Antimicrobic susceptibility testing subsets of reporting range concentrations

Abstract

It is possible to use a machine learning model to determine the minimum inhibitory concentration of an antimicrobial agent with respect to a microorganism across concentrations for a complete reporting range based on information gathered for only a subset of those concentrations. This may be done using a set of minimum inhibitory concentration reporting acts which comprises receiving a plurality of sets of test well evaluation values, determining a first set of machine learning inputs based on the plurality of sets of test well evaluation values, and determining the minimum inhibitory concentration based on providing the first set of machine learning inputs to the machine learning model. Corresponding systems comprising processers and non-transitory computer readable media having instructions operable to perform such methods when executed by the processor can also be implemented.

Description

[0001] ANTIMICROBIC SUSCEPTIBILITY TESTING SUBSETS OF REPORTING RANGE CONCENTRATIONS

[0002] Cross Reference to Related Applications

[0003] This is a PCT International application of, and claims the benefit of priority to, provisional patent application 63 / 612,529 for “Antimicrobic Susceptibility Testing Subsets of Reporting Range Concentrations” filed on December 20, 2023 in the U.S. patent office, the disclosure of which is hereby incorporated by reference in its entirety.

[0004] BACKGROUND

[0005] An important function of many microbiological analyzers is to determine amounts of antimicrobics which are effective in controlling the growth of microorganisms. Analyzers which perform this function may place a small sample to be tested into a plurality of small sample test wells in panel that contain different antimicrobics in serial dilutions of clinical interest. The wells may be incubated, and characteristics of the different wells may be used to determine the effectiveness of the antimicrobics after incubation (e.g., in some cases, the turbidity will be increased or unchanged in test wells where growth has not been inhibited by the antimicrobics in those test wells). The minimum inhibitory concentration (MIC) of each antimicrobial agent can then be determined by identifying the lowest concentration test well where growth was inhibited for each antimicrobic is measured by lack of growth with respect to each concentration of antimicrobial agent.

[0006] While current approaches may be able to determine MIC, they have a number of drawbacks which limit their effectiveness. For example, creating a panel, incubating it and then waiting to see the effectives of the tested antimicrobics at each dilution can be a time consuming process, leading to antimicrobic susceptibility testing taking 16-24 hours to provide a result. Additionally, providing information for a required reporting range of dilutions (which reporting ranges may vary based on the region where testing is performed and the microbe present in a patient sample) can limit the number of antimicrobics tested on any particular panel. For example, in a panel which has only 96 wells, testing on a reporting range from 0.06 pg / ml to 128 pg / ml can require 12 serially diluted wells per antimicrobic (e.g., 0.06, 0.125, 0.25, 0.5, 1, 2, 4, 8, 16, 32, 64 and 128 pg / ml), limiting the number of antimicrobics which can be tested per 96 well panel to only 8. Accordingly, there is a need for technology which allows antimicrobic susceptibility testing to be performed more quickly and / or in a manner which requires fewer wells per antimicrobic.

[0007] SUMMARY

[0008] In some aspects, this disclosure relates to methods which comprise one or more performances of a set of minimum inhibitory concentration acts. This set of minimum inhibitory concentration acts may include receiving a plurality of sets of test well evaluation values. In such a method, each of those sets may correspond to a concentration which is different from the concentration for each other set of test well evaluation values, and may be based on a fluorescence value for a test well inoculated with a biological sample. Such a method may also include determining a first set of machine learning inputs based on the plurality of sets of test well evaluation values. Such a method may also include determining a minimum inhibitory concentration based on providing the first set of machine learning inputs to a machine learning model. Such a machine learning model may be configured to provide minimum inhibitory concentration values which are equal to concentrations corresponding to the plurality of sets of test well evaluation values, and to provide minimum inhibitory concentration values which are not equal to any of the concentrations corresponding to the test well evaluation values.

[0009] Corresponding systems, machines, computer program products, and computer readable media can also be implemented, and further aspects, implementations and embodiments of the disclosed technology are also possible. Accordingly, the examples provided in this summary should be understood as being illustrative only, and should not be treated as limiting on the protection provided by this document or any related document.

[0010] BRIEF DESCRIPTION OF THE DRAWINGS

[0011] While the specification concludes with claims which particularly point out and distinctly claim the invention, it is believed the present invention will be better understood from the following description of certain examples taken in conjunction with the accompanying drawings, in which like reference numerals identify the same elements and in which: FIG. 1A depicts a portion of a diagrammatic view of an exemplary biological testing system;

[0012] FIG. IB depicts another portion of the diagrammatic view of the biological testing system of FIG. 1A;

[0013] FIG. 2 depicts a perspective view of an exemplary incubator system and an exemplary measuring system of the biological testing system of FIG. IB;

[0014] FIG. 3 depicts a perspective view of the measuring system of FIG. 2;

[0015] FIG. 4 depicts another perspective view of the optics system of FIG. 2 showing an XY stage of the measuring system;

[0016] FIG. 5 depicts a diagrammatic view of an exemplary computer system;

[0017] FIG. 6 depicts a process which can be used to report the minimum inhibitory concentration of an antimicrobial;

[0018] FIGS. 7A-7C depict how data gathered using a measuring system can be transformed and reformatted into a form which can be provided to a machine learning model to make MIC determinations;

[0019] FIGS. 8A-8B depict a machine learning model which can be used to make minimum inhibitory concentration determinations;

[0020] FIG. 9 depicts a process which can be used in creating outputs which would be processed internally by a machine learning model;

[0021] FIG. 10 depicts a process which may be used to determine a minimum number of tests for establishing a minimum inhibitory concentration for an antimicrobial with respect to a microorganism;

[0022] FIG. 11 depicts a set of minimum inhibitory concentration reporting acts;

[0023] FIG. 12 depicts a method which can be used to determine an optimized set of concentrations to include in a test panel;

[0024] FIG. 13 depicts a method which may be used for determining minimum inhibitory concentration for a plurality of antimicrobials;

[0025] FIG. 14 depicts a table with data gathered during testing of a system implemented based on the disclosed technology; and

[0026] FIG. 15 depicts a table with data gathered during testing of a system implemented based on the disclosed technology.

[0027] The drawings are not intended to be limiting in any way, and it is contemplated that various embodiments of the invention may be carried out in a variety of other ways, including those not necessarily depicted in the drawings. The accompanying drawings incorporated in and forming a part of the specification illustrate several aspects of the present invention, and together with the description serve to explain the principles of the invention; it being understood, however, that this invention is not limited to the precise arrangements shown.

[0028] DETAILED DESCRIPTION

[0029] The following description of certain examples should not be used to limit the scope of the protection provided by this document or any related document. Other examples, features, aspects, embodiments, and advantages of the disclosed technology will become apparent to those skilled in the art from the following description, which is by way of illustration, one of the best modes contemplated for applying the described technology. Accordingly, the drawings and descriptions should be regarded as illustrative in nature and not restrictive.

[0030] It will be appreciated that any one or more of the teachings, expressions, versions, examples, etc. described herein may be combined with any one or more of the other teachings, expressions, versions, examples, etc. that are described herein. The following-described teachings, expressions, versions, examples, etc. should therefore not be viewed in isolation relative to each other. Various suitable ways in which the teachings herein may be combined will be readily apparent to those of ordinary skill in the ail in view of the teachings herein. Such modifications and variations are intended to be included within the scope of the claims.

[0031] I. Biological Testing System Hardware

[0032] FIGS. 1A and IB depict a diagrammatical example of various hardware components available in a biological testing system 1. Biological testing system 1 facilitates an optimized antimicrobial susceptibility testing (AST) method 101 (FIG. 17). Biological testing system 1 broadly includes a consumable preparation system 3, an inoculating system 5, an incubator system 7 (also referred to as incubator), and a measuring system 9. The various systems within biological testing system 1 coordinate with each other and work automatically once loaded with adequate material by a user. The measuring system may be or comprise a fluorometer.

[0033] To operate biological testing system 1, the user first acquires an appropriate microbe sample. As shown in FIG. 1 A, a microbe sample may be obtained from an agar plate 11 or, under certain circumstances, from a blood sample. Next, the user prepares an inoculum suspension by transferring the microbes into a tube containing a suitable liquid medium or broth. One such tube is shown in FIG. 1 A as an inoculum 13. In some versions of biological testing system 1 , the liquid medium or broth may be an approximately 0.5 mM phosphate buffered solution with small amounts of sodium and potassium chloride to aid in maintaining the viability of the microbes introduced into solution without adversely interfering with the MIC determination or other associated testing. Each inoculum 13 is placed into an inoculum rack 15 and the entire inoculum rack 15 is placed into inoculating system 5. Once in inoculating system 5, the inoculum in each inoculum 13 is adjusted if necessary to a standard turbidity value of 0.5 McFarland to create an inoculum 17. In some versions of biological testing system 1, a 1 microliter plastic loop or swab may be provided to the user to easily pick colonies from the agar plate and to minimize the amount of adjustment needed to bring the inoculum to the desired turbidity value. Once adjusted to the desired turbidity value, the inoculum is finalized. The finalized inoculum will be referred to hereinafter as inoculum 17, as depicted in FIG. IB. Inoculum 17 may be further diluted into a 1:250 dilution and converted into an inoculum 18. The inoculum contained in each inoculum 18 is applied to a panel such as an AST array holder 23. AST array holder 23 is assembled by consumable preparation system 3 and provided to inoculating system 5 for use with inoculum 17 and inoculum 18.

[0034] Consumable preparation system 3 is loaded with magazines of test arrays 19, which may contain various antimicrobials or other agents required by biological testing system 1 disposed in a series of test wells 20. For example, test array 19 may comprise an antimicrobic dilution array or an identification array. Consumable preparation system 3 may also be loaded with bulk diluents (not shown) and / or various other elements for preparing and finalizing AST array holder 23 and the inoculate therein. Primarily, consumable preparation system 3 operates to retrieve test arrays 19 as required and combine each retrieved test array 19 into an AST array holder 23. Test arrays 19 may be selected and assembled by a robotic gripper (not shown) or other mechanical features as dictated by the prescribed testing. For example, a physician may order biological testing using the antibiotic amoxicillin. Test arrays 19 relating to amoxicillin testing are therefore retrieved and assembled into the appropriate AST array holder 23. All or some portions of test array 19 may be formed of a styrene material to aid in reducing fluorescent crosstalk, fallout, and / or bubbles when digitally examining each test well 20.

