Multiple time windows with associated calibration curves for extending the range of an assay

Abstract

Disclosed is a use of reaction kinetics to generate multiple dose-response curves from a single reaction, thus eliminating the need to run a second experiment with additional sample, reagents, and time to cover a broader measuring range than is available in a standard assay. Using a single protocol, the differences in the reaction kinetics for different sample concentrations yield different responses at different measurement times. Selection of the appropriate dose-response curve cross-section increases the measuring range and accuracy of the assay from a single reaction without substantially increasing imprecision. Several overlapping dose-response curves are pieced together to provide a standard curve to ensure continuity throughout the expanded measuring range.

Claims

1. A method for extending the range of a single assay and for measuring an analyte level of the assay based on a time-variant signal reflecting a dynamic response level in an instrument, the method comprising: using a first calibration curve representative of analyte level and response level of the assay at a first time point subsequent to initiation of the assay to generate the analyte level if a signal strength corresponds to a predefined signal level for the first time point; and if the signal strength fails to correspond to the predefined signal level for the first time point, then using a second calibration curve representative of analyte level and response level of the assay at a second time point subsequent to the first time point to estimate the analyte level if the signal strength corresponds to a predefined signal level for the second time point.

2. A method for measuring a level of an analyte using a plurality of calibration curves, each calibration curve associated with at least one threshold in regard to a single reaction at a corresponding plurality of fixed time points after initiation of the single reaction, the method comprising: determining whether a first condition is met based on comparing a measured signal level with a first predetermined threshold associated with a first calibration curve from the plurality of calibration curves is satisfied at a first time point subsequent to initiation of the reaction, in which the first calibration curve is representative of signal levels and corresponding analyte levels at the first time point, and if the first condition is satisfied then using the first calibration curve to generate a first measured value for the level of the analyte; and determining whether a second condition is met based on comparing a measured signal level at a second time of the reaction with a second predetermined threshold associated with a second calibration curve from the plurality of calibration curves at the second time point.

3. The method of claim 2, wherein if more than one calibration curve from the plurality of calibration curves are available for measuring the level of the analyte, due to thresholds corresponding to each of the more than one calibration curve being satisfied at corresponding time points of the single reaction, then using an average of analyte levels corresponding to each of the available calibration curves as the measured level of the analyte.

4. The method of claim 3, wherein the average is a weighted average of the analyte levels of the available calibration curves.

5. The method of claim 2, wherein the first condition requires that a signal corresponding to the level of the analyte exceed the first threshold.

6. The method of claim 2, wherein the first condition requires that a signal corresponding to the level of the analyte be less than the first threshold.

7. The method of claim 2, wherein the first condition requires that a signal corresponding to the level of the analyte be equal to the first threshold.

8. The method of claim 2, wherein the second condition requires that a signal corresponding to the level of the analyte exceed the second threshold.

9. The method of claim 2, wherein the second condition requires that a signal corresponding to the level of the analyte be less than the second threshold.

10. The method of claim 2, wherein the second condition requires that a signal corresponding to the level of the analyte be equal to the second threshold.

11. A method for scheduling a test in a clinical analyzer supporting an extended range, the method comprising the steps of: initiating the test using a reaction mix; determining a first signal strength from the reaction mix at a first predefined time point after initiating the test; determining whether there is a suitable first calibration curve corresponding to the first predefined time point by comparing the determined first signal strength with a first predetermined threshold signal level of the first calibration curve, wherein the first calibration curve is suitable if the determined first signal strength at least equals the first predetermined threshold signal level; scheduling, if there is no suitable first calibration curve, a second predefined time point after initiating the test for determining a second signal strength from the reaction mix; determining, at the second predefined time point, the second signal strength; identifying a second calibration curve corresponding to the second predefined time point and determining if the second calibration curve is suitable by comparing the determined second signal strength with a second predetermined threshold signal level of the second calibration curve; and determining a level of an analyte from one or more of the first signal strength and second signal strength.

12. A diagnostic clinical analyzer supporting an extended range comprising a scheduler for implementing the reading of a signal at a time point following initiation of a single assay, the time point of the assay having a corresponding calibration curve selected from a plurality of calibration curves, wherein a determination is made by the analyzer as to the suitability of at least one said calibration curve based on the read signal as compared to a threshold signal of the corresponding calibration curve at time points of the assay.

