following describes a setup 50 for testing and training a device in a laboratory to determine an automatic learning model. In the learning tests performed, the environmental conditions were not changed; therefore, the measurements were taken at room temperature. However, it should be noted that, in general, the ambient temperature can be changed; this requires using the setup 50 in an environmental chamber.

[0393] Set 50 in Fig. 9 , and comprises two circuits along which the extracorporeal blood circulates. In more detail, there is a blood preparation circuit 51 and a main blood circuit 52, in which the blood suitably prepared by means of the blood preparation circuit 51 circulates in order to perform parameter measurements. Fig. 9 The loops 51, 52 are depicted differently in FIG. by means of corresponding and different lines. Fig. 9 The main blood circuit 52 is shown in the upper part of Fig. 9 The left side extends to the lower part of the figure), while Fig. 9 In the lower part of the , a blood preparation circuit 51 is shown. In the blood preparation circuit 51 and the main blood circuit 52, blood circulates through corresponding processing components 53, 54; in particular, for this purpose, the arrangement shown provides a main centrifugal pump 54 for the main blood circuit 52 and a secondary centrifugal pump 53 for the blood preparation circuit 51.

[0394] The blood preparation circuit 51 allows blood to be prepared in a desired state. To this end, a blood oxygenator 55 and a plasma filter 56 arranged along the blood preparation circuit 51 are provided.

[0395] The plasma filter 56 is a component whose function is to extract plasma from the blood so that the blood can be "concentrated" to increase the hematocrit value. The plasma filter 56 is connected to the extracted plasma discharge container 57, and the plasma removed by the plasma filter 56 is poured into the container 57. The removal of plasma occurs due to the pressure gradient established between the blood preparation circuit 51 and the extracted plasma discharge container 57; in order to interrupt the removal, the line connecting the plasma filter 56 and the extracted plasma discharge container 57 is closed.

[0396] As for the blood oxygenator 55, it is a component of the arrangement 50, which has a function in the blood preparation circuit 51 that is equivalent to the function of the human lungs; as described below, it is configured to oxygenate the blood, remove carbon dioxide, and also heat or cool the blood. To this end, the blood oxygenator 55 includes a heat exchanger and is connected to a thermostatic bath 58. The arrangement 50 also includes a water pump 59, which is configured to circulate water between the thermostatic bath 58 and the blood oxygenator 55. The thermostatic bath 58 is configured to heat or cool water, and the water circulates in the heat exchanger in the blood oxygenator 55 with the help of the water pump 59, so that the thermostatic bath 58 cools or heats the blood to the desired temperature. In order to change the saturation state of the blood and in particular the oxygen saturation state of the blood, the blood oxygenator 55 is connected to a gas mixer 60, which is in turn connected upstream to an oxygen tank 61, a nitrogen tank 62 and a carbon dioxide tank 63. The gas mixer 60 is configured to mix the gases from the three tanks 61, 62 and 63 just described in different proportions. As Fig. 9 As shown, the tanks may be in the form of gas cylinders 61 , 62 , 63 configured to deliver respective gases to a gas mixer 60 .

[0397] The blood preparation circuit 51 and the main blood circuit 52 share common elements including a blood container 64, which is Fig. 9 53 is shown as a blood bag. The blood container 64 acts as a reservoir from which blood is extracted by the secondary centrifugal pump 53, and since the blood preparation circuit is a closed circuit, it returns the appropriately prepared blood to the reservoir. The blood container 64 is connected to an isotonic solution source 65, which is configured to provide an isotonic solution. Since the isotonic solution does not contain red blood cells, its supply to the blood container 64 allows the blood contained therein to be diluted. The isotonic solution (on the other hand, called physiological saline) enters the main blood circuit 52 by gravity and performs an action opposite to that performed by the plasma filter 56. The blood container 64 can feed the main blood circuit 52 with appropriately prepared blood from the blood preparation circuit 51 just described by means of the main centrifugal pump 54.

[0398] The arrangement 50 also provides a plurality of devices 1L, 1L' (wherein "L" stands for learning, indicating that these devices are subjected to automatic learning) along the main blood circuit 52, which are associated with corresponding containers 25, 25' of the above-mentioned type. In more detail, device 1L is of a type corresponding to the first embodiment (and is associated with a corresponding test tube 25), while device 1L' is of a type corresponding to the second embodiment (and is associated with a tube 25'), and differs from devices 1, 1' in that devices 1L, 1L' are subjected to machine learning to develop artificial intelligence (computational methods, firmware, related logic and code) that will then be implemented in devices 1, 1', while devices 1, 1' are already equipped with artificial intelligence and only require simple calibration in the laboratory before they can be delivered and used. The containers associated with devices 1L, 1L' are respectively test tubes 25 and tubings 25' arranged along the main blood circuit 52. The tubings 25' have different formulations from each other (and therefore usually have different colors) and are connected by means of connecting elements 77 (see Fig. 9 ); the color and / or hue of the individual tubing 25' depends on the chemical formulation of the tube material and / or the tube sterilization process. A plurality of devices 1L' connected to tube portions of the same type (i.e. the same formulation and the same color or hue) may be provided; in this case, a single tube of the same color or hue may be provided instead of the individual tubing. It is provided that it is advantageous for the device 1L' to learn to recognize and distinguish between tubes of different types, colors and / or hues in order to provide accurate measurements of the parameters; the artificial intelligence derived from the machine learning provides the device 1' with the necessary information to prepare it to recognize tubes of different types, colors and / or hues and to adjust the measurements accordingly. The test tubes 25 and tubing 25' respectively coupled to the devices 1L, 1L' are of disposable type. Fig. 9 The settings 50 in allow the probe to learn the behavior of different tube types. For learning experiments, the same color tubes 25' can be set at the same time (i.e., in the same learning session or round) to collect more statistically valid data; then, the type of tube is changed (replacing the tube 25' with another different color tube) and more data is collected, etc. to have a large number of example history for training the neural network.

