WO2023025388A1

WO2023025388A1 - System and method for measuring and device for determining a required drug dosage

Info

Publication number: WO2023025388A1
Application number: PCT/EP2021/073620
Authority: WO
Inventors: Thomas BENNINGER; Markus REICHHARTINGER
Original assignee: Technische Universität Graz
Priority date: 2021-08-26
Filing date: 2021-08-26
Publication date: 2023-03-02

Abstract

A system (5) and a method (24) for measuring from a patient's blood sample (6) a required dosage (u) of a drug for treating hyperfunction of the patient's thyroid (2) the system comprising a sensing device (7) for sensing a level of free thyroxine, FT4 (LFT4) in the blood sample (6) and a processing device (8) having an electronic processor (9) an electronic controller (10) and an electronic memory (11) storing a state- space model (14) of the thyroid (2) wherein the processor (9) is configured to initialise the state-space model (14) and to adjust parameters (k*) of the controller (10) controlling the state-space model (14) in order to approach a target level of FT4 (TFT4) and wherein the controller (10) is configured to compute the required dosage in dependence on the sensed and the target level of FT4 (LFT4, TFT4).

Description

System and Method for Measuring and Device for Determining a Required Drug Dosage The present invention relates to a system for measuring, from a patient’s blood sample, a required dosage of a drug for treating hyperfunction of the patient’s thyroid. The invention further relates to a processing device of the system and a method employing the system. The thyroid hormones thyroxine (“T4”) and triiodothyronine (“T3”) stimulate the metabolism of almost every tissue of ver- tebrates; a balance of these hormones is, thus, essential for a healthy life. To illustrate the interdependency of the thy- roid’s function and the most influential hormones, Fig. 1 shows the so called hypothalamic-pituitary-thyroid (“HPT”) axis 1 which provides for the balance: The production of T3 and T4 in the thyroid 2 is stimulated by thyroid-stimulating hormone (“TSH”) which itself is produced in the pituitary gland 3; the production of TSH is stimulated by thyrotropin-releasing hor- mone (“TRH”) which is produced in the hypothalamus 4. T3 and T4, in turn, attenuate the production of TSH in the pituitary gland 3 and of TRH in the hypothalamus 4, which is symbolised by the minus sign “-” in Fig. 1, such that the levels of T3 and T4 are balanced by negative feedback control in the body. When a patient, be it human or animal, suffers from hyper- thyroidism, i.e., hyperfunction of the thyroid, the HPT axis is disturbed and the thyroid overproduces T3 and T4 causing symp- toms like tremor, palpitations, nervousness, hyperactivity, anxiety, heat intolerance, hair loss, muscle aches, weakness, etc. A common cause of hyperthyroidism is Graves’ autoimmune disease, wherein the production of T3 and T4 is stimulated by TSH receptor antibodies (“TRAb”) while at the same time the TSH level is significantly reduced, often even down to zero. In order to achieve healthy levels of the thyroid hormones and to alleviate symptoms, patients may take antithyroid drugs for treating hyperthyroidism, such as methimazole, carbimazole or propylthiouracil. Applying a proper, i.e., the required, dosage of an antithyroid drug helps to avoid the abovementioned symptoms as well as fluctuating or oscillating hormone levels and drug side effects. Thus, patients have to frequently visit their physician who, each time, measures and determines the re- quired drug dosage from the patient’s most recent blood sample, the levels of hormones sensed therein and his experience. How- ever, also for the physicians the proper determination of the required dosage is non-trivial especially during the initial phase of the disease and during phases of a fluctuating anti- body stimulation. Hence, most patients under drug treatment still face a reduced life quality. It is an object of the present invention to provide a sys- tem and a method which measure and a processing device which determines the required dosage of a drug for treating hyper- function of the patient’s thyroid, in each case automatically and accurately. In a first aspect, this object is achieved with a system for measuring, from a patient’s blood sample, the required dos- age of the drug, the system comprising a sensing device for sensing a level of free, i.e., un- bound, T4 (“FT4”) in the blood sample, and a processing device having an electronic processor, an electronic controller and an electronic memory storing a state- space model of the patient’s thyroid, which state-space model describes the FT4 level over time in dependence on a drug lev- el, a level of thyrotropin receptor antibodies, TRAb, and a level of thyroid stimulating hormone, TSH,; wherein the processor is configured to receive the sensed FT4 level from the sensing device, to initialise the state- space model with the sensed FT4 level, an initial drug level, an initial TRAb level and an initial TSH level, and to adjust parameters of the controller such that, in a closed control loop of the controller controlling the state-space model, the FT4 level described by the state-space model approaches a pre- determined target FT4 level over time; and wherein the controller is configured to receive the sensed FT4 level from the sensing device and to compute, by means of the adjusted controller parameters, the required dos- age in dependence on the sensed FT4 level and the target FT4 level. The inventive system employs control theory to automati- cally measure the required dosage of an antithyroid drug for the patient, which dosage can then be administered. Thereby, the controller controls the level of FT4 in the patient’s blood by means of its parameters that have been adjusted on the basis of the state-space model. The initial levels of the drug, of TRAb and of TSH are de- termined before initialisation of the state-space model as de- scribed below. The FT4 level over time, i.e., its temporal change or evolution, is obtained from the state-space model which may be linear or nonlinear. Using the initialised state- space model, the processor adjusts the parameters of the con- troller applying control theory to the control loop, e.g., by pole placement, by eigenvalue assignment, by a linear-quadratic regulator (LQR) approach, by simulating the control loop etc., to eventually approach the predetermined target FT4 level cor- responding to a healthy reference of, e.g., 15 pmol/l. Thereby, large hormone oscillations or fluctuations, overshooting FT4 levels or drug overdosages are avoided. Thus, the system auto- matically and accurately measures the dosage required for treating hyperfunction of the patient’s thyroid. Being based on levels of FT4, the drug, TRAb and TSH, the state-space model facilitates an accurate dosage measurement. Moreover, sensing of FT4 allows for a simple, small size sens- ing device, such that the whole system may be miniaturised and, e.g., kept by the patient, potentially as a lab-on-a-chip im- planted in or attached to the patient’s body. In a favourable embodiment, the state-space model compris- es a first Michaelis-Menten term with TRAb acting as a sub- strate and said drug acting as an inhibitor, and a second Mich- aelis-Menten term with TSH acting as a substrate and said drug acting as an inhibitor. Therein, each of the Michaelis-Menten terms has one of the following structures