13. A method for scheduling a test in a clinical analyzer supporting an extended range, the method comprising the steps of: initiating the test using a reaction mix; measuring a signal strength from the reaction mix at a first time point subsequent to the initiation of the test selected from a plurality of time points subsequent to initiating the test; determining analyte level using a first calibration curve corresponding to the first time point for measuring the signal strength, the first calibration curve being representative of signal strength corresponding to analyte values at the first time point and in which the measured signal strength is compared to a predetermined threshold signal strength value; scheduling, if there is no corresponding calibration curve, a second time point subsequent to the initiation of the test for determining a second signal strength from the reaction mix; and determining analyte level using a second calibration curve corresponding to the second time point for measuring the signal strength.

14. The test scheduling method of claim 13 further comprising the step of scheduling multiple time points subsequent to the initiation of the test for measuring analyte levels using corresponding calibration curves.

15. The test scheduling method of claim 14, wherein a reported analyte level is a mean of measured analyte levels.

16. The test scheduling method of claim 15, wherein the reported analyte level is a weighted mean of measured analyte levels.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 (Prior Art) depicts a prior art clinical diagnostic analyzer and its major subsystems.

(2) FIG. 2 (Prior Art) shows an exemplary dose-response curve.

(3) FIG. 3 (Prior Art) shows the challenges in fitting a calibration curve to data covering a broad measurement range.

(4) FIG. 4 shows a family of dose-response curves differentiated by the value of one of the Logit/Logit4 parameters.

(5) FIG. 5 shows a family of dose-response curves with the time at which the reaction results are read level as the parameter.

(6) FIG. 5B shows the dose-response curve from FIG. 5 corresponding to 1200 seconds and covering a wider range of analyte levels.

(7) FIG. 6 shows a dose-response curve corresponding to FIG. 4 with time for reading the reaction results of 360 seconds scaled for relatively low levels of the analyte.

(8) FIG. 7 shows a dose-response curve corresponding to FIG. 4 with time for reading the reaction results of 360 seconds scaled for relatively high levels of the analyte.

(9) FIG. 8 shows the two dose-response curves that can be used together to span a greater analyte concentration range, each corresponding to different time windows post initiation of the reaction (at 20 sec (solid line) and 3000 sec (dashed line)).

(10) FIG. 9 depicts an illustrative flowchart showing how a particular dose-response curve is selected based on the signal strength. As shown, if the signal corresponding to the unknown analyte at the 20 sec window is greater than 0.065 then the 20 seconds dose-response curve is used. Else, the 3000 seconds dose-response curve is used.

(11) FIG. 10 shows three calibration curves corresponding to measurements taken at 9.5 seconds after reagent addition (the Early dose), at 161.5 seconds after reagent addition (the Standard dose), and at 275.5 seconds (the Late dose) to illustrate the effect of the selection of the time window on the measurement accuracy. The illustrative curves share two calibrators in common (0.449 g/dL and 0.84 g/dL) to assist in the cross-over region between the early and late dose-response curves.

(12) FIG. 11 shows calibration curves corresponding to measurements taken at 9.5 seconds after reagent addition (the Early dose) and at 275.5 seconds (the Late dose) combined to get a dual calibration curve implementation.

(13) FIG. 12 illustrates these cross-over rules in operation by way of a diagram showing the decision making logic to guide the choice of the calibration curve to use in the manner similar to that in FIG. 9.

(14) FIG. 13 shows the mean bias between the predicted analyte level for each sample and the reference analyte level. The early dose-response curve deviates significantly from the calibration curve, so the bias is large for all sample concentrations greater than 0.84 g/dL. Both the standard and late dose-response curves significantly deviate from zero bias at lower concentration due to the flattening of the dose-response curve.

(15) FIG. 14 shows the standard deviation for the dual dose-response model is favorably below that for the standard dose-response curve even though the dual dose-response curve covers a larger range.

(16) FIG. 15 shows a flow diagram illustrating implementations of a method using the dual calibration curve for extending the range and improving the accuracy in measuring an analyte of interest.

(17) FIG. 16 shows a flow diagram illustrating implementations of a method using the multiple calibration curves for extending the range and improving the accuracy in measuring an analyte of interest.

