WO2022165071A1 - Moteur pour applications d'apprentissage machine à effets mixtes - Google Patents

Moteur pour applications d'apprentissage machine à effets mixtes Download PDF

Info

Publication number
WO2022165071A1
WO2022165071A1 PCT/US2022/014150 US2022014150W WO2022165071A1 WO 2022165071 A1 WO2022165071 A1 WO 2022165071A1 US 2022014150 W US2022014150 W US 2022014150W WO 2022165071 A1 WO2022165071 A1 WO 2022165071A1
Authority
WO
WIPO (PCT)
Prior art keywords
group
value
target variable
population
computer
Prior art date
Application number
PCT/US2022/014150
Other languages
English (en)
Inventor
Sourav Dey
Rajendra KOPPULA
Ramesh Sridharan
Original Assignee
Manifold Inc.
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Manifold Inc. filed Critical Manifold Inc.
Publication of WO2022165071A1 publication Critical patent/WO2022165071A1/fr

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/20—Ensemble learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound

Definitions

  • the present disclosure is related to modeling a target variable response from sampled data, wherein the input variable may have known or unknown group effects on the target variable. More specifically, embodiments as disclosed herein provide a method to identify and account for group variation to predict a target variable with improved accuracy even for unknown groups.
  • FIG. 1 illustrates an example architecture suitable for providing a mixed effects machine learning engine for healthcare management, according to some embodiments.
  • FIG. 2 is a block diagram illustrating an example server and client from the architecture of FIG. 1, according to certain aspects of the disclosure.
  • FIG. 3 is a chart illustrating the reduction of a loss function following a repeated alternate optimization boosting schedule, according to some embodiments
  • FIGS. 4A-4C illustrate charts to compare the performance of various mixed effects gradient boosting schedules, according to some embodiments
  • FIG. 5 illustrates a dataset splitting with five groups, according to some embodiments.
  • FIG. 6 is a flowchart illustrating steps in a method for modeling a mixed effect statistical dataset, according to some embodiments.
  • FIG. 7 is a flowchart illustrating steps in a method for managing healthcare with a mixed effects model of a patient population, according to some embodiments.
  • FIG. 8 is a block diagram illustrating an example computer system with which the client and server of FIGS. 1 and 2 and the methods of FIGS. 6-7 can be implemented.
  • a computer-implemented method includes receiving data for a population that includes a value for one or more input variables, and a value for a target variable, identifying a group index for the data based on one or more categories, determining a common function and one or more group functions based on the input variables, wherein the common function associates a value to the target variable that is common across the population and each of the group functions associates a value to the target variable for a group in the population, modeling the target variable as a combination of the common function and the one or more group functions, and determining a new value for the target variable based on a new set of values for the input variables.
  • a system includes one or more processors and a memory storing instructions which, when executed by the one or more processors, cause the system to perform operations.
  • the operations include to receive data for a population that includes a value for one or more input variables, and a value for a target variable and to identify a group index for the data based on one or more categories.
  • the operations also include to determine a common function based on the input variables and one or more group functions, wherein the common function associates a value to the target variable that is common across the population and each of the group functions associates a value to the target variable for a group in the population, to model the target variable as a combination of the common function and the one or more group functions, and to determine a new value for the target variable based on a new set of values for the input variables.
  • a computer-implemented method includes receiving health data for a population, the health data including a value for one or more health monitoring variables, and a value for a health outcome, identifying groups in the population based on one or more categories, determining a common function and one or more group functions based on the health monitoring variables, wherein the common function associates a value to the health outcome that is common across the population and each of the group functions associates a value to the health outcome for a group in the population, and updating a database to associate the health outcome to a new set of values for the health monitoring variables according to the common function.
  • categorical data can provide valuable information about the interaction between numerical quantities. For example, when predicting outcomes of treatment in a hospital, the outcome may vary by hospital, the type of hospital, or region where the hospital is located. As another example, when predicting the likelihood of a disease, the likelihood may vary by genotype or demographics, including sex, gender, age, race, or ethnicity. Moreover, the likelihood of a disease may depend on comorbidity conditions, such as asthma, cancer, dementia, diabetes, glaucoma, hypothyroidism, hypertension, or other diseases.
  • enrollment in a clinical trial, response to a treatment, or adherence to a treatment may depend on a patient’s medical conditions, medications, medical history, demographics, genetic characteristics, or socioeconomic characteristics, such as income, education, or employment. More generally, a model’s predictions could be more accurate if it accounts for the effect of patient variability.
  • the categorical variable imposes a grouping structure upon the data.
  • a group may be any subset of a population, including an individual member of the population.
  • a mixed effects machine learning framework for modeling nonlinear effects in data with group structure.
  • Embodiments as disclosed herein enable modeling of nonlinear interactions at a population-level and within each of several groups in a population of data.
  • the methods disclosed herein may capture population-level and group-level nonlinear effects using any existing model for regression.
  • a family of tree- and forest-based nonlinear methods is used in supervised learning (regression and classification) on data with a grouping structure.