[0399] Thus, the device 1L, 1L' provided by the arrangement 50 is a photoelectric probe configured to excite blood and detect the associated electromagnetic and / or optical response; its photoelectric excitation unit (excitation means) and its electromagnetic and / or optical response detection means (electromagnetic radiation detection means) are similar to those previously described with reference to the corresponding embodiments of the devices 1, 1'. Essentially, the devices 1L, 1L' of the arrangement are of the type previously described, except for the fact that they lack an algorithm capable of autonomously measuring blood parameters in clinical use, since this algorithm has been precisely developed by the training to which the devices 1L, 1L' have been subjected as described herein. In more detail, the devices 1L, 1L' and the devices 1, 1' differ in their calibration coefficients, which are described below. After preparation of the devices 1L, 1L' by means of learning, which is precisely performed by learning in a laboratory, the devices 1L, 1L' will be configured to calculate the values ​​of the parameters for all operating states expected to occur during the clinical use of the devices 1, 1' using a suitably trained neural network. It is reiterated that this learning / calibration is carried out in the laboratory, i.e. before the clinical use of the devices 1, 1'. As Fig. 9 As shown, the test tubes and corresponding devices 1L may be two in number, and the devices 1L' may be three in number; it should be understood that a plurality of devices 1L and corresponding test tubes greater than two in number, and / or a plurality of devices 1L' greater than three in number may be provided. In an embodiment, at least five devices 1L and at least five devices 1L' may be provided. As the number of devices 1L, 1L' increases, the amount of "raw" data collected also increases, and includes greater variability between devices.

[0400] The device 1L, 1L' during learning "interrogates" the blood by emitting electromagnetic radiation of different wavelengths and detects the response in the electromagnetic radiation by deriving calibration curves as described below, in particular at least one calibration curve for hematocrit at fixed oxygen saturation values ​​and at least one calibration curve for oxygen saturation at fixed hematocrit values.

[0401] The arrangement 50 also includes a blood flow meter 66 and a temperature detector 67 along the main blood circuit 52; the temperature detector 67 serves as a temperature verification and reference instrument.

[0402] The blood flow meter 66 is configured to measure the blood flow rate in the main blood circuit 52. The blood flow meter 66 can measure the blood flow rate downstream of the main centrifugal pump 54. Fig. 9 , a blood flow meter 66 is shown connected to the main centrifugal pump 54.

[0403] The temperature detector 67 is configured to measure the temperature of the blood, preferably in the vicinity of a container coupled to a corresponding device. Fig. 9In FIG. 6 , the temperature detector 67 is shown in the form of a digital thermometer having a temperature sensor immersed in the blood and arranged upstream and in the vicinity of a container coupled to the device 1L, 1L′.

[0404] The main blood circuit 52 also includes a blood sampling point 68 and a reference blood analyzer 69. The blood sampling point 68 is a point in the circuit at which an extracorporeal blood sample is collected; the collected sample is analyzed by a reference blood analyzer 69, which is a reference instrument to which the "raw" measurements of the parameters obtained from the plurality of devices 1L, 1L' (and the related "raw" data) are referenced. The reference blood analyzer 69 is a clinically accepted reference instrument and therefore provides measurements of the parameters with the necessary accuracy. In essence, the reference blood analyzer 69 provides reference values ​​of the parameters under the specific conditions (actual blood temperature and, for example, blood oxygenation) in which the measurements are made. In the tests performed, the blood reference analyzer 69 measures at least the oxygen saturation value of the blood, the hemoglobin concentration of the blood and the hematocrit value. Fig. 9 The blood reference analyzer shown in FIG. 6 is a laboratory blood chemistry analyzer 69 .

[0405] use Fig. 9 The setting 50 is then changed to the previously indicated state and, for each change, the response of the electromagnetic radiation resulting from the excitation of the blood is acquired. Thus, the ratio between the optical counts is acquired and a calibration curve is generated; the curve thus acquired is digitized by means of components built into the device 1L, 1L'.

[0406] In this way, analog information of the electromagnetic and / or optical response of the blood to a given change of state is transformed into digital data that can be processed. The value of the parameter corresponding to the changed state, the analog information and the corresponding digital data are associated with the value of the parameter measured by the reference blood analyzer (in the tests performed, oxygen saturation and hematocrit). By extensively simulating every possible state that may occur in clinical use, thousands of associated parameters / digitized curve data are obtained for each of the states set up in the laboratory, covering all states for which you want the device 1, 1' to measure the parameter (oxygen saturation and hematocrit).

[0407] The arrangement further comprises a data collector 70 and a data processor 71. The data collector 70 is connected to the plurality of devices 1L, 1L' to collect data from the devices 1L, 1L'; the collected data is digital data related to the electromagnetic and / or optical response of the blood. Fig. 9 As shown, for collecting data, the data collector 70 may have a plurality of input terminals corresponding in number to the devices 1L, 1L' of the arrangement 50, and at least one output terminal for connection to a computer 71. The data collector 70 is essentially an electronic instrument; Fig. 9The data collector used in the experiment shown is a multiplexer 70. After collecting the data, the data collector 70 transmits the data to the processor, which Fig. 9 Shown in FIG. 7 is a personal computer 71. The computer 71 collects "raw" data from the devices 1L, 1L' for all operating states intended to be tested; this "raw" data constitutes the data set used to train the neural network.

[0408] The arrangement 50 may also provide an external archive 72 connected to the computer 71 and configured to store data received from the computer 71. In essence, the external archive is an electronic archive 72, which may be remote, for storing data.

[0409] Having described a setup 50 for implementing the training, a machine learning method based on a neural network NN is now described.

[0410] Machine learning methods based on one or more neural networks

[0411] If with the help of Fig. 9 As mentioned in the settings of , the conditions of temperature, oxygen saturation and blood hematocrit can be changed in discrete steps in all expected measurement ranges (13% to 55% for blood hematocrit, 35% to 99.9% for blood oxygen saturation, 9°C to 42°C for blood temperature), as well as each of the multiple combinations of these parameters. The process of learning and designing a device for correctly measuring Sat% and Hct% is completed by performing the following steps.

[0412] 1) Perform the setting of the gain, intensity of the individual LED elements, i.e. the maximization of the optical counts of each of the individual LED elements in the conditions at the extreme values ​​of the measurement range of Hct% and Sat%. The gain is the ability of the photodiode to convert the light signal into a current signal, while the optical count is an indicative measure of the signal transmitted through the tube and detected by the photodiode. The maximum count value is about 4000, beyond which the signal saturates and therefore no longer indicates the measurement; therefore, in extreme measurement conditions (i.e. for the measurement of parameters at the end of their measurement range), it is necessary to set the counts so as not to exceed this maximum threshold.

[0413] 2) The calibration curve for Hct% is constructed by means of a "best fit" technique, having a 4th order curve of the ratio of the values ​​of optical counts of electromagnetic radiation read by an InGaAs photodiode 9 for excitation at 805 nm to the values ​​of optical counts of electromagnetic radiation read by the same photodiode 9 for excitation at 1450 nm, relative to those obtained by the reference instrument. The oxygen saturation value Sat% of the blood is kept fixed at 80%, the blood temperature is kept at 37°C and the hematocrit value is increased in steps of 3 percentage points within the range of 12Hct% to 55Hct% (see Fig.10 ; Ratio = counts 805 / counts 1450). Fig.10 Each point indicated on the graph corresponds to a measurement taken.