[0035] Once inoculum 17, inoculum 18, and AST array holder 23 arc assembled, inoculating system 5 dispenses the diluted inoculum from inoculum 18 into test wells 20 of AST array holder 23. The time between applying inoculum 18 to AST array holder 23 and the start of logarithmic growth of the microbes disposed therein is known as “lag time.” Lag time may be decreased by using enhanced broth such as a broth with yeast extract, vitamins, and / or minerals. Lag time may also be decreased by increasing the inoculum. In some versions of biological testing system 1, the amount of inoculum may be doubled to decrease the lag time by approximately 30 minutes without affecting the accuracy of the MIC determination. The dispensing may be accomplished via an elevator assembly 26 having an XY robot or XYZ robot (not shown) with a gripper (not shown) and pipettor (not shown), along with various circuitry, channels, and tubing as necessary. The XYZ robot is tasked with retrieving inoculum from inoculum racks 15 and dispensing the inoculum into test wells 20 of AST array holder 23. Once AST array holder 23 is sufficiently loaded with inoculum, AST array holder 23 is moved into incubator system 7 by way of an elevator assembly 26.

[0036] As shown in FIG. 2, incubator system 7 includes slots 27 for holding a large number of AST array holders 23. Each array holder is placed into a corresponding slot 27 by an XYZ robot 29 using a gripper 31. XYZ robot 29 operates to move in any portion of the XYZ plane and position gripper 31 proximate the desired AST array holder 23. While in incubator system 7, each array holder incubates in specific desired environmental conditions. For example, incubator system 7 may be set to incubate array holders at thirty-five degrees Celsius. At certain time intervals during the incubation, XYZ robot 29 retrieves a particular AST array holder 23 and move the selected array holder into the measuring system 9. As shown in FIG. 2-4, measuring system 9 includes features that are configured to observe, monitor, review, and / or capture measurements for each test well 20 of an AST array holder 23. Specifically, each AST array holder 23 may be monitored by a fluorimeter 33. To accomplish the monitoring, XYZ robot 29 retrieves the particular array holder with gripper 31 and places the selected array holder onto an XY-stage 37. The XY-stage 37 moves in the XY plane to position the array holder under the associated monitoring element. XY-stage 37 includes finely tuned motor control to allow each test well 20 of the associated array holder to be positioned accurately within the observation frame of the fluorimeter 33.

[0037] Referring now to FIG. 5, the various components of biological testing system 1 may incorporate one or more computing devices or systems, such as exemplary computer system 49. For example, any one of consumable preparation system 3, inoculating system 5, incubator system 7, and / or measuring system 9 may incorporate one or more computing systems such as exemplary computer system 49. Alternatively, each of these subsystems of biological testing system 1 may function via commands from one overall computing system such as exemplary computer system 49.

[0038] Computer system 49 may include a processor 51, a memory 53, a mass storage memory device 55, an input / output (VO) interface 57, and a Human Machine Interface (HMI) 59. Computer system 49 may also be operatively coupled to one or more external resources 61 via a network 63 or I / O interface 57. External resources may include, but are not limited to, servers, databases, mass storage devices, peripheral devices, cloud-based network services, or any other suitable computer resource that may used by computer system 49.

[0039] Processor 51 may include one or more devices selected from microprocessors, microcontrollers, digital signal processors, microcomputers, central processing units, field programmable gate arrays, programmable logic devices, state machines, logic circuits, analog circuits, digital circuits, or any other devices that manipulate signals (analog or digital) based on operational instructions that are stored in memory 53. Memory 53 may include a single memory device or a plurality of memory devices including, but not limited, to read-only memory (ROM), random access memory (RAM), volatile memory, non-volatile memory, static random access memory (SRAM), dynamic random access memory (DRAM), flash memory, cache memory, or any other device capable of storing information. Mass storage memory device 55 may include data storage devices such as a hard drive, optical drive, tape drive, non- volatile solid state device, or any other device capable of storing information.

[0040] Processor 51 may operate under the control of an operating system 65 that resides in memory 53. Operating system 65 may manage computer resources so that computer program code embodied as one or more computer software applications, such as an application 67 residing in memory 53, may have instructions executed by the processor 51. In an alternative embodiment, processor 51 may execute application 67 directly, in which case the operating system 65 may be omitted. One or more data structures 69 may also reside in memory 53, and may be used by processor 51 , operating system 65, or application 67 to store or manipulate data.

[0041] The I / O interface 57 may provide a machine interface that operatively couples processor 51 to other devices and systems, such as network 63 or external resource 61. Application 67 may thereby work cooperatively with network 63 or external resource 61 by communicating via I / O interface 57 to provide the various features, functions, applications, processes, or modules comprising embodiments of the invention. Application 67 may also have program code that is executed by one or more external resources 61, or otherwise rely on functions or signals provided by other system or network components external to computer system 49. Indeed, given the nearly endless hardware and software configurations possible, persons having ordinary skill in the ait will understand that different versions of the invention may include applications that are located externally to computer system 49, distributed among multiple computers or other external resources 61, or provided by computing resources (hardware and software) that are provided as a service over network 63, such as a cloud computing service.

[0042] HMI 59 may be operatively coupled to processor 51 of computer system 49 in a known manner to allow a user to interact directly with the computer system 49. HMI 59 may include video or alphanumeric displays, a touch screen, a speaker, and any other suitable audio and visual indicators capable of providing data to the user. HMI 59 may also include input devices and controls such as an alphanumeric keyboard, a pointing device, keypads, pushbuttons, control knobs, microphones, etc., capable of accepting commands or input from the user and transmitting the entered input to the processor 51. A database 71 may reside on mass storage memory device 55, and may be used to collect and organize data used by the various systems and modules described herein. Database 71 may include data and supporting data structures that store and organize the data. In particular, database 71 may be arranged with any database organization or structure including, but not limited to, a relational database, a hierarchical database, a network database, or combinations thereof. A database management system in the form of a computer software application executing as instructions on processor 51 may be used to access the information or data stored in records of the database 71 in response to a query, where a query may be dynamically determined and executed by operating system 65, other applications 67, or one or more modules.

[0043] II. Optimized MIC Reporting

[0044] Turning next to FIG. 6, that figure illustrates a process which can be used to report the minimum inhibitory concentration of an antimicrobial, for example in a biological testing system 1. The method comprising one or more performances of a set of minimum inhibitory concentration reporting acts according to the present disclosure may comprise at least some of the below features. In particular', the reporting acts may be carried out as described below.

[0045] Initially, in the process of FIG. 6, a set of test mixtures would be created 601. To illustrate how this could be done, consider a case where the disclosed technology is used to report minimum inhibitory concentration of an antimicrobial across a reporting range from pg / ml to 128 pg / ml. In this scenario, the test mixtures may be created using the sample being tested (e.g., a sample collected from a patient who would be prescribed an antimicrobial based on the MIC) and serial dilutions ranging from 0.06 pg / ml to 128 pg / ml of the antimicrobial, with each serial dilution being approximately twice as concentrated as the preceding dilution. However, using the disclosed technology, it is possible that MIC may be determined without creating a set of serial dilutions over a complete reporting range of concentrations. For example, in some cases, the test mixtures may include the extrema of the reporting range (i.e., 0.06 pg / ml and 128 pg / ml in the example scenario described above), but may include only a subset of the sequence of serial dilutions between those extrema. Alternatively, in some cases, a set of test mixtures may be created 601 which do not include one or both of the extrema of the reporting range (i.e., in the scenario described above, a set of test mixtures may be created which does not include 0.06 pg / ml and / or 128 pg / ml). Accordingly, the description of creating 601 test mixtures comprising a serial sequence of dilutions across a complete reporting range of concentrations should be understood as being illustrative only, and should not be treated as limiting.

[0046] Continuing with the discussion of FIG. 6, once the test mixtures had been created, a panel (e.g., AST array holder 23) which includes those mixtures may be incubated 602 (e.g., using incubator system 7). This panel may also include mixtures other than the test mixtures described above. For example, a panel which is incubated 602 with the test mixtures may also include a growth well (i.e., a well which is inoculated with the sample but no antimicrobial treatment) and a control well (i.e., a well which is inoculated without any kind of fluorogenic substrate). Additionally, it should be noted that the panel will preferably also include test wells inoculated with antimicrobial mixtures of other antimicrobials, thereby allowing MIC to be determined for multiple antimicrobials in parallel. However, for the purpose of clarity, the discussion of FIG. 6 will address only a single antimicrobial, as this discussion can easily be extended, and explicitly addressing the potential multiple antimicrobials which may be analyzed on a panel would unnecessarily complicate the disclosure. It should also be understood that, while the discussion below describes how test well evaluation values may be received 603 during incubation (e.g., at thirty minute intervals until MIC is determined), it is also possible that the receipt of test well evaluation values may occur only after incubation is complete (e.g., a complete set of test well evaluation values captured at thirty minute increments over a several hour period may be sent to a computer after incubation which would then receive 603 them for processing).

[0047] While the panel is incubating 602 (or after incubation, as mentioned above), a system performing a method such as shown in FIG. 6 may receive 603 sets of test well evaluation values, for example a test well evaluation value comprising a fluorescence value for a test well inoculated with a biological sample and an imaging cycle during which the test well evaluation value’s fluorescence value was captured. This may be done by capturing 604 a fluorescence value from one of the wells in the panel, and then continuing to capture fluorescence values until a value had been captured from all of the test wells, as well as the growth and control well in the panel. According to the present disclosure, for each set of test well evaluation values, that set of test well evaluation values may correspond to a concentration which is different from the concentration for each other set of test well evaluation values. Once the test well evaluation values (as well as any associated values, such as growth and control well values) had been received 603, they could be used to determine 605 a (c.g. first) set of machine learning inputs - c.g., the values could be transformed and reformatted into a form which could be provided to a machine learning model operable to make MIC determinations. For example, the method of the present disclosure may comprise determining, e.g. determining 607 described below, a minimum inhibitory concentration based on providing the (first) set of machine learning inputs to a machine learning model, wherein the machine learning model may be configured to provide minimum inhibitory concentration values which are equal to concentrations corresponding to the plurality of sets of test well evaluation values, and to provide minimum inhibitory concentration values which are not equal to any of the concentrations corresponding to the test well evaluation values. To illustrate what this determination may include, consider FIGS. 7A-7C, which are discussed below, and which illustrate how raw fluorescence values may be transformed into a tensor to be input into a machine learning model.

