Machine learning-based prediction of migraine attacks

Abstract

Onset and / or continuation of a migraine attack is predicted in a subject using a machine learning model. Subject health data are accessed with a computer system, where the subject health data include clinical test or measurement data received from the subject and / or subject symptom data received from the subject. A trained machine learning model is accessed with the computer system, where the trained machine learning model has been trained on training data to predict a likelihood of migraine attack occurring within a specified timeframe based on features in subject health data. The subject health data are input to the trained machine learning model using the computer system, generating classified feature data as an output. The classified feature data indicate a likelihood of the subject having a migraine attack within the specified timeframe.

Description

MACHINE LEARNING-BASED PREDICTION OF MIGRAINE ATTACKSCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of U.S. Provisional Patent Application Serial No. 63 / 570,519, filed on March 27, 2024. and entitled “MACHINE LEARNING-BASED PREDICTION OF MIGRAINE ATTACKS,” which is herein incorporated by reference in its entirety.BACKGROUND

[0002] Migraine is a chronic neurological disorder with recurrent attacks of neurological symptoms including head pain, cognitive impairment, hypersensitivities to sensory stimuli, nausea, and vomiting. A migraine includes overlapping phases of prodrome (commonly referred to as premonitory), aura, headache, and postdrome. Prodrome and postdrome symptoms are common, affecting up to 80% of the migraine population. Prodrome symptoms can occur up to 48 hours before a migraine headache attack; however, the most prominent prodromal change exists within the last 12 hours before attack onset. Current standard of care for treating migraine attacks is a reactive treatment strategy, treating once full symptoms are present. Migraine pain peaks within 30 minutes in most subjects, and acute treatment outcomes are suboptimal as pain intensity increases. Suboptimal acute treatment has been shown to be an independent risk factor for progression from episodic to chronic migraine. Proactive treatment, treating prior to full symptom onset, would likely provide superior outcomes by: reducing symptom burden and associated disability; and improving treatment outcomes as early intervention in an attack has been shown to significantly improve treatment response.

[0003] The ability to forecast migraine attacks could offer subjects with migraine opportunities for early intervention and activity planning.SUMMARY OF THE DISCLOSURE

[0004] According to an aspect of the present disclosure, a method for predicting a migraine attack in a subject is provided. The method includes accessing subject health data with a computer system, wherein the subject health data comprise at least one of clinical test or measurement data received from the subject or subject symptom data received from the subject. The method also includes accessing a trained machine learning model with the computer system, wherein the trained machine learning model has been trained on training data to predict a likelihood of migraine attack occurring or continuing within a specified timeframebased on features in subject health data. The method further includes inputting the subject health data to the trained machine learning model using the computer system, generating classified feature data as an output, wherein the classified feature data indicate a likelihood of the subject having a migraine attack within the specified timeframe. The method also includes outputting the classified feature data using the computer system.

[0005] According to another aspect of the present disclosure, a method for predicting a migraine attack in a subject is provided. The method includes accessing subject health data with a computer system, wherein the subject health data comprise rapid automatized naming (RAN) test score data received from the subject and subject symptom data received from the subject. The method also includes accessing a trained machine learning model with the computer system, wherein the trained machine learning model has been trained on training data to predict a likelihood of migraine attack occurring or continuing within a specified timeframe based on features in subject health data. The method further includes inputting the subject health data to the trained machine learning model using the computer system, generating classified feature data as an output, wherein the classified feature data indicate a likelihood of the subject having a migraine attack within the specified timeframe. The method also includes outputting the classified feature data using the computer system.

[0006] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing instructions that, when executed by one or more processors, cause the one or more processors to perform operations is provided. The operations include accessing subject health data, wherein the subject health data comprise at least one of clinical test data received from a subject or subject symptom data received from the subject. The operations also include accessing a trained machine learning model, wherein the trained machine learning model has been trained on training data to predict a likelihood of migraine attack occurring or continuing within a specified timeframe based on features in subject health data. The operations further include inputting the subject health data to the trained machine learning model, generating classified feature data as an output, wherein the classified feature data indicate a likelihood of the subject having a migraine attack within the specified timeframe. The operations also include outputting the classified feature data.