[0414] 3) Then, a calibration curve of the oxygen saturation Sat% of the blood is constructed by means of a "best fit" technique, which has a 3rd order curve of the ratio of the value of optical counts of electromagnetic radiation read by the silicon photodiode 8 for excitation at 805 nm to the value of optical counts of electromagnetic radiation read by the same photodiode 8 for excitation at 660 nm. The blood hematocrit value Hct% is kept fixed at 15%, the blood temperature is kept at 37°C, and the step size of Sat% is increased by 5 points in the range from 35% to 98% (see Fig.11 ; Ratio = counts 805 / counts 660). Fig.11 Each point indicated on the graph corresponds to a measurement taken.

[0415] 4) Then, proceed to construct a temperature calibration curve (linear relationship between the fluid temperature detected by the device under data acquisition at an ambient temperature of 24±2°C and the fluid temperature of the reference instrument).

[0416] 5) In this step, artificial intelligence using a neural network NN of the type described below is used to create a weight matrix in order to accurately calculate Sat% and Hct% including their interdependencies. This step requires the acquisition of a data set for each tube or test tube with which the device is to operate. The data set is acquired by setting the Hct% and blood temperature and scanning the Sat% value for each established Hct%; the scan is performed by changing the blood state to obtain the Sat% value within its entire range, and the process is repeated for each Hct% value within its entire range. According to the method used, after setting a specific temperature of the blood, the specific value of Hct% remains fixed and the entire measurement range of Sat% is scanned; then, at the same temperature of the blood, the value of Hct% is changed, and Sat% is scanned again, etc., to cover the entire measurement range of Hct%. Next, another blood temperature is set, and the above two scans of Hct% and Sat% are repeated. The purpose is to cover substantially all possible values ​​of Sat% and Hct% within the desired temperature range, that is, to cover all values ​​of the parameters of the blood temperature, Hct% and Sat% of interest within their respective ranges. A method for achieving this goal has just been described, which has proven to be advantageous; it will be appreciated that other methods may achieve the same goal in other ways (e.g., by setting the temperature after the value of Hct% or otherwise using another sequence of setting and variation of the parameters blood temperature, Hct% and Sat%).

[0417] 6) In this step, artificial intelligence is used to calculate the correlation between Hct% and Sat%; this step can be performed by using the same neural network NN used in step 5) or another neural network (second neural network) having substantially the same structure.

[0418] 7) In this step, artificial intelligence is used to calculate the correlation between Hct% and blood temperature, and this step can be performed by using the same neural network as in step 5) or another neural network (third neural network) having substantially the same structure.

[0419] Preferably, only one neural network NN is used in learning. If higher accuracy is desired, two other neural networks, ie, a second neural network and a third neural network, may be combined to improve the results from the output of the first neural network (the neural network of step 5).

[0420] Each neural network preferably has the following structure and calculation method.

[0421] In the test performed, the Fig.12A multilayer perceptron type neural network NN (MLP or multilayer perceptron) is shown in . A multilayer perceptron type neural network NN is an artificial neural network model that maps an input data set into an appropriate output data set. Moreover, such a neural network NN is of the feedforward type, i.e. it is an artificial neural network in which the connections between the nodes do not form a loop. It should be understood that different types of machine learning neural networks can be selected depending on the machine learning method that one wishes to implement; the type of neural network selected will be described below with reference to the tested application, which involves the direct measurement of oxygen saturation and hematocrit parameters with the aid of artificial intelligence. Obviously, the following applies mutatis mutandis to the measurement of one or more other blood parameters.

[0422] Fig.12 The neural network NN presents an input or input layer L0 with six neurons, a first layer L1 fully connected by twelve neurons, a second layer L2 fully connected by ten neurons and a third layer L3 fully connected by two neurons. The two neurons constituting this last layer L3 of the output layer or output correspond to the output values ​​of the neural network, namely: Sat% value and Hct% value. Regardless of the number of neurons of the output layer L3, the other layers L0, L1, L2 may present a number of neurons and / or a number of layers different from that described here, depending on the specific optimization performed for the chosen neural network model. The number of neurons in each layer may be partially constrained by the computing power of the microprocessor MP of the control unit 3 of the device 1L, 1L'. All neurons in the network are S-shaped activations; the activation of a neuron corresponds to the activation of the neuron and the operation of its information transfer function (in Fig.14 , and hereinafter referred to as sigmoid), which enables the transmission of incoming stimulus information.

[0423] The input data (input data) of the neural network NN are described below; see Fig.12 Two of the six neurons in the input layer L0 correspond to the Sat% and Hct% values ​​measured by the devices 1L, 1L' (respectively Fig.12 Sat% IN and Hct% IN ), while three of the six neurons correspond to values ​​of ratios 805 / 660, 805 / 1450, 805 / 1550 or alternatively 940 / 1050; the 805 / 1550 ratio is used to improve the measurement accuracy of Hct%, while the 940 / 1050 ratio is used because at a wavelength of 1050 nm, both the 805 / 1550 ratio and the 940 / 1050 ratio have good variation dynamics over the entire measurement range of Sat%. Another neuron in layer L0 corresponds to the blood temperature value detected by the temperature detector 67 ( Fig.12 T IN)correspond.

[0424] The input parameters are scaled to normalize in magnitude to prevent very large values ​​from being weighted more heavily than small values ​​in the neural network, and also to allow the neural network to be trained more efficiently. Basically, based on the collected learning set, the minimum and maximum values ​​of each parameter input to the network are calculated, and the minimum and maximum values ​​are normalized before they are passed to the network, and then denormalized to obtain the output data. These values ​​must be rescaled before being passed to the network in this way: Normalized Value = (Collected Value - Minimum Value) / (Maximum Value - Minimum Value). Similarly, the values ​​output from the network must be rescaled in this way to obtain the actual value: Denormalized Value = Value * (Maximum Value - Minimum Value) + Minimum Value.

[0425] The neural network NN is then able to calculate the Sat% and Hct% values ​​(output data) of the blood flowing into the container based on the input of the above mentioned data. In the neural network selected and tested, this calculation, which enables the measurement performed by the device, is performed as follows.

[0426] The calculation method used is Fig.13 is shown in and involves the following formula, where the symbol * indicates a scalar product:

[0427]

[0428] a out =g(z) (Formula 2)

[0429] In Formula 1, you should also specify:

[0430] -w i =weight i

[0431] -b i =bias i

[0432] - "wi" and "bi" are matrices of weights and biases respectively; weights and biases are calibration coefficients.

[0433] By providing a matrix, the computational model is a mathematical matrix model.

[0434] For Formula 2, specifying a out is the output of each neuron, and therefore also the output of the inner layers L1 and L2.