[0048] Initially, FIG. 7A illustrates that a set of raw test well fluorescence values 701, represented as Fo - Fw can be captured for each of the test wells in a panel. These values can be supplemented with fluorescence values for the growth and control wells, represented respectively as FG and Fc, which were captured contemporaneously with the test well fluorescence values, to provide a single increment set of raw fluorescence values 702. For example, if the sets of test well fluorescence values were captured every thirty minutes until the MIC was determined, then the growth and test well values which were used to supplement them may be the growth and test well values which were captured during the same thirty minute increment as the test well values. This single increment set of raw fluorescence values 702 may then be combined with other single increment sets of raw fluorescence values captured previously during incubation to provide an extended set of raw fluorescence values 703. For example, if fluorescence values are captured every thirty minutes, then an extended set of raw fluorescence values could comprise the most recently captured single increment set of raw fluorescence values, plus the single increment sets of raw fluorescence values from each preceding thirty minute increment.

[0049] Turning now to Fig. 7B, the values in the extended set of raw fluorescence values 703 could be normalized, thereby providing a set of normalized fluorescence values 704. For example, if the fluorescence values were stored as numeric values from 0 to 0 to 65535, the values in the extended set of raw fluorescence values could each be divided by 65535, resulting in fluorescence values between 0 and 1, and avoiding errors which might be caused by systems which report fluorescence values on other scales (e.g., from -32,768 to 32768). This set of normalized fluorescence values 704 could then be supplemented with any values which were necessary to ensure that the dimensions of the machine learning input would match the dimensions of the input used to train the machine learning model which would be used to make the MIC determination. For example, if the machine learning model had been trained on fluorescence data gathered from 0.06, 0.125, 0.25, 0.5, 1, 2, 4, 8, 16, 32, 64 and 128 pg / ml dilutions but the set of normalized fluorescence values 704 only included data gathered from dilutions of 0.125, 0.5, 1 and 32 pg / ml dilutions, placeholder values (e.g., random values, all zeros, all ones) could be added for the dilutions 0.06, 0.25, 2, 4, 8, 16, 64 and 128 pg / ml, thereby providing an imputed set of normalized values 705 with values extending from the beginning (Ro) to the end (RN) of the expected input range.

[0050] Turning next to FIG. 7C, after the imputed set of normalized values 705 had been created, the values from that set of values 705 could be projected into a higher dimensionality embedding space to generate a set of projected tensors 706 comprising a tensor for each of the values in the imputed set of normalized values. Each of the values in the set of projected tensors 706 can then be modified to incorporate positional information reflecting the time and concentration information absent from the values of the set of projected tensors 706, thereby providing a set of embedded tensors 707. For example, each of the values of P(N(Fo), 0) to P(N(Fo), n) which were based on values of Fo gathered during a third data gathering cycle could be combined with positional information in the form of a two dimensional sinusoidal positional embedding in a space where one dimension was the data gathering cycle (in this case the third data gathering cycle) and the other dimension was the concentration (0.125 pg / ml, to continue the example in which the test wells were only inoculated with the antimicrobial at dilutions of 0.125, 0.5, 1 and 32 pg / ml). This set of embedded tensors 707 could then, along with instructions to mask or otherwise disregard the values not derived from actual measurements (e.g., values derived from Ro or RN, rather than values derived from Fo, Fi, or other fluorescence readings), be treated as the machine learning input for purposes of MIC determination.

[0051] Returning now to the discussion of FIG. 6, after the machine learning inputs had been determined 605, a check 606 may be performed to determine if those inputs could be used to determine the MIC. This may be done, for example, by providing the machine learning inputs to a model which had been trained to make MIC determinations, and seeing if it was able to provide a MIC with sufficient confidence given the information it was provided. If a MIC determination could not be made (e.g., if the machine learning model could not provide an MIC with sufficient confidence given the inputs provided), then a new set of test well evaluation values could be received 603 on a further data gathering cycle, and the determination of inputs 605 and the check 606 on whether a MIC could be determined could be repeated based on this new information. Alternatively, if a MIC could be determined based on the machine learning inputs, then that determination 607 would be performed - e.g., by specifying that an MIC which had been determined to have a sufficient confidence during the check 606 should be treated as the MIC for the antimicrobial and sample under analysis.

[0052] It should be understood that, while the above discussion of FIGS. 6 and 7A-7C illustrated how minimum inhibitory concentration may potentially be reported in a system implemented based on this disclosure, that discussion is intended to be illustrative only, and the disclosed technology can be implemented in manners which differ from the particular examples given above. For instance, while the above disclosure described a process in which machine learning inputs were determined 605 before checking 606 whether MIC could be determined, this order is not a requirement, and may be varied from in some implementations. For example, in some cases, MIC determinations may be made using a machine learning model which had been validated as being able to determine MIC after 12 data gathering cycles. In such a case the check 606 may simply be whether 12 data gathering cycles had taken place, and the machine learning inputs may be determined 605 only after sufficient data collection cycles were complete. Similarly, in some cases, the acts of creating 601 and incubating 602 test mixtures and checking 606 if MIC could be determined may be omitted entirely. For example, if the present disclosure was used to implement a system which made MIC determinations based on images captured by a separate biological analyzer, then the system making the MIC determination may simply receive 603 the test well evaluation values, use them to determine 605 machine learning inputs, and then determine 607 MIC based on providing those inputs to a machine learning model. Additional variations are also possible, and will be immediately apparent to those of skill in the art in light of this disclosure. Accordingly, the above exemplary variations, like the discussion of FIGS. 6 and 7A-7C, should not be treated as implying limits on the scope of protection provided by this document or any related document.

[0053] III. Machine Learning for MIC Determination

[0054] FIGS. 8A-8B provide a high level architecture of a machine learning model that systems and methods implemented based on this technology may use for making an MIC determination. In that architecture, FIG. 8A illustrates a first part of the machine learning model which would encode a set of inputs (e.g., the embedded tensors 707 from FIG. 7C, illustrated in FIG. 8A as inputs Io, Ii to IM) SO that they could be processed by a second part of the model, illustrated in FIG. 8B, which could provide a MIC as its output. Starling with FIG. 8A, that part of the machine learning model is made up of a plurality of encoders 801-1 to 801-e, each of which would itself comprise a self-attention layer 802 and a feed forward layer 803 (only illustrated in FIG. 8 A for the first encoder 801-1). The self attention layer 802 includes a set of weight matrices, labeled as WQ, WKand Wvin FIG. 8A, which could be used to create a corresponding output that could be provided to that encoder’ s feed forward layer 803. This may be done, for example, using a process such as shown in FIG. 9, discussed below.

[0055] In the method of FIG. 9, initially, a set of vectors would be determined 901 from the inputs to the encoder which includes the self-attention layer (i.e., Io, Ii and IM, in the case of encoder 801- 1 in FIG. 8 A) and that self-attention layer’s weight matrices (i.e., matrices WQ, WKand Wvin the self-attention layer 802 of encoder 801-1 illustrated in FIG. 8A). This may be done, for example by, for each input, multiplying that input by each of the weight matrices, resulting in each input having one vector for each of the encoder’s weight matrices. For instance, in the case of input Io in FIG. 8A, three vectors, Qo, Ko and Vo could be created by multiplying Io and the weight matrices WQ, WKand Wv. After these vectors had been determined 901, a check 902 could be performed to determine if further vector scores were needed. This check 902 may be performed, for example, by checking if, for each input, a score had been created for both that input and each other input reflecting the relevance of the various inputs to each other (and to themselves). If new vector scores were needed (and, the first time the check 902 was performed, they would be, since no scores would have been created at that time), the process could proceed with identifying 903 an input for which scores still needed to be created (which, on the first iteration, could be any of the inputs).

[0056] Once an input needing vector scores had been identified 903, the necessary vector scores could be created 904 for that input. This may be done, for example, selecting a first vector for the input whose scores are being created (e.g., the Qo vector, when scores are being created for input Io) and taking the dot product of that vector and a second vector for both that input and each other input (e.g., taking the dot product of Qo and Ko, the dot product of Qo and Ki, through the dot product of Qo and KM). This could then be repeated until, for each input, a set of scores had been determined which had the same cardinality as the inputs to the self-attention layer, at which point those scores and the remaining vectors could be used to create 905 the self- attention layer’s outputs. This could be done, for example, by normalizing the scores (e.g., applying a softmax function so that the scores for each input would all be positive and sum to 1), multiplying them by the vector which had not been used in creating the scores (e.g., the Vo vector, in the case of input Io), and summing the values of the vectors obtained by that multiplication to obtain outputs corresponding to each of the self-attention layer’s inputs. Additionally, when creating 905 the output of the self-attention layer, any masking instructions which had been provided to the machine learning model could be applied (e.g., by setting scores corresponding to masked inputs to 0) so that inputs which were not actually based on measured values would not impact the ultimate MIC determination.

[0057] Returning now to the discussion of FIG. 8 A, after the self-attention layer’s outputs had been determined (e.g., using a process such as discussed above in the context of FIG. 9), those outputs could be provided to a feed forward layer 803. This layer could, for each of the selfattention layer’s outputs, apply each of the values in that output to an input node in a feed forward neural network, thereby generating an output for the feed forward layer comprising a number of values equal to the number of output nodes in the feed forward neural network. This number may be equal to the number of values in each input (e.g., the number of dimensions in the embedding space used in creating the projected tensors 706), thereby allowing each of the encoders to have an identical structure (albeit with different weight values in their weight matrices and the feed forward neural networks). Alternatively, it is also possible that the number of values in each output of the feed forward layer (and therefore in the encoder which contains it) may be different, with the specific numbers of values and relationships between inputs and outputs potentially varying from case to case. Finally, once the last encoder in the machine learning model had processed its inputs and created its outputs, those outputs could be provided to a set of decoders, discussed below in the context of FIG. 8B.

