Method for titrating drug

Abstract

A method of titrating an anti-diabetic drug to a patient and a system for performing the method are disclosed. The method includes providing a database having anonymous personal parameters for a plurality of previously managed subjects; and generating a queue from the plurality of previously managed subjects based on similarities in the anonymous personal parameters. The method further includes receiving a patient-specific personal parameter; and identifying a patient-similar queue corresponding to the patient-specific personal parameter from the generated queue. The titration output parameters to be optimized may be entered or calculated. Based on the titration output parameters to be optimized and the patient similar queue, a customized titration protocol for the patient may then be derived.

Description

[0001] Related applications

[0002] This application claims priority to PCT / US2023 / 70698, filed July 21, 2023, the entire contents of which are incorporated herein by reference. Technical Field

[0003] The teachings of this disclosure generally relate to a system and method for titrating pharmaceuticals. Specifically, this disclosure relates to a system and method for titrating antidiabetic drugs. Background Technology

[0004] Generally speaking, titration is the process of adjusting drug dosage to achieve optimal therapeutic benefit and minimal adverse reactions. Titration is particularly important for drugs with a narrow therapeutic index, because the difference between the therapeutic dose and the dose that could cause significant side effects is relatively small. Antidiabetic drugs (including insulin and biosimilar insulin) are among the medications that typically require titration to achieve adequate glycemic control without causing hypoglycemic events.

[0005] Existing antidiabetic drug titration systems largely rely on pre-set titration schedules, which may be based on drug manufacturer guidelines and / or the fundamental characteristics of the therapy (i.e., different levels of settings corresponding to titration zeitgeist), but these systems do not take into account the individual patient characteristics. Alternatively, titration systems can be manually customized by the healthcare provider before protocol initiation, allowing for customization of the starting dose, total dose, dose tier adjustments, adjustment timing, and target range before titration begins. Regardless of the chosen titration protocol, if failure occurs, the healthcare provider adjusts the parameter settings and restarts with another titration protocol.

[0006] A typical procedure (10) for titrating antidiabetic drugs is shown in Figure 1. The steps include:

[0007] a) Diagnostic need for antidiabetic medication (12).

[0008] b) Prescribe specific antidiabetic drugs (14).

[0009] c) Obtain the general titration instructions for the prescribed antidiabetic drugs (16).

[0010] d) Establish application schemes based on general instructions (18).

[0011] e) Teach the patient the administration method (20).

[0012] f) Follow the application protocol (22).

[0013] g) Assess patient outcomes (24).

[0014] h) If the titration is successful, continue with the administration protocol (28).

[0015] i) If titration is unsuccessful, the healthcare provider manually adjusts the administration protocol (30).

[0016] In practice, such universal titration procedures are suboptimal. This is because patients starting any particular antidiabetic drug (such as insulin) may respond differently to treatment due to many patient-specific factors that universal titration procedures fail to consider. Universal titration protocols are drug-specific, not patient-specific. Healthcare providers have few tools to determine a patient's likely response and adjust the chosen antidiabetic drug or universal titration parameters to better suit the specific patient. While healthcare providers may adjust certain parameters after a titration protocol has been in place for some time, this process increases the provider's cognitive burden, and in many cases, adjustments are made through merely educated guesswork.

[0017] Existing titration methods typically fail to produce truly patient-specific protocols, thus reducing the likelihood of successful and timely glycemic control. Delays due to inadequate titration protocols can lead to delays in optimizing treatments to meet patient needs, which in turn increases the risk of progression of chronic comorbidities and / or adverse side effects. Therefore, improved tools are needed to create personalized, reliable titration protocols. Summary of the Invention

[0018] A method for customizing titration protocols based on previously successful titration protocols in patients with similar demographics and health records is disclosed.

[0019] As used below, the terms “have,” “contain,” or “include,” or any of their arbitrary grammatical variations, are generally open-ended terms. Thus, these terms can refer either to a situation where no other features exist in the entity described in this context besides those introduced by these terms, or to a situation where one or more other features exist. For example, the statements “A has B,” “A contains B,” and “A includes B” can refer either to a situation where no other element exists in A besides B (i.e., where A consists solely and exclusively of B), or to a situation where one or more other elements (such as element C, element C and D, or even further elements) exist in entity A besides B.

[0020] Additionally, it should be noted that the terms "at least one," "one or more," or similar expressions (if any) indicating that a feature or element may exist once or more will generally be used only once when introducing the corresponding feature or element. In the following text, in most cases, when referring to a corresponding feature or element, the expressions "at least one" or "one or more" will not be repeated, even though the corresponding feature or element may indeed exist once or more. It should also be understood that, for the purposes of this disclosure and the appended claims, regardless of whether the phrase "one or more" or "at least one" appears before an element or feature appearing in this disclosure or claims, such element or feature should not be considered a single interpretation unless expressly stated herein. By way of non-limiting example, the terms "antidiabetic drug," "personal parameter," and "titration protocol parameter" (to name only) wherever they appear in this disclosure and claims should be interpreted as meaning "at least one" or "one or more," regardless of whether they are introduced using the expressions "at least one" or "one or more." All other terms used herein should be interpreted similarly unless expressly indicated as intended to be a single interpretation.

[0021] Furthermore, as used below, the terms “preferredly,” “more preferably,” “specifically,” “more specifically,” “more concretely,” “more specifically,” or similar terms are used in combination with optional features without limiting the possibility of alternatives. Therefore, features introduced by these terms are optional features and are not intended to limit the scope of the claims in any way. As those skilled in the art will recognize, the invention can be practiced by using alternative features. Similarly, features introduced by phrases such as “in embodiments of the invention” or similar expressions are intended to be optional features, without any limitation on alternative embodiments of the invention, without any limitation on the scope of the invention, and without any limitation on the possibility of combining features introduced in this manner with other optional or non-optional features of the invention.

[0022] The terms “patient” and “subject” are used interchangeably in this article. Both refer to a person with diabetes or a person in the prediabetic stage.

[0023] the term

[0024] As used herein, the term “titration” is a broad term and should be given its common and customary meaning to a person skilled in the art, and should not be limited to a particular or customary meaning. The term “titration” may specifically refer to, but is not limited to, a procedure, system, or method for adjusting the dosage and / or timing of a particular drug to achieve a therapeutic effect on a patient while minimizing adverse drug reactions.

[0025] As used herein, the term "antidiabetic drug" is a broad term and should be given its common and customary meaning to those skilled in the art, and should not be limited to a particular or customary meaning. The term "antidiabetic drug" may specifically refer to, but is not limited to, insulin as defined below, as well as, for example, amyloid injections, alpha-glucosidase inhibitors, biguanides, dopamine-2 agonists, dipeptidyl peptidase-4 (DPP-4) inhibitors, glucagon-like peptide-1 receptor agonists (GLP-1 receptor agonists), meglitinides, sodium-glucose transporter (SGLT) 2 inhibitors, sulfonylureas, tetrahydrothiazolidinediones, and other drugs with similar therapeutic effects.

[0026] As used herein, the term "insulin" is a broad term and should be given its common and customary meaning to those skilled in the art, and should not be limited to a particular or customary meaning. The term "insulin" may specifically refer to, but is not limited to, naturally occurring human or animal insulin, partially or fully biosynthesized insulin, such as biosimilar insulin, long-acting insulin, rapid-acting insulin, etc. Insulin can be delivered orally, by inhalation, or by injection.

[0027] As used herein, the term "personal parameter" is a broad term and should be given its common and customary meaning to those skilled in the art, and should not be limited to a specific or customary meaning. The term "personal parameter" may specifically refer to, but is not limited to, parameters describing the health, demographic, or other attributes of a particular person, subject, or patient.