(18) FIG. 17 shows a flow diagram illustrating implementations of a method using more than one calibration curve to allow measurement of the same reaction at different timesand possibly averaging the results to improve accuracy or track errors.

(19) FIG. 18 shows a flow diagram illustrating implementations of a method using the dual calibration curve for extending the range and improving the accuracy in measuring an analyte of interest.

(20) FIG. 19 shows a scheduler implement reading a signal at a time point.

DETAILED DESCRIPTION

(21) When measuring an analyte, one obtains a signal from a reaction mix incubated for a predefined time. This signal, measured during a specified time window, is converted into the level or concentration of the analyte using a calibration curve. Instead of using a physical curve, many implementations provide parameters defining the calibration curvesuch as the slope and intercept of a line together with the range over which such parameters should be relied upon. One approach for accurately estimating the analyte is to use a linear calibration curve or a piecewise linear calibration curve. However, many regions of the calibration curve are still not suitable for inferring the analyte concentration with sufficient accuracy. As a result a range of measurements possible on an instrument is defined by the accuracy with which a measurement can be made using its calibration curve.

(22) Varying the sample concentration, for instance by diluting it, may allow readings to be obtained that are within the range of the instrument. However, this requires carrying out another reaction with measurements associated with it.

(23) This disclosure of techniques for extending the range of an instrument without requiring another reaction by employing two or more calibration curves also includes a procedure to dynamically select the appropriate calibration curve.

(24) This limited measuring range is typically due to the shape of the response function, or due to the lack of fit of the calibration model, or due to another restrictive means. The end result is that the possible measuring range is limited unless another reaction is carried out with some varied parameters to change the characteristic response shape to allow measurement over a different range in the customary time window. The current disclosure provides two or more response curves from a single reaction to expand the possible range of measurements. This is accomplished by making measurements from a single reaction in two or more time windows.

(25) The disclosed model concept is described using simulated reaction kinetics (Table 1, FIG. 4). Simulated reaction kinetics are assumed to be fit by the four parameter Logit/Log 4 model in accordance with Equation 1, where R is the response, C is the concentration, and .sub.0-.sub.3 are the four Logit/Log 4 parameters. It should be noted that alternative mathematical models could be used instead of the Logit/Log 4 method with no loss of generality.

(26) $\begin{array}{cc}R={}_0+\frac{{}_1}{1+{\exp }^{-\left({}_2+{}_{3}Xl\mathrm{nC}\right)}}& \mathrm{Equation}1\end{array}$
In practice, the mathematical curve fitting model selected does not detract from the teachings of the disclosure. In the simulation shown in FIG. 4, .sub.0-.sub.2 are kept constant while .sub.3 is varied to yield a family of simulated kinetic curvesas is shown in the legend for FIG. 4. A family of analyte-concentration/instrument-signal curves (also known as calibration curves or dose-response curves) from the simulated kinetics are shown in FIG. 4 with the analyte concentration corresponding to the different .sub.3 values shown as parameters on the right.

(27) The curves in FIG. 5 were generated using Eq. 1 where the time for which the reaction is carried out is the parameter defining a series of plots between the observed signal on the y-axis and the initial analyte concentration along the x-axis. In FIG. 5 the time post-reaction for which the calibration curve was generated is shown in the Figure legend for each curve.

(28) For instance, the kinetic curve in FIG. 5 corresponding to measurements made at 360 seconds after the initiation of the reaction show that there is seen a fairly flat response for all concentrations below 7 au (concentrations are shown on along the horizontal axis). This curve is reproduced in FIG. 6, which shows a flat response above 500 au as is illustrated in FIG. 7 for measurements made at 360 seconds after the initiation of a reaction. Because of the inherent shape of the calibration curve, the measuring range would be limited to approximately 10-500 au if a measurement window at 360 seconds was used with a single reaction mix. Earlier measurements would allow more accurate measurements at the lower end, but would not be suitable for higher analyte levels. For instance, a measurement range of 2-50 au is possible from a single reaction mix if the measurements are made during a time window that corresponds to 20 seconds post initiation of the reaction. Similarly, a late measurement would be better suited to detect higher analyte concentrations, but at the cost of accuracy in detecting relatively lower analyte concentrations. As is readily seen in FIG. 5 and FIG. 5B, for an exemplary time window corresponding to 1200 seconds post initiation of the reaction, the measurement range is 15-greater than 1000 au. At less than 15 au, the 1200 second curve is too flat as is illustrated in FIG. 5. Thus, no single calibration curve spans the entire measuring range between 2-2500 au using a single reaction mix. Customarily this limitation requires that multiple reactions be set up with different concentrations of reagents or test substanceseven a dilution series to make accurate measurements.