for uncompetitive inhibition, or

for competitive inhibition, or

for mixed inhibition, with P_A, P_B, P_C1, P_C2...parameters of the state-space model, L_sub........ the level of the respective substrate, and L_inh,1, L_inh,2...the levels of the respective inhibitors. State-space models based on Michaelis-Menten kinetics are very well suited to accurately describe the production of FT4 and of free, i.e., unbound T3 (“FT3”), if required, at a low complexity. In a particularly favourable variant of this embodiment, the state-space model is given by the equations

with

The processing device can determine at least one of said parameters P₁, …, P₈ of the state-space model itself by fitting the state-space model to levels of FT4, the drug, TSH or TRAb, e.g., when previous blood samples of the specific patient have been taken and corresponding levels of FT4, the drug, TRAb and TSH over time have been sensed therein. In a beneficial embodi- ment, however, the parameters P₁, …, P₈ of the state-space mod- el are predetermined such that the parameter P₁ is between 0.01/day and 2/day, preferably between 0.0347/day and 1.3863/day, particularly preferably 0.7/day; the parameter P₂ is between 1 pmol/(day∙1) and 30 pmol/ (day∙1), preferably between 2.8 pmol/(day∙1) and 27.72 pmol/ (day∙1), particularly preferably 10.05 pmol/(day∙1); the parameter P₃ is between 0.1 U/l and 600 U/l, prefera- bly between 1 U/l and 500 U/l, particularly preferably 50 U/l; the parameter P₄ is between 0.1 mg and 1000 mg, preferably between 0.21 mg and 12.36 mg, particularly preferably 3 mg; the parameter P₅ is between 1 pmol/(day∙1) and 15 pmol/ (day∙1), preferably between 2.8 pmol/(day∙1) and 9 pmol/(day∙1), particularly preferably 7.5 pmol/(day∙1); the parameter P₆ is between 0.1 U/l and 600 U/l, prefera- bly between 1 U/l and 500 U/l, particularly preferably 250 U/l; the parameter P₇ is between 0.05 µU/ml and 100 µU/ml, preferably between 1.6 µU/ml and 19.8 µU/ml, particularly pref- erably 10.7 µU/ml; and the parameter P₈ is between 0.01/day and 0.3/day, prefera- bly between 0.07/day and 0.17/day, particularly preferably 0.12/day. Thereby, the parameters are determined from data of a mul- titude of patients, e.g., from levels of FT4, TRAb, TSH, drug levels of other patients, which is particularly suitable in the initial phase of drug treatment when little or no information about the specific patient and his condition is available. The unit “U”, therein, denotes the enzyme unit. In an embodiment, the controller computes the required dosage only in dependence on the sensed and the target FT4 lev- el, particularly, in dependence on their difference. In an al- ternative and preferred embodiment, however, the controller is configured to compute the required dosage in further dependence on at least one of said initial levels of the drug, of TRAb, of TSH, and of FT3. Such a controller can, e.g., penalise higher levels of TRAb or FT3 and/or reward a higher level of TSH. In the latter case, the controller can guide the thyroid towards the target FT4 level while simultaneously reducing the drug dosage. Reducing the drug dosage - eventually to zero - results in a healthy thyroid and a cure of its hyperfunction. In a preferred variant of this embodiment the controller is configured to compute the required dosage according to