[0058] Turning now to FIG. 8B, as shown in that figure, the outputs of the final encoder (illustrated as Ei, E2 to EM) can be provided to the decoders (of which only a single decoder 804-1 has its internal structure depicted in detail in FIG. 8B) as a second machine learning input, potentially along with some additional information such as the antimicrobial and microorganism strain for which MIC is being evaluated. This may be done, for example, by creating embeddings for the antimicrobial and microorganism strain (e.g., by projection into an embedding space, or through generating sinusoidal positional embeddings as described previously in the context of FIGS. 7A- 7C), adding those embeddings to the outputs of the final encoder, and then providing those combined values as inputs to a first decoder 804-1 from a set of decoders 804-1, 804-2 to 804- f. This decoder may then process those inputs using a self-attention layer 805 in a manner similar to that described for the self- attention layer 802 illustrated in FIG. 8A. These outputs may then be combined with the original outputs from the final encoder and provided to an encoder-decoder attention layer 806. This encoder-decoder attention layer 806 could process its inputs in the same manner as described for the self attention layers 802, 805, and provide its outputs to a feed forward layer 807 which could generate the outputs of the decoder in the same manner as the feed forward layer 803 would generate the outputs of the encoders which comprised it. This process could be repeated for each of the decoders in the machine learning model, and the outputs of the final decoder could be provided as inputs to a neural network 808 which could determine the MIC for the strain and antimicrobial in question, e.g., by providing it as a value on a final output node. a. Variations on, and different approaches than, the approaches described above and illustrated in FIGS. 8 A, 8B and 9 are also possible when applying machine learning to determine minimum inhibitory concentration. For example, in some cases an architecture following FIGS. 8 A and 8B may include additional layers, such as normalization layers after the various self attention, encoder-decoder attention and feed forward layers. Similarly, while the illustrations of FIGS. 8 A and 8B depicted single level neural networks in the feed forward layers and as the final neural network which would provide a MIC determination, it is possible that networks with other structures, such as with one or more hidden layers, could be used instead of single layer neural networks as shown. Similarly, in some cases, as an alternative to, or in addition to, a neural network which provides a MIC as a value on a sole output node, it is possible that there may be a neural network which provides growth / inhibition predictions for each of a set of concentrations (e.g., serial concentrations from the minimum to the maximum of a reporting range), and these predictions may be reported directly or used to provide a single MIC value (e.g., the lowest concentration for which the prediction is inhibition rather than growth). Indeed, other types of machine learning architectures may also be used, such as transformers as described in Ashish Vaswani, et. al., Attention Is All You Need, available g / abs / 1706.03762; and / or recurrent neural networks as described in Robin M. Schmidt, Recurrent Neural Networks (RNNs): A gentle Introduction and Overview, available at each of which is hereby incorporated by reference in its entirety. Additional variations are also possible, and will be immediately apparent to those of skill in the art in light of this disclosure. Accordingly, the above exemplary variations, like the discussion of FIGS. 8A-8B and 9, should not be treated as implying limits on the scope of protection provided by this document or any related document.

[0059] IV. Machine Learning Training

[0060] To train a machine learning model such as could be implemented based on this disclosure to make MIC determinations, a universal dilution model - e.g., one capable of receiving inputs for, and to provide MIC predictions across a complete range of concentrations for a broadest possible reporting range - could be built. This may be done by providing training data made up of fluorescence values gathered from a complete range of dilutions across the desired reporting range, asking the machine learning model to make MIC predictions (and / or growth / inhibition predictions, depending on the implementation) and then updating the model based on the differences between its predictions and the actual measured minimum inhibitory concentrations associated with those fluorescence values. These differences may be simple comparisons between predicted and actual MIC, but may alternatively be based on the following types of loss evaluations, depending on what is seen as most important in a particular scenario:

[0061] Variations on the described training are also possible. For instance, in some cases, more than one type of loss may be considered simultaneously, such as by using an objective function comprising a weighted sum of penalties for each of the losses to be considered. Similarly, to address the potential that actual data used for making MIC determinations may not cover the entire scope of concentrations used in a universal model, during training concentrations from the training data may be skipped (c.g., randomly based on a binomial distribution) to ensure that this docs not prevent the model from making accurate MIC determinations (and / or predictions of growth / inhibition). Various types of validations, such as k-fold cross validations across the sweep of all possible dilution combinations may also be performed. Other variations on how such a machine learning model may be trained are also possible and will be immediately apparent to those of skill in the art. Accordingly, the above exemplary variations, like the discussion of training which preceded them, should not be treated as implying limits on the scope of protection provided by this document or any related document.

[0062] V. Optimal Test Well Determination

[0063] Other applications of the disclosed technology beyond determining MIC for a given biological sample are also possible. For example, the disclosed technology may also be used to determine minimum numbers of test wells which are needed for data collection in order to provide MIC determinations across a required reporting range. For example, by applying the disclosed technology, it is possible to generate MICs across a reporting range including 0.06, 0.125, 0.25, 0.5, 1, 2, 4, 8, 16, 32, 64 and 128 pg / ml concentrations using only four (and in some cases fewer) test wells, thereby allowing significantly more antimicrobials to be evaluated in a single panel. A process which may be used to determine such minimum numbers of test wells is described below in the context of FIG. 10.

[0064] As shown in FIG. 10, a method for determining optimized sets of test wells could begin with training 1001 a machine learning model able to make MIC determinations across a full reporting range. This may be done, for example, by using the techniques described in the preceding section for training machine learning models such as those shown in FIGS. 8A-8B. Once the trained machine learning model was available, a check 1002 could be performed of whether there were subsets of concentrations from the full reporting range that needed to be checked for whether they could accurately predict MIC. This could be done, for example, by determining if there were any sets of concentrations which (i) had been established as suitable for determining MIC across the entire reporting range; and (ii) had at least one subset of concentrations with one lower cardinality that had not been evaluated for suitability for determining MIC across the entire reporting range. If there were no subsets that still needed to be evaluated, then the process could be finished 1003, with the smallest (c.g., lowest cardinality) set of concentrations which had been established as being suitable for determining MIC across the full reporting range for the strain and antimicrobial under consideration being treated as the optimal set of concentrations to use for that strain and antimicrobial.

[0065] Alternatively, if there were subsets that still needed to be evaluated, the process of FIG. 10 could proceed with going 1004 to the next subset so that that subset could be evaluated. For example, a system performing a process such as shown in FIG. 10 could determine that the next subset of concentrations to test is a subset created by removing one of the concentrations from a larger set which had previously been determined to be suitable. The performance of this subset could then be evaluated 1005. For example, sets of test well measurements could be reduced by omitting concentrations not present in the subset being evaluated and the MIC predictions made based on those reduced sets could be compared with the MIC predictions created based on the full set of test measurements (or on ground truth MIC determinations, if available). These comparisons could then be repeated until it was possible to say that the subset could (or couldn’t) be used for making MIC determinations across the entire reporting range at whatever level of statistical confidence was needed (e.g., based on regulatory requirements) and, when the subset had been evaluated 1005 the process could return to the check 1002 for further subsets so that the evaluation could be continued until all potentially viable subsets had been evaluated 1005.

[0066] It should be understood that, as with the other aspects of this disclosure, variations are possible on how optimal sets of test wells can be determined. For example, when going 1004 to the next subset, some embodiments implementing the approach described above could determine that the next subset to be evaluated should be a subset of the smallest set which had previously been evaluated as suitable (i.e., depth first search), while other embodiments may determine that the next subset to be evaluated should be a subset of the largest set which had previously been evaluated (i.e., breadth first search). As another example, it is possible that an embodiment may start with smallest reasonable subsets (e.g., starting by evaluating all subsets with at least three concentrations), and would only continue to test further subsets until at least one subset with suitable performance was identified. Other variations are also possible, and will be immediately apparent to those of ordinary skill in the art in light of this disclosure. Accordingly, the above exemplary variations, like the discussion of FIG. 10, should not be treated as implying limits on the scope of protection provided by this document or any related document.

[0067] VI. Performance Data

[0068] Technology such as described in the preceding sections can be used to determine the MIC for antimicrobials such as Amikacin, Amoxicillin / K Clavulanate (2:1), Amoxicillin / K Clavulanate (constant K Clavulanate), Ampicillin, Ampicillin / Sulbactam (2:1), Ampicillin / Sulbactam (constant Sulbactam), Aztreonam, Cefaclor, Cefazolin, Cefepime, Cefepime / Taniborbactam, Cefoperazone / Sulbactam, Cefoxitin, Cefotaxime, Ceftazidime, Ceftazidime / Avibactam, Ceftolozane / Tazobactam, Ceftriaxone, Ciprofloxacin, Colistin, Ertapenem, Fosfomycin, Gentamicin, Imipenem, Imipenem / Relebactam, Levofloxacin, Meropenem, Meropenem / Vaborbactam, Minocycline, Nitrofurantoin, Penicillin, Piperacillin / Tazobactam, Tetracycline, Tigecycline, Tobramycin, V ancomycin, and Trimethoprim / Sulfamethoxazole when used for the treatment of microbial strains such as Acinetobacter spp., Citrobacter freundii complex, Citrobacter koseri, Enterobacter cloacae, Enterobacter cloacae spp. complex, Escherichia coli, Klebsiella aerogenes, Klebsiella oxytoca, Klebsiella pneumoniae, Morganella morganii, Proteus mirabilis, Pseudomonas aeruginosa, Serratia marcescens, Stenotrophomonas maltophilia, and Gram positive bacteria such as Staphylococcus spp. and Enterococcus spp.