(29) TABLE-US-00001 TABLE 1 Logit/Log4 Parameters for the Simulated Reaction Kinetics .sub.0 .sub.1 .sub.2 .sub.3 Conc. Arbitrary units (au) 0 0.75 5 4 2.19 0 0.75 5 2 4.78 0 0.75 5 1.9 5.19 0 0.75 5 1.8 5.69 0 0.75 5 1.6 7.06 0 0.75 5 1.4 9.34 0 0.75 5 1.2 13.56 0 0.75 5 1 22.83 0 0.75 5 0.9 32.32 0 0.75 5 0.8 49.91 0 0.75 5 0.7 87.06 0 0.75 5 0.6 183.77 0 0.75 5 0.5 521.34 0 0.75 5 0.45 1044.77 0 0.75 5 0.4 2491.14

(30) In accordance with this disclosure, combining two or more calibration curves, each corresponding to a different time window post initiation of the reaction allows measurements to be made from the same reaction mix to cover a far broader range of analyte measurements than what was possible otherwise. For the simulated curves in the above example, the desire to cover the measuring range from 2-2500 au cannot be accomplished by a single dose-response curve. However, by collecting readings at 20 sec and 3000 sec from the same reaction mix after the reaction has been initiated and combining the two dose-response curves to evaluate any unknown within the measuring range between 2-2500 au extends the range.

(31) FIG. 8 shows the two dose-response curves corresponding to time windows post initiation of the reaction at 20 sec (solid line) and 3000 sec (dashed line). These curves are used together to span the entire desired measuring range between 2-2500 au.

(32) Specifically, the 20 sec calibration curve spans the range from 2-40 au, and the 3000 sec calibration curve spans the range from 30-2500 au. The two curves overlap between 30-40 au. The combination of these two calibration curves spans the entire range with an overlap region between 30-40 au. The overlap of the calibration curves facilitates a cross-over from one calibration curve to the other since when creating the calibration curves, test samples in this range can be read post initiation of the reaction at 20 seconds and 3000 seconds to ensure consistency and continuity.

(33) The cross-over region provides a link for switching between the appropriate calibration curves. A desirable cross-over region ensures that the calibration curves consistently cover the entire measuring range.

(34) In addition, the process of selecting the appropriate calibration curve can be automated. This makes the use of multiple calibration curves and measurement time windows no different than a single measurement time and a single calibration curve to an operator of a clinical diagnostic analyzer. A suitable clinical analyzer may be programmed to perform like a machine that is using a single calibration curve even though in practice multiple calibration curves corresponding to different observation time points are being used. Such a machine includes computer executable instructions that allow suitable incubation times, queuing, resource allocation in general to make possible the use of multiple calibration curves in evaluating an unknown sample of interest.

(35) In another aspect, the time windows for observation may be selected to ensure a desired degree of accuracy. Thus, by combining several calibration curves, a desired degree of accuracy may be obtained in many instances where the signal does not vary as a linear function of the starting analyte concentration.

(36) There are multiple ways to generate two overlapping calibration curves. Table 2 shows one such possible method using the Logit/Log 4 function, which requires a minimum of four calibrators to define the calibration function. In a preferred embodiment, there is at least one calibrator (also called a standard) common to both calibration curves to assist with a smooth transition from one calibration curve to another.