[0028] As used herein, the term "health parameter" is a broad term and should be given its common and customary meaning to a person skilled in the art, and should not be limited to a specific or customary meaning. The term "health parameter" may specifically refer to, but is not limited to, HbA1c, comorbidities, age, height, weight, body mass index, other medications, hypoglycemia risk level, blood glucose level, vital signs, etc.

[0029] As used herein, the term "demographic parameter" is a broad term and should be given its common and customary meaning to those skilled in the art, and should not be limited to a specific or customary meaning. The term "demographic parameter" may specifically refer to, but is not limited to, race, age, sex, socioeconomic status, preferred mode of communication, etc.

[0030] As used herein, the term "digital twin" is a broad term and should be given its common and customary meaning to those skilled in the art, and should not be limited to a specific or customary meaning. The term "digital twin" may specifically refer to, but is not limited to, a digital representation of a patient used as a digital counterpart for purposes such as statistical analysis. For the purposes of this disclosure, the digital twin need not be identical. A digital twin can be created independently using continuous glucose monitoring data, activity data from, for example, smart devices, electronic medical records, etc. Alternatively, the twin can be selected from an existing set of common records. The common record that most closely resembles the patient's data can be selected and used as the digital twin.

[0031] As used herein, the term "cohort" is a broad term and should be given its common and customary meaning to those skilled in the art, and should not be limited to a particular or customary meaning. The term "cohort" can specifically refer to, but is not limited to, a group of people who share one or more common characteristics of interest (also referred to herein as "commonality" or "similarity"). More specifically, a "cohort" can refer to a group of patients who share one or more personal parameters, such as health and / or demographic parameters.

[0032] As used herein, the term "patient-similar cohort" is a broad term and should be given its common and customary meaning to those skilled in the art, and should not be limited to a specific or customary meaning. The term "patient-similar cohort" may specifically refer to, but is not limited to, a cohort that shares common characteristics of interest with patients for whom custom titration protocols are being developed.

[0033] As used herein, the term "titration protocol" is a broad term and should be given its common and customary meaning to those skilled in the art, and should not be limited to a specific or customary meaning. The term "titration protocol" may specifically refer to, but is not limited to, the procedure used to titrate a drug, including various input parameters such as the selection of the antidiabetic drug, the starting dose, the total dose, dose step adjustments, adjustment time, target range, etc.

[0034] As used herein, the term "titration protocol parameters" is a broad term and should be given its common and customary meaning to those skilled in the art, and should not be limited to a specific or customary meaning. The term "titration protocol parameters" may specifically refer to, but is not limited to, various input parameters such as the selection of antidiabetic drugs, starting dose, total dose, dose step adjustment, adjustment time, target range, etc.

[0035] As used herein, the term "titration output parameter" is a broad term and should be given its common and customary meaning to those skilled in the art, and should not be limited to a specific or customary meaning. The term "titration output parameter" may specifically refer to, but is not limited to, parameters generated by performing a titration protocol, such as, for example, the time to achieve glycemic control, the number of hypoglycemic events, the number of hyperglycemic events, HbA1c, fasting blood glucose level, and the percentage of time within the target glycemic range during the titration period.

[0036] As used herein, the term "successful titration" is a broad term and should be given its common and customary meaning to those skilled in the art, and should not be limited to a specific or customary meaning. The term "successful titration" may specifically refer to, but is not limited to, titrations that meet target or threshold levels of titration output parameters.

[0037] As used herein, the term "unsuccessful titration" is a broad term and should be given its common and customary meaning to those skilled in the art, and should not be limited to a specific or customary meaning. The term "unsuccessful titration" may specifically refer to, but is not limited to, titrations that do not meet the target or threshold levels of the titration output parameters.

[0038] As used herein, the term "delivery medium" is a broad term and should be given its common and customary meaning to a person skilled in the art, and should not be limited to a particular or customary meaning. The term "delivery medium" may specifically refer to, but is not limited to, various devices or modes that can be used alone or in combination to communicate instructions for a customized titration protocol to a patient. For example, "delivery medium" may refer to SMS, lightweight titration service applications, integration with other applications (e.g., mySugr), voice skills Alexa, Siri, Cortana, Google Home (smartphones and home devices), robots or assistants that call patients, personal computers, enterprise computers, dumb terminals, television screens, robotic assistants, network communication devices, tablet computers, smartphones, smartwatches, etc.

[0039] method

[0040] The method steps disclosed herein can be performed in the order shown. However, alternative orders are also possible. Furthermore, individual or multiple method steps can be performed individually or in groups, in parallel, simultaneously, or repeatedly. For example, steps for determining titration protocol parameters do not need to be performed in the precise order described below. Additionally, the method may include additional method steps not shown. Regardless of the use of the term "method step," the term "step" does not refer to the duration of a method step. Therefore, specified method steps can be performed individually or in groups for short periods, or over longer time periods (e.g., time intervals of minutes, hours, days, weeks, or even months) (e.g., continuously or repeatedly).

[0041] This disclosure relates to a method for custom titration protocols, comprising the following steps:

[0042] a) Provide a database containing anonymized personal parameters for multiple previously managed subjects;

[0043] b) Generate cohorts from multiple previously managed subjects based on commonalities in anonymous personal parameters;

[0044] c) Identify successful and unsuccessful titrations within each queue;

[0045] d) Receive patient-specific personal parameters;

[0046] e) Identify patient-like queues from the generated queues that correspond to patient-specific personal parameters;

[0047] f) Receive the titration output parameters to be optimized;

[0048] g) Determine which titration parameters contribute to the success of a successful titration and which contribute to the failure of a failed titration; and

[0049] h) Based on titration output parameters and a patient similarity cohort, derive a customized titration protocol for each patient.

[0050] In some embodiments, the method may further include a step of diagnosing the need for administration of antidiabetic medication prior to performing step (a).

[0051] In some embodiments, the method may further include:

[0052] i) Delivering customized titration protocols to patients

[0053] j) Follow the titration protocol

[0054] k) Evaluate the titration results

[0055] l) Once titration is successful, maintain the administration protocol.

[0056] m) If titration fails, adjust the application plan.

[0057] diagnosis:

[0058] Healthcare providers can usually diagnose the need for medication to treat diabetes. Diagnosis may occur based on routine blood tests or during standard office visits where fasting blood glucose is measured. Diagnosis may also occur when a patient presents with symptoms such as frequent urination, thirst, fatigue, unexplained weight loss, slow healing of incisions or wounds, or blurred vision. Diagnosis may also be derived through tests such as an oral glucose tolerance test.

[0059] Identify patient-specific cohorts:

[0060] The titration system may include anonymized personal parameters of previously managed subjects, such as, for example, health and demographic parameters. Previously managed subjects may be grouped into cohorts based on similarity analysis, and any technique or combination of techniques may be used to perform the similarity analysis to determine the statistical or learned similarity of the datasets. Examples of techniques suitable for use in the disclosed methods include, but are not limited to:

[0061] Cosine similarity By modeling previously managed subjects as vectors of defined health and demographic parameter data points and comparing these vectors using similarity measures, the similarity between a patient and a database of previously managed subjects can be determined. The vectors are feature embeddings consisting of binary or numerical features representing the health and demographic parameter data points, such as, but not limited to, existing conditions, medications received, vital signs, laboratory observations, and the temporal relationship between these data points and clinical events.