[0069] That the disclosed technology can provide MIC determinations significantly faster than in the prior art is demonstrated in FIG. 14, which provides effective agreement between actual MIC and MIC predicted for various combinations of antimicrobials and microorganisms by a system implemented based on the disclosed technology after 7 hours (13 imaging cycles), which is significantly faster than the 16-24 hours that would be expected in the prior art. Table 2, below, indicates the antimicrobials and antimicrobial combinations indicated by the abbreviations in FIG. 14, as well as the dilution sequences which were used in the testing which generated the data shown in that figure. The abbreviation NA, not included in table 2, is used in FIG. 14 to indicate results which were not therapeutically relevant to report.

[0070] Table 2

[0071] Additionally, it was also found that in many cases MIC predictions could be made in four hours or less, though for some microorganisms a longer time, such as 7 hours, was found to be necessary. That the disclosed technology can allow fewer dilutions to be used in determining MIC is demonstrated in FIG. 15 which shows effective agreement between MIC determined for Klebsiella pneumoniae, Enterobacter cloacae, Pseudomonas aeruginosa, Acinetobacter spp., Proteus mirabilis and Stenotrophamonas maltophilia using reduced dilution sequences and MIC determined for those microorganisms using complete dilution sequences of the indicated antimicrobials or antimicrobial combinations. This use of fewer dilutions for determining MIC has also been tested with and found to be suitable for generating MIC predictions for other microorganisms, including Citrobacter freundii complex, Citrobacter koseri, Enterobacter cloacae spp. complex, Escherichia coli, Klebsiella aerogenes, Klebsiella oxytoca, Morganella morganii, Serratia marcescens, and Gram positive bacteria such as Staphylococcus spp. and Enterococcus spp.

[0072] VII. Further Illustrations

[0073] As a further illustration of potential variations on manners in which the disclosed technology can be implemented and applied, consider FIGS. 11-13, each of which is discussed below.

[0074] Starting with FIG. 11, that figure illustrates a set of minimum inhibitory concentration reporting acts which comprise receiving 603 a plurality of sets of test well evaluation values. A first set of machine learning inputs could then be determined 1101. This determination 1101 may include determining 1102 an embedding vector (e.g., one of the projected tensors 706 created by projecting either imputed values or normalized fluorescence values into an embedding space). The determination 1101 of the first set of machine learning inputs may also include determining 1103 a cycle vector and determining 1104 a concentration vector. This may be done, for example, by creating sinusoidal position vectors for the cycles and concentrations associated with the concentration values for an antimicrobial susceptibility test. These vectors - i.e., the embedding vector, the cycle vector and the concentration vector - could then be combined 1105, such as by performing elementwise addition on the vectors in a manner similar to that described previously for creating the set of embedded tensors 707 in the context of FIG. 7C. In an example, as also mentioned in that context, the first set of machine learning inputs may comprise instructions to mask values corresponding to concentrations from the complete set of concentrations which are not comprised by the subset of the complete set of concentrations for that set of times. a. In the process of FIG. 11, after the first set of machine learning inputs has been determined 1101, a minimum inhibitory concentration can be determined 1106 based on providing the first set of machine learning inputs to a machine learning model. As shown in FIG. 1 1 , this can include obtaining 1107 a first machine learning output, for example, by providing the first set of machine learning inputs to an encoder to obtain a set of outputs values such as those illustrated as Ei, E2 to EM in FIGS. 8 A and 8B. For example, in the present disclosure, determining the minimum inhibitory concentration may include determining a second set of machine learning inputs based on the first machine learning outputs. For each performance of the set of minimum inhibitory concentration reporting acts, the respective reporting acts corresponding to a respective microorganism and an antimicrobial, the second set of machine learning inputs may be based on the microorganism and the antimicrobial. Second machine learning output may be obtained, for example based on, particularly by, providing a second set of machine learning inputs to a decoder. Minimum inhibitory concentration may be based on the second machine learning output. In the process illustrated by FIG. 11 , in addition to obtaining 1107 the first machine learning output, determining the minimum inhibitory concentration 1106 may also include determining an antimicrobial 1108 vector and determining 1109 a microbe vector, such as by creating embeddings for antimicrobial and microorganism strain as described in the context of FIG. 8B. These vectors may be used to determine 1110 a second set of machine learning inputs by combining them with vectors from the first set of machine learning outputs (e.g., by elementwise addition). A second machine learning output may then be obtained 1111 (e.g., by providing the second set of machine learning inputs to a decoder, as described in the context of FIG. 8B), and the minimum inhibitory concentration for the microbe and antimicrobial under consideration could be determined 1112 based on that second machine learning output.

[0075] In some examples, for each performance of the set of minimum inhibitory concentration reporting acts, the first machine learning output may comprise a set of output vectors, each of which has a dimensionality which is the same for all vectors from the set of output vectors and determining the minimum inhibitory concentration based on providing the first set of machine learning inputs to the machine learning model comprises: determining an antimicrobial vector based on the antimicrobial corresponding to that performance of the set of minimum inhibitory concentration reporting acts, wherein the antimicrobial vector has the same dimensionality as the vectors from the set of output vectors, determining a microorganism vector based on the microbe corresponding to that performance of the set of minimum inhibitory concentration reporting acts, wherein the microorganism vector has the same dimensionality as the vectors from the set of output vectors, and determining the second set of machine learning inputs based on, for each vector from the set of output vectors, combining that vector with the antimicrobial vector and the microorganism vector.

[0076] In some examples, for each performance of the set of minimum inhibitory concentration reporting acts, determining the first set of machine learning inputs based on the plurality of sets of test well evaluation values comprises, for each test well evaluation value: determining an embedding vector (having a dimensionality) based on that test well evaluation value, determining a cycle vector based on that test well evaluation value’s imaging cycle and having the same dimensionality as the embedding vector based on that test well evaluation value, determining a concentration vector based on the concentration corresponding to the set of test well evaluation values which comprises that test well evaluation value, and combining the embedding vector corresponding to that test well evaluation value with the cycle vector and the concentration vector. The embedding vector “corresponding to” a test well evaluation value may be the embedding vector determined based on the test well evaluation value.

[0077] Turning next to FIG. 12, that figure illustrates a method which can be used to determine an optimized set of concentrations to include in a test panel. As shown in the figure, such a method may include performing 1201 the set of minimum inhibitory concentration acts (e.g., from FIG. 11) repeatedly (e.g., a plurality of sets of times, wherein each set of times corresponds to a subset of a complete set of concentrations, where a complete set of concentrations can be a serial set of concentrations across a reporting range, such as a sequence of concentrations starting with a minimum reportable concentration and doubling with each concentration thereafter). For example, the machine learning model is configured to receive machine learning inputs corresponding to a complete set of concentrations.

[0078] For each set of times, that set of times may correspond to a subset of the complete set of concentrations which is different from the subset of the complete set of concentrations for any other set of times from the plurality of sets of times. On each performance of the set of minimum inhibitory concentration reporting acts in that set of times, the concentrations corresponding to the plurality of sets of test well evaluation values may be the same as the subset of the complete set of concentrations corresponding to that time.

[0079] The method may also include, for each set of times corresponding to sets of performances of the set of minimum inhibitory concentration reporting acts, determining 1202 an effectiveness value. This may be done, for example, as described previously for evaluating 1005 the performance of subsets of concentrations in the context of FIG. 10. For example, the effectiveness value may be determined based on an agreement between ground truth minimum inhibitory concentrations and the minimum inhibitory concentrations determined in that set of times. Based on this, a recommended set of test panel concentrations can be determined 1203, such as using techniques described previously in the context of FIG. 10 for identifying a subset of concentrations when evaluation of the potential subsets was finished 1003.

[0080] For example, the recommended set may be determined based on the performance of the set of minimum inhibitory concentration reporting acts the plurality of sets of times. The recommend set of test panel concentrations may be the subset of the complete set of concentrations which corresponds to a selected set of times from the plurality of sets of times. In an example, for each set of times from the plurality of sets of times, other than the selected set of times, at least one of the following statements is false:

[0081] (i) the effectiveness value for that set of times is higher than the effectiveness value for the selected set of times; and

[0082] (ii) the subset of the complete set of concentrations for that set of times has a cardinality less than a cardinality for the subset of the complete set of concentrations for the selected set of times.

[0083] In an example, for each set of times (except the selected set of times), the effectiveness value is equal to or lower than the effectiveness value for the selected set of times and / or the cardinality of the subset of the complete set of concentrations is equal to or higher than for the selected set of times. In such an example, the selected set of times may be one with highest effectiveness value and / or lowest cardinality of the subset. Turning finally to FIG. 13, that figure illustrates a method which may be used for determining minimum inhibitory concentration for a plurality of antimicrobials, for example, using a system such as the biological testing system 1 as described herein. Initially, in the method of FIG. 13, test mixtures are created 1301 in a set of test wells, comprising a set of test mixtures for each antimicrobial agent from a set of antimicrobial agents. A panel comprising those test wells could then be incubated 1302, and the set of MIC reporting acts (e.g., from FIG. 11) could be performed 1303 for each of the antimicrobial agents. This performance may include, for each of the antimicrobial agents, capturing 1304 fluorescence values from test wells inoculated using that antimicrobial agent. These values may then be used, for example, with techniques described in the context of FIGS. 6-9, to make MIC determinations for each of the antimicrobials included in the panel.

[0084] The above procedures described in the context of FIGS. 11 to 13, including the steps they refer to from the other FIGS, may be carried out by a system according to the present disclosure, such as system comprising a processor configured to carry out the above steps, particularly perform the set of minimum inhibitory concentration reporting acts. In particular, a system may be used that comprises a processor and a non-transitory computer readable medium having stored thereon instructions operable to, when executed using the processor, perform the above steps, particularly perform the set of minimum inhibitory concentration reporting acts. E.g., the instructions, when executed by the processor, may cause the processor to perform the above steps, particularly perform the set of minimum inhibitory concentration reporting acts The system may further comprise a fluorometer, which may be configured to capture, as part of receiving the test well evaluation values, fluorescence values for the test wells, and / or an incubator, which may be configured to be used for the above-described incubating. For example, a system as described in the context of the figures may be used.