[0435] In more detail:

[0436] -For the first layer L1:

[0437] a1=x_input*w1+b1

[0438] z1=sigmoid(a1)

[0439] -For the second L2 layer:

[0440] a2=z1*w2+b2

[0441] z2=sigmoid(a2)

[0442] -For the third layer L3:

[0443] a3=z2*w3+b3

[0444] z3=sigmoid(a3)

[0445] Output = z3

[0446] In the output layer L3 there are two neurons, one of which has as its output Sat % and the other has as its output Hct %, ie the desired output parameter related to the blood circulation in the vessel to which the device is coupled.

[0447] It should also be noted that:

[0448] - "i" ranges from 1 to N,

[0449] - "N" is the number of nodes (neurons) in each layer of the neural network,

[0450] - "x_input" is a vector of input values / parameters (a vector containing parameters related to optical counts and blood temperature),

[0451] - "a", "b", "w" are derived from the learning phase,

[0452] - "g(z)" is the activation function of the neuron and is Fig.14 is shown in .

[0453] The learning process using the sample measurements optimizes the values ​​of the matrices of weights and biases and derives the matrices of weights and biases for each layer, so w1 and b1 are the matrices of weights and biases for the first layer L1, w2 and b2 are the matrices of weights and biases for the second layer L2, and w3 and b3 are the matrices of weights and biases for the third layer L3. Optimization, i.e. finding the "optimal" values ​​of weights and biases, means finding the values ​​of weights and biases that minimize the cost function, which represents the error between the actual measured values ​​and the predicted values ​​of the network.

[0454] The algorithm implemented in the device 1, 1' executes Fig.13Thousands of training epochs were run (1 epoch corresponds to a full cycle of all examples) with an average batch size of 4 examples per batch. In the tests performed, learning was performed up to 4*10 5 epochs, with an adaptive learning rate starting from a value of 0.001 to avoid overfitting. It should be noted that the learning rate is one of several parameters set during the learning phase of a neural network; it modifies the optimizer step. An optimizer called "Adam" was used, which is a derivative of the gradient method. In neural network terminology:

[0455] - Episode = forward pass and backward pass of all training examples. Episode describes the number of times the algorithm has seen the entire dataset. So, each time the algorithm has seen all the samples in the dataset, an episode is completed.

[0456] - Batch size = number of training examples in the forward pass / backward pass. The larger the batch size, the more memory space is required.

[0457] - Number of iterations = number of passes, using number of examples per pass [batch size]. One pass = one forward pass + one backward pass (not counting forward and backward passes as two different passes).

[0458] For less detailed aspects of machine learning, please refer to the literature on neural network theory.

[0459] For each link in the network, there is a function that is transposed into C language code in the microprocessor MP, called sigmoid, which is a mathematical function that connects neurons (sublayers). The input of each neuron in the network is multiplied by the network parameter matrix, and the result is passed to the sigmoid function using the above formula. The result of the sigmoid function represents the intermediate output of each layer or the input to the neurons of the next layer.

[0460] During learning, a non-linear cost function (mean absolute error) was chosen, which was modified to weight examples quadratically at low Sat% and Hct% (which experience higher percentage errors).

[0461] The final output of the neural network NN model is represented by a set of parameters, namely the matrix w of weights and biases i and b i Essentially, the neural network model learns through a large number of laboratory tests and measurements which are the matrices w of weights and biases that are able to provide two values ​​of blood parameters (Sat% and Hct%) as outputs for the input vector x_input i and b i The mathematical model of the multilayer perceptron network is then implemented in the microprocessor MP of the device 1, 1' using C code.

[0462] In summary, the procedure is as follows: A neural network model is selected, i.e., the number of layers, the number of neurons in each layer and the activation functions of the neurons are defined.

[0463] The program continues the learning process, using the example measurements and optimizing the matrix w of weights and biases. i and b i For each layer, there is a matrix w of weights and biases i and b i , thus w1 and b1 are the matrices of weights and biases of the first layer, w2 and b2 are the matrices of weights and biases of the second layer, and w3 and b3 are the matrices of weights and biases of the third layer (and so on if further layers are provided). By "optimization" or "optimal values" it is meant finding those values ​​of weights and biases that minimize a cost function representing the error between actual measurements made by the network and predicted measurements. After defining x_input, i.e. a vector containing parameters either from ratios of optical counts or directly from optical counts and blood temperature, the algorithm implemented in the device performs Fig.13 The operations shown in and summarized above are described below.

[0464] The developed calculation method as a result of machine learning is essentially an algorithm coded and implemented in the microprocessor MP of the device to perform the measurement of blood parameters. The calculation method and therefore the algorithm are preferably coded in a program code, in particular in C code, to be executed by the microprocessor during clinical use of the device 1, 1'. The implementation of the calculation method and the algorithm in the microprocessor MP of the device 1, 1' makes the device 1, 1' according to the invention independent in clinical use; all the calculations required for measuring blood parameters are performed by a control unit built into the device and can be transmitted via a serial cable to a medical machine docked therewith, which only adopts the "finished data" of the measurement (i.e., data ready for use by medical personnel) without performing calculations.

[0465] The device 1, 1' has inside it all the information to perform the parameter measurement. The firmware code is written in such a way that each of the two electromagnetic wavelength ratios calculated by the device 1, 1' by acquiring the relative intensity, together with the blood temperature value also measured by the device, becomes an input parameter of the model. Each device 1, 1' is then programmed so that the matrix containing the weights and biases is stored (in the EEPROM memory), while the algorithm based on the calculation method is written (coded) in the firmware.

[0466] It has been verified that the neural model of the device trained in this way has the ability to measure / estimate the output value of Sat% with an accuracy of ±6% over the entire measurement range and for all states, and the output value of Hct% with an accuracy of ±3% over the entire measurement range and for all operating states; this accuracy is achieved using a matrix containing weights and biases (derived from the neural network model based on what the device learned during the learning phase).

[0467] More details about neural networks

[0468] Neural networks are constructed based on the following assumptions. Three neural networks are described below; the devices are called "probes". Each network has a number of neurons in the input layer L0 equal to the number of input values ​​and a number of neurons in the output layer L3 equal to the number of output values.

[0469]

[0470]

[0471] Here is the reference photodiode used to read the optical counts:

[0472] -Optical counting ratio 805 / 660: Si photodiode,

[0473] - Optical counting ratio 805 / 1450: InGaAs photodiode,

[0474] -Optical counting ratio 940 / 1050: Si photodiode,

[0475] -660 optical counts: Si photodiode,

[0476] -805 Optical Count: Si Photodiode,

[0477] -805 Optical Count: InGaAs photodiode,

[0478] - Optical counts from 1450: InGaAs photodiodes,

[0479] - Optical counts from 1550: InGaAs photodiodes.