  • a mixed effects gradient boosting extends the gradient boosting framework to model both population- and group-level nonlinear effects for classification and regression.
  • FIG. 1 illustrates a system architecture 100 for providing a mixed effects machine learning engine for healthcare management, according to some embodiments.
  • Architecture 100 includes servers 130 communicatively coupled with client devices 110 and at least one database 152 over a network 150.
  • One of the many servers 130 is configured to host a memory including instructions which, when executed by a processor, cause the server 130 to perform at least some of the steps in methods as disclosed herein.
  • the processor is configured to control an application for the user of one of client devices 110 accessing a healthcare management server or a mixed effects engine.
  • the application may include a healthcare management application (accessed by a healthcare provider and/or a healthcare professional), a patient application (accessed by a patient), or a patient messaging application (to send messages, reminders, and alerts to a patient, according to a healthcare intervention strategy).
  • the server may be configured to train a machine learning model for solving a specific healthcare management problem.
  • the processor may include a dataset tool, configured to display components and graphic results to the user via a graphical user interface (GUI) in client devices 110.
  • GUI graphical user interface
  • multiple servers 130 can host memories including instructions to one or more processors, and multiple servers 130 can host a history log in database 152, including multiple training archives used for healthcare management.
  • multiple users of client devices 110 may access the same server to run one or more machine learning models.
  • a single user with a single client device 110 may train multiple machine learning models running in parallel in one or more servers 130. Accordingly, client devices 110 may communicate with each other via network 150 and through access to one or more servers 130 and resources located therein.
  • Servers 130 may include any device having an appropriate processor, memory, and communications capability for hosting the healthcare management server or a mixed effects engine, including multiple tools associated with it.
  • the healthcare management server may be accessible by various clients 110 over network 150.
  • Client devices 110 can be, for example, desktop computers, mobile computers, tablet computers (e.g., including e-book readers), mobile devices (e.g., a smartphone or PDA), or any other device having appropriate processor, memory, and communications capabilities for accessing a healthcare management server on one or more of servers 130.
  • Network 150 can include, for example, any one or more of a local area tool (LAN), a wide area tool (WAN), the Internet, and the like.
  • LAN local area tool
  • WAN wide area tool
  • the Internet and the like.
  • network 150 can include, but is not limited to, any one or more of the following tool topologies, including a bus network, a star network, a ring network, a mesh network, a star-bus network, tree or hierarchical network, and the like.
  • FIG. 2 is a block diagram 200 illustrating an example server 130 and client device 110 from architecture 100, according to certain aspects of the disclosure.
  • Client device 110 and server 130 are communicatively coupled over network 150 via respective communications modules 218- 1 and 218-2 (hereinafter, collectively referred to as “communications modules 218”).
  • Communications modules 218 are configured to interface with network 150 to send and receive information, such as data, requests, responses, and commands to other devices via network 150.
  • Communications modules 218 can be, for example, modems or Ethernet cards.
  • a user may interact with client device 110 via an input device 214 and an output device 216.
  • Input device 214 may include a mouse, a keyboard, a pointer, a touchscreen, a microphone, and the like.
  • Output device 216 may be a screen display, a touchscreen, a speaker, and the like.
  • Client device 110 may include a memory 220- 1 and a processor 212-1.
  • Memory 220- 1 may include an application 222 and a GUI 225, configured to run in client device 110 and to couple with input device 214 and output device 216.
  • Application 222 may be downloaded by the user from server 130, and may be hosted by server 130. Accordingly, application 222 may be installed by server 130 and perform scripts and other routines provided by server 130 through any one of multiple tools. Execution of application 222 may be controlled by processor 212-1.
  • client device 110 and server 130 may transmit data packets 227-1 and 227-2 (hereinafter, collectively referred to as “data packets 227”) between each other, via communication modules 218 and network 150.
  • client device 110 may provide a data packet 227-1 to server 130 including a dataset with information about a patient population or a segment thereof.
  • server 130 may provide to client device 110 a data packet 227-2 including a model of the patient population for efficient healthcare management thereof.
  • Server 130 includes a memory 220-2, a processor 212-2, and communications module 218-2.
  • processors 212-1 and 212-2, and memories 220-1 and 220-2, will be collectively referred to, respectively, as “processors 212” and “memories 220.”
  • Processors 212 are configured to execute instructions stored in memories 220.
  • memory 220-2 includes a mixed effects engine 232.
  • Mixed effects engine 232 may share with, or provide features and resources to, application 222, including multiple tools associated with training and using a mixed effects model for managing a healthcare application. The user may access mixed effects engine 232 through application 222.
  • mixed effects engine 232 may be configured to create, store, update, and maintain a nonlinear model (NLM) 240, as disclosed herein.
  • NLM 240 may include a dataset tool 242, a residual learning tool 244, and a gradient boost tool 246.
  • Mixed effects engine 232 may also include a statistics tool 248.
  • mixed effects engine 232 may access one or more machine learning models stored in a training database 252.
  • Training database 252 includes training archives and other data files that may be used by mixed effects engine 232 in the training of NLM 240, according to the input of the user through application 222.
  • at least one or more training archives or machine learning models may be stored in either one of memories 220.
  • the common function, F may be a nonlinear model (e.g., random forest, deep neural network, GBT, and the like) that accounts for “fixed” effects.
