Invented by Patrick White, Somadev Pasala, Donald Johnson, Sreedhar Potturi, Dhaval Shah, Anthony Degliomini, Michael Brennan, Optum Inc

The market for digital representations of past and future health using vectors is rapidly expanding as technology continues to advance. Vectors, which are mathematical representations of magnitude and direction, are being utilized to create detailed and accurate digital models of an individual’s health history and potential future health outcomes. In the past, health records were primarily stored in physical files, making it difficult for healthcare professionals to access and analyze patient data efficiently. However, with the advent of digital health records, the storage and retrieval of health information have become much more streamlined. Vectors take this a step further by providing a visual representation of health data, allowing for easier interpretation and analysis. One of the key benefits of using vectors in digital health representations is the ability to capture complex relationships between various health factors. For example, a vector model can depict how a person’s diet, exercise routine, and genetic predispositions interact to influence their overall health. By visualizing these relationships, healthcare professionals can gain a better understanding of the underlying causes of certain health conditions and develop more targeted treatment plans. Furthermore, vectors can also be used to predict future health outcomes based on historical data. By analyzing patterns and trends in an individual’s health history, predictive models can be created to estimate the likelihood of developing certain diseases or conditions. This information can be invaluable in preventive healthcare, allowing individuals to make informed decisions about their lifestyle choices and take proactive measures to maintain their health. The market for digital representations of past and future health using vectors is not limited to healthcare professionals alone. Individuals are increasingly interested in monitoring and managing their own health, and digital health tools that utilize vectors can empower them to do so effectively. With the help of user-friendly interfaces and visualizations, individuals can easily track their health data, identify areas of concern, and make informed decisions about their well-being. Moreover, the potential applications of digital health representations using vectors extend beyond individual health management. Researchers and policymakers can utilize these models to study population health trends, identify risk factors, and develop targeted interventions. By harnessing the power of big data and advanced analytics, vectors can enable a more comprehensive understanding of health at both the individual and population levels. As the demand for personalized healthcare continues to grow, the market for digital representations of past and future health using vectors is poised for significant expansion. The ability to visualize and analyze health data in a meaningful way has the potential to revolutionize healthcare delivery and improve patient outcomes. With ongoing advancements in technology and data analytics, the possibilities for leveraging vectors in digital health representations are endless.

The Optum Inc invention works as follows

A computing entity retrieves medical information instances corresponding a population and generates medical sentences by generating, for each of the patients in the population, one medical code and a timestamp corresponding with each instance, such that each medical sentence contains one or more codes in chronological order. The computing entity creates a multi-dimensional dictionary based on the vector generation model that was trained with machine learning. Each multi-dimensional is corresponding to one of the medical codes. The computing entity creates a digital image of a patient in the population by using the anagram model, the vector dictionary and determining the patient’s health status based on this digital image. It then provides a result indicating that patient’s health status.

Background for Digital representations of past and future health using vectors

Medical codes are used to encode medical information, including diagnoses, procedures, drug or prescription codes, equipment codes and revenue codes. These medical codes can be non-interpretable alpha-numeric string that are quite specific. A fracture in the left forearm, for example, may have a code different from that of a right forearm fracture. It can be difficult to identify groups of patients with similar histories or cohorts. A key word search can identify groups of patients who share a common element in their medical history, but have a different health history or condition. Two patients, for example, may have both type 2 diabetes but vastly differing underlying conditions. Type 2 diabetes and hypertension are highly related, but according to key word searches, people with hypertension will be considered a separate population.

Various embodiments” provide methods, devices, computer program products or systems that provide a virtual representation of the medical history of a patient. In one embodiment, the vector of the digital representation is the patient’s history. In some embodiments, the medical data/information corresponding to an entire population (e.g. a plurality) of patients is accessed, and medical sentences are generated for each of those patients. In some embodiments, medical sentences are composed of a plurality of codes (e.g. diagnosis codes or procedure codes; prescription codes or drug codes; equipment codes and revenue codes. Each patient in the population is represented by a medical sentence, and each sentence has a predetermined length (e.g. a certain number of medical code). In one embodiment, if a medical code count of a medical sentence for a patient is less than the predetermined number, dummy codes can be added so that it contains the desired number. In one embodiment, medical codes are arranged in chronological order according to the dates of medical events. These medical sentences are used to train a model for vector generation. The vector generation model can be trained, for example, using machine learning with a data set that includes at least some medical sentences. The vector generation models may include a digital representation for one or more of the patients in the population and/or vector dictionary that corresponds to the vocabulary (e.g. unique medical codes) used to train the vector model.

In various embodiments, a vector generation module and/or an additional module (e.g. using a vector diction generated by the vector creation model) create a digital representation of patients in a population of patients. In some embodiments, a digital representation of a patient is used to determine the current health status of the patient. In some embodiments, a digital image of the patient is used, along with an anagram, multi-dimensional vectors from the vector dictionary or a digital image of a group of patients who have similar medical histories, to determine a future possible health state of the patient. The possible future state can be an optimal state for the patient based on the medical history of the first patient, a future state expected of the patients (e.g. determined using digital representations of cohorts of patients with similar medical histories to the patient), or a predicted outcome of a decision made by the patient. The current health status and/or possible future health states for the patient may be provided via an interface interactive of a computing entity, and can influence one or several decisions about medical treatment.