[0085] VIII. Exemplary Combinations

[0086] The following examples relate to various non-exhau stive ways in which the teachings herein may be combined or applied. It should be understood that the following examples are not intended to restrict the coverage of any claims that may be presented at any time in this application or in subsequent filings of this application. No disclaimer is intended. The following examples are being provided for nothing more than merely illustrative purposes. It is contemplated that the various teachings herein may be arranged and applied in numerous other ways. It is also contemplated that some variations may omit certain features referred to in the below examples. Therefore, none of the aspects or features referred to below should be deemed critical unless otherwise explicitly indicated as such at a later date by the inventors or by a successor in interest to the inventors. If any claims are presented in this application or in subsequent filings related to this application that include additional features beyond those referred to below, those additional features shall not be presumed to have been added for any reason relating to patentability.

[0087] Example 1

[0088] A method comprising one or more performances of a set of minimum inhibitory concentration reporting acts, wherein the set of minimum inhibitory concentration reporting acts comprises: (a) receiving a plurality of sets of test well evaluation values, wherein, for each set of test well evaluation values: (i) that set of test well evaluation values corresponds to a concentration which is different from the concentration for each other set of test well evaluation values from the plurality of sets of test well evaluation values; and (ii) each of the test well evaluation values in that set of test well evaluation values has: (A) a fluorescence value for a test well inoculated with a biological sample and an imaging cycle; and (B) an imaging cycle during which that test well evaluation value’s fluorescence value was captured; (b) determining a first set of machine learning inputs based on the plurality of sets of test well evaluation values; and (c) determining a minimum inhibitory concentration based on providing the first set of machine learning inputs to a machine learning model, wherein the machine learning model is configured to provide minimum inhibitory concentration values which are equal to concentrations corresponding to the plurality of sets of test well evaluation values, and to provide minimum inhibitory concentration values which are not equal to any of the concentrations corresponding to the test well evaluation values.

[0089] Regarding (A) the fluorescence value may, for example, be seen as a fluorescence value for the test well for the imaging cycle. The imaging cycle in (B) may correspond to the imaging cycle in (A).

[0090] Example 2

[0091] The method of example 1 , wherein: (a) the machine learning model is configured to receive machine learning inputs corresponding to a complete set of concentrations; (b) the method comprises: (i) performing the set of minimum inhibitory concentration reporting acts a plurality of sets of times, wherein for each set of times from the plurality of sets of times: (A) that set of times corresponds to a subset of the complete set of concentrations which is different from the subset of the complete set of concentrations for any other set of times from the plurality of sets of times; (B) on each performance of the set of minimum inhibitory concentration reporting acts in that set of times, the concentrations corresponding to the plurality of sets of test well evaluation values are the same as the subset of the complete set of concentrations corresponding to that time; (ii) for each set of times from the plurality of sets of times, determining an effectiveness value for that set of times based on an agreement between ground truth minimum inhibitory concentrations and the minimum inhibitory concentrations determined in that set of times; (iii) determining a recommended set of test panel concentrations based on the performance of the set of minimum inhibitory concentration reporting acts the plurality of sets of times, wherein: (A) the recommend set of test panel concentrations is the subset of the complete set of concentrations which corresponds to a selected set of times from the plurality of sets of times; (B) for each set of times from the plurality of sets of times, other than the selected set of times, at least one of the following statements is false: (i) the effectiveness value for that set of times is higher than the effectiveness value for the selected set of times; and (ii) the subset of the complete set of concentrations for that set of times has a cardinality less than a cardinality for the subset of the complete set of concentrations for the selected set of times.

[0092] Regarding (b)(iii), determining a recommended set of test panel concentrations based on the performance of the set of minimum inhibitory concentration reporting acts the plurality of sets of times may be performed by making that determination based on the outcome of performing reporting acts.

[0093] Example 3

[0094] The method of example 2, wherein for each set of times from the plurality of sets of times, the first set of machine learning inputs comprises instructions to mask values corresponding to concentrations from the complete set of concentrations which are not comprised by the subset of the complete set of concentrations for that set of times.

[0095] Example 4

[0096] The method of any of examples 1-3, wherein in each performance of the set of minimum inhibitory concentration reporting acts, determining the minimum inhibitory concentration based on providing the first set of machine learning inputs to the machine learning model comprises: (a) obtaining a first machine learning output based on providing the first set of machine learning inputs to an encoder; (b) obtaining a second machine learning output based on providing a second set of machine learning inputs to a decoder, wherein the second set of machine learning inputs is based on the first machine learning output; and (c) determining the minimum inhibitory concentration based on the second machine learning output.

[0097] Example 5

[0098] The method of example 4, wherein, for each performance of the set of minimum inhibitory concentration reporting acts: (a) that performance of the set of minimum inhibitory concentration reporting acts corresponds to a microorganism and an antimicrobial; and (b) the second set of machine learning inputs is based on the microorganism and the antimicrobial.

[0099] Example 6

[0100] The method of example 5, wherein, for each performance of the set of minimum inhibitory concentration reporting acts: (a) the first machine learning output comprises a set of output vectors, each of which has a dimensionality which is the same for all vectors from the set of output vectors; (b) determining the minimum inhibitory concentration based on providing the first set of machine learning inputs to the machine learning model comprises: (i) determining an antimicrobial vector based on the antimicrobial corresponding to that performance of the set of minimum inhibitory concentration reporting acts, wherein the antimicrobial vector has the same dimensionality as the vectors from the set of output vectors; (ii) determining a microorganism vector based on the microbe corresponding to that performance of the set of minimum inhibitory concentration reporting acts, wherein the microorganism vector has the same dimensionality as the vectors from the set of output vectors; and (iii) determining the second set of machine learning inputs based on, for each vector from the set of output vectors, combining that vector with the antimicrobial vector and the microorganism vector.

[0101] Example 7

[0102] The method of any of examples 4-6, wherein, for each performance of the set of minimum inhibitory concentration reporting acts, determining the first set of machine learning inputs based on the plurality of sets of test well evaluation values comprises, for each test well evaluation value: (a) determining an embedding vector having a dimensionality and based on that test well evaluation value; (b) determining a cycle vector based on that test well evaluation value’s imaging cycle and having the same dimensionality as the embedding vector based on that test well evaluation value; (c) determining a concentration vector based on the concentration corresponding to the set of test well evaluation values which comprises that test well evaluation values; and (d) combining the embedding vector corresponding to that test well evaluation value with the cycle vector and the concentration vector.

[0103] Example 8

[0104] The method of any of examples 1-7, wherein: (a) the method comprises, for each antimicrobial agent from a set of antimicrobial agents creating a set of test mixtures in a set of test wells, wherein each test mixture from the set of test mixtures is inoculated using: (i) that antimicrobial agent; and (ii) a biological sample; (b) incubating a panel comprising, for each antimicrobial agent from the set of antimicrobial agents, the set of test wells comprising the set of test mixtures created for that antimicrobial agent; (c) performing the set of minimum inhibitory concentration reporting acts a set of times corresponding to the set of antimicrobial agents, wherein, for each performance of the set of minimum inhibitory concentration reporting acts from the set of times corresponding to the set of antimicrobial agents: (i) that performance of the set of minimum inhibitory concentration reporting acts corresponds to a single antimicrobial agent; and (ii) receiving the plurality of sets of test well evaluation values comprises, for each test well from the set of test wells inoculated using the antimicrobial agent corresponding to that performance, capturing a fluorescence value for that test well.

[0105] Example 9

[0106] The method of example 8, wherein the set of antimicrobial agents has a cardinality which is no less than one fifth of a number of test wells comprised by the panel.

[0107] Example 10

[0108] A system comprising a processor and a non-transitory computer readable medium having stored thereon instructions operable to, when executed using the processor, perform a method comprising performing a set of minimum inhibitory concentration reporting acts comprising: (a) receiving a plurality of sets of test well evaluation values, wherein, for each set of test well evaluation values: (i) that set of test well evaluation values corresponds to a concentration which is different from the concentration for each other set of test well evaluation values from the plurality of sets of test well evaluation values; and (ii) each of the test well evaluation values in that set of test well evaluation values has: (A) a fluorescence value for a test well inoculated with a biological sample and an imaging cycle; and (B) an imaging cycle during which that test well evaluation value’s fluorescence value was captured; (b) determining a first set of machine learning inputs based on the plurality of sets of test well evaluation values; and (c) determining the minimum inhibitory concentration based on providing the first set of machine learning inputs to a machine learning model, wherein the minimum inhibitory concentration is from a complete range of concentrations, wherein the concentrations corresponding to the plurality of sets of test well evaluation values are a subset of the complete range of concentrations.

[0109] Example 11

[0110] The system of example 10, wherein: (a) the machine learning model is configured to receive a machine learning inputs corresponding to a complete set of concentrations; (b) the method comprises: (i) performing the set of minimum inhibitory concentration reporting acts a plurality of sets of times, wherein for each set of times from the plurality of sets of times: (A) that set of times corresponds to a subset of the complete range of concentrations which is different from the subset of the complete range of concentrations for any other set of times from the plurality of sets of times; (B) on each performance of the set of minimum inhibitory concentration reporting acts in that set of times, the concentrations corresponding to the plurality of sets of test well evaluation values are the same as the subset of the complete set of concentrations corresponding to that time; (ii) for each set of times from the plurality of sets of times, determining an effectiveness value for that set of times based on an agreement between ground truth minimum inhibitory concentrations and the minimum inhibitory concentrations determined in that set of times; (iii) determining a recommended set of test panel concentrations based on the performance of the set of minimum inhibitory concentration reporting acts the plurality of sets of times, wherein: (A) the recommend set of test panel concentrations is the subset of the complete set of concentrations which corresponds to a selected set of times from the plurality of sets of times; (B) the plurality of sets of times does not include any set of times for which both: (I) the effectiveness value for that set of times is higher than the effectiveness value for the selected set of times; and (II) the subset of the complete set of concentrations for that set of times has a cardinality less than a cardinality for the subset of the complete set of concentrations for the selected set of times.

[0111] Example 12

[0112] The system of example 11, wherein for each set of times from the plurality of sets of times, the first set of machine learning inputs comprises instructions to mask values corresponding to concentrations from the complete set of concentrations which are not comprised by the subset of the complete set of concentrations for that set of times.