[0062] Knowledge Graph Database / Algorithm Knowledge graph databases can be used to determine the similarity between patients and databases of previously managed subjects. Health and demographic parameters can be vertices in the knowledge graph. For example, age, HbA1c, comorbidities, medications, and body mass index can form vertices in the knowledge graph. Edges can be defined accordingly; for example, a patient with diabetes and taking metformin would have edges leading to the vertices of diabetes and metformin. Similarity measures or vertex similarity can be used to determine the similarity between patients. Similarity measures can be determined based on normalized distance measures such as Euclidean distance, Manhattan distance, Leavenstein distance, Maharanobis distance, Minkowski distance, Hamming distance, etc. Vertex similarity can be determined using, for example, neighborhood count, neighborhood selectivity, neighborhood rarity, SimRank, etc.

[0063] Artificial intelligence / machine learning modelingAlternatively, machine learning-based AI models can be used to determine the similarity between one or more patients and a previously managed database of subjects. Non-limiting examples of such learning algorithms include: K-nearest neighbors, support vector machines, Naive Bayes, decision trees such as random forests, logistic regression such as multinomial logistic regression, neural networks, decision trees, and Bayesian networks. Exemplary methods are described below (Sammut et al., Encyclopedia of Machine Learning, Vol. 1, Springer Publishing Company, Incorporated, 2011). Table 1 summarizes the advantages and disadvantages of these methods.

[0064] Clustering Clustering uses one or more analytical techniques to group a set of objects such that objects in the same group are more similar to each other than objects in other groups are more similar to each other.

[0065] K's nearest neighbor This is an example of a clustering method. Other deep clustering methods can also be used with this disclosure. The goal of this method is to place objects into categories that have similar objects. The category of a particular object is determined based on the category in which objects with similar parameter values ​​most frequently occur. To determine the proximity between objects, a similarity metric, such as Euclidean distance, is used. This method is well-suited for significantly larger datasets.

[0066] Support Vector Machine In this method, a hyperplane is computed that classifies objects into categories. To compute the hyperplane, the distance around the category boundaries is maximized, which is why Support Vector Machines (SVMs) are considered "large-margin classifiers." A key assumption of this method is the linear separability of the data, but this can be extended to higher-dimensional vector spaces using the kernel trick. For classification with low overfitting, a large amount of data is required.

[0067] Naive Bayes The naive assumption presupposes that the current variables are statistically independent. In most cases, this assumption does not hold. Even when attributes are slightly correlated, Naive Bayes can still achieve high classification accuracy in many situations. Naive Bayes analysis is relatively simple to perform.

[0068] return In regression analysis, statistical procedures are used to determine the relationship between the dependent variable and one or more independent variables.

[0069] Logistic RegressionThis is an example of a regression method. Other regression methods may also be used with this disclosure. In logistic regression, the likelihood that the value of the independent variable can be attributed to the value of the dependent variable is calculated.

[0070] Deep learning Deep learning is part of a broader family of machine learning methods that are based on artificial neural networks with representation learning capabilities. The term "deep" refers to the use of multiple layers in the network. Deep learning methods can be supervised, semi-supervised, or unsupervised. Deep learning architectures can include, but are not limited to, deep neural networks, deep belief networks, deep reinforcement learning, recurrent neural networks, convolutional neural networks, and transformers.

[0071] Neural networks This is an example of a deep learning approach. Other deep learning approaches may also be used with this disclosure. Artificial neural networks are based on the biological structure of neurons in the brain. A simple neural network consists of neurons arranged in three layers. These layers are the input layer, hidden layer, and output layer. Between each layer, all neurons are interconnected via weights that are optimized during the training phase.

[0072] Decision Tree Decision trees are hierarchical trees that rank objects, characterized by their simple appearance and ease of understanding. Nodes near the root are more meaningful for classifying objects than those near the leaves. Decision trees often suffer from problems caused by overfitting. Therefore, random forest methods can be useful. Random forests consist of multiple decision trees, where each tree represents a subset of variables.

[0073] Bayesian networks A Bayesian network is a directed graph that illustrates a multivariate likelihood distribution. The nodes of the network correspond to random variables, and the edges show the relationships between these random variables. For developing Bayesian networks, it is helpful to describe the dependencies between variables in as much detail as possible.

[0074] Table 1: Summary of the advantages and disadvantages of AI / machine learning methods

[0075]

[0076] Data capture:

[0077] Personal parameters, such as patient health and demographic parameters, can be received by the titration system. For example, parameters can be obtained from an electronic health record system by the titration system, or healthcare providers can enter patient personal parameters using a healthcare provider interface. Non-limiting examples of healthcare provider interfaces include personal computers, enterprise computers, dumb terminals, network communication devices, tablets, smartphones, smartwatches, etc. Non-limiting examples of health parameters include HbA1c, comorbidities, age, height, weight, body mass index, other medications, etc. Non-limiting examples of demographic parameters include race, age, sex, socioeconomic status, preferred communication methods, etc.

[0078] A patient's personal parameters can be used to create a digital twin for that patient. The patient can then be placed into a previously managed cohort of subjects by analyzing the digital twin and the cohort using similarity analysis techniques (such as those discussed above). Alternatively, the patient can be placed into a cohort by identifying a digital twin from a pre-existing set of digital archives using similarity analysis techniques. The patient can then be assigned to a cohort based on the selected digital twin.

[0079] Customize the type of antidiabetic drug and titration regimen based on the patient cohort and one or more selected titration parameters:

[0080] Healthcare providers can use the healthcare provider interface to input one or more titration output parameters to be optimized. Non-limiting examples of the healthcare provider interface include personal computers, enterprise computers, dumb terminals, network communication devices, tablets, smartphones, smartwatches, etc. Non-limiting examples of titration output parameters to be optimized may include: the shortest time to achieve the target range; the maximum number of patients within the expected range over a predefined time period; the maximum number of patients who achieved the target range in the first titration cycle; the highest percentage of time within the range; the minimum number of hypoglycemic events, etc.

[0081] It can separate titrate data from previous patients who met the titration optimization objectives and determine relevant protocol parameters. Simultaneously, it can also separate data from previous patients who were unsuccessfully titrated according to the optimization objectives and determine the relevant protocol parameters that led to the failure.

[0082] For example, each titration protocol parameter can be evaluated based on its statistical correlation with the expected / unexpected titration result and all possible outcomes. This analysis can be performed on the entire patient pool, a pool of similar patient cohorts, or some combination thereof. Within acceptable confidence and significance levels, titration parameters equal to or higher than a minimum statistical correlation threshold (e.g., 0.7, 0.8, 0.9, or 0.95) can be considered for adjustment in custom titration protocols.

[0083] Alternatively, regression and cluster analysis of parameters used to compare successful titrations can be used to identify which characteristics determine success. Simultaneously, regression and cluster analysis of parameters used to compare unsuccessful titrations can be used to identify which characteristics determine failure. For example, logistic regression methods may include:

[0084] 1. Collect patient-specific parameters, including success variables, which can be, for example, binary.

[0085] 2. Use success variables as response / target variables.

[0086] 3. The initial assumption is that all titration parameters are predictors of the regression.

[0087] 4. Identify titration parameters with significant p-values, meaning that such parameters have a significant impact on the results and, to some extent, determine success.

[0088] 5. Take the titration protocol parameters identified in step (4) that meet the significance criteria and bring them into the next step for analysis with a patient-similar cohort.

[0089] To determine which parameters need to be adjusted to form a customized titration protocol, significant / relevant parameters can be evaluated based on the corresponding protocol results within the patient cohort. This may include:

[0090] 1. Identify the patient cohort (discussed above).

[0091] 2. Identify relevant / significant titration protocol parameters.