[0480] The first neural network has six neurons in the input layer L0 and two neurons in the output layer L3. As indicated above, the values ​​of Hct% and Sat% at the output of the two calibration curves are two of the six input parameters of the neural network. As mentioned previously, the first neural network described above (see also Fig.12) may be sufficient to achieve the desired accuracy of measurement of the parameters Sat% and Hct%; in this case, the artificial intelligence of the device 1, 1 'provides a single neural network NN. However, if this is not sufficient and / or a higher accuracy is desired, the following second and third neural networks may also be used, or possibly the number of neural networks is equal to two or the number of neural networks is greater than three.

[0481]

[0482]

[0483] The second neural network has: two neurons in the input layer L0 (corresponding to the input values ​​of the calculated Sat% output and Hct% output from the first neural network); and two neurons in the output layer L3 (corresponding to the detected output values ​​of Sat% and Hct%).

[0484]

[0485] The third neural network has: two neurons in the input layer L0 (corresponding to the input values ​​of the calculated Sat% output and Hct% output from the second neural network and the blood temperature); and two neurons in the output layer L3 (corresponding to the detected output values ​​of Sat% and Hct%).

[0486] When a neural network is developed / designed and the coefficients / weights of the computational matrix are identified, the input / output of the network is as follows.

[0487]

[0488] Therefore, in the case of using multiple neural networks, the structure of the neural network is the same except for the number of neurons of the input layer L0 and the corresponding input values, which vary depending on the type of network to be designed. The output data calculated by the neural network through artificial intelligence are the Sat% and Hct% values ​​for each neural network, while the Hb (hemoglobin) value is calculated mathematically because it is derived from the hematocrit value. The Hb value (relative to the reference blood analyzer 69) can be calculated according to the following formula: Hb (g / dL) = Hct% / approximately 3.

[0489] Preparation of the device after training (calibration)

[0490] Each device 1 , 1 ′ produced after the training of the above-described device 1L, 1L′ is equipped with a firmware taking into account the above-described algorithm and is therefore equipped with artificial intelligence functions.

[0491] For each of these devices 1, 1', only two curves are acquired, a fixed Sat% calibration curve and a fixed Hct% calibration curve, without having to perform the full training described above. Therefore, in the laboratory, it is necessary to acquire these two curves so that the probe has all the necessary inputs to provide the required output. They can be considered as curves that allow the probe to provide the initial saturation and hematocrit, which the probe uses as input together with the ratio of light counts for each measurement (see Fig.12 , the two neurons in the L0 input layer are dedicated to these two input values). The above two calibration curves are similar to Fig.10 The curve shown in (hematocrit calibration curve) and Fig.11 Saturation calibration curve. Artificial intelligence allows the device 1 , 1 ′ to be calibrated according to these two calibration curves; the device is then ready for measurements of multiple parameters.

[0492] In other words, each device 1, 1' produced after the training of the probe 1L, 1L' is calibrated in the laboratory (using the aforementioned calibration curves), so that the end user of the device 1, 1' does not have to perform a calibration and therefore has an instrument immediately ready for use. Such calibration in the laboratory is relatively fast (e.g., tens of minutes or hours); in essence, as described above, two calibration curves are preferably acquired at a blood temperature of 37°C, so that the device 1, 1' has all the necessary inputs to output the desired parameter values. In essence, only this step of acquiring two curves serves as the only calibration operation for each newly manufactured probe and allows to take into account the variability between probes (hardware, component variability and their geometrical position, even infinitesimal in the probe housing) and gives a solid starting point for the Sat% and Hct% measurements expected in the calibrated state.

[0493] Identify the relevance of the type (color / hue) of tubes associated with the device according to the second embodiment

[0494] The purpose is to identify the type of tube associated with the device 1 ' using LED elements of wavelengths 525nm and 940nm. The optical counts of these wavelengths are obtained by having saline in the tube 25' (ie before blood flows into the tube), and then the factors to be used (coefficients of the neural network NN, ie weights and biases, which are as many as the number of neurons) are derived.

[0495] The different hues of the tubes have an effect on the optical counts detected at these wavelengths. Each tube 25' is identified according to a mathematical relationship based on these two optical counts.

[0496] Identifying correlations with blood flow

[0497] In order to identify the correlation of saturation with blood flow, the variation of the optical counts of the 940 nm wavelength LED elements as a function of flow is used. The blood flow is preferably the volumetric flow rate of the blood flowing into the extracorporeal blood circuit and thus into the tube 25', and can be measured in liters / minute. Basically, since the optical counts of the 940 nm wavelength LED elements are correlated with Sat%, compensation for this correlation is required. The counts caused by the excitation of the 940 nm wavelength LED elements are affected by both blood flow and saturation; in order to be able to use these counts for flow detection, it must be ensured that the counts are independent of saturation.

[0498] By introducing an additional LED excitation element of wavelength 1050nm (which varies with Sat%) mounted on the same excitation element as the 940nm LED element (first excitation element) and then detected with the same photodiode (first photodetector), a "second" saturation (ratio of counts for 940nm excitation to counts for 1050nm excitation) is obtained to compensate for the variation of the 940nm wavelength LED element and therefore make it immune to saturation dependence for flux detection.

[0499] The flow information thus obtained is used to develop an algorithm for correcting Sat% and Hct% according to blood flow.

[0500] Clinical use of the device

[0501] The invention also relates to the use of the device 1, 1' described previously. The device 1, 1' is used to measure a plurality of blood parameters; these parameters are of the type described above.

[0502] The use does not require and therefore does not include an initial calibration of the device. The use is preferably a clinical use, wherein the device 1, 1' can be used in conjunction with a medical machine 90', 90', such as a heart-lung machine 90' ​​or an extracorporeal membrane oxygenation (ECMO) machine 90' ​​or other machine. Due to training, the device does not require any initial calibration and is therefore ready to use.

[0503] Fig. 8A and Figure 8B A corresponding device 100 and a possible clinical use of the apparatus 1 , 1 ′ according to the invention are shown. Fig. 8A The device 1, 1' is shown in clinical use in an operating room, where the device 1, 1' is used in conjunction with a heart-lung machine 90' ​​to which the device 1, 1' is connected. Figure 8B, which shows clinical use of the device in an intensive care unit, where the device 1, 1' is used in conjunction with an extracorporeal membrane oxygenation machine 90' ​​to which the device 1, 1' is connected. An example of intensive care in which the device 1 according to the invention may be appropriately used is intensive care due to Covid, in which case monitoring one or more parameters related to the presence or concentration of oxygen in the blood may be crucial.