[0113] Example 13

[0114] The system of any of examples 10-12, wherein the instructions are operable to, each time the method is performed, determining the minimum inhibitory concentration based on providing the first set of machine learning inputs to the machine learning model by performing acts comprising: (a) obtaining a first machine learning output based on providing the first set of machine learning inputs to an encoder; (b) obtaining a second machine learning output based on providing a second set of machine learning inputs to a decoder, wherein the second set of machine learning inputs is based on the first machine learning output; and (c) determining the minimum inhibitory concentration based on the second machine learning output. Example 14

[0115] The system of example 13, wherein the method comprises, each time the set of minimum inhibitory concentration reporting acts is performed: (a) maintaining data identifying a microorganism and an antimicrobial corresponding to that performance of the set of minimum inhibitory concentration reporting acts; and (b) determining the second set of machine learning inputs based on the microorganism and the antimicrobial corresponding to that performance of the set of minimum inhibitory concentration reporting acts.

[0116] Example 15

[0117] The system of example 14, wherein, for each performance of the set of minimum inhibitory concentration reporting acts comprised by the method: (a) the first machine learning output comprises a set of output vectors, each of which has a dimensionality which is the same for all vectors from the set of output vectors; (b) determining the minimum inhibitory concentration based on providing the first set of machine learning inputs to the machine learning model comprises: (i) determining an antimicrobial vector based on the antimicrobial corresponding to that performance of the set of minimum inhibitory concentration reporting acts, wherein the antimicrobial vector has the same dimensionality as the vectors from the set of output vectors; (ii) determining a microorganism vector based on the microbe corresponding to that performance of the set of minimum inhibitory concentration reporting acts, wherein the microorganism vector has the same dimensionality as the vectors from the set of output vectors; and (iii) determining the second set of machine learning inputs based on, for each vector from the set of output vectors, combining that vector with the antimicrobial vector and the microorganism vector.

[0118] Example 16

[0119] The system of any of examples 13-15, wherein, for each performance of the set of minimum inhibitory concentration reporting acts comprised by the method, determining the first set of machine learning inputs based on the plurality of sets of test well evaluation values comprises, for each test well evaluation value: (a) determining an embedding vector having a dimensionality and based on that test well evaluation value; (b) determining a cycle vector based on that test well evaluation value’s imaging cycle and having the same dimensionality as the embedding vector based on that test well evaluation value; (c) determining a concentration vector based on the concentration corresponding to the set of test well evaluation values which comprises that test well evaluation values; and (d) combining the embedding vector corresponding to that test well evaluation value with the cycle vector and the concentration vector.

[0120] In particular, in the above examples, the instructions, when executed by the processor, may cause the processor to carry out the methods of the above examples.

[0121] The processor of the system of the examples above may be configured to carry out the methods mentioned in said examples.

[0122] Example 17

[0123] The system of any of examples 10-16, wherein: (a) the system comprises an incubator and a fluorometer; (b) the method comprises: (i) incubating a panel which comprises, for each antimicrobial agent from a set of antimicrobial agents, a set of test wells inoculated using that antimicrobial agent and the biological sample, in particular by means of the incubator; and (ii) performing the set of minimum inhibitory concentration reporting acts a set of times corresponding to the set of antimicrobial agents, in particular by means of the processor, wherein, for each performance of the set of minimum inhibitory concentration reporting acts from the set of times corresponding to the set of antimicrobial agents: (A) that performance of the set of minimum inhibitory concentration reporting acts corresponds to a single antimicrobial agent; and (B) receiving the plurality of sets of test well evaluation values comprises, for each test well from the set of test wells inoculated using the antimicrobial agent corresponding to that performance, capturing a fluorescence value for that test well using the fluorometer.

[0124] Example 18

[0125] The system of example 17, wherein the set of antimicrobial agents has a cardinality which is no less than one fifth of a number of test wells comprised by the panel.

[0126] Example 19

[0127] A machine comprising: (a) a fluorometer; and (b) means for reporting minimum inhibitory concentrations based on fluorescence values from the fluorometer.

[0128] Example 20 The machine of example 19, wherein the machine further comprises means for optimizing antimicrobial dilution using the means for reporting minimum inhibitory concentrations.

[0129] Example 21

[0130] The machine of example 19, wherein the machine comprises the system of any of examples 10 to 18.

[0131] Example 22

[0132] Computer program product comprising instructions which, when the program is executed by a computer, cause the computer to carry out the method of any of Examples 1 to 7.

[0133] Computer-readable medium comprising instructions which, when executed by a computer, cause the computer to carry out the method of any of Examples 1 to 7.

[0134] Miscellaneous

[0135] It should be understood that, in the above examples and the claims, a statement that something is “based on” something else should be understood to mean that it is determined at least in part by the thing that it is indicated as being based on. To indicate that something must be completely determined based on something else, it is described as being “based EXCLUSIVELY on” whatever it must be completely determined by.

[0136] It should be understood that, in the claims “means for reporting minimum inhibitory concentrations based on fluorescence values from the fluorometer” should be understood as a limitation set forth in means plus function form as set forth in 35 U.S.C. § 112(f), in which the function is “reporting minimum inhibitory concentrations based on fluorescence values from the fluorometer” and the corresponding structure is a computer configured with algorithms as illustrated in FIGS. 6-9 and 11, and described in section III.

[0137] It should be understood that, in the claims, “means for optimizing antimicrobial dilution” should be understood as a limitation set forth in means plus function form as set forth in 35 U.S.C. § 112(f), in which the function is “optimizing antimicrobial dilution” and the corresponding structure is a computer configured with algorithms as illustrated in FIGS. 10 and 12, and described in section VI. It should be understood that, in the claims, “set” should be understood as referring to one or more thing of similar nature, design or function.

[0138] It should be understood that any of the examples described herein may include various other features in addition to or in lieu of those described above. By way of example only, any of the examples described herein may also include one or more of the various features disclosed in any of the various references that are incorporated by reference herein.

[0139] It should be understood that any one or more of the teachings, expressions, embodiments, examples, etc. described herein may be combined with any one or more of the other teachings, expressions, embodiments, examples, etc. that are described herein. The above-described teachings, expressions, embodiments, examples, etc. should therefore not be viewed in isolation relative to each other. Various suitable ways in which the teachings herein may be combined will be readily apparent to those of ordinary skill in the art in view of the teachings herein. Such modifications and variations are intended to be included within the scope of the claims.

[0140] It should be appreciated that any patent, publication, or other disclosure material, in whole or in part, that is said to be incorporated by reference herein is incorporated herein only to the extent that the incorporated material does not conflict with existing definitions, statements, or other disclosure material set forth in this disclosure. As such, and to the extent necessary, the disclosure as explicitly set forth herein supersedes any conflicting material incorporated herein by reference. Any material, or portion thereof, that is said to be incorporated by reference herein, but which conflicts with existing definitions, statements, or other disclosure material set forth herein will only be incorporated to the extent that no conflict arises between that incorporated material and the existing disclosure material.

[0141] Having shown and described various versions of the present invention, further adaptations of the methods and systems described herein may be accomplished by appropriate modifications by one of ordinary skill in the art without departing from the scope of the present invention. Several of such potential modifications have been mentioned, and others will be apparent to those skilled in the art. For instance, the examples, versions, geometries, materials, dimensions, ratios, steps, and the like discussed above are illustrative and are not required.

[0142] Accordingly, the scope of the present invention should be considered in terms of the following claims and is understood not to be limited to the details of structure and operation shown and described in the specification and drawings.

Claims

CLAIMS1. A method comprising one or more performances of a set of minimum inhibitory concentration reporting acts, wherein the set of minimum inhibitory concentration reporting acts comprises:(a) receiving (603) a plurality of sets of test well evaluation values, wherein, for each set of test well evaluation values:(i) that set of test well evaluation values corresponds to a concentration which is different from the concentration for each other set of test well evaluation values from the plurality of sets of test well evaluation values; and(ii) each of the test well evaluation values in that set of test well evaluation values has:(A) a fluorescence value for a test well inoculated with a biological sample and an imaging cycle; and(B) an imaging cycle during which that test well evaluation value’s fluorescence value was captured;(b) determining (605, 1101) a first set of machine learning inputs based on the plurality of sets of test well evaluation values; and(c) determining (607, 1106) a minimum inhibitory concentration based on providing the first set of machine learning inputs to a machine learning model, wherein the machine learning model is configured to provide minimum inhibitory concentration values which are equal to concentrations corresponding to the plurality of sets of test well evaluation values, and to provide minimum inhibitory concentration values which are not equal to any of the concentrations corresponding to the test well evaluation values.

2. The method of claim 1, wherein:(a) the machine learning model is configured to receive machine learning inputs corresponding to a complete set of concentrations;(b) the method comprises:(i) performing (120l )the set of minimum inhibitory concentration reporting acts a plurality of sets of times, wherein for each set of times from the plurality of sets of times:(A) that set of times corresponds to a subset of the complete set of concentrations which is different from the subset of the complete set of concentrations for any other set of times from the plurality of sets of times;(B) on each performance of the set of minimum inhibitory concentration reporting acts in that set of times, the concentrations corresponding to the plurality of sets of test well evaluation values are the same as the subset of the complete set of concentrations corresponding to that time;(ii) for each set of times from the plurality of sets of times, determining (1202) an effectiveness value for that set of times based on an agreement between ground truth minimum inhibitory concentrations and the minimum inhibitory concentrations determined in that set of times;(iii) determining (1203) a recommended set of test panel concentrations based on the performance of the set of minimum inhibitory concentration reporting acts the plurality of sets of times, wherein:(A) the recommend set of test panel concentrations is the subset of the complete set of concentrations which corresponds to a selected set of times from the plurality of sets of times;(B) for each set of times from the plurality of sets of times, other than the selected set of times, at least one of the following statements is false:(i) the effectiveness value for that set of times is higher than the effectiveness value for the selected set of times; and(ii) the subset of the complete set of concentrations for that set of times has a cardinality less than a cardinality for the subsetof the complete set of concentrations for the selected set of times.

3. The method of claim 2, wherein for each set of times from the plurality of sets of times, the first set of machine learning inputs comprises instructions to mask values corresponding to concentrations from the complete set of concentrations which are not comprised by the subset of the complete set of concentrations for that set of times.