[0092] 3. Interpolate relevant / significant titration protocol parameters to determine which parameters have the maximum number of patients who successfully / unsuccessfully completed the titration protocol.

[0093] 4. Adjust the default titration protocol based on parameter values ​​that have the highest success rate and / or the fewest failures, and present customized titration protocols to healthcare providers.

[0094] The above analysis can generate one or more recommended adjustments to standard titration protocol parameters to tailor titration protocols to specific patients. Healthcare providers may be given recommendations in various contexts. For example, recommendations may include patient cohorts, success rates and / or success ranges for the patient cohorts, initial starting doses for successful titrations, number of days to achieve the target glucose range, recommended doses, etc.

[0095] Success rates can be calculated for various individual parameters, depending on the desired titration output parameters. For example, success rate = number of successful patients with individual parameters / number of patients with individual parameters.

[0096] For each significant / relevant titration protocol parameter, recommendations can be made to healthcare providers to adjust the corresponding titration protocol parameters within a custom titration protocol based on values ​​from analyses with the highest success rates.

[0097] In a non-restrictive example, the appropriate parameters for the patient can be selected along the following logical path:

[0098] - There are 100 similar patients in the cohort.

[0099] - 50 / 100 Starting dose: 10 mg / dL => Success rate: 80%

[0100] - 50 / 100 Starting dose: 15 mg / dL => Success rate: 90%

[0101] => Select 15 mg / dL as the initial dose for this patient.

[0102] Before making a customized titration recommendation, the titration system may validate the recommendation against contraindications based on specific factors relevant to the patient or the titration service itself. For example, the recommended basal insulin type might negatively interact with another medication the patient is currently taking. Alternatively, the recommended initial starting dose might cause the patient to reach their maximum daily insulin dose prematurely. If such contraindications exist, the recommendation is not presented to the healthcare provider, or it may be presented to the healthcare provider along with a disregardable warning.

[0103] Delivery of titration protocol to patient:

[0104] In addition to the customized titration protocol as disclosed above, the titration system can also determine the preferred delivery medium for delivering the titration protocol to patients. Similarity analysis of demographic parameters within similar patient cohorts can be performed to determine the titration success rate for different types of delivery media. The delivery medium with the highest success rate for that patient cohort can be recommended. Optimizing the titration protocol and delivery medium increases the likelihood of patient adherence to the titration protocol, which in turn increases the probability of successful titration results.

[0105] For example, patients in one cohort might be more successful using a mobile app to manage their titration regimen because members of that cohort use smartphones more than average. Another cohort might prefer less technical / functional delivery methods, such as SMS-based services that only require patients to respond to simple prompts.

[0106] Information on the success rates of various delivery media for a given patient cohort can be found along the following routes:

[0107] - There are 100 similar patients in the cohort.

[0108] - 70 / 100 Using mobile app-based services => Success rate: 90%

[0109] - 20 / 100 SMS-based services => Success rate: 70%

[0110] - 10 / 100 Smart Home Assistant-Based Services => Success Rate: Not Statistically Significant

[0111] => Select a mobile application-based service for this patient

[0112] Possible delivery media for patient cohort-based titration protocols, monitoring, and data entry may include: SMS, lightweight titration service apps, integration with other apps (e.g., mySugr), voice skills such as Alexa, Siri, Cortana, Google Home (smartphones and home devices), and call-to-patient robots or assistants.

[0113] Titration system

[0114] The titration system includes a database containing anonymized data from multiple previously managed subjects. The anonymized data may include personal parameters such as demographic and health parameters for each previously managed patient, the titration protocol used for each previously managed patient, and the titration output parameters for each previously managed patient.

[0115] The titration system may further include a healthcare provider interface configured to receive patient-specific parameters. The healthcare provider interface may be further configured to receive titration output parameters to be optimized when selecting a titration protocol. Non-limiting examples of healthcare provider interfaces include personal computers, enterprise computers, dumb terminals, network communication devices, tablet computers, smartphones, smartwatches, etc.

[0116] In some embodiments, the titration system may include a data processing device configured to generate cohorts from multiple subjects, identify the cohort most relevant to the patient, and determine the appropriate antidiabetic medication and titration protocol for the patient.

[0117] The titration system may also include a patient user interface configured to receive administration instructions from a data processing device. Devices that can be used as a patient user interface are non-exclusive. Suitable patient user interfaces may include personal computers, enterprise computers, dumb terminals, television screens, robotic assistants, network communication devices, tablet computers, smartphones, smartwatches, etc.

[0118] Advantageously, the titration system and method according to this disclosure provide a customized titration protocol tailored to each patient, which increases the likelihood of achieving successful titration results (e.g., timely titration, minimal adverse side effects, and sustained normal blood glucose levels).

[0119] Example 1: A method for titrating antidiabetic drugs for a patient, the method comprising:

[0120] a) Provide a database containing anonymized personal parameters for multiple previously managed subjects;

[0121] b) Receive patient-specific personal parameters;

[0122] c) Identify patient-like cohorts from a database with anonymized personal parameters that correspond to patient-specific personal parameters;

[0123] d) Receive the titration output parameters to be optimized for the patient; and

[0124] e) Based on titration output parameters and a patient similarity cohort, derive a customized titration protocol for each patient.

[0125] Example 2: The method according to Example 1, wherein the patient similarity cohort is selected from a pre-computed cohort generated from multiple previously managed subjects based on similarity in anonymous personal parameters.

[0126] Example 3: The method described in Example 1, wherein the patient similarity cohort is calculated in real time.

[0127] Example 4: According to the method described in Example 1, the personal parameter is a health parameter, which includes at least one of the following: HbA1c, comorbidity, age, height, weight, body mass index, other medications, hypoglycemia risk level, blood glucose level, and vital signs.

[0128] Example 5: The method according to Example 1, wherein the personal parameters are demographic parameters, which include at least one of the following: race, age, sex, socioeconomic status, social determinants of health, and preferred communication methods.

[0129] Example 6: The method according to Example 1, wherein the queue is generated using one or more of the following: cosine similarity, knowledge graph, artificial intelligence, machine learning, clustering, k nearest neighbors, support vector machine, naive Bayes, Bayesian network, regression, logistic regression, deep learning, neural network, decision tree, random forest.

[0130] Example 7: The method described in Example 1, wherein a digital twin of a patient is used to identify a patient similarity cohort.

[0131] Example 8: The method described in Example 1, wherein the titration output parameters to be optimized are entered by a healthcare provider.

[0132] Example 9: The method described in Example 1, wherein the titration output parameters to be optimized are determined by the titration system.

[0133] Example 10: According to the method described in Example 1, the titration output parameters to be optimized are selected from the group consisting of the following: time to reach the desired blood glucose level, percentage of time the blood glucose level is within the target range, number of hyperglycemic events, number of hypoglycemic events, titration success rate and combinations thereof.

[0134] Example 11: The method described in Example 1, wherein the customized titration scheme is based on the correlation between the titration scheme parameters and the titration output parameters to be optimized.

[0135] Example 12: The method described in Example 1, wherein the customized titration scheme is determined by regression and cluster analysis.

[0136] Example 13: The method according to Example 1, wherein step e) includes: i) identifying one or more titration protocol parameters associated with the titration output parameter to be optimized among a plurality of previously managed subjects.

[0137] Example 14: The method described in Example 13, wherein the titration protocol parameters associated with the titration output parameters to be optimized are identified within a patient-specific cohort.

[0138] Example 15: The method according to Example 13, wherein step e) further includes: ii) identifying one or more titration protocol parameters that are negatively correlated with the titration output parameter to be optimized among a plurality of previously managed subjects.