According to an aspect of the invention, there is provided a method of providing a state of health of a patient. In one embodiment, the method includes accessing by a computing device comprising of a processor, and a memory that stores computer program codes, a plurality instances of medical data encoded with medical codes, and corresponding to an entire population of patients, which comprises a plurality patients. The method also comprises, by the computing device, generating a plurality medical sentences that correspond to the population by generating for each patient, one medical statement based on the one or multiple instances of the plurality instances of medical information associated with the patient, and the timestamps associated with the one or several instances, such that the medical sentence that corresponds to the particular patient includes one or many medical codes in chronological order. The method also includes generating by the computing entity a dictionary of multi-dimensional multi-dimensional vectors based on the machine-learning-trained vector generation model and the plurality medical sentences. Each multi-dimensional code corresponds to each multi-dimensional vector. The computing entity generates a digital image of a patient using the anagram model, the multi-dimensional vectors, and the vector dictionary. It then determines the health status of that patient by analyzing the digital image.

A device is also provided in another aspect of this invention. The apparatus can include at least one computer and a memory containing computer program codes for one or several programs. The computer program code and at least one memory are configured to cause the apparatus, in conjunction with the atleast one processor, to: “at least access a multitude of instances corresponding medical information encoded with medical codes, and correspond to a patient population comprising multiple patients; generate, for each of these patients, one medical statement based upon one or several instances of each of those instances, as well as a timestamp that is associated with each of them, such that the medical sentence corresponding the patient includes one or two medical codes in chronologically;

A computer program product according to a further aspect of the invention is provided. In one embodiment, the computer product includes at least one nontransitory computer readable storage medium that contains computer-executable code portions. The computer-executable portions of the program code are program code instructions. When executed by a computer processor, the instructions are configured to: “cause the computing entity to access a number of instances of encoded medical information using medical codes, which correspond to a group of patients including a number of patients; generate, for each of those patients, one medical statement based upon one or multiple instances, with a timestamp, such that each medical sentence is based in chronological order on the medical code associated with the instance; generate, in a multidimensional space, having a configurable dimension, a

BRIEF DESCRIPTION ABOUT THE VIEWS OF MANY DRAWINGS

Having described the invention in general terms we will now refer to the accompanying drawings. These drawings are not necessarily drawn at scale and contain:

FIG. “FIG.

FIG. “FIG.

FIG. “FIG.

FIG. “FIG.

FIG. “FIG.

FIG. “FIG.

FIGS. “FIGS.

FIG. “FIG. 8 is a flowchart that illustrates example procedures, processes and/or operations carried out by a computing entity in order to determine a potential future health state of a first patient according to an example embodiment.

The following paragraphs will describe the various embodiments, with the help of the drawings that accompany the description. Some, but not all, embodiments are shown. These inventions can be implemented in many different ways and are not limited to those described here. They are rather provided for the purpose of satisfying applicable legal requirements. The term ‘or’ is used to describe the underlying concept. The term ?/?) The term is used in this article both in the alternative and conjunctive meaning, unless indicated otherwise. The term ‘illustrative’ is used to describe a particular example. The terms ‘illustrative’ and ‘exemplary? ?exemplary? “Like numbers refer to similar elements throughout.

I. “General Overview

In various embodiments, digital representations of one or more patients are provided by methods, systems and apparatuses. Computer program products may also be used. In one embodiment, the digital representation of a person is a vector in a multidimensional space. The digital representation of a patient can be, for example, an ordered list that encodes their current health status. In some embodiments, a vector-generation model is used to generate the digital representation. In one embodiment, the vector-generation model is configured for receiving a plurality input medical sentences, and providing output that comprises a multidimensional vector corresponding each vocabulary item (e.g. each unique medical code). In one embodiment, a medical sentence is used to represent each patient. The medical code order is based on the chronological order in which the events that correspond to the codes occurred. The medical sentences that have fewer medical codes than the specified number are expanded by adding dummy code to the sentence. The vector generation model can then be trained using the medical sentences. The digital representation is an ordered combination of vectors that corresponds to the medical sentence of the patient.

Various embodiments also include use of an Anagram model to determine or simulate a future possible health state of a patients. In one embodiment, an anagram is used to combine the digital representations of a patient with the multi-dimensional vectors of a vector dictionary in order to simulate a possible future state for the patient. It may also be applied to the digital image of the person to apply one or several anagram relationships found within the vector dictionaries. In one embodiment, the future health of the patient can be predicted based on digital representations from a cohort of similar patients, vectors in the vector dictionary or anagram relationships within the vector database. In different embodiments, a possible future state for a patient can be an optimal state of health based on the medical history of the person, an expected health state derived from digital representations of other patients with similar histories, or a predicted outcome of a decision made by a physician. It is possible to provide the patient’s present health status and/or a possible future state (e.g. via an interactive user-interface of a computing entity). In one embodiment, a decision regarding the current and/or future treatments of a patient can be made based on his or her current or future health status.

II. “II.

Click here to view the patent on Google Patents.