with u.......... the required dosage of the drug to be computed, k₁, k₂, k_p, k_i, k_TRAb, k_TSH...parameters of the controller, L_dg,0 ....... the initial drug level, L_FT4........ the sensed FT4 level, e.......... the difference between the sensed and target FT4 levels, L_TRAb,0...... the initial TRAb level, and L_TSH,0....... the initial TSH level. Such a linear controller allows to achieve the abovemen- tioned advantages while at the same time the processing device can be kept simple due to the linearity of the controller. For example, when the state-space model is linear or linearized around its initial state the processor can employ simple linear control theory to algebraically obtain the FT4 level over time and the controller parameters. The abovementioned initial levels of the drug, of TRAb, of TSH and of FT3 may be determined in different ways prior to the initialisation of the state-space model. In an advantageous em- bodiment, the sensing device is further configured to determine at least one of said initial levels of the drug, of TRAb, of TSH and of FT3 by sensing. The sensing results in particularly accurate initial levels and, consequently, in a particularly accurately measured required dosage. In an alternative or additional advantageous embodiment, the processing device further comprises an electronic state ob- server configured to determine at least one of said initial levels of the drug, of TRAb, of TSH and of FT3 by observation. By employing a state observer such as a Luenberger observer, a sliding mode observer or a Kalman filter, the corresponding in- itial levels can be determined without sensing and, thus, more efficiently. In a favourable embodiment, the system further comprises an administration device for administering the drug to the pa- tient and the controller is further configured to control the administration device to administer the measured required dos- age of the drug to the patient. The administration device, e.g., an infusion pump, automatically applies the determined dosage to the patient resulting in a fully automated drug dos- age measurement and administration system. In a second aspect, the invention provides for a pro- cessing device for determining, from a sensed FT4 level, a re- quired dosage of a drug for treating hyperfunction of a pa- tient’s thyroid, the processing device having an electronic processor, an electronic controller and an electronic memory storing a state-space model of the patient’s thyroid, which state-space model describes the FT4 level over time in depend- ence on a drug level, a level of thyrotropin receptor antibod- ies, TRAb, and a level of thyroid stimulating hormone, TSH,; wherein the processor is configured to receive the sensed FT4 level, to initialise the state-space model with the sensed FT4 level, an initial drug level, an initial TRAb level and an initial TSH level, and to adjust parameters of the controller such that, in a closed control loop of the controller control- ling the state-space model, the FT4 level described by the state-space model approaches a predetermined target FT4 level over time; and wherein the controller is configured to receive the sensed FT4 level and to compute, by means of the adjusted con- troller parameters, the required dosage in dependence on the sensed FT4 level and the target FT4 level. The processing device carries out the determination part of the measuring system. With respect to the advantages and variants of the processing device reference is made to the abovementioned embodiments and variants of the system. In a third aspect, the invention provides for a method em- ploying the system, i.e., a method for measuring, from the pa- tient’s blood sample, the required dosage of a drug for treat- ing hyperfunction of the patient’s thyroid, by means of the system, the method comprising the steps of: sensing, by means of the sensing device, a level of FT4 in the blood sample; receiving, in said processor and in said controller, the sensed FT4 level; storing, in said memory, a state-space model of the pa- tient’s thyroid, which state-space model describes the FT4 lev- el in dependence on a drug level, a TRAb level, and a TSH lev- el; initialising, by means of the processing device, the state-space model in the memory with the sensed FT4 level, an initial drug level, an initial TRAb level and an initial TSH level; adjusting, by means of the processing device, parameters of the controller such that, in a closed control loop of the controller controlling the state-space model, the FT4 level de- scribed by the state-space model approaches a predetermined target FT4 level over time; and computing, by means of the controller and the adjusted controller parameters, the required dosage in dependence on the sensed FT4 level and the target FT4 level. It shall be understood that the steps of the method may be carried out in any order, even in parallel, unless the comple- tion of one step is required for carrying out another step. In a beneficial embodiment of the inventive method, in the step of computing, the required dosage is computed to lie with- in a predetermined range. The benefit of such a limitation of the dosage is a twofold. The risk of an overdosage and thyroid hormone fluctuations is reduced. In a preferred embodiment of the method, the steps of sensing and computing are repeated for at least one further blood sample of the patient to obtain a sequence of required dosages. In this way, the required dosage is measured repeated- ly in regular or irregular intervals, which results in a more accurate measurement based on the most current condition of the patient; when the state-space model is not re-initialised and the controller parameters are not re-adjusted in each repeti- tion, the controller is a non-adaptive controller. In a partic- ularly accurate variant, the state-space model is initialised and the controller parameters are adjusted in each repetition, rendering the controller an adaptive controller. In a further alternative or additional variant of this em- bodiment, the required dosage is computed such that in the se- quence of required dosages the difference between subsequent required dosages lies within a predetermined range, e.g., in the case of methimazole, in a range between +50% and -50% of the respective previous required dosage in the sequence. There- by, the risk of overdosage and thyroid hormone fluctuations is further reduced. Relating to further variants and embodiments of the method and the advantages thereof, reference is made to the above statements on the system. The invention shall now be described in further detail by means of exemplary embodiments thereof under reference to the enclosed drawings, in which: Fig. 1 shows a hormone flow in a hypothalamic-pituitary- thyroid, HPT, axis including a patient’s thyroid in a biologic control loop diagram; Fig. 2 shows a system according to the invention in a block diagram; Fig. 3 shows two exemplary levels of FT4 over time in a level-time-diagram; Fig. 4 shows an optional observer of the system of Fig. 2 in a block diagram; and Fig. 5 shows several embodiments of a method according to the present invention in a flow diagram. As already described in the outset, Fig. 1 shows a hypo- thalamic-pituitary-thyroid (“HPT”) axis 