[0504] The measured parameter values ​​are available to the responsible medical personnel, for example, at the user interface of the medical machine 90', 90". The parameter values ​​can be provided to the medical machine 90', 90" continuously or at a certain rhythm and additionally or alternatively at the request of the medical machine 90', 90".

[0505] Obviously, the device 1 is suitable for other clinical uses than the one described here, in particular in combination with any additional medical machine and / or in any treatment or area requiring extracorporeal blood circulation.

[0506] equipment

[0507] The invention also relates to an apparatus 100 comprising a device 1, 1' and a medical machine 90', 90". The device 1, 1' is configured to cooperate and communicate with the medical machine 90', 90".

[0508] The medical machine 90', 90" may be a heart-lung machine 90' ​​or an extracorporeal membrane oxygenation machine 90' ​​or other medical machine. The medical machine 90', 90" may include a user interface configured to make available the values ​​of parameters measured by the apparatus 1, 1'. The user interface may include a display device 91, 91', such as a screen 91, for displaying the values. The display device 91 is operatively connected or connectable to the apparatus 1, 1'. The display device 91 may be part of the medical machine or be associated with or connectable to the medical machine. Fig. 8A and Figure 8B In FIG. 1 , both a device 1 according to a first embodiment and a device 1 ′ according to a second embodiment are shown; in FIG. Fig. 8A In the embodiment, they are connected to a screen 91 integrated in a medical machine 90', 90" and a screen 91' capable of displaying physiological parameters, respectively. Figure 8B In the embodiment, they are connected to the same screen 91 integrated in the medical machine 90 ′, 90 ″. Alternatively, the user interface may make the parameter values ​​available or communicate the parameter values ​​to the medical staff in another way.

[0509] The device 100 comprises an extracorporeal blood circuit 92 configured to circulate blood. The extracorporeal blood circuit 92 connects a medical machine 90 ′, 90 ″ to a patient; according to a first embodiment, a tube can be displaced along the circuit associated with the device 1 . Fig. 8A , Figure 8B Each of them also shows a device 1 ' according to a second embodiment directly associated with an extracorporeal blood circuit tube 92. In essence, the device 1, 1 'can constitute an accessory to a medical machine 90', 90", which functionally acts as a sensor intended to measure a plurality of parameters of the blood circulating in the extracorporeal blood circuit 92.

[0510] Method for measuring multiple blood parameters

[0511] The invention also relates to a method for measuring a plurality of blood parameters, which parameters are of the type described above. The method is preferably performed with the aid of the device 1 , 1 ′ described above.

[0512] The method comprises at least the following steps:

[0513] - using electromagnetic radiation of a plurality of defined wavelengths to stimulate blood flow,

[0514] - detecting a plurality of electromagnetic responses, in particular optical responses, of the blood, the plurality of electromagnetic responses comprising electromagnetic radiation, in particular light, reflected or diffused by the blood,

[0515] - receiving analog information about a plurality of electromagnetic and / or optical responses of blood, the plurality of electromagnetic and / or optical responses including light reflected or diffused by the blood,

[0516] - converting electromagnetic and / or optical response analog information into electromagnetic and / or optical response digital data,

[0517] - processing said electromagnetic and / or optical response digital data and the actual temperature value of the blood by one or more neural networks NN,

[0518] - As a result of the processing operations performed by means of one or more neural networks NN, a value for each of a plurality of blood parameters is determined.

[0519] The latter two steps involve the following steps:

[0520] o determining a plurality of ratios, each ratio being defined between a magnitude indicative of radiation retroreflected or diffused by the blood due to excitation at a determined wavelength and a magnitude indicative of radiation retroreflected or diffused by the blood due to excitation at another determined wavelength,

[0521] o providing the plurality of ratios and temperature values ​​as input to one or more neural networks NNN,

[0522] o processing said plurality of ratios by means of one or more neural networks NN taking into account a plurality of data of previous measurements of said blood parameters performed during previous training,

[0523] o As an output of the one or more neural networks NN, a value for each of the plurality of blood parameters is provided.

[0524] The method may provide steps corresponding to one or more of the operations described above with respect to the control unit.

[0525] The method involves measuring the oxygen saturation, hematocrit and hemoglobin of the blood with the aid of artificial intelligence, wherein the oxygen saturation and hematocrit are measured directly with the aid of the artificial intelligence and the hemoglobin is measured indirectly, i.e. as a measured value derived from the hematocrit value.

[0526] The method also provides for measuring the actual temperature of the blood, preferably with the aid of a suitable temperature sensor 12; this temperature constitutes both an input parameter for one or more neural networks and an output parameter of the device, making the output parameter of the device available and viewable by an end user of the device on a suitable display device.

[0527] The method may include the step of taking into account values ​​related to blood flow, in particular the volumetric flow rate of blood, when measuring blood oxygen saturation and / or hematocrit. This step may involve correcting the measured values ​​of oxygen saturation and / or hematocrit according to blood flow.

[0528] Prior to exciting the blood flow with electromagnetic radiation of a plurality of determined wavelengths, the method comprises the step of coupling the device 1, 1' to a corresponding container 25, 25'.

[0529] Specific method steps for the first embodiment of the device

[0530] The step of coupling the device 1 to the cuvette 25 may be performed by coupling one or more coupling elements 17e , 17f of the device 1 with one or more corresponding coupling elements 25b , 25c of the cuvette 25 .

[0531] Specific method steps for the second embodiment of the device

[0532] The step of coupling the device 1' to the container 25' can be performed by accommodating a portion of the tube 25' of the extracorporeal blood circuit at the seat 19 of the cassette 2 of the device 1' and bringing the covering elements 18, 18' into the operative configuration (closing the covering elements 18, 18' on the cassette 2). Said step consists in moving the covering elements 18, 18' close to the constraining elements 22, 23 displaced on the cassette until the engagement of the movable elements 22a, 23a with the corresponding seats defined on the opposite operative ends 21a, 21b of the closing portion 21 of the covering elements 18, 18' is determined.

[0533] Closure of the covering elements 18, 18' results in compression of the central portion 25a' of the tube 25' within which the blood flows and flattening of the opposing surfaces of the portion of the tube 25' that engages at the base 19 by compression of these surfaces (see Fig. 6A ). The compression step results in a reduction of the blood passage section of the tube 25' by between 9% and 17% for a length between 15 mm and 30 mm or equal to one of these values.

[0534] In an embodiment in which an oscillating compression element 18o is provided, the step of closing the covering element 18' comprises oscillating the compression element 18o relative to the body 18b of the covering element 18', and thus oscillating the compression element 18o relative to the tube 15' through which the blood flows. Such a closing step comprises gradually and gently compressing the tube 25', thereby gently altering the blood flow flowing in the tube 25'.