(37) TABLE-US-00002 TABLE 2 Calibrator Level 20 sec. Calibrators 3000 sec. Calibrators 1 2 au 2 5 au 3 12 au 12 au 4 35 au 35 au 5 250 au 6 1000 au 7 2500 au

(38) Some exemplary methods for selecting which calibration curve to use for the determination of an unknown analyte concentration are illustrated next. The following exemplary example illustrates possible rules for using the two calibration curves shown in FIG. 8. As is shown in the flowchart of FIG. 9, when the response corresponding to the unknown analyte at the 20 sec window is greater than 0.065 (corresponding to a concentration of approximately 35 au), the signal from the 20 sec window will be analyzed on the 20 sec dose-response curve. If the response corresponding to the unknown analyte at the 20 sec window is less than 0.065, then instead of the signal from the 20 sec window, the signal from the 3000 sec window is analyzed using the 3000 sec dose-response curve (FIG. 9).

EXAMPLE

(39) An inversely proportional clinical chemistry assay has a standard protocol to measure the response 161.5 seconds after the addition of the last reagent. In this example, the protocol was altered to measure responses at 9.5 seconds, 161.5 seconds, and 275.5 seconds in a single reaction cuvette after the addition of the last reagent to enable the evaluation of both the dual dose-response curve model and the standard assay protocol model. In the description that follows, the response measured at 9.5 seconds after reagent addition is called the Early dose, the response measured at 161.5 seconds after reagent addition is called the Standard dose, and the response measured at 275.5 seconds after reagent addition is called the Late dose. The early dose and the late dose calibration curves are combined in to form the Dual dose-response.

(40) Seven calibrators were run in the experiment and used to calibrate the three separate dose-response curves as specified in Table 3. As described above in the necessary requirements for the multiple dose-response model, the early and late dose-response curves share at least one calibrator in the cross-over region to cover the entire measuring range in a continuous manner. In this particular example, the early and late dose-response curves share two calibrators in common (0.449 g/dL and 0.84 g/dL) to assist in the cross-over region between the early and late dose-response curves.

(41) TABLE-US-00003 TABLE 3 Calibrator Calibrator Level Concentration (g/dL) Std. Assay Early Dose Late Dose 1 0 X X 2 0.217 X 3 0.449 X X X 4 0.84 X X X 5 1.447 X X 6 2.083 X X 7 2.604 X X

(42) In addition to the calibrators, multiple fluids that span the measuring range from 0.2-2.6 g/dL were run in triplicate to demonstrate the true shape of the dose-response curve for each of the three different protocols, which are shown in FIG. 10. The standard dose-response curve (diamonds) shows a flattening of the dose-response curve at 0.217 g/dL on the low end and at 2.411 g/dL on the high end. This flattening shape limits the effective measuring range of the assay based on the response function alone to approximately 0.2-2.4 g/dL. The early dose-response curve (squares) shows no flattening on the low end and flattening on the high end at 1.302 g/dL, providing an effective measuring range of the assay based on the response function alone to approximately 0-1.3 g/dL. The late dose-response curve (triangles) shows a flattening of the dose-response curve at 0.449 g/dL on the low end and no flattening on the high end, providing an effective measuring range of the assay based on the response function alone to approximately 0.5-2.6 g/dL.

(43) Each of the three dose-response curves was calibrated using the Logit/Log 4 calibration model (Equation 1) with calibrator levels indicated in Table 3. The Logit/Log 4 calibration curves for the three response times are shown with fine broken lines corresponding to the Early calibration curve, the solid line to Standard calibration curve and the rough broken line corresponding to the Late calibration curve in FIG. 11. Both the Standard and Late calibration curves show the expected flattening of the calibration curves at early times, limiting the effective measuring range more than by the dose-response curve shape alone as described above. The Early calibration curve shows a deviation from the data at higher concentrations since not all calibrator levels were used in the calibration.

(44) FIG. 11 shows a dual dose calibration curve for the data of FIG. 10. The dual dose calibration curve of the disclosure is shown in black and transitions from the early calibration curve to the late calibration curve at the cross-over response of 0.14 OD on the early calibration curve (corresponding to a concentration of 0.84 g/dL). FIG. 12 illustrates these cross-over rules in operation by way of a diagram showing the decision making logic to guide the choice of the calibration curve to use in the manner similar to that in FIG. 9.

(45) A benefit of the current disclosure is the extended measuring range and the enhanced precision and accuracy. The extended measuring range has been qualitatively described above based on the curvature of the calibration curves seen in FIG. 8 and FIG. 11. A dual dose calibration curve has no or little flattening at either the low or high concentrations. The extended measuring range can be easily seen in the accuracy of the test fluids' predicted concentration from the dual dose calibration curve verses the early, standard, or late calibration curve analysis (FIG. 13).