[0139] Example 16: The method described in Example 15, wherein the titration protocol parameters negatively correlated with the titration output parameters to be optimized are identified within a patient-specific cohort.

[0140] Example 17: The method according to Example 15, wherein step e) further includes adjusting the default titration protocol based on the correlations identified in steps i) and ii) to derive a customized titration protocol for the patient.

[0141] Example 18: The method according to Example 1, wherein step e) includes

[0142] i) Collect binary success variables reflecting positive or negative titration results for multiple previously managed subjects;

[0143] ii) Use logistic regression to determine titration protocol parameters that have significant p-values ​​for positive titration results;

[0144] iii) Use logistic regression to determine titration protocol parameters that have significant p-values ​​for negative titration results;

[0145] iv) Adjust the default titration protocol based on the titration protocol parameters identified in steps ii) and iii) to derive a customized titration protocol for the patient.

[0146] Example 19: The method described in Example 18, wherein the binary success variable is collected from a patient-specific cohort.

[0147] Example 20: The method described in Example 1, wherein the titration scheme includes one or more of the following: drug type, drug subtype, initial drug dose, dose adjustment frequency, and dose adjustment increment.

[0148] Example 21: The method according to Example 1 further includes:

[0149] f) Determine the delivery medium most likely to produce successful titration for each patient based on the patient cohort; and

[0150] g) Use the delivery medium determined in step (f) to deliver titration protocol instructions to the patient.

[0151] Example 22: The method according to Example 18, wherein the delivery medium is one or more of SMS, an application, voice skills Alexa, Siri, Cortana, Google Home, a smartphone, email, a home device, a robot or assistant that calls the patient.

[0152] Example 23: The method according to Example 21, the method further includes administering an antidiabetic drug to the patient.

[0153] Example 24: The method according to Example 23, wherein instructions are delivered to the insulin pump and the antidiabetic drug is automatically administered to the patient.

[0154] Example 25: A system for titrating antidiabetic drugs for a patient, the system comprising:

[0155] A database containing anonymized personal parameters for multiple previously managed subjects;

[0156] A healthcare provider interface configured to receive patient-specific parameters and titration output parameters to be optimized; and

[0157] The processor is configured to:

[0158] (i) Generate a patient cohort from multiple previously managed subjects based on commonalities in anonymized personal parameters.

[0159] (ii) Identify patient-like queues from the generated queues that correspond to patient-specific personal parameters, and

[0160] (iii) Based on the titration output parameters and the patient similarity cohort, derive a customized titration plan for the patient.

[0161] Example 26: The system according to Example 25, wherein the health parameters include at least one of the following: HbA1c, comorbidity, age, height, weight, body mass index, other medications, hypoglycemia risk level, blood glucose level, and vital signs.

[0162] Example 27: The system according to Example 25, wherein demographic parameters include at least one of the following: race, age, sex, socioeconomic status, and preferred communication method.

[0163] Example 28: The system according to Example 25, wherein the processor is configured to generate queues using one or more of the following: cosine similarity, knowledge graph, artificial intelligence, machine learning, clustering, k nearest neighbors, support vector machine, naive Bayes, Bayesian network, regression, logistic regression, deep learning, neural network, decision tree, random forest.

[0164] Example 29: The system according to Example 25, wherein the processor is configured to use the patient's digital twin to identify patient-similar queues.

[0165] Example 30: The system according to Example 25, wherein the titration output parameters to be optimized are entered using a healthcare provider interface.

[0166] Example 31: The system according to Example 25, wherein the processor is configured to determine the titration output parameters to be optimized.

[0167] Example 32: The system according to Example 25, wherein the titration output parameters to be optimized are selected from the group consisting of: time to reach the desired blood glucose level, percentage of time the blood glucose level is within the target range, number of hyperglycemic events, number of hypoglycemic events, and combinations thereof.

[0168] Example 33: The system according to Example 25, wherein the processor is configured to create a customized titration scheme based on the correlation between titration scheme parameters and titration output parameters to be optimized.

[0169] Example 34: The system according to Example 25, wherein the processor is configured to create a customized titration scheme based on regression and cluster analysis.

[0170] Example 35: The system according to Example 25, wherein the processor is configured to identify one or more titration protocol parameters related to the titration output parameter to be optimized among a plurality of previously managed subjects.

[0171] Example 36: The system according to Example 35, wherein the titration protocol parameters associated with the titration output parameters to be optimized are identified within a patient-specific cohort.

[0172] Example 37: The system according to Example 35, wherein the processor is further configured to identify one or more titration protocol parameters that are negatively correlated with the titration output parameter to be optimized among a plurality of previously managed subjects.

[0173] Example 38: The system according to Example 37, wherein the titration protocol parameters negatively correlated with the titration output parameters to be optimized are identified within a patient-specific cohort.

[0174] Example 39: The system according to Example 37, wherein the processor is further configured to adjust the default titration scheme based on the identified correlations and negative correlations to derive a customized titration scheme for the patient.

[0175] Example 40: The system according to Example 25, wherein the processor is configured to

[0176] i) Collect binary success variables reflecting positive or negative titration results for multiple previously managed subjects;

[0177] ii) Use logistic regression to determine titration protocol parameters that have significant p-values ​​for positive titration results;

[0178] iii) Use logistic regression to determine titration protocol parameters that have significant p-values ​​for negative titration results;

[0179] iv) Adjust the default titration protocol based on the titration protocol parameters identified in steps ii) and iii) to derive a customized titration protocol for the patient.

[0180] Example 41: The system according to Example 25, wherein the binary success variable is collected from a patient-specific cohort.

[0181] Example 42: The system according to Example 25 further includes a patient user interface configured to receive a customized titration scheme from a data processing device.

[0182] Example 43: The system according to Example 25, wherein the processor is further configured to:

[0183] (iv) Determine the delivery medium most likely to produce successful titration for each patient based on the patient cohort; and

[0184] (v) Deliver titration protocol instructions to the patient interface.

[0185] Example 44: The system according to Example 43, wherein the delivery medium is one or more of SMS, applications, voice skills Alexa, Siri, Cortana, Google Home, smartphones, email, home devices, televisions, robots or assistants that call patients.

[0186] Example 45: The system according to Example 43 further includes an insulin pump, wherein instructions are delivered to the insulin pump and antidiabetic drugs are automatically administered to the patient. Attached Figure Description

[0187] The above aspects of the exemplary embodiments will become more apparent and better understood by referring to the following description of the embodiments in conjunction with the accompanying drawings, wherein:

[0188] Figure 1 is a schematic diagram of a typical titration process;

[0189] Figure 2 shows a schematic diagram of the custom titration system;

[0190] Figure 3 illustrates a schematic diagram of a method for creating a custom titration protocol;

[0191] Figure 4 shows a schematic diagram of a method for using a customized titration scheme according to the embodiment of Figure 3;

[0192] Figure 5 shows a schematic diagram of the process for placing a patient into a queue according to the embodiments of Figures 3 and 4;

[0193] Figures 6A and 6B illustrate schematic diagrams of a method for customizing a titration protocol for a patient according to the embodiments of Figures 3 and 4;

[0194] Figure 7 shows a schematic diagram of a titration scheme for delivery according to an embodiment of Figure 4.

[0195] illustrate

[0196] The embodiments described below are not intended to be exhaustive or to limit the invention to the precise forms disclosed in the detailed description below. Rather, the embodiments have been chosen and described so that others skilled in the art can understand and comprehend the principles and practice of this disclosure.