[0535] The method may enable the detection of the type of tube 25'. The step of detecting the type of tube 25' is performed before the step of exciting the blood flow at a plurality of determined wavelengths. The method may prepare the control unit 3 of the device 1 for measurements based on the detected type of tube 25'. The latter step may involve adapting the mode of measurement of one or more parameters to the detected type of tube 25'. Preferably, the step of detecting the type of tube 25' comprises detecting the color of the tube. The step of detecting the type of tube 25' is performed when a fluid other than saline or blood flows in the container. Preferably, the step of adapting the mode of measurement of one or more parameters to the detected type of tube 25' comprises: selecting a determined matrix from a plurality of matrices that can be used by the neural network (each matrix can correspond to a specific color or hue of the tube). This selection can be performed by consulting a memory (EEPROM memory) in which information related to a plurality of matrices is stored.

[0536] Examples of possible modifications

[0537] The invention is susceptible to modifications and / or further improvements.

[0538] For example, the apparatus 1, 1', in particular the apparatus 1' according to the second embodiment, may be provided with a display device, such as a display, on which the blood parameters are shown. A preliminary calculation of such blood parameters may be performed under the assumption that a standard tube (i.e. a tube of the most commonly used color or hue) is coupled to the apparatus 1'. The apparatus 1' may provide that there is a possibility on the display device to enter as input the type of tube 25' actually coupled to the apparatus 1' (in particular the type of tube 25' in terms of color or hue; for example, it may be necessary to enter 1 for a "blue" tube, 2 for a "yellow" tube, etc.), so that the apparatus 1' will automatically load the coefficient matrix of the associated neural network and recalculate the blood parameters to take into account the tube actually coupled to the apparatus 1'. The visual support and this interaction possibility make the "on-the-go" measurement option even more advantageous and faster.

[0539] According to a second embodiment, the device 1' can be provided with a sensor for verifying the position of the covering elements 18, 18', in particular for verifying that the covering elements 18, 18' are in the closed position (operating configuration). The control unit 3 can provide that the measurement of the parameter can only be performed after verifying that the covering elements 18, 18' are correctly in the closed position.

[0540] The control unit 3 described above provides a possible implementation of the artificial intelligence according to the invention, which the applicant has verified to be particularly efficient; however, possible modifications / improvements or further embodiments are not excluded. Similarly, different neural networks can be used to implement the artificial intelligence features of the invention.

[0541] Further modifications are conceivable according to specific needs or contingencies related to the implementation of the invention.

[0542] Other advantages and conclusions

[0543] In summary, the main advantages of the present invention are as follows:

[0544] - parameter measurement without the user having to perform any calibration of the device 1, 1'; in essence, in clinical use, the user has the device 1, 1' immediately ready for use, since the device 1, 1' has built-in all the information necessary to measure the blood parameter of interest,

[0545] the device 1 , 1 ′ is autonomous in terms of measurement, i.e. it does not rely on other instruments external to the device 1 , 1 ′ itself to perform parameter measurements; the external instruments may only serve to make the measured parameter values ​​available,

[0546] - the measurements made by the artificial intelligence are reliable over the entire measuring range of the parameter; in more detail, the measurements made by the artificial intelligence show an accuracy of ±6% over the entire measuring range of the parameter Sat% and for all operating states, and an accuracy of ±3% over the entire measuring range of the parameter Hct% and for all operating states,

[0547] The measurements are carried out with the aid of a compact, space-saving, lightweight device 1 , 1 ′ having appropriately miniaturized components.

[0548] The protection conferred by the claims extends to every element, component and / or step of the invention that is equivalent to the claimed element, component and / or step. Thus, in accordance with the present invention, every element, component and / or step of the product / process may be replaced by an equivalent element, component and / or step (hereinafter "equivalent"); such equivalents may have existed at the filing or priority date of this patent document or subsequent conception / development.

Claims

1. A device (1, 1') for measuring a plurality of blood parameters by means of artificial intelligence, comprising: at least one excitation member (4, 5) configured to excite the blood, in particular the blood flow, by electromagnetic radiation of a plurality of determined wavelengths; at least one electromagnetic radiation detection member (8, 9), in particular at least one photodetector, configured to detect a plurality of electromagnetic responses, in particular optical responses, of the blood, said plurality of electromagnetic responses comprising electromagnetic radiation, in particular light, reflected or diffused by the blood, said electromagnetic radiation being reflected or diffused by the blood upon excitation of said excitation member (4, 5) in an operating state of said device (1, 1'); - a control unit (3) configured to perform the following operations: o commanding the at least one excitation member (4, 5) during an excitation step in which the at least one excitation member (4, 5) excites the blood by electromagnetic radiation of a plurality of determined wavelengths, o receiving analog information about a plurality of electromagnetic and / or optical responses of the blood, the plurality of electromagnetic and / or optical responses comprising electromagnetic radiation, in particular light, reflected or diffused by the blood, oConverting electromagnetic and / or optical response analog information into electromagnetic and / or optical response digital data; o processing the electromagnetic and / or optical response digital data and the actual temperature value of the blood by one or more neural networks (NN), o determining a value for each of the plurality of blood parameters as a result of processing operations of the one or more neural networks (NNs), The control unit (3) is configured to process the electromagnetic and / or optical response digital data and the actual temperature value of the blood through one or more neural networks (NN) and determine the value of each of the plurality of blood parameters by: o determining a plurality of ratios, each ratio being defined between a value indicative of the amount of radiation retroreflected or diffused by the blood as a result of excitation at a determined wavelength and a value indicative of the amount of radiation retroreflected or diffused by the blood as a result of excitation at another determined wavelength, o providing said plurality of ratios and said temperature values ​​as input to said one or more neural networks (NN), o processing said plurality of ratios by said one or more neural networks (NN) by taking into account a plurality of data of previous measurements of said blood parameters performed during previous training, o providing as an output from the one or more neural networks (NNs) a value for each of the plurality of blood parameters, Wherein, the device (1, 1') further comprises: - a housing (2), in which the at least one or each of the excitation members (4, 5), the control unit (3) and the at least one or each of the electromagnetic radiation detection members (8, 9) are accommodated; - a coupling portion (17, 19) configured to allow coupling of the device (1, 1') to a container (25, 25') in which the blood can flow, the coupling portion (17, 19) being connected to the box (2).

2. The device according to claim 1, wherein: The control unit (3) is configured to detect a value of each of the plurality of blood parameters that the artificial intelligence considers to correspond to a predetermined ratio based on the plurality of data previously measured.

3. The device according to claim 1 or 2, further comprising firmware, and the control unit (3) comprises artificial intelligence information, such as one or more matrices that can be used by the one or more neural networks (NN), the artificial intelligence information being encoded in the firmware and being suitable for enabling the calculation of the plurality of parameters by the one or more neural networks (NN).