  • a fixed effect may be any effect of the input variable Xi on the target value, y, that is common across the population.
  • each Gd is any model (e.g., random forest, deep neural net, GBT, and the like) that accounts for group effects introduced by each of multiple grouping structures, a, in the dataset.
  • the variables Xi and Zi may be different.
  • the variables Zi may be a subset of the variables Xi. In some embodiments, specific knowledge of the different sets of variables Xi and Zi may not be necessary.
  • the variables Xi and Zi may include the same set of variables.
  • the factor may be a normalizing factor, or a sigmoidal function to link the model in Eq. 1 to a given classification model.
  • the functions F( «) and G(*) may be learned from dataset tool 242.
  • Residual learning tool 244 may be referred to as a “mixed effects residual learning” (MERL) tool 244.
  • MERL tool 244 uses an iterative coordinate descent algorithm to fit nonlinear fixed effects model F and nonlinear random effects model G c (cf. Eq. 1).
  • MERL tool 244 initializes functions F and G c to 0.
  • MERL tool 244 fits function F on a difference (y - G) using Xi as the dependent variable in the dataset (e.g., keeping the value of G fixed).
  • MERL tool 244 fits a separate G Ci function on (y - ) using Zi as the dependent variable (and keeping the common function, F, and all the other group functions G c j fixed).
  • NLM 240 iterates MERL tool 244 to optimize functions F and G c alternating between learning F and learning G.
  • MERL tool 244 is iterated until a condition is satisfied. The condition may be the minimization of a cost function, when a change in the value of the target variable is less than a pre-selected threshold, or when the error does not improve after a certain number of iterations.
  • Gradient boost tool 246 includes a boosting schedule to optimize the convergence of functions F and G (cf. Eq. 1).
  • gradient boost tool 246 includes a series of gradient boosted trees for each of functions F and G. Accordingly, functions F and G may be a sum of weak fixed effect learners / and a series of weak group effects learners, g ? c , respectively:
  • NLM 240 may fit a series of weak learners (Eq. 2) in dataset tool 242, using any one of a pre-selected loss function, L, to minimize or reduce its gradients relative to the target variable, y.
  • a loss function, L as used in some embodiments, are illustrated in Table 1, below:
  • an alternate optimization boosting schedule may include adding fixed effect weak learners / f while fitting the fixed function F until a loss measure does not improve after a pre-selected number of iterations (e.g., a “patience” factor), then switching to adding group effect weak learners gf while fitting the group functions G (e.g., one group function G Ci at a time, or all together) until a loss measure does not improve after a pre-selected number of iterations (e.g., a “patience” factor).
  • NLM 240 is particularly effective when strong nonlinear effects affect different groups in dataset tool 242.
  • Statistics tool 248 performs error analysis based on the variance of data samples and the values obtained for target variable, y (cf. Eq. 1).
  • statistics tool 248 determines a signal-to-noise (SNR) as a ratio of the variance of the signal F + G to the variance of the added noise. In some embodiments, statistics tool 248 determines a percent random effects variance (PREV), as the ratio of the variance of the group effect G to the variance of the total signal F + G, or the percent of the variance that is due to the group effect G. In some embodiments, statistics tool 248 measures the percent of variance in the data that is unexplained by a fit to a linear model. In some embodiments, statistics tool 248 may provide to NLM 240 a fixed effects component (FE-NLM) and a group effects component (RE-NLM), separately.
  • SNR signal-to-noise
  • PREV percent random effects variance
  • statistics tool 248 measures the percent of variance in the data that is unexplained by a fit to a linear model.
  • statistics tool 248 may provide to NLM 240 a fixed effects component (FE-NLM) and a group effects component (
  • Nonlinear model 240 may include algorithms trained for the specific purposes of the engines and tools included therein.
  • the algorithms may include machine learning or artificial intelligence algorithms making use of any linear or non-linear algorithm, such as a neural network algorithm, or multivariate regression algorithm.
  • the machine learning model may include a neural network (NN), a convolutional neural network (CNN), a generative adversarial neural network (GAN), a deep reinforcement learning (DRL) algorithm, a deep recurrent neural network (DRNN), a classic machine learning algorithm such as random forest, k- nearest neighbor (KNN) algorithm, k-means clustering algorithms, or any combination thereof.
  • the machine learning model may include any machine learning model involving a training step and an optimization step.
  • training database 252 may include a training archive to modify coefficients according to a desired outcome of the machine learning model.
  • nonlinear model 240 is configured to access training database 252 to retrieve documents and archives as inputs for the machine learning model.
  • mixed effects engine 232 and one or more of dataset tool 242, residual learning tool 244, gradient boost tool 246, statistics tool 248, and at least part of training database 252 may be hosted in a different server that is accessible by server 130.
  • FIG. 3 is a chart 300 illustrating the reduction of a loss function following a repeated alternate optimization boosting schedule, according to some embodiments.
  • the abscissae 301 (X- axis) indicate the number of iterations of the gradient boost tool (e.g., gradient boost tool 246), and the ordinates (Y-axis) 302 indicate a value of the loss function as the iterations progress.
  • Loss curve 310-1 illustrates a validation loss
  • loss curve 310-2 illustrates a training loss (hereinafter, collectively referred to as “loss curves 310”).
  • the different transitions between a common function, F, minimization step and a group function, G, minimization step are illustrated.
  • more weak learner functions ( t and g ? c , respectively) are added to the description to reduce the loss (c/. Eq. 2)).
  • FIGS. 4A-4C illustrate charts 400A, 400B, and 400C (hereinafter, collectively referred to as “charts 400”) to compare the performance of various mixed effects gradient boosting schedules, according to some embodiments.