1 of a patient’s thy- roid 2 producing the thyroid hormones triiodothyronine (“T3”) and thyroxine (“T4”). Once released into the blood circuit most T3 and T4 binds to plasma proteins, whereas some T3 and T4, called free triiodothyronine (“FT3”) and free thyroxine (“FT4”), respectively, remains unbound. In a healthy (human or animal) patient, the production of T3 and T4 is stabilised by the HPT axis 1. In case of hyper- function of the patient’s thyroid 2, the thyroid 2 produces too much T3 and T4. This is caused, e.g., by Graves’ autoimmune disease, symptomatic of which is the presence of antithyroid autoantibodies, e.g., TRAb, in the thyroid 2 and a significant- ly lowered TSH level, often lowered to zero. Herein, the term “TRAb” denotes either the sum of TSH- Stimulation Blocking Antibodies (“TSBAb”) and Thyroid Stimulat- ing Antibodies (“TSAb”), or TSAb only. Moreover, the term “lev- el” covers an absolute amount, e.g., a particle number or mass etc., on the one hand, and a relative amount, e.g., a number or mass per volume, per thyroid, per person, etc., on the other hand. Fig. 2 shows a system 5 for measuring, from a blood sample 6 of the patient, a drug dosage u that is required for treating hyperfunction of a patient’s thyroid 2. The drug is an antithy- roid agent such as methimazole, carbimazole, propylthiouracil, potassium perchlorate, etc. By administering the required dos- age u to the patient, a target FT4 level T_FT4 in the patient’s blood shall be reached over time; the target level T_FT4 is un- derstood to correspond to a healthy thyroid. As shown in Fig. 2, the system 5 comprises a sensing de- vice 7 and a processing device 8. The sensing device 7 which senses an FT4 level L_FT4 in the blood sample 6 is, e.g., a la- boratory device or a small sized device such as a lab-on-chip directly attached above or under the patient’s skin. By its sensing, the sensing device 7 performs an FT4 measurement in the blood sample 6, e.g., by an equilibrium analysis, a 2-step immunoextraction, an ultrafiltration, an HPLC-MS, a surface plasmon resonance, a radio immunoassay, an electrochemical im- munoassay, another immunoassay method etc. The processing device 8 is an electronic computing device, e.g., a programmable computer executing a corresponding soft- ware, or a specific hardware device, e.g., an ASIC, ASIP, FPGA, etc., which determines the required dosage u of the drug from the sensed FT4 level L_FT4. To this end, the processing device 8 comprises an electronic processor 9, an electronic controller 10, and an electronic memory 11. The processing device 8 of the example of Fig. 2 also has an input 12 connected to the proces- sor 9 and to the controller 10 for receiving the sensed FT4 level L_FT4 from the sensing device 7. The sensing device 7 may be connected directly to the input 12 (line 13 in Fig. 2) or the sensed FT4 level L_FT4 may be read from the sensing device 7 and, e.g., entered into a keyboard connected to the input 12 (not shown). The memory 11 stores a state-space model 14 (also called “plant model” in control theory) of the patient’s thyroid 2. The state-space model 14 describes the FT4 level L_FT4 over time in dependence on a drug level, a TRAb level and a TSH level, i.e., the FT4 level’s L_FT4 temporal change or evolution starting from an initial state of the thyroid 2 as will be exemplified in greater detail below. Before the state-space model 14 is initialised, an initial drug level L_dg,0, an initial TRAb level L_TRAb,0 and an initial TSH level L_TSH,0 are determined either as general estimations or as patient specific estimations; the estimations are, e.g., pro- vided by the attending physician and received via the input 12, or they are determined by the system 5 itself as will be de- scribed further below. The processor 9 then initialises the state-space model 14 with the sensed FT4 level L_FT4 received from the sensing device 7 and with the initial drug level L_dg,0, the initial TRAb level L_TRAb,0, and the initial TSH level L_TSH,0 by setting them as initial values in the state-space model 14. When the state-space model 14 is initialised, the proces- sor 9 adjusts parameters k* of the controller 10 which is, e.g., a linear controller such as a P-, PI-, PD-, PID- controller or a non-linear controller, and may be implemented in hardware or in software. To this end, the controller 10 and the state-space model 14 form a closed control loop 15 in which the controller 10 controls the state-space model 14 by means of the required dosage u and receives the FT4 level L_FT4 from the state-space model 14. In the course of this, the processor 9 adjusts the controller parameters k* such that the FT4 level L_FT4 approaches the predetermined target FT4 level T_FT4 over time, e.g., a target FT4 level T_FT4 of 15 pmol/l. By means of the adjusted controller parameters k*, the controller 10 computes the required dosage u in dependence on the sensed FT4 level L_FT4 and the target FT4 level T_FT4 as known in the art for controllers, particularly, in dependence on their difference, their respective absolute values and/or their respective evolutions. Thereby, the required dosage u is meas- ured by the system 5 from the patient’s blood sample 6. The required dosage u of the drug may then be administered to the patient, e.g., as one or more pills, infusions, etc., to eventually control the patient’s thyroid 2 (also called the “controlled system” or “plant” in control theory). Therein, the required dosage u is, e.g., an absolute dosage, a dosage per week, per day, per hour, or a continuous dosage, respectively. For administering the drug to the patient, the system 5 optionally comprises an administration device 16, e.g., an in- fusion pump, which is connected to the controller 10 via an output 17 of the processing device 8. In this case, the con- troller 10 controls the administration device 16 to administer the measured required dosage u of the drug to the patient. Al- ternatively, the controller 10 outputs the measured required dosage u via the output 17, e.g., on a display (not shown) for manual administration. Hereinafter, details and exemplary embodiments of the state-space model 14, the initialisation thereof, the adjust- ment of the controller parameters k*, and the computation of the required dosage u shall be explicated. For describing the FT4 level L_FT4 over time, the state- space model 14 reflects the FT4 production in the patient’s body and, optionally, one or more of the ingestion and degrada- tion of the drug, the production and degradation of TRAb, TSH and T3 or FT3, etc. in the patient’s body. In some embodiments, the state-space model 14 is a linear state-space model. In oth- er embodiments, the state-space model 14 is a nonlinear state- space model, e.g., given by a set of nonlinear differential equations or by a neural network. In one embodiment which is exemplified below, the state- space model 14 is nonlinear with respect to the drug level and comprises two Michaelis-Menten terms with substrates and inhib- itors, i.e., a first Michaelis-Menten term with TRAb acting as the substrate and the drug acting as the inhibitor, and a sec- ond Michaelis-Menten term with TSH acting as the substrate and the drug acting as the inhibitor, i.e., the FT4 production is described using