(46) FIG. 13 shows the mean bias between the predicted analyte level for each sample (Table 4) versus the reference analyte level assigned to each sample. The early dose-response curve deviates significantly from the calibration curve, so the bias is large for all sample concentrations greater than 0.84 g/dL. Both the standard and late dose-response curves significantly deviate from zero bias at lower concentration due to the flattening of the dose-response curve and lack of a model fit. Only the dual dose model, the subject of this disclosure, shows excellent agreement with the reference concentration throughout the entire measuring range (0.1 g/dL bias) by transitioning between the Early and Late Dose-Response curves to better utilize each calibration curve and demonstrates the benefit of the extended measuring range with the assay performance.

(47) TABLE-US-00004 TABLE 4 Reference Early Dose Late Dose Std. Assay Dual Dose Conc. Conc. Conc. Conc. Conc. Sample ID (g/dL) (g/dL) (g/dL) (g/dL) (g/dL) 1 0 0.016 0.447 0.209 0.016 2 0.022 0.049 0.423 0.270 0.049 3 0.096 0.125 0.367 0.207 0.125 4 0.217 0.217 ME* 0.316 0.217 5 0.296 0.288 0.323 0.357 0.288 6 0.449 0.449 0.455 0.469 0.449 7 0.566 0.572 0.638 0.576 0.572 8 0.723 0.752 0.739 0.713 0.752 9 0.840 0.840 0.836 0.830 0.840 10 1.042 0.950 1.026 1.035 1.026 11 1.302 1.052 1.291 1.341 1.291 12 1.447 1.080 1.453 1.473 1.453 13 1.730 1.122 1.736 1.725 1.736 14 2.083 1.169 2.048 2.069 2.048 15 2.411 1.192 2.397 2.358 2.397 16 2.604 1.204 2.668 2.538 2.668 *MEMechanical Errorno result reported

(48) Besides the improvement in accuracy throughout the measuring range enabled by the dual dose-response model, see, e.g., FIG. 13, the low end imprecision is also substantially reduced in the dual dose-response model of this disclosure. As is shown in FIG. 14, the SD for the dual dose-response model is 0.01 g/dL for measuring analyte levels below 0.5 g/dL, which is significantly better than the imprecision for standard dose-response curve (>0.035 g/dL) over the same range. This improvement in precision is accompanied with a 3-fold increase in response range for the dual dose model verses the standard model (0.34 OD vs. 0.11 OD) in the range between 0 and 0.5 g/dL.

(49) The dual dose-response model offers a larger signal range throughout the measuring range compared to the signal for the standard model. Table 5 shows the nearly 50% increase in OD range for the dual dose-response model over the standard dose-response model. This causes a dramatic improvement in the precision at low levels of analyte as described above.

(50) The expanded OD range further accompanies improvements in the accuracy of the dose-response curve slope at both lower and higher analyte concentrations, as is shown in FIG. 13 in the dual dose-response model spanning a measuring range beyond the current 2.6 g/dL. The high concentration end slope (between 2.4-2.6 g/dL) is more than twice as large for the dual dose-response model (Table 6) than for the Standard dose-response curve.

(51) TABLE-US-00005 TABLE 5 Concentration Standard Standard Dose Dual Dual Dose Range Dose OD Range Dose OD Range 0-0.84 g/dL 0.73-0.47 0.26 0.61-0.14 0.47 0.84-2.6 g/dL 0.47-0.1 0.37 0.58-0.13 0.45 0-2.6 g/dL 0.63 0.92

(52) TABLE-US-00006 TABLE 6 Single Dose Dual Dose Slope 0.036 0.074 (2.4-2.6 g/dL)

(53) Although there is great flexibility in using the more than two dose-response curves as described herein, the measuring range of most processes of interest are fairly narrow that they can be covered with two different dose-response curves. The preferred mode therefore uses two calibration curves. In addition, in the preferred embodiment calibrators are shared in generating the dose-response curves by the simple modicum of reading the same calibrator at two different times, which limits the total number of calibrators required to generate the dose-response curves. This further reduces the time and cost of the calibration itself.