4. The device according to claim 1 or 2 or 3, further comprising a memory, the control unit (3) being configured to provide, as input to the one or more neural networks (NN), a reference value of a first parameter to be measured and a reference value of a second parameter to be measured, the reference values ​​being acquired during a calibration step prior to use of the device (1, 1') and stored in the memory of the device (1, 1'), the memory therefore comprising information about the reference values.

5. The device according to any one of the preceding claims, the device (1, 1') being configured to measure the oxygen saturation SatO2, the hematocrit Hct and optionally also the hemoglobin Hb content by means of artificial intelligence.

6. The device according to any one of the preceding claims, wherein The control unit (3) is configured to command the at least one excitation member (4, 5) or the two excitation members (4, 5) during an excitation step arranged to excite the blood flow at one wavelength at a time.

7. The device according to any one of the preceding claims, wherein The at least one excitation member (4, 5) is configured to excite blood flow at least at the following wavelengths: 660 nm, 805 nm, 1450 nm and at least one of 525 nm, 940 nm and 1050 nm.

8. The device according to any one of the preceding claims, comprising: a first excitation component (4) configured to excite blood flow at least at a first plurality of wavelengths; A second excitation member (5) is configured to excite blood flow at a second plurality of wavelengths, and the control unit (3) is configured to alternately activate the first excitation member (4) and the second excitation member (5) so as to alternately excite blood flow at the first plurality of wavelengths and at the second plurality of wavelengths.

9. The device according to any one of the preceding claims, wherein The device (1, 1'), in particular the box (2), is substantially pocket-sized.

10. Device according to any of the preceding claims, the device (1, 1') being able to be associated with a container (25, 25'), such as a tube, in which blood can flow, and in an operating state the device (1, 1') being associated with the container (25, 25'), The control unit (3) is configured to perform the following operations: - detecting the type of container (25, 25'), in particular the type of tube, - Based on the detected type of container (25, 25'), in particular the type of tube, a measurement is prepared.

11. Device according to any of the preceding claims, the device (1') being able to be associated with a tube (25'), the control unit (3) being configured to detect the color of the tube (25') and to prepare a measurement based on the color of the tube (25') detected by selecting a determined matrix from a plurality of matrices that can be used by the neural network (NN).

12. Device according to any of the preceding claims, comprising a cover element (18, 18') movable relative to the cassette (2) and a base (19) suitable for accommodating a portion of a container (25, 25'), in particular a tube (25'), in which blood flows in an operating state of the device (1'), the cover element (18, 18') being configured to operate at least between the following configurations: an operating configuration in which the covering element (18, 18') flattens the opposing surfaces of the container (25, 25') received at the seat (19), - Rest configuration.

13. The device according to claim 12, wherein: The covering element (18, 18') comprises a compression element (18a, 18o) adapted to compress the container (25') accommodated at the base (19) in an operating configuration of the covering element (18, 18').

14. The device according to claim 13, wherein: The compression element (18a, 18o) is configured to determine a reduction of the fluid passage section of the container (25') housed at the seat (19) comprised between 9% and 17%.

15. The device according to claim 13 or 14, wherein: The covering element (18') comprises a body (18b), and the compression element (18o) is configured to oscillate relative to the body (18b).

16. The device according to any of the preceding claims, further comprising a temperature sensor (12) housed inside the casing (2) of the device (1, 1') and configured to measure an actual temperature value of the blood.

17. The device according to any one of the preceding claims, wherein The coupling portion (17, 19) is integrated with the box body (2).

18. Use of a device (1, 1') according to any one of the preceding claims for measuring a plurality of blood parameters by means of artificial intelligence.

19. A device (100), comprising: - a device (1, 1') according to any one of claims 1 to 17, - a medical machine (90', 90"), such as a heart-lung machine (90') or an extracorporeal membrane oxygenation machine (90"), - a user interface, such as a display device (91), operatively connected or connectable to the apparatus (1, 1') and configured to provide measured values ​​of the plurality of blood parameters.

20. A method for measuring multiple blood parameters by artificial intelligence, the method comprising the following steps: - exciting the blood, in particular the blood flow, by electromagnetic radiation of a plurality of determined wavelengths, - detecting a plurality of electromagnetic responses, in particular optical responses, of the blood, said plurality of electromagnetic responses comprising electromagnetic radiation, in particular light, reflected or diffused by the blood, - receiving analog information about said plurality of electromagnetic and / or optical responses of blood, said plurality of electromagnetic and / or optical responses comprising light reflected or diffused by the blood, - converting electromagnetic and / or optical response analog information into electromagnetic and / or optical response digital data, - processing said electromagnetic and / or optical response digital data and the actual temperature value of the blood by one or more neural networks (NN), - determining a value for each of said plurality of blood parameters as a result of processing operations of said one or more neural networks (NN), The step of processing the electromagnetic and / or optical response digital data and the actual temperature value of the blood and determining the value of each of the plurality of blood parameters by the one or more neural networks (NN) comprises: o determining a plurality of ratios, each ratio being defined between a value indicative of the amount of radiation retroreflected or diffused by the blood as a result of excitation at a determined wavelength and a value indicative of the amount of radiation retroreflected or diffused by the blood as a result of excitation at another determined wavelength, o providing said plurality of ratios and said temperature values ​​as input to said one or more neural networks (NN), o processing said plurality of ratios by said one or more neural networks (NN) by taking into account a plurality of data of previous measurements of said blood parameters performed during previous training, o Providing as an output from the one or more neural networks (NNs) a value for each of the plurality of blood parameters.

21. The method according to claim 20, further comprising the step of compressing a portion (25a') of the container (25') in which the blood flows or can flow, the step of compressing the portion (25a') of the container (25') comprising: A reduction of the fluid passage section of the container (25') is determined to be comprised between 9% and 17%.

22. The method according to claim 20 or 21, further comprising the step of oscillating the compression element (18o) relative to the container (25') in which the blood flows or can flow.

23. The method according to claim 20, 21 or 22, wherein: The method is performed by means of a device (1, 1') according to any one of claims 1 to 17.

24. The method according to any one of claims 20 to 23, comprising the steps of: - providing a device (1, 1') according to any one of claims 1 to 17, - before the step of exciting the blood, in particular the blood flow, by electromagnetic radiation of a plurality of determined wavelengths, the device (1, 1') is coupled to a container (25, 25').

25. Method according to any one of claims 20 to 24, comprising the step of detecting the actual temperature value of the blood, preferably by means of a temperature sensor (12) housed inside the casing (2) of the device (1, 1').

Images