for uncompetitive inhibition, or using

for competitive inhibition, or using

for mixed inhibition, with

In one exemplary variant of this embodiment, the state- space model 14 is given by

In the second line of equation (3), the first Michaelis- Menten term is a term for competitive inhibition with TRAb act- ing as the substrate and the drug acting as the (here: competi- tive) inhibitor and the second Michaelis-Menten term is a term for mixed inhibition with TSH acting as the substrate, the drug acting as the first (here: competitive) inhibitor and TRAb act- ing as the second (here: uncompetitive) inhibitor. Moreover, the model parameters P₁, …, P₈ can be determined in different ways, e.g., by fitting the state-space model 14 to levels of FT4, the drug, TSH or TRAb that have been sensed ear- lier. At least some of the parameters P₁, …, P₈ may alterna- tively or additionally be determined by experience, e.g., such that the parameter P₁ is between 0.01/day and 2/day, preferably between 0.0347/day and 1.3863/day, particularly preferably 0.7 /day, the parameter P₂ is between 1 pmol/(day∙1) and 30 pmol/ (day∙1), preferably between 2.8 pmol/(day∙1) and 27.72 pmol/ (day∙1), particularly preferably 10.05 pmol/(day∙1), the parame- ter P₃ is between 0.1 U/l and 600 U/l, preferably between 1 U/l and 500 U/l, particularly preferably 50 U/l, the parameter P₄ is between 0.1 mg and 1000 mg, preferably between 0.21 mg and 12.36 mg, particularly preferably 3 mg, the parameter P₅ is be- tween 1 pmol/(day∙1) and 15 pmol/(day∙1), preferably between 2.8 pmol/ (day∙1) and 9 pmol/(day∙1), particularly preferably 7.5 pmol/ (day∙1), the parameter P₆ is between 0.1 U/l and 600 U/l, preferably between 1 U/l and 500 U/l, particularly preferably 250 U/l, the parameter P₇ is between 0.05 µU/ml and 100 µU/ml, preferably between 1.6 µU/ml and 19.8 µU/ml, particularly pref- erably 10.7 µU/ml, and/or the parameter P₈ is between 0.01/day and 0.3/day, preferably between 0.07/day and 0.17/day, particu- larly preferably 0.12/day; these values of the model parameters P₁, …, P₈ are representative for most patients by experience. The unit “U”, therein, denotes the enzyme unit. The processor 9 adjusts the controller parameters k* as known in the art, e.g., by pole placement, eigenvalue assign- ment, a linear-quadratic regulator (LQR) approach, by simulat- ing the control loop 15, etc. In general, the processor 9 ad- justs the controller parameters k* depending on the controller 10 and the state-space model 14 as known in control theory. An adjustment of the controller parameters k* by eigenval- ue assignment shall now be exemplified for the state-space mod- el 14 given by equations (2) and (3); in this example, the con- troller 10 computes the required dosage u according to

with u.......... the required dosage, k₁, k₂, k_p, k_i, k_TRAb, k_TSH... the parameters k* of the controller 10, i.e., k* = {k₁, k₂, k_p, k_i, k_TRAb, k_TSH}, L_dg......... the drug level as described by equation (2), L_FT4........ the FT4 level as described by equation (3), e.......... the difference between the sensed and target FT4 levels L_FT4, T_FT4, L_TRAb,0...... the initial TRAb level, and L_TSH,0....... the initial TSH level. Therein, the processor 9 linearises equation (3) about the initial levels L_dg,0, L_TRAb,0, L_TSH,0 to obtain a linear model given by the differential equation

u.......... the required dosage. From this linear model, the processor 9 first calculates the parameter k_p as

with T denoting transposing. Then, the processor 9 sets up the 3x3 matrix

with denoting the direct product.

The processor 9 further sets up the eigenvalue equation of the matrix D and its characteristic polynomial, and calculates, from predetermined eigenvalues of the matrix D and the charac- teristic polynomial, the parameters k₁, k₂, k_i . The eigenvalues are each predetermined to have a negative real part, e.g. , be- tween -0.1 and -10, for the FT4 level L_FT4 approaching the tar- get FT4 level T_FT4. The parameter k_TSH may be set to a negative number, e.g. , between -0.08 and -9, and the parameter k_TRAb to a positive number, e.g. , between 0.005 and 0.3.

Fig. 3 shows two different FT4 levels L_FT4 over time

(curves 18 and 19) obtained for different assigned eigenvalues and, hence , for different controller parameters k* . As can be seen in the first curve 18 of Fig. 3, by assigning suitable ei- genvalues, an oscillation between hyperthyroidism and hypothy- roidism, large fluctuations of thyroid hormones can be avoided, and a smooth, steady approach of the target FT4 level T_FT4 is achieved . When assigning different eigenvalues, on the other hand, the FT4 level L_FT4 would oscillate between hyperthyroidism and hypothyroidism and the target FT4 level T_FT4 , if ever, is only steadily reached after long time, cf . curve 19.

When the controller parameters k* are adjusted, e.g. , as described above with reference to equations (5) to (7) , the controller 10 computes the required dosage u to be administered to the thyroid 2, e.g. , in dependence on the sensed FT4 level

L_FT4 , the target level T_FT4, the initial drug level L_dg,0, the in- itial TRAb level L_TRAb,0 and the initial TSH level L_TSH,0 accord- ing to

with u.......... the required dosage, k₁, k₂, k_p, k_i, k_TRAb, k_TSH... the parameters k* of the controller 10, L_dg,0 ....... the initial drug level, L_FT4........ the FT4 level, e.......... the difference between the sensed and target FT4 levels L_FT4, T_FT4, L_TRAb,0...... the initial TRAb level, and L_TSH,0....... the initial TSH level. Alternatively, the controller 10 computes the required dosage u differently, for instance, only in dependence on the sensed FT4 level L_FT4 and the target FT4 level T_FT4, e.g., when k₁, k_TRAb and k_TSH are zero in equations (4) and (5), or in fur- ther dependence on an initial level of free T3 (“FT3”), etc.; any dependence on the target, sensed and initial levels T_FT4, L_FT4, L_dg,0, LTRAb_,0, L_TSH,0, L_FT3,0 is possible. Two combinable embodiments for determining one or more of the initial levels of the drug, TRAb, TSH and FT3 L_dg,0, L_TRAb,0, L_TSH,0 and L_FT3,0 for initialising the state-space model 14 and/or for the controller’s computation shall now be described: In a first optional embodiment, the sensing device 7 or another sensing device determines at least one of the initial levels L_dg,0, L_TRAb,0, L_TSH,0 and/or L_FT3,0 by sensing, prior to ini- tialising the state-space model 14. In a second optional embodiment according to Fig. 4, the processing device 8 further comprises an electronic state ob- server 20, in this example a Luenberger observer. The state ob- server 20 determines at least one of said initial levels L_dg,0, L_TRAb,0, L_TSH,0 and/or L_FT3,0 by observation prior to initialising the state-space model 14. The exemplary Luenberger observer em- ploys an observer model 21 between the input 12 and the con- troller 10. The observer model 21, in this example, is given by the equations

as an extension to the state-space model 14 of equations (2) and (3), wherein W = [W₁, W₂, W₃] is a (here: one-dimensional) weighting matrix, L'_FT4 is an FT4 level described by the observ- er model 21 and P₉ is a further model parameter, e.g., chosen between 0.0001 and 0.01, preferably between 0.001 and 0.005. To observe the initial TRAb level L_TRAb,0 according to equations (8) to (10), the state observer 20 subtracts the sensed FT4 level L_FT4 received via the input 12 from the FT4 level L'_FT4 described by the observer model 21 (subtraction 22) and weights the dif- ference with the weighting matrix W (block 23) to determine the initial TRAb level L_TRAb,0 according to equation (10). It goes without saying that, instead of or in addition to the initial TRAb level L_TRAb,0, any other level L_TSH,0, L_FT3,0, L_TSAb,0, L_TSBAb,0 could be determined by observation. Moreover, the state observer 28 may be any other state observer known in the art, e.g., a sliding mode observer, a Kalman filter, etc. Fur- thermore, the initial levels L_dg,0, L_TRAb,0, L_TSH,0 and/or L_FT3,0 may be determined differently, e.g., by extrapolating a respective level over time, by estimating from a previously sensed respec- tive level, by educated guessing, for instance, assuming a TSH level L_TSH,0 of zero in the early phase of treating Graves’ dis- ease etc. With reference to Fig. 5, various embodiments of a method 24 for measuring the required dosage u carried out by the sys- tem 5 of Fig. 2 shall now be illustrated. In step 25, the sensing device 7 senses the FT4 level L_FT4 in the patient’s blood sample 6. In steps 26' and 26" which are carried out in any mutual sequence (including in parallel) but after step 25, the controller 10 and the processor 9 receive the sensed FT4 level L_FT4 from the sensing device 7, respective- ly. In step 27, the state-space model 14 of the patient’s thy- roid 2 is stored in the memory 11; step 27 may be carried out prior to, in parallel with or after steps 25, 26' and 26". In step 28 after steps 25, 26" and 27, the processor 9 initialises the state-space model 14 in the memory 11 with the sensed FT4 level L_FT4, the initial drug level L_dg,0, the initial TRAb level L_TRAb,0 and the initial TSH level L_TSH,0. In step 29 which follows steps 25 to 28, the processor 9 adjusts the parameters k* of the controller 10 such that in the closed control loop 15 of the controller 10 controlling the state-space model 14, the FT4 level L_FT4 described by the state-space model 14 approaches the predetermined target FT4 level T_FT4 over time. In subsequent step 30, the controller 10 computes, by means of the adjusted controller parameters k*, the required dosage u in dependence on at least the sensed and the target FT4 level L_FT4, T_FT4. Following the measurement of the required drug dosage u in steps 25 to 30, the determined drug dosage u is optionally ad- ministered to the patient in an administration step 31. In an embodiment of the method 24 at least some of the steps 25 to 31 are repeated to obtain - and optionally adminis- ter - a sequence of required dosages u: In a first variant, all of the steps 25 to 31 are carried out repeatedly for respective further blood samples 6 of the patient, as symbolised by a loop 32 in Fig. 5. Therein, whenev- er a more recent initial level L_dg,0, L_TRAb,0, L_TSH,0 is available, e.g., a sensed or observed one, the state-space model 14 is re- initialsied and the controller parameters k* is re-adjusted on the basis of the most recent levels available. In a second variant, only steps 25, 26', 30 and 31 are carried out repeatedly for respective further blood samples 6 of the patient as symbolised by a branch 33 of the loop 32 in Fig. 5. Therein, the state-space model 14 is only initialised once and the parameters k* are only adjusted once at the begin- ning of the method 24. In an optional step 34, the state observer 20 determines at least one of the initial levels of the drug, of TRAb, of TSH and FT3 L_dg,0, L_TRAb,0, L_TSH,0, L_FT3,0 by observation as described above. In optional embodiments of the method 24, the required dosage u is computed such that it lies within a predetermined range in step 30. Alternatively or additionally, when the re- quired dosage u is measured repeatedly or a previous dosage is available, the required dosage u can be computed such that in the sequence of required dosages u the difference between sub- sequent required dosages u lies within a predetermined range which may be an absolute or relative range, e.g., in the case of methimazole, a range between +50% and -50% of the respective previous required dosage in the sequence. The present invention is not restricted to the specific embodiments and variants described in detail herein but encom- passes all those variants, combinations and modifications thereof that fall within the scope of the appended claims.

Claims

Claims: 1. A system for measuring, from a patient’s blood sample (6), a required dosage (u) of a drug for treating hyperfunction of the patient’s thyroid (2), the system (5) comprising a sensing device (7) for sensing a level of free thyrox- ine, FT4, (L_FT4) in the blood sample (6), and a processing device (8) having an electronic processor (9), an electronic controller (10) and an electronic memory (11) storing a state-space model (14) of the patient’s thyroid (2), which state-space model (14) describes the FT4 level (L_FT4) over time in dependence on a drug level, a level of thyrotropin receptor antibodies, TRAb, and a level of thyroid stimulating hormone, TSH; wherein the processor (8) is configured to receive the sensed FT4 level (L_FT4) from the sensing device (7), to initial- ise the state-space model (14) with the sensed FT4 level (L_FT4), an initial drug level (L_dg,0), an initial TRAb level (L_TRAb,0) and an initial TSH level (L_TSH,0), and to adjust parameters (k*) of the controller (10) such that, in a closed control loop (15) of the controller (10) controlling the state-space model (14), the FT4 level (L_FT4) described by the state-space model (14) ap- proaches a predetermined target FT4 level (T_FT4) over time; and wherein the controller (10) is configured to receive the sensed FT4 level (L_FT4) from the sensing device (7) and to compute, by means of the adjusted controller parameters (k*), the required dosage (u) in dependence on the sensed FT4 level (L_FT4) and the target FT4 level (T_FT4).

2. The system according to claim 1, wherein the state- space model (14) comprises a first Michaelis-Menten term with TRAb acting as a substrate and said drug acting as an inhibi- tor, and a second Michaelis-Menten term with TSH acting as a substrate and said drug acting as an inhibitor.

3. The system according to claim 2, wherein the state- space model (14) is given by the equations

4. The system according to claim 3, wherein the parame- ters of the state-space model (14) are predetermined such that the parameter P₁ is between 0.01/day and 2/day, preferably between 0.0347/day and 1.3863/day, particularly preferably 0.7/day; the parameter P₂ is between 1 pmol/(day∙1) and 30 pmol/(day∙1), preferably between 2.8 pmol/(day∙1) and 27.72 pmol/(day∙1), particularly preferably 10.05 pmol/(day∙1); the parameter P₃ is between 0.1 U/l and 600 U/l, prefera- bly between 1 U/l and 500 U/l, particularly preferably 50 U/l; the parameter P₄ is between 0.1 mg and 1000 mg, preferably between 0.21 mg and 12.36 mg, particularly preferably 3 mg; the parameter P₅ is between 1 pmol/(day∙1) and 15 pmol/(day∙1), preferably between 2.8 pmol/(day∙1) and 9 pmol/(day∙1), particularly preferably 7.

5 pmol/(day∙1); the parameter P₆ is between 0.1 U/l and 600 U/l, prefera- bly between 1 U/l and 500 U/l, particularly preferably 250 U/l; the parameter P₇ is between 0.05 µU/ml and 100 µU/ml, preferably between 1.

6 µU/ml and 19.8 µU/ml, particularly pref- erably 10.

7 µU/ml; and the parameter P₈ is between 0.01/day and 0.3/day, prefera- bly between 0.07/day and 0.17/day, particularly preferably 0.12/day. 5. The system according to any one of claims 1 to 4, wherein the controller (10) is configured to compute said re- quired dosage (u) also in dependence on at least one of said initial levels of the drug (L_dg,0), of TRAb (L_TRAb,0), of TSH (L_TSH,0), and of free triiodothyronine, FT3. 6. The system according to claim 5, wherein the control- ler (10) is configured to compute the required dosage (u) ac- cording to

8. The system according to any one of claims 1 to 7, wherein the processing device (8) further comprises an elec- tronic state observer (20) configured to determine at least one of said initial levels of the drug (L_dg,0), of TRAb (L_TRAb,0), of TSH (L_TSH,0) and of FT3 (L_FT3,0) by observation.

9. The system according to any one of claims 1 to 8, wherein the system (5) further comprises an administration de- vice (16) for administering the drug to the patient and wherein the controller (10) is configured to control the administration device (16) to administer the measured required dosage (u) of the drug to the patient.

10. A processing device for determining, from a sensed level of free thyroxine, FT4, (L_FT4), a required dosage (u) of a drug for treating hyperfunction of a patient’s thyroid (2), the processing device (8) having an electronic processor (9), an electronic controller (10) and an electronic memory (11) stor- ing a state-space model (14) of the patient’s thyroid (2), which state-space model (14) describes the FT4 level (L_FT4) over time in dependence on a drug level, a level of thyrotropin re- ceptor antibodies, TRAb, and a level of thyroid stimulating hormone, TSH; wherein the processor (8) is configured to receive the sensed FT4 level (L_FT4), to initialise the state-space model (14) with the sensed FT4 level (L_FT4), an initial drug level (L_dg,0), an initial TRAb level (L_TRAb,0) and an initial TSH level (L_TSH,0), and to adjust parameters (k*) of the controller (10) such that, in a closed control loop (15) of the controller (10) controlling the state-space model (14), the FT4 level (L_FT4) de- scribed by the state-space model (14) approaches a predeter- mined target FT4 level (T_FT4) over time; and wherein the controller (10) is configured to receive the sensed FT4 level (L_FT4) and to compute, by means of the ad- justed controller parameters (k*), the required dosage (u) in dependence on the sensed FT4 level (L_FT4) and the target FT4 level (T_FT4).

11. A method for measuring, from a patient’s blood sample (6), a required dosage (u) of a drug for treating hyperfunction of the patient’s thyroid (2) by means of a system (5) according to any one of claims 1 to 9, the method (24) comprising the steps of: sensing (25), by means of the sensing device (7), a level of free thyroxine, FT4, (L_FT4) in the blood sample (6); receiving (26', 26"), in said processor (9) and in said controller (10), the sensed FT4 level (L_FT4) from the sensing device (7); storing (27), in said memory (11), a state-space model (14) of the patient’s thyroid (2), which state-space model (14) describes the FT4 level (L_FT4) over time in dependence on a drug level, a level of thyrotropin receptor antibodies, TRAb, and a level of thyroid stimulating hormone, TSH; initialising (28), by means of the processor (9), the state-space model (14) in the memory (11) with the sensed FT4 level (L_FT4), an initial drug level (L_dg,0), an initial TRAb lev- el (L_TRAb,0) and an initial TSH level (L_TSH,0); adjusting (29), by means of the processor (9), parameters (k*) of the controller (10) such that, in a closed control loop (15) of the controller (10) controlling the state-space model (14), the FT4 level (L_FT4) described by the state-space model (14) approaches a predetermined target FT4 level (T_FT4) over time; and computing (30), by means of the controller (10) and the adjusted controller parameters (k*), the required dosage (u) in dependence on the sensed FT4 level (L_FT4) and the target FT4 level (T_FT4).

12. The method according to claim 11, wherein in said step of computing (30) the required dosage (u) is computed such that the required dosage (u) lies within a predetermined range.

13. The method according to claim 11 or 12, characterized by repeating the steps of sensing (25) and computing (30) for at least one further blood sample (6) of the patient to obtain a sequence of required dosages (u).

14. The method according to claim 13, wherein in said step of computing (30) the required dosage (u) is computed such that in the sequence of required dosages the difference between subsequent required dosages (u) lies within a predetermined range.

15. The method according to any one of claims 11 to 14, wherein the state-space model (14) is given by the equations

16. The method according to any one of claims 11 to 15, wherein the required dosage (u) is computed according to

with u.......... the required dosage of the drug, k₁, k₂, k_p, k_i, k_TRAb, k_TSH...parameters of the controller (10), L_dg,0 ....... the initial drug level, L_FT4........ the sensed FT4 level, e.......... the difference between the sensed and target FT4 levels (L_FT4, T_FT4), L_TRAb,0...... the initial TRAb level, and L_TSH,0....... the initial TSH level.