[0197] Figure 2 is a schematic diagram of a titration system 100 for customizing titration protocols for patients. In the illustrated embodiment, the titration system 100 includes a healthcare provider interface 104 that allows a healthcare provider 102 to interact with the titration system 100. The healthcare provider interface 104 may include both input and output devices, such as, for example, a keyboard, mouse, touchscreen, display, etc. Non-limiting examples of healthcare provider interfaces include personal computers, enterprise computers, dumb terminals, network communication devices, tablet computers, smartphones, smartwatches, etc.

[0198] The healthcare provider interface 104 may include a processor and memory. The processor and memory may be configured to perform a method 106 for creating a custom titration protocol. (See also Figure 4, dashed lines.) Alternatively, a separate device (not shown) having a processor and memory may be configured to perform the method 106 for creating a custom titration protocol. In embodiments having such a separate device, the separate device may be configured to communicate with the healthcare provider interface and perform some or all of the functions of the healthcare provider interface described herein.

[0199] The population titration data pool 110 can be housed within the electronic medical record system 108. The healthcare provider interface 104 can be connected to the electronic medical record system 108 using wired or wireless communication technologies.

[0200] The healthcare provider interface 104 can also connect to the messaging server 112. The messaging server 112 can be configured to transmit instructions for a titration protocol to the patient 116 using one or more delivery media 114. For example, SMS, mobile applications, smart home assistants, robotic assistants, any other suitable delivery media, or combinations thereof, can be used to send instructions to the patient 116.

[0201] Figure 3 illustrates a block diagram of a non-limiting embodiment of a method 120 for creating a custom titration scheme according to the present disclosure. The method steps disclosed herein can be performed in the order shown. However, alternative orders are also possible. Furthermore, individual or multiple method steps can be performed individually or in groups, in parallel, simultaneously, or repeatedly. For example, steps for determining titration scheme parameters do not need to be performed in the precise order described below. Additionally, the method may include additional method steps not shown. Regardless of the use of the term "method step," the term "step" does not refer to the duration of a method step. Therefore, specified method steps can be performed individually or in groups for short periods, but can also be performed over longer time periods (e.g., time intervals of minutes, hours, days, weeks, or even months) (e.g., continuously or repeatedly).

[0202] Initially, at step 121, a database 110 (also referred to herein as a population titration data pool) of previously managed, anonymized personal parameters of the subjects is provided to the titration system. Database 110 may be obtained from, for example, electronic medical records.

[0203] In step 123, a similarity analysis of anonymized personal parameters (e.g., health and demographic parameters of previously managed subjects) can be performed. The previously managed subjects can then be grouped into cohorts based on the results of the similarity analysis. Non-limiting examples of similarity analysis techniques suitable for use in step 123, as discussed above, include cosine similarity, knowledge graphs, artificial intelligence, machine learning, clustering, k-nearest neighbors, support vector machines, Naive Bayes, Bayesian networks, regression, logistic regression, deep learning, neural networks, decision trees, and random forests.

[0204] At step 124, the titration system receives patient-specific personal parameters. For example, healthcare provider 102 can enter the patient's health and demographic parameters into titration system 100. This can be done using healthcare provider interface 104. Alternatively, the titration system can access this information directly from the patient's electronic medical record.

[0205] At step 128, patient similarity cohorts are identified. This can be accomplished by performing a similarity analysis between patients and cohorts of previously managed subjects using techniques such as cosine similarity, knowledge graphs, artificial intelligence, machine learning, clustering, k-nearest neighbors, support vector machines, Naive Bayes, Bayesian networks, regression, logistic regression, deep learning, neural networks, decision trees, random forests, etc. This step will be discussed in more detail below with reference to Figures 4 and 5.

[0206] At step 130, the titration system receives the titration output parameters to be optimized. These parameters can be entered by healthcare provider 102 using healthcare provider interface 104. Alternatively, the titration output parameters can be derived based on statistical analysis of a patient-specific cohort.

[0207] Then, at step 132, the titration system can derive a customized titration protocol for the patient. The customized titration protocol can be based on the results of previously managed subjects in a patient-specific cohort. It can also take into account any patient-specific contraindications. Alternative methods for creating customized titration protocols will be discussed in more detail below with reference to Figures 4, 6A, and 6B.

[0208] Figure 4 shows a block diagram of a non-limiting embodiment of a method for treating a patient using a customized titration protocol. The embodiment depicted in Figure 4 uses the same method as shown in Figure 3 to create a customized titration protocol, and the same steps have the same numbering. The method shown in Figure 4 can also be implemented using the titration system 100 shown in Figure 2.

[0209] In step 122, the healthcare provider diagnoses the need for antidiabetic medication. This may be done during a scheduled medical visit or through examinations performed in response to the patient’s symptoms, such as frequent urination, thirst, unexplained weight loss, or more serious diabetic complications or comorbidities.

[0210] At step 124, the titration system receives patient-specific personal parameters. For example, healthcare provider 102 can enter the patient's health and demographic parameters into titration system 100. This can be done using healthcare provider interface 104. Alternatively, the titration system can access this information directly from the patient's electronic medical record.

[0211] In step 126, a digital twin can be created using the patient's personal parameters. Alternatively, a digital twin can be selected from an existing set of common profiles. The common profile that most closely resembles the patient's data can be selected and used as the digital twin. For the purposes of this disclosure, the digital twin does not need to be exactly the same as the patient. It only needs to adequately represent the patient's health and demographic parameters to perform the disclosed statistical analyses.

[0212] In step 128, shown in more detail in Figure 5, the titration system identifies the patient cohort corresponding to the digital twin. As shown in Figure 5, data regarding previously managed subjects can be mined from the electronic medical record system 108 in step 202. Specifically, the data may include anonymized personal parameters, such as health and demographic parameters of previously managed subjects.

[0213] In step 204, a similarity analysis of the anonymized personal parameters of the previously managed subjects can be performed. The previously managed subjects can then be grouped into cohorts based on the results of the similarity analysis. Non-limiting examples of similarity analysis techniques suitable for use in step 204, as discussed above, include cosine similarity, knowledge graphs, artificial intelligence, machine learning, clustering, k-nearest neighbors, support vector machines, Naive Bayes, Bayesian networks, regression, logistic regression, deep learning, neural networks, decision trees, and random forests.

[0214] Finally, at step 206, the digital twin can be placed into a queue of subjects with whom the previous administration was conducted. This can be accomplished using similarity analysis. Several examples of similarity analysis techniques suitable for use with this disclosure are discussed elsewhere herein.

[0215] Returning to Figure 4, at step 130, one or more titration output parameters can be selected for optimization within a customized titration protocol. In a non-limiting embodiment, the healthcare provider can select the titration output parameters to be optimized and enter them into the titration management system using the healthcare provider interface 104. Other methods for selecting and entering titration parameters are also envisioned. For example, the titration system can derive a set of parameters to be optimized based on statistical analysis of similar patient cohorts.

[0216] At step 132, which is shown in more detail in Figures 6A and 6B, the titration system optimizes the titration protocol based on the selected titration output parameters and results from the previously managed patient cohort to which it has been assigned a digital twin.

[0217] In an alternative embodiment shown in Figure 6A, at step 402, data from previous patients within each cohort who met or failed to meet the titration output optimization target can be separated, and relevant protocol parameters can be determined. For example, each titration protocol parameter can be evaluated based on its statistical correlation with the expected / unexpected titration outcome and all possible outcomes. Within acceptable confidence and significance levels, titration parameters adjusted to be equal to or higher than a minimum statistical correlation threshold (e.g., 0.7, 0.8, 0.9, or 0.95) in a customized titration protocol can be considered.

[0218] Alternatively, at step 402, regression and cluster analysis of parameters used to compare successful titrations can be used to identify which characteristics determined success for a specific cohort of previously managed subjects. Simultaneously, regression and cluster analysis of parameters used to compare unsuccessful titrations can be used to identify which characteristics determined failure. For example, a logistic regression method may include:

[0219] 1. Collect patient health and demographic parameters, including success variables, which may be, for example, binary.

[0220] 2. Use success variables as response / target variables.

[0221] 3. The initial assumption is that all parameters of the scheme are regression predictors.

[0222] 4. Identify parameters with significant p-values, meaning that such parameters have a significant impact on the results and, to some extent, determine success or failure.

[0223] 5. Substitute the parameters that are identified as meeting the significance criteria into step 404.

[0224] The above analysis can also be performed on the entire previously managed patient pool 110, as shown in Figure 6B.

[0225] In the alternative embodiment shown in Figure 6B, at step 402', data from previously managed subjects whose titration met or failed to meet the titration output optimization objective can be separated, and relevant protocol parameters can be determined. For example, each titration protocol parameter can be evaluated based on its statistical correlation with the expected / unexpected titration result and all possible outcomes. Within acceptable confidence and significance levels, titration parameters adjusted to be equal to or higher than a minimum statistical correlation threshold (e.g., 0.7, 0.8, 0.9, or 0.95) in a custom titration protocol can be considered.

[0226] Alternatively, at step 402', regression and cluster analysis of the parameters used to compare successful titrations can be used to identify which characteristics determined success. Simultaneously, regression and cluster analysis of the parameters used to compare unsuccessful titrations can be used to identify which characteristics determined failure. For example, a logistic regression method may include:

[0227] 1. Collect patient health and demographic parameters, including success variables, which may be, for example, binary.

[0228] 2. Use success variables as response / target variables.

[0229] 3. The initial assumption is that all parameters of the scheme are regression predictors.

[0230] 4. Identify parameters with significant p-values, meaning that such parameters have a significant impact on the results and, to some extent, determine success or failure.

[0231] 5. Substitute the parameters that are identified as meeting the significance criteria into step 404.

[0232] At step 404, a set of titration schemes with different optimization parameters can be created for each queue. This can be done as follows:

[0233] a. Interpolate relevant / significant titration parameters to determine which titration protocol values ​​have the maximum number of patients who successfully complete the titration protocol.

[0234] b. Adjust the default titration scheme based on the titration scheme value that has the highest success rate.

[0235] At step 406, the titration system may receive a patient queue and titration output parameters to be optimized. In one example, the titration parameters to be optimized may be entered by healthcare provider 102 using healthcare provider interface 104. In another embodiment, the patient queue and / or titration parameters to be optimized may be derived based on data from previously managed subjects.

[0236] As shown in Figures 6A and 6B, at step 408, before making a customized titration protocol recommendation, the titration system may verify that the recommendation is not contraindicated based on specific factors relevant to the patient or the titration service itself. For example, the type of basal insulin that might be recommended may negatively interact with another medication the patient is currently taking. Alternatively, the initial starting dose to be recommended might cause the patient to reach their maximum daily insulin dose prematurely. If such contraindications exist, the recommendation is not displayed to the healthcare provider, or it may be displayed to the healthcare provider along with a disregardable warning.

[0237] The above analysis can generate one or more recommended adjustments to the standard titration protocol parameters to tailor a titration protocol for a specific patient. At step 410, the success rate of titration protocol parameter adjustments can be calculated for various individual parameters. For example, success rate = number of successful patients with individual parameters / number of patients with individual parameters.

[0238] Finally, at step 412, the titration system returns a customized antidiabetic drug and titration regimen based on the patient cohort and the selected titration output parameters for optimization.

[0239] Returning to Figure 4, once the customized titration protocol has been returned, it must be transmitted at step 134.

[0240] Figure 7 illustrates the process for delivering the titration protocol in more detail. At step 602, a recommendation may be given to the healthcare provider to adjust each significant / relevant titration protocol parameter within the customized titration protocol based on values ​​from the analysis at step 410, which has the highest success rate. The healthcare provider may be given a titration protocol recommendation at 602 in several contexts. For example, the recommendation may include the patient cohort, the success rate and / or success range of the patient cohort, the initial starting dose for successful titration, the number of days to achieve the target glucose range, the recommended dose, etc.

[0241] In a non-limiting example, the appropriate parameters for the patient can be transmitted along the following logical path:

[0242] - There are 100 similar patients in the cohort.

[0243] - 50 / 100 Starting dose: 10 mg / dL => Success rate: 80%

[0244] - 50 / 100 Starting dose: 15 mg / dL => Success rate: 90%

[0245] => Select 15 mg / dL as the initial dose for this patient.

[0246] At step 604, the titration system may also determine a preferred delivery medium for delivering the titration protocol to the patient. Similarity analysis can be performed on demographic parameters within similar patient cohorts to determine the titration success rate for different types of delivery media. The delivery medium with the highest success rate for the patient cohort can be recommended via the healthcare provider interface 104. Optimizing the titration protocol delivery medium increases the likelihood of patient adherence during the protocol, which in turn increases the probability of a successful titration outcome.

[0247] For example, patients in one cohort might be more successful using a mobile app to manage their titration regimen because that cohort uses smartphones more than average. Another cohort might prefer less technical / functional delivery media, such as SMS-based services that only require patients to respond to simple prompts.

[0248] Information on the success rates of various delivery media for a given patient cohort can be found along the following routes:

[0249] - There are 100 similar patients in the cohort.

[0250] - 70 / 100 Using mobile app-based services => Success rate: 90%

[0251] - 20 / 100 SMS-based services => Success rate: 70%

[0252] - 10 / 100 Smart Home Assistant-Based Services => Success Rate: Not Statistically Significant

[0253] => Select a mobile application-based service for this patient

[0254] Possible delivery media for patient cohort-based titration protocols, monitoring, and data entry may include: SMS, lightweight titration service apps, integration with other apps (e.g., mySugr), voice skills such as Alexa, Siri, Cortana, Google Home (smartphones and home devices), and call-to-patient robots or assistants. Because there is always a possibility that a particular patient may not have access to the preferred delivery media for his or her cohort, the healthcare provider and / or patient may optionally confirm the choice of delivery media at step 606.

[0255] At step 608, the customized titration protocol is delivered to the patient, and then the process returns to Figure 4, with the patient following the administration protocol as described at step 136. Results, such as fasting blood glucose, hyperglycemia and hypoglycemia events, and time within range, can then be assessed at step 138. If the titration is successful at step 140, the patient can continue using the protocol at step 142. If the titration is unsuccessful at step 140, the healthcare provider can manually adjust the titration protocol at step 144.

[0256] Although exemplary embodiments have been disclosed above, the invention is not limited to the disclosed embodiments. Rather, this application is intended to cover any variations, uses, or adaptations of the general principles of this disclosure. Furthermore, this application is intended to cover such deviations from this disclosure within the scope of known or customary practices in the art, and such deviations fall within the limitations of the appended claims.

Claims

1. A method for titrating an antidiabetic drug in a patient, comprising: a) Provide a database containing anonymous personal parameters for multiple previously managed subjects; b) Receive patient-specific personal parameters; c) Identify patient-like cohorts from the database containing anonymous personal parameters that correspond to the patient-specific personal parameters; d) Receive the titration output parameters to be optimized for the patient; as well as e) Based on the titration output parameters and the patient similarity cohort, derive a customized titration protocol for the patient.

2. The method of claim 1, wherein the patient similarity cohort is selected from a pre-computed cohort generated from the plurality of previously managed subjects based on similarity in anonymous personal parameters.

3. The method according to claim 1, wherein the patient similarity cohort is calculated in real time.

4. The method according to claim 1, wherein the personal parameter is a health parameter, the health parameter including at least one of the following: HbA1c, comorbidities, age, height, weight, body mass index, other medications, hypoglycemia risk level, blood glucose level, and vital signs.

5. The method of claim 1, wherein the personal parameter is a demographic parameter, the demographic parameter including at least one of: race, age, sex, socioeconomic status, social determinants of health, and preferred communication method.

6. The method of claim 1, wherein the queue is generated using one or more of the following: cosine similarity, knowledge graph, artificial intelligence, machine learning, clustering, k-nearest neighbor, support vector machine, naive Bayes, Bayesian network, regression, logistic regression, deep learning, neural network, decision tree, random forest.

7. The method of claim 1, wherein the digital twin of the patient is used to identify the patient similarity cohort.

8. The method of claim 1, wherein the titration output parameters to be optimized are entered by a healthcare provider.

9. The method according to claim 1, wherein the titration output parameter to be optimized is determined by the titration system.

10. The method of claim 1, wherein the titration output parameter to be optimized is selected from the group consisting of: time to reach the desired blood glucose level, percentage of time the blood glucose level is within the target range, number of hyperglycemic events, number of hypoglycemic events, titration success rate, and combinations thereof.

11. The method of claim 1, wherein the customized titration scheme is based on the correlation between titration scheme parameters and the titration output parameters to be optimized.

12. The method of claim 1, wherein the customized titration scheme is determined by regression and cluster analysis.

13. The method of claim 1, wherein step e) comprises: i) Identify one or more titration protocol parameters related to the titration output parameter to be optimized among the plurality of previously managed subjects.

14. The method of claim 13, wherein the titration protocol parameters associated with the titration output parameters to be optimized are identified within a patient-specific cohort.

15. The method of claim 13, wherein step e) further comprises: ii) Identify one or more titration protocol parameters that are negatively correlated with the titration output parameter to be optimized among the plurality of previously managed subjects.

16. The method of claim 15, wherein the titration protocol parameter negatively correlated with the titration output parameter to be optimized is identified within a patient-specific cohort.

17. The method of claim 15, wherein step e) further comprises adjusting the default titration protocol based on the correlations identified in steps i) and ii) to derive the customized titration protocol for the patient.

18. The method of claim 1, wherein step e) comprises i) Collect binary success variables reflecting positive or negative titration results for the multiple previously managed subjects; ii) Use logistic regression to determine the titration protocol parameters that have a significant p-value for the positive titration result; iii) Use logistic regression to determine the titration protocol parameters that have a significant p-value for the negative titration result; iv) Adjust the default titration protocol based on the titration protocol parameters identified in steps ii) and iii) to derive the customized titration protocol for the patient.

19. The method of claim 18, wherein the binary success variable is collected from a patient-specific cohort.

20. The method of claim 1, wherein the titration scheme comprises one or more of the following: drug type, drug subtype, initial drug dose, dose adjustment frequency, and dose adjustment increment.

21. The method of claim 1, further comprising: f) Determine the delivery medium most likely to produce a successful titration for the patient based on the patient cohort; as well as g) Use the delivery medium determined in step (f) to deliver titration protocol instructions to the patient.

22. The method of claim 18, wherein the delivery medium is one or more of SMS, an application, voice skills Alexa, Siri, Cortana, Google Home, a smartphone, email, a home device, a robot or assistant that calls the patient.

23. The method of claim 21, further comprising administering an antidiabetic drug to the patient.

24. The method of claim 23, wherein the instruction is delivered to the insulin pump and the antidiabetic drug is automatically administered to the patient.

25. A system for titrating an antidiabetic drug in a patient, comprising: A database containing anonymous personal parameters for multiple previously managed subjects; A healthcare provider interface configured to receive individual parameters for the patient and to receive titration output parameters to be optimized; as well as Processor, the processor being configured to: (iv) Generate a patient cohort from the plurality of previously managed subjects based on commonalities in the anonymized personal parameters. (v) Identify patient-like queues from the generated queues that correspond to patient-specific personal parameters, and (vi) Based on the titration output parameters and the patient similarity cohort, derive a customized titration protocol for the patient.

26. The system of claim 25, wherein the health parameters include at least one of the following: HbA1c, comorbidities, age, height, weight, body mass index, other medications, hypoglycemia risk level, blood glucose level, and vital signs.

27. The system of claim 25, wherein the demographic parameters include at least one of the following: race, age, sex, socioeconomic status, and preferred communication method.

28. The system of claim 25, wherein the processor is configured to generate the queue using one or more of the following: cosine similarity, knowledge graph, artificial intelligence, machine learning, clustering, k-nearest neighbor, support vector machine, naive Bayes, Bayesian network, regression, logistic regression, deep learning, neural network, decision tree, random forest.

29. The system of claim 25, wherein the processor is configured to use the patient's digital twin to identify the patient similarity queue.

30. The system of claim 25, wherein the output parameters to be optimized for titration are entered using the healthcare provider interface.

31. The system of claim 25, wherein the processor is configured to determine the titration output parameters to be optimized.

32. The system of claim 25, wherein the titration output parameter to be optimized is selected from the group consisting of: time to reach the desired blood glucose level, percentage of time the blood glucose level is within the target range, number of hyperglycemic events, number of hypoglycemic events, and combinations thereof.

33. The system of claim 25, wherein the processor is configured to create the customized titration scheme based on the correlation between titration scheme parameters and the output parameters of the titration to be optimized.

34. The system of claim 25, wherein the processor is configured to create the customized titration scheme based on regression and cluster analysis.

35. The system of claim 25, wherein the processor is configured to identify one or more titration protocol parameters associated with the titration output parameter to be optimized among the plurality of previously managed subjects.

36. The system of claim 35, wherein the titration protocol parameters associated with the titration output parameters to be optimized are identified within a patient-specific cohort.

37. The system of claim 35, wherein the processor is further configured to identify one or more titration scheme parameters negatively correlated with the titration output parameter to be optimized among the plurality of previously managed subjects.

38. The system of claim 37, wherein the titration protocol parameter negatively correlated with the titration output parameter to be optimized is identified within a patient-specific cohort.

39. The system of claim 37, wherein the processor is further configured to adjust the default titration scheme based on the identified correlations and negative correlations to derive the customized titration scheme for the patient.

40. The system of claim 25, wherein the processor is configured to i) Collect binary success variables reflecting positive or negative titration results for the multiple previously managed subjects; ii) Use logistic regression to determine the titration protocol parameters that have a significant p-value for the positive titration result; iii) Use logistic regression to determine the titration protocol parameters that have a significant p-value for the negative titration result; iv) Adjust the default titration protocol based on the titration protocol parameters identified in steps ii) and iii) to derive the customized titration protocol for the patient.

41. The system of claim 25, wherein the binary success variable is collected from a patient-specific cohort.

42. The system of claim 25, further comprising a patient user interface configured to receive the customized titration scheme from a data processing device.

43. The system of claim 25, wherein the processor is further configured to: (iv) Determine the delivery medium most likely to produce a successful titration for the patient based on the patient cohort; and (v) Deliver titration protocol instructions to the patient interface.

44. The system of claim 43, wherein the delivery medium is one or more of SMS, an application, voice skills Alexa, Siri, Cortana, Google Home, a smartphone, email, a home device, a television, a robot or assistant that calls the patient.

45. The system of claim 43, further comprising an insulin pump, wherein the instruction is delivered to the insulin pump and the antidiabetic drug is automatically administered to the patient.

Images