4. The method of any of claims 1-3, wherein in each performance of the set of minimum inhibitory concentration reporting acts, determining the minimum inhibitory concentration based on providing the first set of machine learning inputs to the machine learning model comprises:(a) obtaining a first machine learning output based on providing the first set of machine learning inputs to an encoder;(b) obtaining a second machine learning output based on providing a second set of machine learning inputs to a decoder, wherein the second set of machine learning inputs is based on the first machine learning output; and(c) determining the minimum inhibitory concentration based on the second machine learning output.

5. The method of claim 4, wherein, for each performance of the set of minimum inhibitory concentration reporting acts:(a) that performance of the set of minimum inhibitory concentration reporting acts corresponds to a microorganism and an antimicrobial; and(b) the second set of machine learning inputs is based on the microorganism and the antimicrobial.

6. The method of claim 5, wherein, for each performance of the set of minimum inhibitory concentration reporting acts:(a) the first machine learning output comprises a set of output vectors, each of which has a dimensionality which is the same for all vectors from the set of output vectors;(b) determining the minimum inhibitory concentration based on providing the first set of machine learning inputs to the machine learning model comprises:(i) determining an antimicrobial vector based on the antimicrobial corresponding to that performance of the set of minimum inhibitory concentration reporting acts, wherein the antimicrobial vector has the same dimensionality as the vectors from the set of output vectors;(ii) determining a microorganism vector based on the microbe corresponding to that performance of the set of minimum inhibitory concentration reporting acts, wherein the microorganism vector has the same dimensionality as the vectors from the set of output vectors; and(iii) determining the second set of machine learning inputs based on, for each vector from the set of output vectors, combining that vector with the antimicrobial vector and the microorganism vector.

7. The method of any of claims 4-6, wherein, for each performance of the set of minimum inhibitory concentration reporting acts, determining the first set of machine learning inputs based on the plurality of sets of test well evaluation values comprises, for each test well evaluation value:(a) determining an embedding vector having a dimensionality and based on that test well evaluation value;(b) determining a cycle vector based on that test well evaluation value’ s imaging cycle and having the same dimensionality as the embedding vector based on that test well evaluation value;(c) determining a concentration vector based on the concentration corresponding to the set of test well evaluation values which comprises that test well evaluation value; and(d) combining the embedding vector corresponding to that test well evaluation value with the cycle vector and the concentration vector.

8. The method of any of claims 1-7, wherein:(a) the method comprises, for each antimicrobial agent from a set of antimicrobial agents creating (1301) a set of test mixtures in a set of test wells, wherein each test mixture from the set of test mixtures is inoculated using:(i) that antimicrobial agent; and(ii) a biological sample;(b) incubating (1302) a panel comprising, for each antimicrobial agent from the set of antimicrobial agents, the set of test wells comprising the set of test mixtures created for that antimicrobial agent;(c) performing (1303) the set of minimum inhibitory concentration reporting acts a set of times corresponding to the set of antimicrobial agents, wherein, for each performance of the set of minimum inhibitory concentration reporting acts from the set of times corresponding to the set of antimicrobial agents:(i) that performance of the set of minimum inhibitory concentration reporting acts corresponds to a single antimicrobial agent; and(ii) receiving the plurality of sets of test well evaluation values comprises, for each test well from the set of test wells inoculated using the antimicrobial agent corresponding to that performance, capturing a fluorescence value for that test well.

9. The method of claim 8, wherein the set of antimicrobial agents has a cardinality which is no less than one fifth of a number of test wells comprised by the panel.

10. A system comprising a processor and a non-transitory computer readable medium having stored thereon instructions operable to, when executed using the processor, perform a method comprising performing a set of minimum inhibitory concentration reporting acts comprising:(a) receiving a plurality of sets of test well evaluation values, wherein, for each set of test well evaluation values:(i) that set of test well evaluation values corresponds to a concentration which is different from the concentration for each other set of test well evaluation values from the plurality of sets of test well evaluation values; and(ii) each of the test well evaluation values in that set of test well evaluation values has:(A) a fluorescence value for a test well inoculated with a biological sample and an imaging cycle; and(B) an imaging cycle during which that test well evaluation value’s fluorescence value was captured;(b) determining a first set of machine learning inputs based on the plurality of sets of test well evaluation values; and(c) determining the minimum inhibitory concentration based on providing the first set of machine learning inputs to a machine learning model, wherein the minimum inhibitory concentration is from a complete range of concentrations, wherein the concentrations corresponding to the plurality of sets of test well evaluation values are a subset of the complete range of concentrations.

11. The system of claim 10, wherein:(a) the machine learning model is configured to receive machine learning input corresponding to a complete set of concentrations;(b) the method comprises:(i) performing the set of minimum inhibitory concentration reporting acts a plurality of sets of times, wherein for each set of times from the plurality of sets of times:(A) that set of times corresponds to a subset of the complete range of concentrations which is different from the subset of the complete range of concentrations for any other set of times from the plurality of sets of times;(B) on each performance of the set of minimum inhibitory concentration reporting acts in that set of times, the concentrations correspondingto the plurality of sets of test well evaluation values are the same as the subset of the complete set of concentrations corresponding to that time;(ii) for each set of times from the plurality of sets of times, determining an effectiveness value for that set of times based on an agreement between ground truth minimum inhibitory concentrations and the minimum inhibitory concentrations determined in that set of times;(iii) determining a recommended set of test panel concentrations based on the performance of the set of minimum inhibitory concentration reporting acts the plurality of sets of times, wherein:(A) the recommend set of test panel concentrations is the subset of the complete set of concentrations which corresponds to a selected set of times from the plurality of sets of times;(B) the plurality of sets of times does not include any set of times for which both:(I) the effectiveness value for that set of times is higher than the effectiveness value for the selected set of times; and(II) the subset of the complete set of concentrations for that set of times has a cardinality less than a cardinality for the subset of the complete set of concentrations for the selected set of times.

12. The system of claim 11, wherein for each set of times from the plurality of sets of times, the first set of machine learning inputs comprises instructions to mask values corresponding to concentrations from the complete set of concentrations which are not comprised by the subset of the complete set of concentrations for that set of times.

13. The system of any of claims 10-12, wherein the instructions are operable to, each time the method is performed, determining the minimum inhibitory concentration based onproviding the first set of machine learning inputs to the machine learning model by performing acts comprising:(a) obtaining a first machine learning output based on providing the first set of machine learning inputs to an encoder;(b) obtaining a second machine learning output based on providing a second set of machine learning inputs to a decoder, wherein the second set of machine learning inputs is based on the first machine learning output; and(c) determining the minimum inhibitory concentration based on the second machine learning output.

14. The system of claim 13, wherein the method comprises, each time the set of minimum inhibitory concentration reporting acts is performed:(a) obtaining data identifying a microorganism and an antimicrobial corresponding to that performance of the set of minimum inhibitory concentration reporting acts; and(b) determining the second set of machine learning inputs based on the microorganism and the antimicrobial corresponding to that performance of the set of minimum inhibitory concentration reporting acts.

15. The system of claim 14, wherein, for each performance of the set of minimum inhibitory concentration reporting acts comprised by the method:(a) the first machine learning output comprises a set of output vectors, each of which has a dimensionality which is the same for all vectors from the set of output vectors;(b) determining the minimum inhibitory concentration based on providing the first set of machine learning inputs to the machine learning model comprises:(i) determining an antimicrobial vector based on the antimicrobial corresponding to that performance of the set of minimum inhibitory concentration reporting acts, wherein the antimicrobial vector has the same dimensionality as the vectors from the set of output vectors;(ii) determining a microorganism vector based on the microbe corresponding to that performance of the set of minimum inhibitory concentration reportingacts, wherein the microorganism vector has the same dimensionality as the vectors from the set of output vectors; and(iii) determining the second set of machine learning inputs based on, for each vector from the set of output vectors, combining that vector with the antimicrobial vector and the microorganism vector.

16. The system of any of claims 13-15, wherein, for each performance of the set of minimum inhibitory concentration reporting acts comprised by the method, determining the first set of machine learning inputs based on the plurality of sets of test well evaluation values comprises, for each test well evaluation value:(a) determining an embedding vector having a dimensionality and based on that test well evaluation value;(b) determining a cycle vector based on that test well evaluation value’ s imaging cycle and having the same dimensionality as the embedding vector based on that test well evaluation value;(c) determining a concentration vector based on the concentration corresponding to the set of test well evaluation values which comprises that test well evaluation values; and(d) combining the embedding vector corresponding to that test well evaluation value with the cycle vector and the concentration vector.

17. The system of any of claims 10-16, wherein:(a) the system comprises an incubator and a fluorometer;(b) the method comprises:(i) incubating a panel which comprises, for each antimicrobial agent from a set of antimicrobial agents, a set of test wells inoculated using that antimicrobial agent and the biological sample; and(ii) performing the set of minimum inhibitory concentration reporting acts a set of times corresponding to the set of antimicrobial agents, wherein, for eachperformance of the set of minimum inhibitory concentration reporting acts from the set of times corresponding to the set of antimicrobial agents:(A) that performance of the set of minimum inhibitory concentration reporting acts corresponds to a single antimicrobial agent; and(B) receiving the plurality of sets of test well evaluation values comprises, for each test well from the set of test wells inoculated using the antimicrobial agent corresponding to that performance, capturing a fluorescence value for that test well using the fluorometer.

18. The system of claim 17, wherein the set of antimicrobial agents has a cardinality which is no less than one fifth of a number of test wells comprised by the panel.

19. A machine comprising:(a) a fluorometer; and(b) means, in particular comprising the system of any of claims 10 to 16, for reporting minimum inhibitory concentrations based on fluorescence values from the fluorometer.

20. The machine of claim 19, wherein the machine further comprises means for optimizing antimicrobial dilution using the means for reporting minimum inhibitory concentrations.

21. Computer program product comprising instructions which, when the program is executed by a computer, cause the computer to carry out the method of any of claims 1 to 7.

22. Computer-readable medium comprising instructions which, when executed by a computer, cause the computer to carry out the method of any of claims 1 to 7.

Images