(54) Preferably, the cross-over point is at or very near one of the common calibrator concentrations to ensure a continuous predicted concentration between the two different dose-response curves. The continuity is also enhanced when there is a substantial cross-over region where both calibration curves provide essentially the same measured levels of the analyte regardless of the time window used.

(55) Many alternative mathematical models (other than Logit/Log 4) could be used to describe the calibration curves. The preferred models need not be changed from those that are known to work well for the single dose-response model already in use in the field. Some examples of mathematical models for describing the multiple dose-response curves are linear, polynomial, cubic spline, Logit/Log 4, and Logit/Log 5.

(56) A Scheduler is the brains making the analyzer subsystems work together. The Scheduler performs scheduling functions, for instance, to allocate resources to samples input in any order, without regard to the type or quantity of tests required, and to maintain or improve the throughput of the analyzer. The Scheduler ensures that samples are accepted from an input queue as resources are reserved for the various expected tests or steps relevant to a particular sample. Unless the required resources are available, a sample continues to be in the input queue. In a preferred analyzer model, the sample is aspirated and then sub-samples are taken from this aspirated volume for various tests. The operation of the Scheduler together with the types of tests supported by the analyzer provides a reasonably accurate description of an analyzer under consideration.

(57) A preferred Scheduler includes the synergistic effects of two-dimensional random access to the input samples while providing access to resources including, for example, multiple platforms, supply of consumables including thin film slides, reaction vessels and cuvettes, along with a plurality of sensiometric devices such as electrometers, reflectometers, luminescence, light transmissivity, photon detection, an incubator for heating the samples, a supply of reagents, and a plurality of reagent delivery subsystems, all of which can be accessed and used readily.

(58) Implementing FIGS. 9 and 12 preferably is with the aid of a Scheduler that allocates tentative times for completion of a test. This is useful since a second time window may need to be allocated together with a specification of the corresponding calibration curve. Such a Scheduler handles samples to optimize throughput and processing of samples to ensure the extended analyte range is utilized by allocating additional resources to a sample as needed. Such a Scheduler preferably converts the clinical laboratory analyzer of FIG. 1 into one that supports measuring far broader ranges of analyte concentrations by programming it to implement the new Scheduler functionality. This programming may be by way of programming languages or graphical programming interfaces and delivered in the form of an update.

(59) For implementing the multiple calibration curves in a clinical diagnostic analyzer, such as the one illustrated in FIG. 1, a preferred method modifies the Scheduler to permit signal measurement at different times based on a prior signal strength measurement. In effect, this is an implementation of the decision logic similar to that illustrated in FIGS. 9 and 12 in a clinical diagnostic analyzer.

(60) Turning to FIG. 9, it shows a method for implementing an extended range. During step 900, a module determines a signal level in a sample of interest during a first time windowhere the exemplary time window is at about 20 seconds. Then control passes to step 910 during which the signal strength is compared to a thresholdsuch as the illustrative threshold of 0.065to transfer control to step 920 if the threshold is exceeded, or, alternatively to step 930 if the threshold is not exceeded. If control is passed to step 920, the 20 second calibration curve is used while if control is passed to step 930, the 3000 seconds calibration curve is used to convert the signal strength into an analyte level.

(61) Similarly, in FIG. 12, during step 1200, a module determines a signal level in a sample of interest using both the Early and Late read windows. Then control passes to step 1210 during which the analyte response based on the Early read window is compared to a thresholdsuch as the illustrative threshold of 0.14to transfer control to step 1220 if the threshold is exceeded, or, alternatively to step 1230 if the threshold is not exceeded. If control is passed to step 1220, the Early calibration curve is used while if control is passed to step 1230, the Late calibration curve is used to convert the signal strength into an analyte level.

(62) FIGS. 15 and 16 show flow diagrams illustrating implementations of the dual and multiple calibration curve based methods for extending the range and improving the accuracy in measuring an analyte of interest. In FIG. 15, during step 1500 the signal strength is measured at a first time point. Herein the term signal strength encompasses various possible units in which a signal may be measured or converted. Control then passes to step 1510, during which if the signal strength is large enough, control passes to step 1520, during which the first calibration curve is used. Alternatively during step 1510 if the signal strength is not large enough, control passes to step 1530. During step 1530 a measurement at a second time point is scheduled and control passed to step 1540. During step 1540 the signal strength is measured at the second time point and control passes to step 1550 for use of the second calibration curve for measuring the level of the analyte of interest.