[0007] According to another aspect of the present disclosure, a system for predicting a migraine attack in a subject is provided. The system includes an input to receive subject health data, wherein the subject health data comprise at least one of clinical test data received from the subject or subject symptom data received from the subject. The system also includes amemory storing a trained machine learning model, wherein the trained machine learning model has been trained on training data to predict a likelihood of migraine attack occurring or continuing within a specified timeframe based on features in subject health data. The system further includes a processor in communication with the input and the memory', the processor configured to receive the subject health data from the input, receive the trained machine learning model from the memory, and to input the subject health data to the trained machine learning model to generate classified feature data as an output, wherein the classified feature data indicate a likelihood of the subject having a migraine attack within the specified timeframe. The system also includes an output to output the classified feature data.BRIEF DESCRIPTION OF THE DRAWINGS

[0008] FIG. 1 is a flowchart of an example method for generating classified feature data that predict an onset and / or continuation of a migraine attack in a subject.

[0009] FIG. 2 is a flowchart of an example data collection protocol.

[0010] FIGS. 3A-3D illustrate examples of rapid automatized naming (RAN) tests administered to a subject.

[0011] FIGS. 4A-4D illustrate an example subject questionnaire workflow.

[0012] FIG. 5 is a flowchart of an example method for training a machine learning model on training data to predict migraine attack onset and / or continuation based on subject health data.

[0013] FIG. 6 shows example RAN test score data.

[0014] FIG. 7 shows example variable importance values for subject health data features.

[0015] FIG. 8 is a block diagram of an example system for predicting onset and / or continuation of a migraine attack in a subject.

[0016] FIG. 9 is a block diagram of example components that can implement the system of FIG. 8.DETAILED DESCRIPTION

[0017] Described here are systems and methods for predicting the onset and / or continuation of a migraine attack in a subject using a machine learning-based framework that analyzes subject health data received from the subject. For instance, the disclosed systems and methods can forecast impending and / or continuation of migraine attacks based on subject health data, which may include clinical test data (e.g., rapid automatized naming (RAN) testscores), wearable data, subject symptom data (e.g., symptoms reported by a subject via a questionnaire and / or diary), subject demographic data, combinations thereof, and the like.

[0018] Referring now to FIG. 1, a flowchart is illustrated as setting forth the steps of an example method for generating classified feature data using a suitably trained machine learning model. As will be described, the machine learning model takes subject health data as input data and generates classified feature data as output data. As an example, the classified feature data can be indicative of a prediction of migraine attack onset and / or continuation in a subject from whom the subject health data have been recorded.

[0019] The method includes accessing subject health data with a computer system, as indicated at step 102. Accessing the subject health data may include retrieving such data from a memory or other suitable data storage device or medium. Additionally, or alternatively, accessing the subject health data can include acquiring subject health data (e.g., directly or indirectly) from a subject and recording the subject health data with the computer system, or otherwise transferring the subject health data to the computer system. As one non-limiting example, subject health data may be retrieved in whole or in part from one or more data sources, which may include electronic health records for the subject. FIG. 2 illustrates an example flowchart for collecting subject health data from a subject at multiple times per day on a daily basis.

[0020] The subj ect health data can include clinical test data, subj ect symptom data, and subject demographic data, amongst other types of subject health data or other relevant data. In some implementations, the subject health data may include only a subset of the data used to train the machine learning model. For instance, the machine learning model may be finetuned, retrained, or otherwise updated based on subject health data and classified feature data generated by the model. As the machine learning model is finetuned or otherwise retrained on these data, the updated machine learning model may be able to generate accurate predictions on fewer or updated inputs (e.g., only a RAN test score, a RAN test score and one or two subject-reported symptoms, with or without wearable data, etc.). Additionally, other machine learning techniques may also be employed to improve the model performance, including unsupervised machine learning techniques to identify subject phenotypic clusters, reinforcement learning techniques designed to improve the accuracy of the prediction for a given subject, incorporating other health foundation models in the model training to optimze the model performance, and / or employing other machine learning or deep learning techniques.