  • the abscissae 401 (X-axis) indicate the percent random effects variance (PERV), and the ordinates (Y-axis) 402 indicate the percent root mean square root error (% RMSE) reduction as the iterations progress.
  • Charts 400 include graphs 410A- 1, 410B-1, and 410C-1 (hereinafter, collectively referred to as “known group charts 410-1”) that illustrate results obtained for known, previously classified groups.
  • Charts 400 include graphs 410A-2, 410B-2, and 410C-2 (hereinafter, collectively referred to as “new group charts 410-2”) that illustrate results obtained for new, unclassified groups.
  • Chart 400A illustrates graphs 410A-1 and 410A-2 (hereinafter, collectively referred to as “graphs 410A”).
  • graphs 410A show a standard alternating boosting schedule that alternates between a single fixed-effects boost and a single group-effects boost (indicated by MEGB curve 411A-3) that has significantly degraded performance in the low PREV regime (e.g., abscissae
  • a repeating alternate optimization boosting schedule (indicated by AMEGB curve 411A-1) performs the best across different PREV regimes 401.
  • a curve 411A-4 illustrates results with a random forest with one-hot encoded group label (RF-OHE) classification strategy having a flat performance across different PERV regimes 401.
  • RF-OHE one-hot encoded group label
  • Chart 400B illustrates graphs 410B-1 and 410B-2 (hereinafter, collectively referred to as “graphs 410B”).
  • graphs 410B show a percent reduction in root-mean squared error (e.g., ordinates
  • AMEGB curve 411B-1 and MERL curve 411B-2 Some embodiments perform similarly to known models, e.g., mixed effects random forest (MERF) curve 41 IB -3, CatBoost (CB) curve 411B-4, or RF-OHE curve 411B-5, when the data is linear (e.g., a linear relationship between Xi and zt for target value y). Some embodiments have a significant improvement over known models when there is greater non-linearity in the relationship between the input variables Xi and Zi and target variable y, (which is a common occurrence). Interestingly, even for “unknown” groups (not identified or observed in the training data, graph 410B-2), embodiments as disclosed herein tend to improve accuracy (e.g., reduce, albeit slightly, % RMSE 402).
  • MEF mixed effects random forest
  • CB CatBoost
  • RF-OHE curve 411B-5 RF-OHE curve 411B-5
  • Chart 400C illustrates graphs 410C-1 and 410C-2 (hereinafter, collectively referred to as “graphs 410C”).
  • graphs 410C show performance results (e.g., % RMSE 402) for various modeling approaches, e.g., AMEGB curve 411C-1, MERL curve 411C-2, MERF curve 411C-3, CB curve 411C-4, or RF-OHE curve 411C-5, across different numbers of groups.
  • Graphs 410C show that embodiments as disclosed herein provide comparable or better accuracy than known models.
  • Each of groups 510 may include three different types of data: training data 520-1, 520-2, and 520- 3 (hereinafter, collectively referred to as “training data 520”); validation data 530-1, 530-2, 530- 3, and 530-4 (hereinafter, collectively referred to as “validation data 530”); and test data 540-1, 540-2, 540-3, and 540-4 (hereinafter, collectively referred to as “test data 540”).
  • training data 520 training data 520-1, 520-2, and 520- 3
  • validation data 530-1, 530-2, 530- 3, and 530-4 hereinafter, collectively referred to as “validation data 530”
  • test data 540-1, 540-2, 540-3, and 540-4 hereinafter, collectively referred to as “test data 540”.
  • the dataset tool may assign an inter-group split such as (0.6, 0.2, 0.2).
  • an inter-group split such as (0.6, 0.2, 0.2).
  • 60% of groups 510 are assigned to be spread across the three types (e.g., training data 520, validation data 530, and test data 540); 20% of groups 510 are assigned to each of the validation (e.g., validation data 530-4) and test sets (e.g., test data 540-4).
  • the dataset tool may assign an intra-group split such as (0.7, 0.2, 0.1), indicating that for groups 510-1, 510-2, and 510-3, 70% of the points are assigned to the training set (e.g., training data 520-1, 520-2, and 520-3).;, 20% of the points are assigned to validation data (e.g., validation data 530-1, 530-2, and 530-3);and 10% of the points are assigned to test data (e.g., test data 540-1, 540-2, and 540-3).
  • an intra-group split such as (0.7, 0.2, 0.1), indicating that for groups 510-1, 510-2, and 510-3, 70% of the points are assigned to the training set (e.g., training data 520-1, 520-2, and 520-3).;, 20% of the points are assigned to validation data (e.g., validation data 530-1, 530-2, and 530-3);and 10% of the points are assigned to test data (e.g., test data 540-1,
  • FIG. 6 illustrates a flowchart with steps in a method 600 for modeling a dataset, according to some embodiments.
  • method 600 may be performed at least partially by one or more processors executing instructions stored in a memory in a client device or server, as disclosed herein (c/. processors 212, memories 220 in client devices 110 and servers 130).
  • the instructions may be part of an application (e.g., a healthcare management application) run in a client device and/or hosted by a server that includes a mixed effects engine that includes a nonlinear model and a statistics tool (e.g., application 222, mixed effects engine 232, nonlinear model 240, and statistics tool 248).
  • an application e.g., a healthcare management application
  • a server that includes a mixed effects engine that includes a nonlinear model and a statistics tool (e.g., application 222, mixed effects engine 232, nonlinear model 240, and statistics tool 248).
  • a statistics tool e.g., application 222
  • the nonlinear model may use a dataset tool, a residual learning tool, and a gradient boost tool (e.g., dataset tool 242, residual learning tool 244, and gradient boost tool 246).
  • methods consistent with method 600 may include one or more steps in method 600, performed in a different sequence, simultaneously, or overlapping in time.
  • Step 602 includes receiving data for a population, the data including a value for one or more input variables, and a value for a target variable.
  • Step 604 includes identifying a group index for the data based on one or more categories.
  • step 604 may include selecting the one or more categories from a common characteristic in the data such as a locality, a medical condition, diagnostic, treatment, a demographic characteristic, and the like.
  • step 604 includes identifying a hospital or clinic, type of hospital or clinic, or locality of the hospital or clinic.
  • step 604 includes identifying a demographic category of an indvidual in the population.
  • step 604 includes identifying a comorbidity condition or medical condition.
  • step 604 includes identifying one or more medications.
  • step 604 includes identifying one or more events in a medical history.
  • step 604 includes identifying one or more socioeconomic categories.
  • Step 606 includes determining a common function and one or more group functions based on the input variables, wherein the common function associates a value to the target variable that is common across the population, and wherein each of the group functions associates a value to the target variable for a group in the population.
  • step 606 includes gradient boosting.
  • step 606 includes applying an alternate optimization boosting schedule.
  • Step 608 includes modeling the target variable as a combination of the common function and the one or more group functions.
  • Step 610 includes determining a new value for the target variable based on a new set of values for the input variables.
  • the value for the target variable is a healthcare outcome for an individual.
  • the value for a target variable is a healthcare cost for an individual.
  • step 610 includes predicting enrollment in a clinical trial.
  • step 610 includes predicting a response to a treatment.
  • step 610 includes predicting adherence to a treatment.
  • step 610 includes determining a new value for an individual who is not associated with the group identified by the group index.
  • FIG. 7 is a flowchart illustrating steps in a method 700 for managing healthcare with a mixed effects model of a population, according to some embodiments.
  • method 700 may be performed at least partially by one or more processors executing instructions stored in a memory in a client device or server, as disclosed herein (c/. processors 212, memories 220 in client devices 110 and servers 130).
  • the instructions may be part of an application (e.g., a healthcare management application) run in a client device and/or hosted by a server that includes a mixed effects engine that includes a nonlinear model and a statistics tool (e.g., application 222, mixed effects engine 232, nonlinear model 240, and statistics tool 248).
  • an application e.g., a healthcare management application
  • a server that includes a mixed effects engine that includes a nonlinear model and a statistics tool (e.g., application 222, mixed effects engine 232, nonlinear model 240, and statistics tool 248).
  • a statistics tool
  • the nonlinear model may use a dataset tool, a residual learning tool, and a gradient boost tool (e.g., dataset tool 242, residual learning tool 244, and gradient boost tool 246).
  • methods consistent with method 700 may include one or more steps in method 700, performed in a different sequence, simultaneously, or overlapping in time.
  • Step 702 includes receiving health data for a population, the health data including a value for one or more health monitoring variables, and a value for a health outcome.
  • the health monitoring variables may include a healthcare therapy, such as medication intake, or disease state.
  • Step 704 includes identifying groups in the population based on one or more categories.
  • Step 706 includes determining a common function and one or more group functions based on the health monitoring variables, wherein the common function associates a value to the health outcome that is common across the population and each of the group functions associates a value to the health outcome for a group in the population.
  • the method further includes modifying a feature in the healthcare therapy to improve the health outcome for a segment of the population.
  • FIG. 8 illustrates a system 800 to implement the architecture and perform the various methods illustrated in FIGS. 1 and 2, and in the flowcharts in FIGS. 6-7, according to some embodiments.
  • the computer system 800 may be implemented using hardware or a combination of software and hardware, either in a dedicated server, or integrated into another entity, or distributed across multiple entities.
  • Computer system 800 (e.g., system 200) includes a bus 808 or other communication mechanism for communicating information, and a processor 802 (e.g., processors 212) coupled with bus 808 for processing information.
  • processor 802 e.g., processors 212
  • the computer system 800 may be implemented with one or more processors 802.
  • Processor 802 may be a general-purpose microprocessor, a microcontroller, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA), a Programmable Logic Device (PLD), a controller, a state machine, gated logic, discrete hardware components, or any other suitable entity that can perform calculations or other manipulations of information.
  • DSP Digital Signal Processor
  • ASIC Application Specific Integrated Circuit
  • FPGA Field Programmable Gate Array
  • PLD Programmable Logic Device
  • Computer system 800 can include, in addition to hardware, code that creates an execution environment for the computer program in question, e.g., code that constitutes processor firmware, a protocol stack, a database management system, an operating system, or a combination of one or more of them stored in an included memory 804 (e.g., databases 152 and 252, and memories 220), such as a Random Access Memory (RAM), a flash memory, a Read-Only Memory (ROM), a Programmable Read-Only Memory (PROM), an Erasable PROM (EPROM), registers, a hard disk, a removable disk, a CD-ROM, a DVD, or any other suitable storage device, coupled to bus 808 for storing information and instructions to be executed by processor 802.
  • the processor 802 and the memory 804 can be supplemented by, or incorporated in, special purpose logic circuitry.
  • the instructions may be stored in the memory 804 and implemented in one or more computer program products, e.g., one or more modules of computer program instructions encoded on a computer-readable medium for execution by, or to control the operation of, the computer system 800, and according to any method well-known to those of skill in the art, including, but not limited to, computer languages such as data-oriented languages (e.g., SQL, dBase), system languages (e.g., C, Objective-C, C++, Assembly), architectural languages (e.g., Java, .NET), and application languages (e.g., PHP, Ruby, Perl, Python).
  • data-oriented languages e.g., SQL, dBase
  • system languages e.g., C, Objective-C, C++, Assembly
  • architectural languages e.g., Java, .NET
  • application languages e.g., PHP, Ruby, Perl, Python.
  • Instructions may also be implemented in computer languages such as array languages, aspect-oriented languages, assembly languages, authoring languages, command line interface languages, compiled languages, concurrent languages, curly-bracket languages, dataflow languages, data-structured languages, declarative languages, esoteric languages, extension languages, fourth-generation languages, functional languages, interactive mode languages, interpreted languages, iterative languages, list-based languages, little languages, logic-based languages, machine languages, macro languages, metaprogramming languages, multiparadigm languages, numerical analysis, non-English-based languages, object-oriented class-based languages, object-oriented prototype-based languages, offside rule languages, procedural languages, reflective languages, rule-based languages, scripting languages, stack-based languages, synchronous languages, syntax handling languages, visual languages, wirth languages, and xml-based languages.
  • Memory 804 may also be used for storing temporary variable or other intermediate information during execution of instructions to be executed by processor 802.
  • a computer program as discussed herein does not necessarily correspond to a file in a file system.
  • a program can be stored in a portion of a file that holds other programs or data (e.g., one or more scripts stored in a markup language document), in a single file dedicated to the program in question, or in multiple coordinated files (e.g., files that store one or more modules, subprograms, or portions of code).
  • a computer program can be deployed to be executed on one computer or on multiple computers that are located at one site or distributed across multiple sites and interconnected by a communication network.
  • the processes and logic flows described in this specification can be performed by one or more programmable processors executing one or more computer programs to perform functions by operating on input data and generating output.
  • Computer system 800 further includes a data storage device 806 such as a magnetic disk or optical disk, coupled to bus 808 for storing information and instructions.
  • Computer system 800 may be coupled via input/output module 810 to various devices.
  • Input/output module 810 can be any input/output module.
  • Exemplary input/output modules 810 include data ports such as USB ports.
  • the input/output module 810 is configured to connect to a communications module 812.
  • Exemplary communications modules 812 e.g., communications modules 218) include networking interface cards, such as Ethernet cards and modems.
  • input/output module 810 is configured to connect to a plurality of devices, such as an input device 814 (e.g., a keyboard, a mouse, a pointer, a touchscreen display, a microphone, a webcam, and the like) and/or an output device 816 (e.g., a display, a touchscreen display, a speaker, and the like).
  • exemplary input devices 814 include a keyboard and a pointing device, e.g., a mouse or a trackball, by which a user can provide input to the computer system 800.
  • Other kinds of input devices 814 can be used to provide for interaction with a user as well, such as a tactile input device, visual input device, audio input device, or brain-computer interface device.
  • feedback provided to the user can be any form of sensory feedback, e.g., visual feedback, auditory feedback, or tactile feedback; and input from the user can be received in any form, including acoustic, speech, tactile, or brain wave input.
  • exemplary output devices 816 include display devices, such as an LCD (liquid crystal display) monitor, for displaying information to the user.
  • the client and server can be implemented using a computer system 800 in response to processor 802 executing one or more sequences of one or more instructions contained in memory 804. Such instructions may be read into memory 804 from another machine-readable medium, such as data storage device 806. Execution of the sequences of instructions contained in main memory 804 causes processor 802 to perform the process steps described herein. One or more processors in a multi-processing arrangement may also be employed to execute the sequences of instructions contained in memory 804. In alternative aspects, hard-wired circuitry may be used in place of or in combination with software instructions to implement various aspects of the present disclosure. Thus, aspects of the present disclosure are not limited to any specific combination of hardware circuitry and software.
  • a computing system that includes a back-end component, e.g., a data server, or that includes a middleware component, e.g., an application server, or that includes a front-end component, e.g., a client computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the subject matter described in this specification, or any combination of one or more such back-end, middleware, or front-end components.
  • the components of the system can be interconnected by any form or medium of digital data communication, e.g., a communication network.
  • the communication network can include, for example, any one or more of a LAN, a WAN, the Internet, and the like. Further, the communication network can include, but is not limited to, for example, any one or more of the following tool topologies, including a bus network, a star network, a ring network, a mesh network, a star-bus network, tree or hierarchical network, or the like.
  • the communications modules can be, for example, modems or Ethernet cards.
  • Computer system 800 can include clients and servers.
  • a client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.
  • Computer system 800 can be, for example, and without limitation, a desktop computer, laptop computer, or tablet computer.
  • Computer system 800 can also be embedded in another device, for example, and without limitation, a mobile telephone, a PDA, a mobile audio player, a Global Positioning System (GPS) receiver, a video game console, and/or a television set top box.
  • GPS Global Positioning System
  • machine-readable storage medium or “computer-readable medium” as used herein refers to any medium or media that participates in providing instructions to processor 802 for execution. Such a medium may take many forms, including, but not limited to, non-volatile media, volatile media, and transmission media.
  • Non-volatile media include, for example, optical or magnetic disks, such as data storage device 806.
  • Volatile media include dynamic memory, such as memory 804.
  • Transmission media include coaxial cables, copper wire, and fiber optics, including the wires forming bus 808.
  • Machine -readable media include, for example, floppy disk, a flexible disk, hard disk, magnetic tape, any other magnetic medium, a CD- ROM, DVD, any other optical medium, punch cards, paper tape, any other physical medium with patterns of holes, a RAM, a PROM, an EPROM, a FLASH EPROM, any other memory chip or cartridge, or any other medium from which a computer can read.
  • the machine-readable storage medium can be a machine-readable storage device, a machine-readable storage substrate, a memory device, a composition of matter affecting a machine -readable propagated signal, or a combination of one or more of them.
  • a method may be an operation, an instruction, or a function and vice versa.
  • a claim may be amended to include some or all of the words (e.g., instructions, operations, functions, or components) recited in either one or more claims, one or more words, one or more sentences, one or more phrases, one or more paragraphs, and/or one or more claims.
  • the phrase “at least one of’ preceding a series of items, with the terms “and” or “or” to separate any of the items, modifies the list as a whole, rather than each member of the list (e.g., each item).
  • the phrase “at least one of’ does not require selection of at least one item; rather, the phrase allows a meaning that includes at least one of any one of the items, and/or at least one of any combination of the items, and/or at least one of each of the items.
  • phrases “at least one of A, B, and C” or “at least one of A, B, or C” each refer to only A, only B, or only C; any combination of A, B, and C; and/or at least one of each of A, B, and C.
  • exemplary is used herein to mean “serving as an example, instance, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments. Phrases such as an aspect, the aspect, another aspect, some aspects, one or more aspects, an implementation, the implementation, another implementation, some implementations, one or more implementations, an embodiment, the embodiment, another embodiment, some embodiments, one or more embodiments, a configuration, the configuration, another configuration, some configurations, one or more configurations, the subject technology, the disclosure, the present disclosure, other variations thereof and alike are for convenience only and do not imply that a disclosure relating to such phrase(s) is essential to the subject technology or that such disclosure applies to all configurations of the subject technology.
  • a disclosure relating to such phrase(s) may apply to all configurations, or one or more configurations.
  • a disclosure relating to such phrase(s) may provide one or more examples.
  • a phrase such as an aspect or some aspects may refer to one or more aspects and vice versa, and this applies similarly to other foregoing phrases.
  • a reference to an element in the singular is not intended to mean “one and only one” unless specifically stated, but rather “one or more.”
  • Pronouns in the masculine include the feminine and neuter gender (e.g., her and its) and vice versa.
  • the term “some” refers to one or more.
  • Underlined and/or italicized headings and subheadings are used for convenience only, do not limit the subject technology, and are not referred to in connection with the interpretation of the description of the subject technology. Relational terms such as first and second and the like may be used to distinguish one entity or action from another without necessarily requiring or implying any actual such relationship or order between such entities or actions.
  • Embodiments as disclosed herein may include any one of the following.
  • Embodiment I A computer-implemented method includes receiving data from a population that includes a value for one or more input variables, and a value for a target variable; identifying a group index for the data based on one or more categories, determining a common function and one or more group functions, wherein the common function associates a value to the target variable that is common across the population and wherein each of the group functions associates a value to the target variable for a group in the population, modeling the target variable as a combination of the common function and the one or more group functions; and determining a new value for the target variable based on a new set of values for the input variables.
  • Embodiment II A computer-implemented method includes receiving health data for a population, the health data including a value for one or more health monitoring variables, and a value for a health outcome, identifying groups in the population based on one or more categories, determining a common function and one or more group functions based on the health monitoring variables, wherein the common function associates a value to the health outcome that is common across the population and each of the group functions associates a value to the health outcome for a group in the population, and updating a database to associate the health outcome to a new set of values for the health monitoring variables according to the common function.
  • Embodiments as disclosed herein may further include the features in Embodiments I and II, combined with any one or more of the following elements:
  • Element 1 wherein the group index identifies a locality, a medical condition, diagnostic, medical intervention, or a demographic category.
  • Element 2 wherein the determining a common function and one or more group functions includes gradient boosting.
  • Element 3 wherein the value for the target variable is a healthcare outcome for an individual.
  • Element 4 wherein the value for a target variable is a healthcare cost for an individual.
  • Element 5 wherein the group index identifies a hospital or clinic, type of hospital or clinic, or locality of the hospital or clinic.
  • Element 6 wherein determining a new value for the target variable includes predicting a healthcare outcome.
  • Element 7 wherein the group index identifies a demographic category.
  • Element 8 wherein determining a new value for the target variable includes predicting the likelihood of a disease for an individual.
  • the group index identifies a comorbidity or medical condition.
  • the group index identifies one or more medications.
  • the group index identifies events in a medical history.
  • the group index identifies a socioeconomic category.
  • the determining a new value for the target variable includes predicting enrollment in a clinical trial.
  • Element 14 wherein the determining a new value for the target variable includes predicting response to a treatment.
  • Element 15 wherein the determining a new value for the target variable includes predicting adherence to a treatment.
  • Element 16 wherein the health monitoring variables include a healthcare therapy or disease state.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Artificial Intelligence (AREA)
  • Computational Linguistics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Measuring And Recording Apparatus For Diagnosis (AREA)

Abstract

L'invention concerne un procédé de modélisation du comportement d'effets mixtes d'une variable cible. Le procédé consiste à recevoir des données d'une population qui comprend une valeur pour une ou plusieurs variables d'entrée, et une valeur pour une variable cible, à identifier un indice de groupe pour les données de santé sur la base d'une ou de plusieurs catégories, à déterminer une fonction commune et une ou plusieurs fonctions de groupe, la fonction commune associant une valeur à la variable cible à travers la population et chaque fonction de groupe associant une valeur à la variable cible pour un groupe dans la population, à modéliser la variable cible avec la fonction commune et la ou les fonctions de groupe, et à déterminer une nouvelle valeur pour la variable cible. L'invention concerne également un système ainsi qu'un support non transitoire lisible par ordinateur stockant une instruction pour amener le système à mettre en œuvre le procédé ci-dessus.
PCT/US2022/014150 2021-01-27 2022-01-27 Moteur pour applications d'apprentissage machine à effets mixtes WO2022165071A1 (fr)

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
US202163142428P 2021-01-27 2021-01-27
US63/142,428 2021-01-27
US202163145945P 2021-02-04 2021-02-04
US63/145,945 2021-02-04

Publications (1)

Publication Number Publication Date
WO2022165071A1 true WO2022165071A1 (fr) 2022-08-04

Family

ID=82653936

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2022/014150 WO2022165071A1 (fr) 2021-01-27 2022-01-27 Moteur pour applications d'apprentissage machine à effets mixtes

Country Status (1)

Country Link
WO (1) WO2022165071A1 (fr)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180096105A1 (en) * 2009-09-24 2018-04-05 Optum, Inc. Data processing systems and methods implementing improved analytics platform and networked information systems
US20200082941A1 (en) * 2018-09-11 2020-03-12 Hitachi, Ltd. Care path analysis and management platform
US20200118691A1 (en) * 2018-10-10 2020-04-16 Lukasz R. Kiljanek Generation of Simulated Patient Data for Training Predicted Medical Outcome Analysis Engine

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180096105A1 (en) * 2009-09-24 2018-04-05 Optum, Inc. Data processing systems and methods implementing improved analytics platform and networked information systems
US20200082941A1 (en) * 2018-09-11 2020-03-12 Hitachi, Ltd. Care path analysis and management platform
US20200118691A1 (en) * 2018-10-10 2020-04-16 Lukasz R. Kiljanek Generation of Simulated Patient Data for Training Predicted Medical Outcome Analysis Engine

Similar Documents

Publication Publication Date Title
US11636847B2 (en) Ontology-augmented interface
US11257579B2 (en) Systems and methods for managing autoimmune conditions, disorders and diseases
US10332631B2 (en) Integrated medical platform
US20210027896A1 (en) Learning platform for patient journey mapping
US20220319646A1 (en) Machine learning engine and rule engine for document auto-population using historical and contextual data
US20230368070A1 (en) Systems and methods for adaptative training of machine learning models
US20230343468A1 (en) Automated individualized recommendations for medical treatment
Panesar et al. Artificial intelligence and machine learning in global healthcare
WO2017007461A1 (fr) Plateforme médicale intégrée
WO2023164312A1 (fr) Appareil de classification de candidats à des publications et son procédé d'utilisation
US11275903B1 (en) System and method for text-based conversation with a user, using machine learning
WO2022165071A1 (fr) Moteur pour applications d'apprentissage machine à effets mixtes
US11314488B2 (en) Methods and systems for automated screen display generation and configuration
WO2022146930A1 (fr) Système et procédé de prédiction de résultats cliniques et d'optimisation d'interventions de santé
La Cava et al. A flexible symbolic regression method for constructing interpretable clinical prediction models
Ratan Applied Machine Learning for Healthcare and Life Sciences Using AWS: Transformational AI implementations for biotech, clinical, and healthcare organizations
US20240062885A1 (en) Systems and methods for generating an interactive patient dashboard
US20220375549A1 (en) Method and system of generating, delivering and displaying cross platform personalized digital software applications
US11996196B2 (en) Intelligent computer application for diagnosis suggestion and validation
US11847616B2 (en) Apparatus for wage index classification
US20230368148A1 (en) Apparatus for automatic posting acceptance
Reyana et al. Emergence of decision support systems in healthcare
US20220172837A1 (en) Intelligent Computer Application For Diagnosis Suggestion And Validation
Marmolejo-Ramos et al. Distributional regression analysis of learning analytics and educational data
WO2022032096A1 (fr) Systèmes et procédés de normalisation d'ensembles de données d'apprentissage machine

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22746628

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 231123)