(63) FIG. 16 illustrates a more general scheme. In FIG. 16, during step 1600 the signal strength is measured. Control then passes to step 1610, during which if the signal strength is large enough, control passes to step 1620, during which the corresponding calibration curve is used. In effect, the method determines if there is a suitable corresponding first (or second or third and the like) calibration curve. In the absence of a suitable calibration curve, a later time point for determining the signal strength from the reaction mix is scheduled. Accordingly, during step 1610 if the signal strength is not large enough, control passes to step 1630. During step 1630 a measurement at a later (which could be a second or third and the like) time point is scheduled and control passed to step 1640. During step 1640 the signal strength is measured at the later time point and control passes back to step 1610 and then onto identification of a suitable calibration curve. If there are multiple calibration curves corresponding to different signal magnitudes or time points, that may be used, then the level of the analyte may be determined from just one or more than one calibration curve.

(64) In FIG. 17 is illustrated a method for using more than one calibration curve to allow measurement of the same reaction at different timesand possibly averaging the results to improve accuracy or track errors. For such averaging advantageously a weighted mean of analyte levels measured at different times may be used to further refine the results. In FIG. 17, during step 1700 the signal strength is measured. Control then passes to step 1710, during which if the signal strength is large enough, control passes to step 1720, during which the corresponding calibration curve is used. Control passes from step 1720 to step 1730 to determine if additional measurements are desired at later time points. The method terminates if no additional measurements are desired with control passing to step 1740. However, if additional measurements are desired at later time points, then control passes to step 1750, during which a later measurement is scheduled. Control passes to step 1760 for making the scheduled measurement, following which control loops back to step 1710. In effect, the method also determines if there is a suitable corresponding first (or second or third and the like) calibration curve. In the absence of a suitable calibration curve, a later time point for determining the signal strength from the reaction mix is scheduledas well as for just determining a second or more measurements from the same reaction mix. Multiple measurements may be averaged.

(65) In FIG. 18 is illustrated a method for using more than one calibration curve to allow measurement in a clinical diagnostic analyzer. In FIG. 18, during step 1800 the reaction is initiated. Control then passes to step 1810, during which a signal strength from the reaction mix at a time point selected from a plurality of time points subsequent to initiating the test is measured. Control then passes to step 1820, during which the analyte level is determined using a calibration curve corresponding to the time point for measuring the signal strength the signal strength is large enough. Then control passes to step 1830, during which a second time point for determining a second signal strength from the reaction mix is scheduled if there is no corresponding calibration curve. Control passes from step 1830 to step 1840 to determining the analyte level using a second calibration curve corresponding to the second time point for measuring the signal strength.

(66) FIG. 19 depicts a scheduler 1900 for use in a diagnostic clinical analyzer supporting an extended range. The scheduler implements reading a signal at a time point by checking if the time point has been reached during step 1910 and then scheduling the signal during step 1920. If during step 1910 the time point is not reached, the control loops back. This looping back is show as being a direct loop for clarity. In practice, a preferred embodiment will have the scheduler attend to other tasks before testing the time point again. In a preferred embodiment, the selected time point corresponds to a calibration curve selected from a plurality of calibration curves based on the signal. The signal so read during step 1920 is mapped into a measured value of an analyte of interest using the selected calibration curve.

(67) Many of the limitations on the measuring range of diagnostic and other tests can be overcome by implementing the multiple dose-response model of this disclosure. The description herein shows the enhancements to extend the measuring range, and the example shows the theoretical model put into practice with a real improvement for increased measuring range for the assay described. In addition to the extended measuring range, the disclosure also demonstrates an improvement in the test method precision due to the increased response range and in the test method accuracy due to the improved fitting of the calibration curve. By the improvements offered from the multiple dose-response model, the shortcomings on the measuring range and precision from current model have been eliminated.

(68) One skilled in the art will appreciate that the above disclosure is susceptible to many variations and alternative implementations without departing from its teachings or spirit. The scope of the claims appended below includes such modifications. Further, each reference discussed and cited herein is hereby incorporated herein by reference in its entirety.