[0021] As a non-limiting example, the subject health data may include one or more of detailed baseline questionnaire data. RAN test scores, and headache diary symptoms recorded at multiple timepoints by the subject. For example, a subject can complete one or more baseline headache questionnaires and may perform one or more recordings to get mean or baseline RAN test scores. Additionally or alternatively, at a timepoint when the subject wants to know whether they will have a migraine attack within a specified timeframe (e.g., in the next 6-12 hours), the subject can record headache symptoms they are experiencing at that time and perform one or more RAN tests. To obtain RAN-related data, such as RAN test scores at the first trial, the number of errors made in the first trial, the number of trials attempted before a trial without any errors, and / or the RAN test score of the trial in which no errors were made. These subject health data may then be inputted in the machine learning model to generate a prediction migraine onset and / or continuation within the specified timeframe.

[0022] As noted above, the clinical test data can include RAN test scores recorded at one or more different time points by the subject. A RAN test may evaluate saccadic eye movements, attention, and information processing. RAN tests may be highly sensitive to altered cerebral function. The RAN test may be a rapid number naming test, a rapid letter naming test, a rapid color naming test, and / or a rapid object naming test. As one non-limiting example, the RAN test may be a rapid number naming test, such as a King-Devick test (King- Devick Technologies, Inc.; Oak Brook, IL), a rapid letter naming test, such as an Acadience RAN (Acadience Learning, Inc.; Massachusetts, MA), or the like. In a RAN test, participants are asked to read the numbers, letters, colors, and / or objects on test cards from left to right as quickly as possible without making errors. FIGS. 3A-3F illustrate examples of acquiring RAN test score data from a subject. As shown in FIG. 3 A, a RAN test can be presented to a user on computing device, such as a tablet computer, computer system, or the like. FIGS. 3B-3D are examples of different RAN tests that may be presented to a user. Other clinical test data include wearable data that records saccadic eye movement, heart rate, or speech with or without asking subjects to conduct a test.

[0023] Clinical test data related to the RAN test can include, for example: time with zero errors, time of first trial, errors of first trial, number of trials to zero errors, and so on, as illustrated in Table 1. Multiple clinical test scores can be recorded each day for a subject. For example, the clinical test data may include RAN test scores recoded multiple times a day on a daily basis, such as in the morning, in the afternoon, and in the evening, and also if there was an impending migraine attack outside of these three time windows. The clinical test data canadditionally or alternatively include data from one or more clinical tests, such as migraine subjective cognitive impairment (Mig-SCog) test and / or Utah Photophobia Symptom Impact Scale (UPSIS-17) test, as described in Cortez MM, Digre K, Uddin D, Hung M, Blitzer A, Bounsanga J, Voss MW, Katz BJ. Validation of a photophobia symptom impact scale. Cephalalgia. 2019 Oct;39(ll): 1445-1454, which is herein incorporated by reference in its entirety.Table 1: Example RAN test dataRAN Test ResultsDate / Time of RAN TestTime with zero errors Time of first trial Errors of first trialNumber of trials to zero errors

[0024] The subject symptom data can include self-reported symptoms and / or symptoms recorded in, or derived from, the subject’s electronic medical record (EMR) data, electronic health record (EHR) data, or the like. For example, subject symptom data may include questionnaire response data included symptoms reported by the subject when answering one or more questionnaires. As a non-limiting example, the subject symptom data can include baseline headache characteristics reported by the subject, including headache frequency, headache duration, headache location, headache quality, headache intensity, associated symptoms, aura, number of years with migraine, current preventive and acute therapies, and the like. Headache frequency can, for example, include a qualitative or quantitative measurement of how frequently the subject suffered headaches symptoms over an interval of time. As an example, headache frequency can be computed as the number of days with a headache captured over the trailing 28 days and compared to baseline. A non-limiting list of example questions that can be presented to a subject as part of a headache symptom questionnaire is provided in Table 2. The subject symptom data may include answ ers to one or more of these questions, or other questions not listed in Table 2. An example questionnaire workflow is illustrated in FIGS. 4A-4D.Table 2: List of Subject Symptom Data PointsBaseline Headache CharacteristicsHow many years have you had migraines for?How many days per month (28 days) with headache of any kind / severity?How many days per month (28 days) with complete headache freedom?How many days per month (28 days) with migraine headache?How long do headaches last if untreated / inadequately treated?Where are they usually located?Unilateral locationQualityIntensityAverage PainMaximum PainHeadaches worse with physical activityHeadaches worse with mental activity NauseaVomitingSensitivity to lightSensitivity to soundConjunctival injection TearingNasal congestion / rhinorrheaEyelid drooping Auras with headachesWhen was your first headache (month and year) Family history of migraines If yes, whoCurrently taking medications If yes, list medications to prevent headachesDays per month (28 days) taking abortive medicationsList Abortive MedicationBaseline Prodrome Do you get premonitory / prodrome symptoms (i.e., symptoms beforeCharacteristics headache)?With what percentage of your migraines do you get prodrome symptoms?Tired / weary / fatigue / loss of energyHyperactiveMood changesAnxiousIrritable / IntolerantDifficulty with thoughtsDifficulty with reading or writingDifficulty speakingDifficulty with concentration Light sensitivityBlurred vision / vision changesSound sensitivityOdor sensitivity / distortionsSkin sensitivity Temperature changes (e.g. chills, sweats)Facial flushing or pale faceNeck stiffnessMuscle PainDizziness / vertigo / lightheadedYawningConstipation / DiarrheaIncreased urinationAbdominal painHunger / food cravingsLack of appetiteThirstWhich are your most common prodrome symptoms?Baseline Do you have postdrome symptoms (i.e. symptoms after headache isPostdrome over)?CharacteristicsWith what percentage of your migraines do you have postdrome symptoms?Tired / weary / fatigue / loss of energyHyperactiveMood changesAnxiousIrritable / IntolerantDifficulty with thoughtsDifficulty with reading or writingDifficulty speakingDifficulty with concentrationLight sensitivity'Blurred vision / vision changes Sound sensitivityOdor sensitivity / distortionsSkin sensitivityTemperature changes (e.g. chills, sweats)Facial flushing or pale faceNeck stiffnessMuscle PainDizziness / vertigo / lightheadedYawningConstipation / DiarrheaIncreased urinationNausea / vomitingAbdominal painHunger / food cravingsLack of appetiteThirstWhich are your most common prodrome symptoms?Personal Medical Current Medical ConditionsHistoryPast MedicalCondi tions / SurgeriesCurrent Medication / TreatmentMedications and (Include dose, frequency.Other Treatments start date, end date)Prior Migraine Medication / Treatment (Include dose, frequency, months of use)TreatmentsMigraine Scog Do you feel confused?During your Do you have trouble performing tasks at your normal speed? headaches...Is it difficult to follow a route or path (diving or walking)?Do you have trouble thinking?Do you have trouble maintaining a line of thought?Is it difficult to understand words spoken to you?Is it difficult to organize a sentence or a conversation?Do you have trouble speaking out other people's names?Is it difficult to remember the correct name of objects?Symptom DiaryState of Health Current Date / TimeMark your current state of healthAre you on your period now?Is this the first time you accessed the questionnaire TODAY?Sleep State Rate the quality7of sleep last nig< ? .ht?What time did you go to bed(Date / Time)What time did you wake up(Date / Time)?How many hours of sleep did you get?Did you have a headache last evening / night?If you had a headache last evening / night, did it end before you went to sleep?Headache State Do you have a headache?Is it a migraine?Indicate your Feeling a migraine coming current state: onRecovering from a migraineDo not feel a migraine coming onDuring the headache:Having a Migraine If yes: What time did the headache start?What is the maximum headache severity?Where is the headache located:Did you have any7of the Tired / weary / fatigue / loss of energy' following symptoms?HyperactiveMood changesAnxiousIrritable / IntolerantDifficulty7with thoughtsDifficulty with reading or writingDifficulty speakingDifficulty' with concentrationLight sensitivityBlurred vision / vision changesSound sensitivity’Odor sensitivity / distortionsSkin sensitivityTemperature changes (e.g. chills, sweats)Facial flushing or pale faceNeck stiffnessMuscle PainDizziness / vertigo / lightheadedYawningConstipation / Diarrhea Increased urination Nausea / vomiting Abdominal painHunger / food cravings Lack of appetite ThirstDuring the headache: Was it throbbing or pulsing?Do lights bother you more than usual? Do sounds bother you more than usual? Are you nauseated?Have you vomited?Does routine physical activity (e.g. walking, climbing stairs) make your headache worse?Did you have an aura?Did you take medication to treat this headache?Having a Headache If yes: What time did the headache start?What is the maximum headache severity?Where is the headache located:Did you have any of the Tired / weary / fatigue / loss of energy following symptoms?Hyperactive Mood changes AnxiousIrritable / Intolerant Difficulty with thoughtsDifficulty with reading or writingDifficulty speakingDifficulty with concentrationLight sensitivity Blurred vision / vision changes Sound sensitivityOdor sensitivity / distortions Skin sensitivityTemperature changes (e.g. chills, sweats)Facial flushing or pale face Neck stiffnessMuscle PainDizziness / vertigo / lightheadedYawningConstipation / Diarrhea Increased urination Nausea / vomiting Abdominal painHunger / food cravings Lack of appetite ThirstDuring the headache: Was it throbbing or pulsing?Do lights bother you more than usual? Do sounds bother you more than usual? Are you nauseated?Have you vomited?Does routine physical activity (e.g. walking, climbing stairs) make your headache worse?Did you take medication to treat this headache?Feeling a Migraine How likely do you feel you are to get a migraine? coming onDo you have any of the Tired / weary / fatigue / loss of energy’ following symptoms?Hyperactive Mood changes AnxiousIrritable / IntolerantDifficulty with thoughts Difficulty’ with reading or writingDifficulty' speaking Difficulty with concentrationLight sensitivity Blurred vision / vision changes Sound sensitivityOdor sensitivity / distortions Skin sensitivity’Temperature changes (e.g. chills, sweats)Facial flushing or pale face Neck stiffnessMuscle PainDizziness / vertigo / lightheaded YawningConstipation / Diarrhea Increased urinationNausea / vomitingAbdominal pain Hunger / food cravings Lack of appetite ThirstWhat time did the symptoms begin?Do you believe these symptoms indicate a migraine is coming on? Did you take medication to treat these symptoms?Recovering from a When did your headache end: migraineDo you have symptoms Tired / weary / fatigue / loss of energy relating to recovering from a migraine?Hyperactive Mood changes AnxiousIrritable / Intolerant Difficult)' with thoughtsDifficulty with reading or writingDifficulty speaking Difficulty with concentrationLight sensitivity Blurred vision / vision changes Sound sensitivityOdor sensitivity / distortions Skin sensitivityTemperature changes (e.g. chills, sweats)Facial flushing or pale face Neck stiffnessMuscle PainDizziness / vertigo / lightheaded YawningConstipation / Diarrhea Increased urination Nausea / vomiting Abdominal painHunger / food cravings Lack of appetite ThirstDid you take medication to treat these symptoms?

[0025] In some instances, the subject symptom data may also include symptoms reported by the subject in a headache symptom diary, or the like. Examples of additional subject symptom data may include state of health, current state of health, menstruation history or status,sleep state, quality of sleep, bed time, wake time, hours of sleep, headache in evening / ov emight, headache state, presence of headache, presence of migraine, headache questionnaire response data, migraine questionnaire response data, prodrome questionnaire response data, and / or postdrome questionnaire response data. These subject symptom diary response can be recorded multiple times a day on a daily basis, such as in the morning, in the afternoon, and in the evening, and also if there was an impending migraine attach outside of these three time windows.

[0026] The demographic data can include subject age, subject gender, subject race, subject ethnicity, and / or subject handedness.

[0027] Additionally or alternatively, the subject health data may include other types of subject health data, including data stored in, retrieved from, extracted from, or otherwise derived from the subject’s EMR and / or EHR. The subject health data can include unstructured text, subject symptom data (e.g., physical symptoms, mental health symptoms), questionnaire response data, clinical test data, clinical laboratory data, histopathology data, demographic data, genetic sequencing, medical imaging, data from wearable devices (e.g., physiological measurements or other data recorded with a wearable device), and other such clinical data ty pes. Physiological measurements that may be recorded with a wearable device include heart rate, heart rate variability', temperature, electrophysiological signals (or brain waves), or other physical parameters. Electrophysiological signals may include electroencephalography signals, electromyography signals, electrocardiography signals, electrooculography signals, and the like. Other data that may be recorded with a wearable device may include measurements of stress, measurements of sleep state and / or sleep quality', video recording or other measurements of eye movement, and the like. Additionally, the subject health data may include location data and / or environmental factors.

[0028] As one non-limiting example, subject health data can include clinical features associated with information derived from clinical records of a subject, which in some instances can also include records from family members of the subject. These clinical features and data may be abstracted from unstructured clinical documents, EMR, EHR. or other sources of subject history. Such data may include subject symptoms, diagnosis, treatments, medications, therapies, responses to treatments, laboratory testing results, medical history7, geographic locations, environmental factors, demographics, or other features of the subject which may be found in the subject's EMR and / or EHR.

[0029] Additionally or alternatively, subject health data can include clinical features derived from structured, curated. EMR and / or EHR data, such as diagnoses; symptoms; therapies; outcomes; subject demographics, such as subject name, date of birth, gender, and / or ethnicity; diagnosis dates for , illness, disease, or other physical or mental conditions; personal medical hi story ; family medical history; clinical diagnoses, such as date of initial diagnosis, ; and the like. Additionally, the subject health data may also include features such as treatments and outcomes, such as line of therapy, therapy groups, clinical trials, medications prescribed or taken, surgeries, , imaging, adverse effects, and associated outcomes.

[0030] Examples of clinical laboratory data and / or histopathology data can include genetic testing and laboratory information, such as performance scores, lab tests, pathology results, prognostic indicators, date of genetic testing, testing method used, and so on.

[0031] In some embodiments, the subject health data can include a collection of data and / or features including all of the datatypes disclosed above. Alternatively, the subject health data may include a selection of fewer data and / or features.

[0032] In some implementations, the subject health data may be preprocessed before being input to the machine learning model. As an example, RAN test scores, headache symptoms diary data, and baseline demographics data can be mapped according to the subjects and the time at which the data were recorded.

[0033] A trained machine learning model is then accessed with the computer system, as indicated at step 104. In general, the machine learning model is trained, or has been trained, on training data in order to generate classified feature data that indicate a prediction of migraine attack onset and / or continuation in a subject. The machine learning model may include a gradient boosting machine (GBM) model, a distributed random forest (DRF) model, a generalized linear model (GLM) model, an XGBoost model, a stacked ensemble model, a convolutional neural network, a residual neural network, or the like. