WO2001087402A1 - Pharmacokinetics-algorithms for controlling and regulating the consumption of various drugs - Google Patents

Abstract

The invention relates to an accurate computer-based method (the 'Method'), related algorithms and apparatus to help diabetic individuals achieve near euglycemic blood glucose levels, as recommend based on Diabetes Control and Complications Trial (DCCT), and in general to help physicians and medical practitioners to devise a dynamic and interactive drug regimen to manage and control various medical conditions and illness. The method uses artificial intelligence to learn about the drug behavior and assimilation for each individual and devises an ideal drug regiment for each such individual, according to the individual's lifestyle and medicals needs. When applied to management of diabetes, the Method achieves the DCCT goals despite diet variations, meal times and other day-to-day changes, with minimum number of injections.

Description

Pharmacokinetics- Algorithms for Controlling and Regulating tne consumption 01 various Drugs

To date, the medical practitioners have been prescribing drugs and medications based on gut feel and a qualitative approach. As used in this document, the term "drug" refers to but is not limited to any chemical compound that may be used on or administered to humans or animals as an aid in the diagnosis, treatment, or prevention of disease or other abnormal condition, for the relief of pain or suffering, or to control or improve any physiologic or pathologic condition. As separate fields, medicine (specifically, education and the course of study followed physicians), pharmacist and biochemists have a tremendous level of education, and their respective fields are very scientific. However, the interdisciplinary fields, particularly that of pharmacokinetics,¹ are treated in a qualitative, trial and error approach. Thus, Dr. Roger Jelliffe, USC School of Medicine,² expresses his frustration in how the medical field prescribed drugs: "Clinicians look at serum [drug] concentrations to see if they are in the so-called therapeutic range, and often adjust the dose intuitively if they are not. . .[they] often do this without looking carefully to see if the patient is tolerating that concentration or not, or if a higher or a lower level might actually be more desirable. Often, if the level is in the "therapeutic " range . . . no further consideration is given as to whether or not a different serum level might be better for the patient. . . .

In our desire to classify such relationships [therapeutic ranges of serum concentrations], we have thrown away the really important data of these quantitative relationships in order simply to develop a qualitative cutoff on the bottom and the top. As a result, we have lost the ability to think of giving a gentle regimen to a patient who needs it, and a more aggressive one to an- other patient who needs a firmer approach . . . looking only at populations, not individuals, without any consideration of what is really going on with each individual patient

To date, the medical community has not had adequate tools to determine proper drug dosages. While the pharmaceutical companies conduct extensive and sophisticated research and clinical trial to study the behavior of drugs and optimize the recommended regimens, the medi- cal community continues to prescribe them based on generalized rules and trial and error, following qualitative approach, without "any consideration of what is really going on with each individual patient."

The following discussion shows why, even the most dedicated medical doctor would be unable to calculate, a priori, and to prescribe correct or ideal drug or serum dosages to treat a given medical condition. For the purpose of simplicity, this discussion will often be illustrated with examples of insulin, as applicable to diabetes; though the concepts discuss herein apply to the entire field of pharmacokinetics, as related to drug or dosages.

The prior art - the way things are being done today. The applicant will refer to several publications, most fairly recent as to the date of this application, to illustrate the prior art. The illustrations are not meant as a criticism of the particular references or the corresponding author; but rather as an illustration of the limitations in the prior art.

A. FIXED MEASUREMENTS AND OVΈRSΓMP DJIED RECOMMENDATIONS FOR ALL PATIENTS Often, the medical community, and the corresponding books and articles written for the medical community, uses fixed amounts of a drug or serum as a recommendation to treat a certain condition. For example, for diabetic patients who must take insulin, to help the physician determine how much regular insulin to prescribe, the Intensive Diabetes Management³ handbook recommends: "If blood glucose is:

• 200-250 mg/dl

• Increase premeal dose of regular insulin by 3 U.

• Consider delaying meal an extra 15 min (to 45 min after injection)

• 250-300 mg/dl Increase premeal dose of regular insulin by 4 U.

• Consider delaying meal an extra 20-30 min (to 40-60 min after injection) "⁴ The same reference recommends how to determine the amount of lispro insulin:

"If blood glucose is:

• 200-250 mg/dl • Increase premeal dose of lispro by 3 U

• 250-300 mg/dl

• Increase premeal dose of regular insulin by 4 U.

• Consider delaying meal to 10-20 min after injection "^{s 6}

Based on^* the above recommendations, the physician may conclude that both insulins be- have roughly the same, and the physician may pass-on such recommendations to his or her patient. However, there are several limitations resulting from such simplistic recommendations:

Not all patients weight the same. Just as one patient may weigh 10 pounds, another patient may weigh 300 pounds. The blood glucose for a given individual is a function of many factors, including the person's caloric intake and weight. Thus, all other factors being the same, one gram of glucose would raise the blood glucose level of a 10 pound patient 30 times as much as is would raise the glucose level of a 300 pound patient. Therefore, it is not likely that patients from these extremes can derive the same benefit from the same recommendations.

Even for patients within the same weight category, say 120 pounds, the insulin require- ments of one patient may be completely different than the requirements for another patient. Thus, there are patients that must consume 10 units of exogenous insulin per day, to maintain adequate blood glucose, while another patient of the same weigh may require 100 units of exogenous insulin per day to maintain adequate glucose. There are several reasons why this may happen. For example, the individuals may have different levels of insulin resistance, hence dif- ferent levels of efficiency to absorb exogenous insulin. Also, just because these individuals are diabetic, does not mean that they do not secrete endogenous insulin; its just that they do not secrete an adequate amount of such insulin, thus these individuals complement their endogenously secreted insulin with exogenous insulin. As such, two seemingly like individuals may require completely different levels of insulin. The two insulins mentioned above, regular and lispro, behave significantly different.

Thus, Dr. Walsh⁷ states "Once injected, Regular insulin takes 30 minutes to begin working, peaks between 2 and 4 hours and hangs on for 6 to 8 hours. Humalog [a trade name for Lispro], on the other hand, begins working in 10 minutes, peaks at 1 to 2 hours and is gone in about 3 and 1/2 hours." FIGURE 1 shows the time activity of human insulins. The first two figures show the activity profiles of lispro and regular respectively (referred to as Humalog and Humulin R). The distribution (time-activity profile) of regular and lispro are seen to be completely different.

However, FIGURE 1, though published by the pharmaceutical company, does not show an accurate picture of the drug's behavior. Every unit of insulin will counteract the same amount of carbohydrates, though at variable rates. Therefore, the area under the curve for one unit of lispro must be the same as the area under the curve for one unit of regular. The published graphs, such as FIGURE 1, are adequate to determine the time-activity profiles; but they give little guidance about the serum strength achieved by each type of insulin. To be meaningful, the area under the curve for each unit of insulin must be the same, only then, can one estimate (or calculate) the serum-strength achieved by the drug. Currently, the medical field groups lispro and regular, as fast acting insulins, and treats them as behaving substantially similar. This oversimplification is incorrect. At best, it results in inadequate management of the glycemia for diabetic patients. And, at worst, the misuse of these insulins may result in serious complications, hypoglycemia, coma and even death. To illustrate the reason for these statements, the reader is referred to FIGURE 2, which shows the amplitude (serum strength) and time-activity profile of one unit of regular and one unit of lispro. Both being one unit, the area under the amplitude-time curve is the same for both. The vertical axis shows the serum strength achieved by the insulin, while the horizontal axis shows its action time. FIGURE 2 shows that lispro is absorbed and is dissipated much faster, within the body, than regular. In doing so, lispro achieves a serum strength that is over twice as high (0.425 vs. 0.195) that achieved by regular. Therefore, the recommendations of 3 units of regular, or alternatively 3 units of lispro, does not make sense, as they result in totally different serum concentrations of insulin in the body.

As a further example of the limitations of the prior art, a reputable physician's reference states:⁸

In typical patients . . . the total daily insulin dose required for meticulous elvcemic control is 0.5-1.0 Uper kg body weight per day.

But, when referring to drugs that may have a tremendous adverse effect on an individual, a 2:1 difference in the amount of a hormone is far from meticulous and may actually be danger- ous.

B. WHAT IS NEEDED TO CORRECT THE PROBLEMS OF "FIXED MEASUREMENTS AND OVERSIMPLIFIED RECOMMENDATIONS FOR ALL PATIENTS"

What is needed is a method of calculating the required dosage of a drug. The required drug dosage must be such that: 1. The recommended amount should be related to the patient's needs based on physiologically logical parameters (such as weight), at least as applicable to such situations as when the dosage based on such parameters, e.g., a low weight patient, such as a small child, would not be the same as the dosage recommendation for a patient exhibiting different magnitude of such parameters, e.g., a heavier patient (e.g., a 300 lb. adult). 2. The recommended amount should be related to the patient's needs when there is medically significant inteφatient variation of recommended amounts; e.g., when two patients of substantially similar weight require substantially different amounts of the drug or medication.

3. A method that takes into consideration the absoφtion rates or other parameters of the drug. Being that the two drugs of the same family, may result in substantially different time- concentration relationships, the recommended dosage should take into consideration such relationship to achieve a desirable drug concentration in the individual.

4. A method that automatically calculates, for the patient or for his or her treating health care professional, the amount of the drug; rather than having to refer to recommended tables. To achieve optimum control, such method should correct the above-mentioned problems and should be on a variable scale. For example, rather than having to put the patient in the category of 200- 250 mg/dl, the method should recommend a drug amount for any glucose level. So, if the glucose level is 214 mg/dl, the recommendation would be different than if the glucose level is 244 mg/dl. 5. A method that does away with the cumbersome approach of having to consider other circumstances, such as having to wait a predetermined delay of time after injection before the meal. C. OVERSIMPLIFIED AND MATHEMATICALLY INACCURATE MODELING OF DRUG BEHAVIOR.

Often, the medical community, and the corresponding written references used by the community, makes recommendations or prescriptions based on fictitious and or idealized drug response/behavior. Here, the applicant discusses the drug response, as graphical representations of the actual drug behavior. In reality, the mathematical behavior of a drug, or its graphical representation is, for our puφoses, the same thing.

For example, FIGURE 3 shows a schematic representation of idealized insulin effect, as shown in Intensive Diabetes Management.⁹ FIGURE 4 shows the mathematical representation, via normalized curves, of the time of activity and serum strength profile for several types of insulin. The simplistic recommendations of FIGURE 3 have several limitations:

The difference between the idealized curves and a mathematically accurate representation are significant, which difference could represent errors in excess of several hundred percent in the estimate of serum strength. For example, comparing a mathematical representation of various types of insulin with the idealized curves show that there is little resemblance between them. Compare the Humalog (lispro) and regular, as shown in the idealized representation of FIGURE 3, with the more mathematically accurate activity profiles of FIGURE 4. One unit of Humalog (lispro) results in twice the serum strength of insulin (albeit, for a shorter period of time) than regular. Similarly, one unit of regular results in over three times the serum strength of Lente or Ultralente.

An idealized curve may be adequate to discuss, from a qualitative point of view, the administration of a given amount of a drug. But when the effect of such a drag dosage superimpose onto previous dosages, and when the consequences of such drugs are such that they directly affect the long term and short term health on the patient, then a much more accurate mathematical representation is required.

By using idealized representations, the medical community is unable to calculate the effect of superimposing different dosages, or the effect of administering dosages before the effects of the previous dosages has worn out. D. WHAT IS NEEDED TO CORRECT THE PROBLEMS OF "OVERSIMPLIFIED AND MATHEMATICALLY INACCURATE MODELING OF DRUG BEHAVIOR"

What is needed is a method of:

1. Generating a mathematically accurate representation of the time of activity and serum strength and other important behavior characteristics of a drug.

2. Calculating the mathematical supeφosition of the effect of one or more drugs and the effect and supeφosition of administering these drags at different times; and calculating whether or not the effect of the previous administration has worn out, or the effects of continuous drug administration, such as by a continuous infusion pump. 3. Mathematically calculating the effects of previous administration of the drugs and projecting the anticipated effect of the drug or drugs, after its administration and the effects for anticipated events, e.g., additional administration of the drug, or behavioral characteristics. For example, in the case of diabetic individuals, anticipating that certain period of time later, the individual will consume meals, or engage in exercise. E. COMPLEX DOSAGE REQUIREMENTS.

Often, the drug dosages; that is, the amounts of the drug, types of drugs and the time administration of the drug, are complex. For example, in the treatment of diabetes, the object of the insulin is to counteract the glucose in the blood stream of the patient to bring and maintain such concentrations within an acceptable level. The reader may refer to US Patent 6,051,551 for a discussion of Diabetes, the DCCT trials, and benefits of a well-managed glycemia.

The regulatory mechanisms of a normal individual help the individual maintain their glucose levels between 80-120 mg/dl. By contrast, the glucose levels of an uncontrolled diabetic may reach 1000 mg/dl. The glucose levels for diabetics under the control of insulin may fluctuate to levels as high as 300-500 mg/dl and sometimes as low as 50 mg/dl. The high glucose (hyperglycemia) results in many detrimental long-term health effects. If the levels get too high, the patient may find him or herself hospitalized, often with serious and fatal consequences (e.g., diabetic ketoacidosis). At the low end, the diabetic may find him or herself in a hypoglycemia, often with serious or fatal consequences.

FIGURE 5, illustrates a glucose distribution for a typical (albeit fictitious) diabetic indi- vidual. In the figure, the top curve illustrates a fictitious glycemia level for a diabetic. The lower curve shows an idealized glucose level after treatment with insulin.

In the prior art, the medical community has been using idealized insulin distributions to tailor an insulin regimen for the patient. However, such effort to tailor an insulin regimen (or regimen for other drugs, to accommodate a patient's specific needs) have not been successful. For example, a well-known physician's reference handbook for internal medicine recommends the following insulin administration:¹⁰

"One approach to initiation of therapy involves administration of 25 percent of the previous daily insulin dose in the patient's conventional regimen at bedtime as intermediate insulin (NPH or lente insulin), with the other 75 percent given as regular insulin divided such tat 40, 30, and 30 percent is given 30 min before brealfast, lunch, and supper, respectively. "

FIGURE 6, shows that the resulting glycemic distribution of such a regimen, based on careful mathematical program of the above-recommended regimen of four injections per day.

The final glycemic distribution (the lower curve) is far from ideal. This fictitious patient, in this case, is taking four injections of insulin, and yet this patient's glycemia ranges from 97 to about

260 mg/dl (normal range is 80-120 mg/dl).

Another reputable reference recommends:¹¹

At least twice-daily insulin injections are necessary to adequately control blood glucose levels. In general, two-thirds of the total dose for the day is given in the morning, in a 2:1 ratio of intermediate-to short-acting insulin (e.g., NPH:regular). Tlie remaining third is given before dinner in a 1:1 ratio. "

It is important to note that management of the glycemia of a pregnant woman is much more important than with other individuals, as improper glycemic management results is high rates of fetal morbidity and mortality. FIGURE 6, shows that the resulting glycemic distribution of such a regimen. A careful mathematical program of the above-recommended regimen shows that this fictitious expecting mother's glycemia ranges from about 75 to about 295 mg/dl (normal range is 80-120 mg/dl).

Determining the correct dosage for a given patient can indeed be very complicated. In a popular book, Dr. Richard S. Baeser and the Staff at the Joslin Diabetes Center states:¹² You may need to be hospitalized to establish insulin doses and set up your personal algorithm, especially if you are going on an insulin pump. Not everyone needs to be hospitalized, however, and intensive outpatient programs are often recommended. Regardless of how you get started, the final treatment program is more effectively established over time as an outpatient. F. WHAT IS NEEDED TO CORRECT THE PROBLEMS OF "COMPLEX DOSAGE REQUIREMENTS" What is needed is a method, preferably automated through machine (computer) interface, which can tailor a drag recommendation, including such matters as amounts of the drug, combinations of the drug and corresponding times of injection or administration, such that the result is as close to the ideal or recommended levels. Such a computer program should do away with the need of hospitalization and the need of extensive and lengthy outpatient programs. G. EXCESSIVE NUMBER OF INJECTIONS

Because of the benefits of an intensive insulin therapy, the ADA and medical authorities now recommend an intensive insulin therapy. Thus, a well known author recommends:¹³

With flexible insulin therapy, two injections of a long-acting insulin are used to minimize the rise and fall of the blood sugar when you are not eating. Then you take injections of Regular before each meal to cover carbohydrates in the food and, if needed, some extra Regular to reduce high blood sugars. Although this regimen may require three to five injections a day [and three to five blood-glucose samples], each injection now has a specific purpose. In turn, this allows each insulin that is being injected to be individually tested and adjusted.

The health advantages of an intensive therapy have been amply demonstrated.¹⁴ However, one may assume that three to five injections per day and three to five finger pricks to obtain blood samples per day a more than a mere inconvenience; but rather have a significant impact in the patient's quality of life. Furthermore, the prospects of such procedures may discourage a substantial number of individuals from adopting an intensive therapy. H. WHAT IS NEEDED TO CORRECT THE PROBLEMS OF "EXCESSIVE NUMBER OF INJECTIONS"

What is needed is a method that achieves optimum benefits with the least number of injections. Also, a method that is adaptable to the individual's life style. While some individuals may be willing to undergo tliree to five injections per day, others may draw the line at three injections per day; others may be willing to submit to two injections per day during week days and three per day on weekends; etc. An adequate method would be one that accommodates the physician's request with the patient's ability and willingness to live with such request, and still offers the optimum therapy under the circumstances. I. CHANGES IN THERAPY AS FUNCTION OF TIME, E.G., A CHILD'S GROWTH, AGING OF AN INDIVIDUAL OR DURING PREGNANCY (ADD A SIMILAR SECTION FOR CLINICAL MANAGEMENT).

Often, a drug therapy varies with time. For example, in the administration of a hormone, during the initiation of therapy the body itself may be generating some of the required hormone

(endogenous hormone), so the need for exogenous hormone may be small. However, with time, such endogenous secretions may decrease, thus the patient may need additional exogenous amounts of the drug.

Another factor that may cause changes in the amount of the drug may be the growth of the individual (e.g., a child) or weight gain or loss of the individual; or factors such a pregnancy. Walsh, in Stop the Roller coaster, gives us an idea on how complex the changes in the drug requirements may be during pregnancy:

"Insulin requirements rise steadily over the course of the pregnancy, and usually double during the pregnancy. This rise is caused by several factors - weight gain, increased caloric in- take, creation of new tissue, and an increase in hormones that conflict with the actions of insulin . . . The one exception to this general rise in the need for insulin occurs during the last four weeks of pregnancy. Then, insulin needs may drop! During the last month, the fetus starts drawing out more glucose from the mother's blood for its needs. . . However, if your need for insulin suddenly droops, not caused by these obvious reasons, this can be an ominous sign. Contact vour obstetrician for consideration of immediate delivery. "

Another reference proposes that "doses show progressive increase during pregnancy, increasing in units per kilogram. Because the patient's weight also increases, the total dose may even triple."

J. WHAT IS NEEDED TO CORRECT THE PROBLEMS OF "CHANGES IN THERAPY AS FUNCTION OF IME"

What is needed is a method that can gradually make decisions based on measurements of one or more parameters and continuously change the dosage to accommodate the patient's changing requirements. Such a method should be able to detect gradual increase and or decrease requirements and make recommendations accordingly. Also, such a method should detect ab- normal changes and issue warnings (such as recommendations to see a doctor).

K. LACK OF FORESEEABILITY IN THE ANTICIPATED BEHAVIOR OF A DRUG OR COMBINATION

OF DRUGS.

Often, in cases of emergency such as admission to a hospital because of a serious condition, the medical community has insufficient information about the behavior of a drug or combination of drags, and insufficient tools to predict how a given dosage or dosages will behave, or how to correct certain dosages based on information of how the patient is responding to the drug therapy (by drug therapy, I mean the response to one or more drugs and the response to one or more dosages of the same).

For example, Dr. Mark Sperling et al recommends ¹⁵ managing a diabetic ketoacidosis ¹⁶ as follows:

Fluid repair should extend over 36 hours, rather than 24 to achieve a slower correction of serum hyperosmolarity (to avoid rapid shift of water from extracellular to intracellular compartment - implicated in the development of cerebral edema. If severe hyperosmolarity is present (> 340 mOstn), correction should occur over 48 hours.

. . . Initial hydration should be normal saline and the glucose concentration should be maintained at 200-300 mg/dl for the initial 36-48 hours. The initial rate of fall of glucose con- centration should, if possible, be limited to 100-150 mg/dl. Simple hydration, however, frequently causes 180-240 mg/dl drops in glucose.

. . . Potassium replacement should begin slowly, after 1 h intravenous treatment, if urine flow is established. Rate of potassium replacement should not exceed 6 meq/kg/day, unless plasma concentration remains abnormally low (<3.5 meq/l). Some potassium can be given as potassium phosphate and/or acetate, thereby limiting the inevitable excess provision of chloride. Usual potassium concentration is 30-40 meq/l in replacement fluids, but careful attention to this variable is necessary because some children will require up to 80 meq/l.

Bicarbonate therapy should riot generally be used. If the blood pH is <7.2, bicarbonate (HCOf ) or lactate may be used. Careful clinical evaluation is required. Bicarbonate, after combining with H+, dissolves to C02 + H20. Whereas the HCOf^~ diffuses poorly across the blood-brain barrier, CO2 diffuses freely into cerebrospinal fluid. Hence, inappropriate use of HCOf may worsen cerebral acidosis while serum acidosis improves [thus giving a false sense of improvement and exacerbating cerebral acidosis]. Also, the use ofHCOf may overcorrect acidosis, and the resultant alkalosis favors intracellular movements of potassium , thereby predisposing further to the development of hypolcalemia.

HCOf should not be given as a bolus because it may precipitated cardiac arrhythmia.

IfHCOf is used, it should be given at a rate of 40-80 mmol/l over at least 2 hours. Insulin is essential to fully correct the metabolic derangements ofDKA, although fluid treatment alone has some corrective effects. After an initial bolus dose of 0.15 U/lzg, a continuous low-dose insulin infusion ofO.1 U/lcg/h is the recommended method of insulin delivery.

Before analyzing the author's concerns about this approach, one should compare the above with the insulin treatment recommendations for D A in adults:¹⁷ Whenever possible, short-acting insulin should be administered as a continuous intravenous infusion in a starting dose of 10U or 0.1 U/kg/h via pump. If this is not feasible, 10 U/h short-acting insulin i.m. or s.c. [intramuscular or subcutaneous] is a satisfactory substitute. An initial bolus of 10 U i.v., although not essential, guarantees an immediate therapeutic level of insulin while the rest of the treatment is being prepared. Lower doses than these should not be used at the outset, and higher initial does offer no advantage.

In response to insulin, plasma glucose should fall at an average rate of 75 mg/dl.h. If no response has occurred by 4 h, an unusual degree of insulin resistance may be present. The dose of short-acting insulin should then be raised to 20-100 U/h i.v., with the higher doses being given for more extreme hyperglycemia and sicker patients. The dose should be further doubled every 2 h until plasma glucose definitely declines.

There are several limitations resulting from the above recommendations:

As given above, the information appears to be insufficient to carefully guide the physi- cian (specially one who is not an expert in the required field) about the proper procedure to bring the subject patient to a recovery.

For a DKA in children, the above procedure does not help the physician establish or regulate the following parameters or conditions: (1) over how many hours should fluid repair extend to achieve the desired rate of correction of serum hyperosmolarity, (2) how to administer the hydration (at what rate) and how to maintain the levels of glucose concentration at the desired 200-300 mg/dl for the initial 36-48 hours; (3) how to predict the initial rate of fall of glucose concentration and how to limit said fall to 100-150 mg/dl. Simple hydration, however, frequently causes 180-240 mg/dl drops in glucose; (4) predict when potassium replacement should start, rates of such replacement and when to discontinue the administration of potassium; (5) predict if and when bicarbonate therapy should be used and, given the serious danger of cerebral edema, particularly in children, anticipate when to discontinue the bicarbonate therapy.

For the treatment of adult DKA, the reference recommends "short-acting insulin should be administered as a continuous intravenous infusion in a starting dose of 10U or 0.1 U/kg/h via pump. If this is not feasible, 10 U/h short-acting insulin i.m. or s.c." However, it is not reason- able to use the same dosages for patients who exhibit low insulin resistance (e.g., those who manage their illness with 10 U per day) and those patients who exhibit high insulin resistance (e.g., those who manage their illness with 100 U per day).

The reference recommends "an initial bolus of 10 U i.v., although not essential, guarantees an immediate therapeutic level of insulin while the rest of the treatment is being prepared. Lower doses than these should not be used at the outset, and higher initial does offer no advantage." It is not reasonable that the same bolus would be used with 90 lb. patients and with 300 lb. patients. How should this vary according to the patient's level of insulin resistance?

The above reference states "In response to insulin, plasma glucose should fall at an average rate of 75 mg/dl.h. If no response has occurred by 4 h, an unusual degree of insulin resistance may be present." It is probably not reasonable to weight four hours to see if response begins to occur. Corrective action on such a critically ill patient should have been taking place long before expiration of the four hours.

The above reference states "the dose of short-acting insulin should then be raised to 20- 100 U/h i.v., with the higher doses being given for more extreme hyperglycemia and sicker patients. The dose should be further doubled every 2 h until plasma glucose definitely declines." Theoretically, several hours have elapsed by the time this takes place. By this time, ideally, the physician should have had a mechanism to allow him or her to determine the proper dosage to achieve a gradual, yet meaningful, decrease in glucose. In summary, the above-mentioned procedure is prone to errors, it does not adequately guide or remind the physician on how to achieve the desired results; and the physician must wait many hours to see if the therapy is working.

L. WHAT IS NEEDED TO CORRECT THE PROBLEMS OF "LACK OF FORESEEABILITY IN THE ANTICIPATED BEHAVIOR OF A DRUG OR COMBINATION OF DRUGS." What is needed is a method that can gradually make recommendations based on measurements of one or more parameters and continuously change the dosage recommendations to accommodate the patient's changing requirements.

Based on information about the patient (including prior diabetes history), the method should recommend initial bolus insulin and recommend i.v. rates. The method should be able to predict the anticipated behavior of such insulin, determine how well is working and adjust its recommendations. Also, the method should be able to establish the desired rates of hydration, desired rates of glucose correction, anticipate the need for potassium (or other electrolyte) replacement and rates of such replacement, recommend when bicarbonate replacement is recommended and when it should be discontinued. Also, such a method should detect abnormal changes and issue warnings to the medical staff.

In short, such a method should be an interactive and dynamic assistant to the medical staff, to free the staff of mental clutter and to give the staff visibility on how the patient is responding and the anticipated event that will take place later; e.g., based on pH levels, recommend whether bicarbonate treatment should be initiated and, based on the changes in pH level warn the physician, with anticipation, when such bicarbonate treatment should be discontinued. SUMMARY OF THE INVENTION

The present invention is a method, including algorithms and apparatus, that overcome the problems discussed above, with respect to the prior art. In practice, the various problems discussed below don't always arise for a given patient; thus the method and this invention contains algorithm that may be used independent of each other or in a combination with one or more of the other algorithms.

However, the proposed scientific approach to the treatment of a given medical condition or, more specifically, the pharmacokinetic management (time-dependent management of pharmaceuticals) of a medical condition, may require the entire family of algorithms disclosed in this invention to be available to the physician, or to the individual, in the case of conditions managed directly by the individual, such as diabetes. In practical use, the entire family of algorithms will probably form part of a computer program, which program may provide feedback directly to the health care professional, or the patient, or feedback to other machines or equipment for the treatment of the medical condition under consideration. These algorithms provide a method to determine an optimum dosage of a drag for a person or live being, such as a pet. The algorithms are based on sophisticated mathematics, rather than a generalized quantitative/qualitative approach, and use the actual time dependent absoφ- tion and other characteristics of the drag. The actual formula used to calculate the time of activity and serum strength profile for the drug is a mathematical model of the information available in the literature. These formulae may be updated as more information and data is obtained on each drug, or as new drugs are developed, thus continuing to improve the accuracy with which the present invention is used to treat a given medical condition.

The optimum dosage proposed herein is not a static (i.e., it is not a non-time varying) dosage. It would not constitute optimum use of the invention described here to calculate a static dosage for use day-in and day-out by a patient. Nor would it constitute an optimum application of this invention to derive a fixed algorithm to be used by a physician for all circumstances.

An optimum application of this invention would be to use the method disclosed herein interactively (as programmed in a computer), on a case-by-case basis to determine an optimum treatment for a specific individual. Said optimum treatment could be that treatment necessary when an individual in the hospital, or that necessary for the individual in the management of his or her illness or medical condition to accommodate his or her day-to-day activities.

The Author does not mean to limit the words optimum treatment (hereinafter "Optimum Treatment") to the clinically optimum treatment for an individual, without considerations of the individual's personal wishes or personal circumstances. For example, a physician may advise his patient that he should take tliree insulin shots a day. But, the individual may, in turn, decide that he or she is only willing to take two shots per day. Under such circumstances, the method disclosed herein may be used to devise an Optimum Treatment to satisfy the individual's requirement. The rationale being, that if the individual does not want to fully accept the physi- cian's recommendations, then it would still be desirable to devise a treatment that tries to achieve recommended goals subject to the constraints imposed by the individual.

As additional examples, an individual may not want to follow the physician- recommended diet, or the individual may indulge in too-much alcohol consumption, or otherwise an improper diet. Changing the individual's habits may be too difficult. So, the method of this invention, takes the individual as he or she is, and endeavors to find an optimum treatment for that individual, despite the individual's less-than-ideal habits.

Below, the inventor discloses the invention, via examples when necessary, on how the algorithms and method work. Though the examples are based on the approach for managing diabetes, it must be understood that on an application-by-application basis, the approach de- scribed herein may be adjusted to accommodate other needs, including use in the clinical setting, or emergency room setting, for a variety of drugs where the administration of the drug, time variations of the same and response of the individual to the drugs become critical.

LIST OF FIGURES AND TABLES FIGURE 1. Time of activity of human insulins, published by the Eli Lilly and Company. FIGURE 2. Time of activity and serum strength profile for lispro and regular. FIGURE 3. Schematic representation of idealized insulin effect.

FIGURE 4. Mathematical representation of the time of activity and serum strength profile for several types of insulin. FIGURE 5. Hypothetical glucose distribution of a diabetic individual. FIGURE 6. Glycemic distribution based on the prior art's regimen of four injections/day. FIGURE 7. Glycemic distribution based on the prior art's regimen of two injections/day. FIGURE 8. Caloric requirements for male and female individuals. FIGURE 8. Glycemic profile for a diabetic individual. FIGURE 9. Effect if the insulin on the glycemia of the above-described individual. FIGURE 10. Glycemic profile for a diabetic individual based on an insulin regimen calculated using the method of this invention.

FIGURE 11. Glycemic profile for a 19 year old male, before undergoing treatment based on the claimed invention. FIGURE 12. Glycemic profile for a 19 year old male, after initiation and conclusion of treatment based on the claimed invention.

TABLE 1A - 19 year-old male patient. Insulin dosages before and after treatment using the Method of this invention. TABLE IB - 63 year-old female patient, diabetic for over 20 years.

DETAILED DESCRIPTION OF THE INVENTION

IE. PROPOSED ALGORITHMS - EXPLANATION AND ALGORITHM DISCUSSION

The explanation and algorithm discussion offered below show how the proposed algorithms and method of their usage overcome the problems and limitations in the prior art. Much of the discussions in this document have revolved on the issue of insulin and diabetics. However, the methods, algorithms and apparatus disclosed in this invention are applicable to the administration of other drags where, as is generally the case, the drug's behavior is time dependent and/or individual dependent. This invention also applies to those cases where the individual must take two or more drugs and where one or more parameters must controlled in the individ- ual

EXAMPLES A. GENERAL EXPLANATION, EXAMPLE 1, BASED ON APROFTLΓNG OF AN INDIVIDUAL.

Assume our subject individual has diabetes, and his/or her (hereinafter, "his") blood glucose levels deviate significantly from the desirable levels. One would start by preparing a profile of the individual's glucose distribution as a function of time. For the profiling, the individual would be asked to follow his "typical" day to day routine, following his typical meal consumption and other regular habits. Thus, to obtain a reliable profile, it would not be recommended that the individual observes a clinically desirable pattern or routine, unless such pattern is already a part of the individual's daily routine. An ideal way to obtain a profile would be to measure the glucose level as a continuous function of time, or at periodic, yet frequent time intervals. Though ideal, this may not be practical with today's technology (though equipment is now available that offers near-continuous glucose monitoring). However, one could obtain an adequate time-dependent profiling by measuring the glucose at certain critical intervals. For example, measuring the blood glucose at morning fast (just before brealcfast) and before each meal ("pre-prandial' measurements). Also, measuring the glucose at the one-hour and two-hour interval after each meal ("post-prandial" measurements), at bedtime and one measurement in the middle of the night (2:00 AM). The above-suggested profiling may require a blood sample (by a prick of the finger) for each measurement, at least until such time as non-invasive methods are developed. This number of blood-glucose measurements is not a pleasant perspective for most individuals. So, one basic premise in the formulation of the Method, has been to make do with whatever information is available, even if it is not the optimum amount of information. Realizing; however, that the Method's results will not be as accurate or as optimum as they could otherwise be. Note that the glucose readings are based on the individual's then-current routine; which includes the individual's routine insulin injections. Also, as part of the profile, it will be necessary to know the time of injection, and the type and amounts of insulin(s) injected.

Having obtained the above glucose readings, the Method generates a graph or a table of values as a function of time, each value generated via inteφolation/extrapolation. The inteφola- tion can be linear, or use more advanced methods, such as quadratic, polynomial, exponential, logarithmic, etc.

Thus far, one has obtained the individual's actual glucose distribution for the day when the profile was taken. The difference between the actual glucose_'distribution and the physiologically ideal distribution (or the goal of the therapy), provides the level of correction that is necessary. Thus, it will probably be necessary to adjust the insulin dosage by increasing and/or decreasing the amounts of insulin and the types of insulin to be administered with each injection to change the individual's glucose distribution to approximate the ideal distribution. These same concepts apply to insulin administered by insulin pump, inhaled, or as later may be developed, by a means of close-loop feedback system. So far, we are relying on glucose measurements to determine amount of insulin required.

So, one basic precept of the Method, is that one does not need to rely on measurements of the actual drug or serum that will be administered. Instead, the Method can calculate the amount of insulin required relying on measurements of the individual's glucose, hi other words, the Method can use direct serum measurements, or indirect measurements, such as in the case of measuring glucose, to determine the necessary insulin. The indirect measurements need not be measurement of another drug, or chemical. Such measurements need only be a measurement of a parameter, chemical or physical, which can somehow be correlated to the drug. Thus, if a given drug is used to treat arrhythmia, and there was a direct correlation between the heart rhythm and the serum concentration of the drug, then the Method of this invention can be used to adjust the required dosages of the drug, to treat the patient's arrhythmia and maintain said patient's heart rhythm within clinically acceptable levels.

The following discussion explains how the Method uses "artificial intelligence" to start planning an Optimum Treatment based on the individual's own needs.

For the subject individual, one has already learned how many units of exogenous insulin this individual consumes per day. If the patient is undergoing his or her initial treatment, and has not been taking exogenous insulin, then the Method starts with a conservatively low amount of insulin, and proceeds to determine the Optimum Treatment, following the procedure discussed herein. The medical literature states that for a given individual, one unit of insulin, such as regular, counteracts the same number of carbohydrates as one unit of another type of insulin, such as lente. The difference between the two insulins is in the time-distribution; that is, the amplitude vs. time with which each type of insulin is absorbed by the body and counteracts the carbohy- drates. Note, however, that from one individual to another the amount of glucose counteracted by one unit of insulin varies (interindividual variations). To a lesser extent, for the same individual, the effectiveness of insulin (that is, the efficiency with which the individual utilizes the insulin) varies from day-to-day (intraindividual variations). Other things may affect the actual or apparent effective utilization of insulin. For example, the individual may secrete more endoge- nous insulin in the morning, therefore, in the morning this individual may need less exogenous insulin. This may appear to be an indication that the morning insulin is utilized more effectively that the evening insulin. In reality, for the method disclosed herein, and for the puφose of managing an individual's condition, the important thing is to determine the actual drug requirements, without placing undue emphasis on why, at that particular time, the individual's needs are dif- ferent than at other times.

Knowing, for example, how much the individual weights, the gender, age, level of activity and other relevant factors, one may estimate how many calories an individual consumes during his typical day and how many such calories may be attributed to carbohydrates, protein and fats. Based on the individual's weight and other characteristics, the medical literature tells us how one can estimate the rise in plasma glucose per gram of carbohydrate. For this puφose, the undersigned has developed an equation to calculate the rise in glucose per gram of carbohydrates. See FIGURE 7.

Based on the rise in glucose per gram of carbohydrate in terms of mg/dl, or mmol/1, or similar units of measure, and the then-current amount of insulin consumed by the individual, the Method can estimate, for this individual, how many grams of carbohydrates are counteracted by each unit of insulin. The prior art recommends to calculate the amount of insulin required based on the weight of the patient. This does not make much sense, since the patient's weight does not offer an indication of the amount of endogenous secreted insulin, nor an indication of the efficiency with which the individual utilizes the insulin. The method disclosed in this application presents a more reliable way to calculate the amount of insulin required.

For example, say that based on a 125 lb individual's physical characteristics and lifestyle, the method calculates that the individual requires 2300 calories per day and the individual consumes approximately 400 grams of carbohydrates per day. Assume further that this individual manages his daily glycemia with 20 units of insulin. This means that, for this individual, each unit of insulin counteracts 20 grams of carbohydrates. On the other hand, if this individual required 50 units of insulin per day, one could estimate that each unit of insulin counteracts 8 grams of carbohydrates. Even at this simple level, the Method can now more reliably estimate the individual's insulin needs to maintain the glycemia throughout the day. Next, one looks at the glucose distribution as a function of time. At this point, the distribution is based on the previous day's profiling. Thus, at this point, this will not be a real-time correction of glucose, but rather, an optimization of the typical day's requirements. Knowing the time dependent characteristics of each type of insulin, theoretically one can calculate the time of injection, amount and type of insulin that is required to adjust an individual's glucose distribu- tion to achieve a recommended distribution. How this Optimum Treatment is devised is a key aspect of the Method.

Below, the inventor discusses how to model the time-amplitude characteristics of insulin, or other drugs. The mathematical characteristics may be obtained and/or approximated by several means, including using published information or making actual measurements. The measurements could be based on direct measurements, i.e., measuring the time-amplitude of the drug within the individual's body, or indirect measurement, that is, measuring other characteristics or substances that can be correlated to the time-amplitude distribution of the drug. Giving someone insulin and measuring the effect of the insulin on the individual's glucose level as a function of time would be an example of indirect measurement. Measuring the actual concentra- tion of insulin in the individual's blood stream would be an example of direct measurement.

For now, assume that one has the time amplitude distribution of insulin (most pharmaceutical companies publish this information, or publish sufficient date that one can use to obtain this information). One must keep in mind that such time-amplitude distribution may vary inter- individually and intra-individually. For a given individual, it may vary from time to time, based on numerous factors, such as site of injection, state of the individual's health, current concentration of insulin, alcohol or interaction with other drugs, etc. One must also keep in mind that the objective here is to make intelligent and calculated decisions to derive the Optimum Treatment. Such decisions cannot take into consideration all factors, as one does not know what they are, and even if one knew them, it may not be practical to do so. But, the Method offers a means for taking into consideration the important factors necessary for deriving the Optimum Treatment. The Method allows the individual, in conjunction with his physician, to determine the level and extent to which additional factors are taken into consideration in achieving the Optimum Treatment. To devise the Optimum Treatment, one can look at the glucose distribution as a function of time, and mathematically calculate how much insulin is required, as a function of time, to counteract the glucose and bring it to the euglycemic levels. One could plot the insulin distribution and see what is required. However, no commercially available insulin exhibits the complex distribution that would be required by the metabolism of the individuals.

An individual with adequate mathematical skills may be able to take the glucose distribution and the time-dependent distribution for each type of insulin, and calculate (or by trial and error estimate) how much insulin is required to "tailor" the insulin needs of the diabetic individual. One drawback of such approach is that such tailoring would probably require direct interaction by the mathematically skilled individual, and may require trial and error. The ideal situation would be to devise a method whereby a computer or other apparatus can be used to automatically determine the ideal insulin levels. A method that can then learn enough about the individual so as to change the insulin requirements on a real-time basis, to accommodate to the individual's requirements, be it because of changes in the meal patters, exercise patters, sick days, growth, aging, pregnancy, etc. The proposed Method does just that.

The above steps will now be illustrated graphically. FIGURE 8, shows a fictitious glycemic profile for a diabetic individual. Based on the insulin distributions shown in figure 4, this individual will be treated with three insulin injections per day, using a combination of Lente, regular and lispro. FIGURE 9 shows the mathematical supeφosition of the effect of the three injections, and FIGURE 10 shows the theoretical glycemia of this individual, based on three insulin injections per day. The theoretical results are significantly better that those obtained based on four injections per day according to the recommendations of current authorities (as shown in FIGURES 5 and 6.

The above discussion is based on theoretical individuals and theoretical glycemia. FIG- URES 11 and 12 illustrate the before and after glycemia of a real patient (19 year old male, diabetic for one year), after undergoing treatment based on the invention herein. The method presented herein achieved near euglycemic levels on this individual, using only two injections per day, the same number of injections as the individual was taking before undergoing treatment under the method presented in this invention. With only two injections per day, this patient achieved much better glycemia that achieved by the intensive control group in the DCCT (discussed above), which group was taking from three to five injections of insulin per day.

TABLE 1A illustrates the insulin regimen derived for this individual, using the method described herein. Note that the total amount of insulin consumed before and after the regimen are substantially similar, within 20 %. Yet, the average glycemia for this individual dropped from a mean of 230 mg/dl (102 % higher than normal glycemia) and a standard deviation of 72 mg/dl under the conventional regimen, to an average of 118 mg/dl (7 % higher than normal) and a standard deviation of 18.5 mg/dl. TABLE IB illustrates the insulin regimen derived for another patient, a this individual 63 year-old female patient, diabetic for over 20 years. With said regimen the patient was able to decrease her glycemia from a mean of 340 mg/dl (309 % higher than normal glycemia) and a standard deviation of 62 mg dl under the conventional regimen, to an average of 128 mg/dl (16 % higher than normal) and a standard deviation of 20.0 mg/dl.

Having derived the Optimum Treatment (by the algorithms discussed in this application), the Method becomes smarter by learning more about the individual. The Method can learn to know its master by a variety of methods, often without the individual knowing that it is teaching, or giving the Method additional information to help the Method become a better manager of the individual diabetic's glucose. A. INSULIN, SYRINGES AND DISPENSING APPARATUS.

The insulin regimens shown in TABLE 1 use broad steps between glycemic ranges be- fore changing the recommended insulin shots. This was done out of necessity for the following reasons: first, the patient had to use a table, rather than direct feedback from a computer. Therefore, refining the steps would have made the table much longer and more complex to use. Second, the prior art insulin formulations and syringes are not adequate for a more refined analysis and treatment. So, while in the theoretical example illustrated in FIGURE 9, the insulin is used in increments of 0.1 unit, the recommended dosages in TABLE 1 use increments of 1 unit. The reason is that in the popular present formulations of insulin lcc of insulin corresponds to 100 units of insulin. Thus, 1 unit corresponds to 0.01 cc. So it is very difficult to measure 1 unit, even with the specialized syringes used by diabetics (generally corresponding to 0.3, 0.5 and 1.0 cc). As such, using the prior art insulin and syringes, it is impossible to measure 1 cc.

FIGURE 3 illustrates the mathematically calculated serum strength achieved by several popular insulin formulations. As shown in the figure, lispro (Humalog) achieves a serum strength that is approximately five times higher than the serum strength achieved by NPH and approximately seven times higher than the serum strength achieved by Lente. Thus, an error of 1 unit of lispro is seven times more dangerous than an error of one unit of Lente.

The inventor herein proposes to dilute the insulins such that it is easier to measure the correct amount and so that a small mistake in the measurement does not represent an undue danger to the patient. One way to dilute the insulins, is to dilute and/or adjust them all, so they all achieve the same serum strength.

Perhaps, a more practical way to dilute the insulins is to dilute the fast acting insulins, lispro and regular for the insulins illustrated. For example, if lispro was diluted to 1/3 strength, and regular diluted to 14 strength, then the recommended dosages would call for tliree times and two times the amounts, respectively, of lispro and regular. Such amounts would be easier to measure, and an error in the measurement would be more forgiving and more obvious to detect. As to the way to dilute them, there are several possible ways. Certainly it would be simple enough to dilute the insulins with distilled water; e.g., 2 parts distilled water to one part lispro, and one part distilled water and one part regular respectively.

Distilled water, however, may not be the preferred approach, as injecting distilled water into the body is painful, partly because distilled water is not a physiologically balanced solution readily acceptable or assimilated by the body. A preferred way to dilute the insulins would be to use an isotonic-pH balanced solution. Such solutions are well know in the prior art. Also, it would be preferable to have a dispensing apparatus which is capable or measuring correct amounts of various insulin and dispense them into a container for subsequent injection or use by the patient.

Having derived the Optimum Treatment (by the algorithms discussed in this application), the Method becomes smarter by learning more about the individual. The Method can learn to know its master by a variety of methods, often without the individual knowing that it is teaching, or giving the Method additional information to help the Method become a better manager of the individual diabetic's glucose.

B. ALGORITHM - MODELING A GENERAL WAVEFORM FROM A SUPERPOSITION AND COMBINATION OF ONE OR MORE DISCRETE AVEFORMS. 1. Find one or more reference parameters in a waveform; i.e., one or more of the following characteristics based on the time of occurrence and amplitude, or otherwise the corresponding characteristics according to the domain and range used for the particular waveform: zero crossings, half-life onset, half-life offset, maxima, minima, area under the curve; using one or more of the following methods: (i) By displacing the original waveform (displaced waveform); that is, take the original waveform and add or subtract the mean from the original waveform from its original position in a graphical plot (or its equivalent graphical position from a table of values) to a more convenient position in the graph. (ii) Same as (i) above, except that instead of the mean, one displaces the waveform based on the median, or the maxima, or the minima, or a combination, ratio or proportion of such parameters, or otherwise displacing the original waveform by adding or subtracting second waveform. (iii) Same as (i) or (ii) above, except the mean, median, minima, maxima, or other function used to displace the original waveform is itself not a constant, but a gradual changing waveform. For example, one could subtract what may be called an instantaneous mean, which, in reality, is obtained from a regression analysis (e.g., linear, logarithmic, exponential or other form of regression). (iv) A combination of two or more of the methods outlined in (i) through (iii) .

C. ARTIFICIAL INTELLIGENCE ALGORITHM FOR THE ADMINISTRATION OF DRUGS OR MEDICATIONS OR OTHERWISE TO MANAGE THE BIOCHEMICAL PROCESS IN A IVING BEING.

Starting either, from a known approximation; i.e., a rational basis for knowing the starting point, or starting without a known approximation, finding the behavior of a given drug or drags within a person or living being. This artificial intelligence approach is based in observing the response of the living being to the drug, said response by open loop or closed loop measurements of the drag serum itself (direct measurements), or measurement of one or more parameters that are affected by the drug (indirect measurements). Based on such affected- parameters measurement; be it on the serum strength (magnitudes), rates of changes, and/or other characteristics, calculating, by means of inteφolation or extrapolation of actual measurements, the new or corrected required level of the drug. The above-mentioned calculations or predictions may be linear or non-linear inteφolation or extrapolation, or otherwise based on a function of the measured parameters. Repeating the process until an ideal or otherwise desirable level is reached. The above-mentioned step of "repeating the process" may sound as trial and error approach. But it is not. It may be considered an iterative approach, but it is one that quickly converges. Above, the applicant discussed the example of a 125 lb. individual who consumes approximately 2300 calories per day which included 400 grams of carbohydrates per day. If the individual, upon initiation of treatment, is consuming 20 units of insulin per day, the method (in its simplified form) estimates that each unit of insulin counteracts 20 grams of carbohydrates. The method then utilizes this information to calculate the amount and timing of various insulins required to manage this individual's glycemia. Based on the algorithms discussed herein, the method calculates a new insulin regime to manage the individual's glycemia. If the new regimen increases the individual's daily insulin to 22 units per day, the new method will now calculate that each unit of insulin counteracts 400÷22=18.2 grams of carbohydrates. Substantially a similar amount as the first try. The substantial difference between the prior art and this invention, is that the method disclosed herein calculates the amounts and timing of the injections to tailor a regimen for the individual. D. ALGORITHM TO DERIVE AN EQUATION THAT FITS OR OTHERWISE MODELS AN EMPIRICAL WAVEFORM.

^■ Once derived, the equation is useful to be used as an equation, or to perform additional mathematical analyses base on the equation, for example, using supeφosition to add or subtract functions, multiplying or even performing calculus operations (derivatives, integrals, etc.). In its final use, it is not necessary to use the mathematical form of the equation. Often, for computerized analyses is more convenient to use look-up tables. Nevertheless, the equation may be convenient for generating such look-up tables. E. ALGORITHM FOR NORMALIZATION OF A FUNCTION.

For use whenever two or more similar functions (e.g. two or more chemical substances) will be used by a method of supeφosition. Note, normalization may be necessary or useful when dealing with different functions related to a similar substance. For example, normalization of regular en lente insulin, such that one can later use supeφosition to add the effects of such functions. Normalization may also be used with respect to one (or more) substances that are used in a different manner. For example, regular insulin has different absoφtion rates when injected in- travenously, subcutaneously or in the muscle. Also, the distribution or absoφtion characteristics may be different in a lean individual than an obese individual; or differences due to different sites of injection, for example, in the abdomen or the thigh.

ARTIFICIAL INTELLIGENCE - ALGORITHMS

1A ARTIFICIAL INTELLIGENCE - DIRECT MEASUREMENT a. Obtain a reading of the concentration (first reading) of substance A in an individual (or otherwise a living being) (CA1). b. Administer a measured amount of A to the individual (MAI). c. Obtain a second reading of the concentration of substance A (CA2). d. Based on inteφolation or extrapolation, as applicable, between CA1 and CA2, calculate the amount of A (MA2) required to increase or decrease the concentration of A to a desired level.

1B ARTIFICIAL INTELLIGENCE - DIRECT MEASUREMENT a. Same as algorithm 1A, except that in step 1 A(d), one calculates the required amount of A based on a reference parameter of the individual; e.g., the individual's weight. For exam- pie, if the initial concentration was CAl, and the administration of CAl increased the concentration to CA2, then one can calculate that amount MAI increased the concentration to CA2-CA1. b. One can then calculate that based on the individual's weight (W) or other chosen refer- ence parameter, an amount MA1/W will change the concentration by: ((CA2-CA1)/CA1) x 100%.

Note 1. Algorithm IB used the individual' s weight to calculate the amount of a substance needed to change the concentration from CAl to CA2. So, based on such an approach, the amount needed to increase a level by a determined amount would be twice as much for a 200 lb. individual than for a 100 lb. individual.

Weight, however, is not the only logical parameter for such calculations. Depending on the situation, the proper baseline parameter may be different. For example, if one is trying to calculate a substance affected by perspiration (e.g., loss of water or electrolytes), one may be more concerned about how such loss is affected by the surface area of the individual's body. Note 2. The above algorithms also apply where the initial concentration is known, so that the initial measurement CAl need not be taken. For example, if the initial concentration is known to be zero or otherwise a constant. 2 ARTIFICIAL INTELLIGENCE - INDIRECT MEASUREMENT

Same as algorithm 1, except that rather than measuring the concentration of substance A (i.e., rather than obtaining a reading of CAl), one obtains a reading of (measures the) concentration of B (CB1), a substance than somehow is affected by the concentration of A, or can somehow be correlated to the concentration of A. Then, one administers a measured amount of A (MAI) and measures or otherwise obtains a reading of the new concentration of B (CB2). Then, following the steps discussed in algorithm 1, one calculates the new amount of A (MA2) that must be administered to achieve the desired level of A.

Note 3. The administration of one substance and the measurement of a second substance, or the indirect measurement approach can be conducted in several ways or for several puφoses; e.g.: a. Measure Bl, administer A, measure B2 -> administer A in sufficient amount (as calcu- lated by linear or non-linear inteφolation or extrapolation) to bring B to the desired level. b. Same as (a), except, in the last step, administer A sufficient to bring A to the desired level, even though A was only ascertained by measuring B. Note 4. The following aspects of the above-disclosed algorithms constitute the artificial intelligence aspect of the same: a. It is possible that the inter-individual variations are negligible such that all individuals require the same amount of a substance or drug. If such was the case, then algorithm la would apply to all individuals and the results could be used for the entire population.

However, this is not likely. b. An algorithm for drug requirement may be developed such that it would apply to the population at large, by accounting for one or more parameters of the individual. Thus, the concentration or amount required of a given drug could be stated as a function of the individual's weight, surface area, age, gender, height, etc. or as a non-linear mathematical function of such variables, e.g., square root of the weight, or a combined function of weight and surface area. Having thus calculated and adjusted the formulae to the various parameters that may affect the drug requirements, the final formula may apply to the population at large or may apply to a larger segment of the population. This is more likely to be the case than (a) above. c. It is possible that too many parameters may affect the amount of a drug required by a given individual. Thus, inter-individual variations may be significant. For example, weight, size, age, physical activity, diet, etc. Theoretically, one could develop an equation with multiple variables to handle such situations; but, at the end, such an equation may be very complex and impractical. d. Instead of a "one complex equation fits all" approach, one could use the algorithms presented above to determine the drug needs for a particular individual. Such an algorithm would, indirectly, account for numerous variables, including unknown variables, thus adjusting the recommendations for the particular needs of the individual. e. Similar situation as described in (d) above, except that other variables, less easy to ascertain, may affect the individual's needs. For example, endocrinological secretions, where the deficiency in one individual is different than the deficiency in another individual. So, by basing each algorithm (or each recommendation) on the specific individual, one indirectly accounts for such other factors, such as the individual's internal secretions. f There may be intra-individual variations (in drug requirements or drug behavior) that may be variable to parameters, such as time of day. This may occur for several reasons:

1. For example, hormone secretions may vary with time of day. Therefore, the amount of a drug required by an individual, or the drag's corresponding behavior, in the morning may be different than at night. The algorithms, as developed and presented above, would develop different values applicable to the drug behavior, which values would be based on time of day. 2. The intra-individual variations may be related to other time-dependent parameters, e.g., menstrual cycle. 3. The intra-individual changes in drag requirements may be related to other time- dependent (or situation-dependent) changes in the individual; for example, diabetic for a pregnant woman, her insulin needs may increase tremendously during the first eight months of her pregnancy (two to three fold); yet, precise dosages are required to maintain the health of the mother and to maintain the health and proper develop- ent of the zygote/embryo/fetus. During the ninth month of pregnancy, the pregnant woman's insulin needs may decrease very fast, since the fetus secretes its own insulin. Based on periodic measurements, the above algorithms continuously adjust the recommended dosages to fit the individual's changing requirements.

3 ARTIFICIAL INTELLIGENCE - DIRECT AND INDIRECT MEASUREMENT Same as algorithms 1 and 2, except that the calculations are based on time-dependent behavior (e.g., absoφtion rates) of the drug, and corresponding calculations based on rates of changes of as the drug is taking effect within the individual

4 ARTIFICIAL INTELLIGENCE - DΓRECT AND INDIRECT MEASUREMENT - MULTIPLE DRUG ADMINISTRATION Same as algorithms 3, except that it uses two or more drags or substances (e.g., different insulins), administered at specific times to the drug response to the individual's constant or variable requirements.

5 ARTIFICIAL INTELLIGENCE - DΠIECT AND INDIRECT MEASUREMENT - SINGLE OR MULTIPLE DRUG ADMINISTRATION - CONTINUOUS INFUSION OR ADMINISTRATION Same as above algorithms, except that the drug or drugs are administered continuously, such as by i.v. or insulin pump; wherein the rate of administration is tailored to meet the individual's requirements.

The invention has been described with reference to various specific and preferred embodiments and techniques. However, it should be understood that many variations and modifications may be made while remaining within the spirit and scope of the invention. All publications and patent applications in this specification are indicative of the level of ordinary skill in the art to which this invention pertains. In the claims, the word "subject" is used to refer to, but not limited to, a living being, such as a person or animal, who is under treatment or otherwise uses or consumes a drug, that is, any chemical compound that may be used on or administered to humans or animals as an aid in the diagnosis, treatment, or prevention of disease or other abnormal condition, for the relief of pain or suffering, or to control or improve any physiologic or pathologic condition.

REFERENCES

1. American Diabetes Association: Report of the Expert Committee on the Diagnosis and Classification of Diabetes Mellitus (Committee Report - Clinical Practice Recommendations). DiabetesCare 22, Vol 22, Suppl. 1. Hereinafter, "ADA - Diagnosis and Classification of Diabetes ".

2. Daniel W. Foster: Part 13: Endocrinology and Metabolism, Section 1: Endocrinology, Chapter 334: Diabetes Mellitus, by. Treatment - Daily Therapy.

3. Humalog ® Insulin lispro Injection (rDNA Origin), PA 9161 FSAMP; Eli Lilly and Company, 1996, 1999. 4. Walsh, et al: "Tips On Humalog ®, Diabetes Services, Inc. 1997.

¹ Pharmacokinetics: "The activity or fate of drugs in the body over a period of time, including the processes of absoφtion, distribution, localization in tissues, biotransformation and excretion." Dorland's Medical Dictionary, 28th Ed. In this document, the term Pharmacokinetics is also used to refer to drug or serum dosage regimens, that is, the prescriptions of medications and the management of a drug/seram/hormone regimen to treat a medical condition.

² "Goal-Oriented, Model-Based drug Regimens: Setting Individualized Goals for Each Patient. By Dr. Roger Jelliffe, USC School of Medicine. Dr. Jelliffe heads the Laboratory of Applied Pharmacokinetics (LAPK) of the University of Southern California School of Medicine. The LAPK is a resource for optimal study and control of pharmacokinetic systems and for individualized drug therapy. The LAPK comprises about 16 members, most of whom hold medical degrees, doctorates in pharmacy and/or Ph.D.s in related sciences. Established in 1973, the LAPK develops improved methods for the optimal study and control of pharmacokinetic systems, designed for optimal analysis of drug behavior and optimal care of patients through optimal management of their drag therapy.

³ Intensive Diabetes Management. 2^nd ed. (hereinafter Intensive Diabetes Management). Published by the American Diabetes Association (hereinafter, the ADA), Clinical Education Series (1998). Table 6.2, p. 94.

The ADA publishes the Clinical Education Series to provide "health care professionals with the comprehensive information needed to give the best possible medical care to patients with diabetes." Intensive Diabetes Management, p. xi.

Intensive Diabetes Management. Table 6.2, p. 94. ⁵ Intensive Diabetes Management. Table 6.3, p. 95.

⁶ Similar recommendations are found in Therapy for Diabetes Mellitus and Related Disorcers, 3^rd ed. (hereinafter, Therapy for Diabetes Mellitus). Published by the ADA, Clinical Education Series (1998). Chapter 23 (by Jay S. Skyler), Tables 23.3 and 23.4, pages 194, 199.

⁷ Tips On Humalog, by John Walsh, P.A., C.D.E. and Ruth Roberts, M.A Copyright © 1997 by Diabetes Services, Inc.

⁸ Intensive Diabetes Management. Gp. 90.

⁹ Intensive Diabetes Management. Figure 6.2.

¹⁰ Harrison's Principles of Internal Medicine, 14^th ed., 1998. Chapter 334: Diabetes Mellitus, by Daniel W. Foster; p. 2067.

¹¹ Therapy for Diabetes Mellitus, Ch. 4, Management of Pregnant Women with Diabetes, by E. A Reece and C. Homko; p. 30.

¹² Richard S. Beaser, Outsmarting Diabetes, 1994, p. 38.

¹³ John Walsh, Ruth Roberts and Lois Jovanovic "Stop the Rollercoaster" 1996, p. 16.

¹⁴ See for example: Diabetes Care, "Clinical Practice Recommendations 2000", American Diabetes Association, 2000. Position Statement S24, "Implications of the Diabetes Control and Complications Trial."

¹⁵ Therapy for Diabetes Mellitus, Ch. 7, Diabetic Ketoacidosis in Children, by Mark A. Sperling, Larry C. Deeb and Nancy M. Wright; p. 53.

¹⁶ Diabetic ketoacidosis (DKA), a serious condition of the body in which there is not enough insulin. Free fatty acids are released from fat cells and produce ketones in the liver. These ketones or acids result in an imbalance in the blood (acidosis). The condition leads to loss of appetite, abdominal pain, nausea and vomiting, rapid and deep respiration, and coma. Death may occur. In the case of a pregnant woman, DKA has been associated with a very high fetus mortality rate.

¹⁷ Therapy for Diabetes Mellitus, Ch. 11, Diabetic Ketoacidosis and Hyperosmolar Hy- perglycemic Nonketotic Syndrome in Adults, by Saul Genuth; p. 85.

¹⁸ Humulin® and Humalog® are registered trademarks of Eli Lilly and Company. LENTE® and ULTRALENTE® are registered trademarks of Novo Nordisk A S.

Claims

CLAIMSI claim

1. A method for regulating the administration of a drug to control the serum concentrations of the drug within a subject, the method comprising the steps of: monitoring the serum concentration of a drug in the subject, over a period of time, to obtain a serum strength curve, said serum strength curve being a representation of the level of the drug over said period of time; and adjusting the concentration of drugs delivered into the subject at particular instances of time, the level of the drug delivered at any one particular instance of time being a function of the particular instance of time and the level represented in the serum strength curve at that particular instance of time.

2. A method for regulating the administration of a drug to improve the serum strength levels of a drug within a subject, the method comprising the steps of: monitoring the serum strength level of a drug in a particular subject, over a period of time, to obtain a serum strength curve, said serum strength curve being a representation of the level of the drug over the period of time; receiving behavior data, the behavior data indicating an event associated with the subject that is likely to alter the serum strength curve during the period of time in which the event occurs; adjusting the concentration of drugs delivered into the subject at particular instances of time, the serum strength of the drug delivered at any one particular instance of time being a function of the particular instance of time, 1he level represented in the level curve at that particular instance of time and the behavior data.

3. A method for regulating the administration of one or more drugs to control the serum concentrations of the drug within a subject, the method comprising the steps of: monitoring the serum concentration of the drug in the subject living being, over a period of time, to obtain a serum strength curve as a function of time for the subject; and comparing the monitored serum strength of the drug in the subject with the physiologically ideal, medically desired or the goal for the serum strength as a function of time, calculating the amount of drugs the subject needs to achieve the desired serum strength, taking into consideration the time dependent behavior of the drug and taking into consideration the amount of the drug that the subject had received, if any, when the serum strength of the drug was monitored, and taking into consideration the supeφosition of the drug effects, from prior dosages and anticipated future dosages, adjusting the concentration of drugs delivered into the subject at particular instances of time, the level of the drug delivered at any one particular instance of time being a function of the particular instance of time and the level represented in the serum strength curve at that particular instance of time.

4. A method for regulating the administration of one or more drugs to control certain physiological or medical conditions in a subject, the method comprising the steps of: monitoring certain parameters in the subject over a period of time, which parameters are affected by the serum concentration of the drug, to obtain a measure of said parameters as a function of time for the subject; and comparing the monitored parameter with the physiologically ideal, medically desired or the goal for the monitored parameter as a function of time, calculating the amount of drugs the subject needs to achieve the desired monitored parameter, taking into consideration the time dependent behavior of the drug and taking into consideration the amount of the drug that the subject had received, if any, when the parameter was monitored, and taking into consideration the superposition of the drug effects, from prior dosages and anticipated future dosages, adjusting the concentration of drugs delivered into the subject at particular instances of time, the level of the drug delivered at any one particular instance of time being a function of the particular instance of time and the level represented in the serum strength curve at that particular instance of time.

5. A method of mathematically calculating and presenting the time-activity data for one or more drugs, the method comprising the steps of: normalizing the time-activity of the drug, such that the area under the curve for the time- activity corresponds to one unit of measure.

6. A method of calculating the effects of one or more drugs on a subject, the method comprising the steps of: using the normalized time-activity profile of the drug or drugs and calculating the cumulative effects of said drug by multiplying the total number of units of said drug time the normalized time-activity profile of the corresponding drug, superimposing the above calculated values to the residue of prior dosages, obtained by similarly calculated values.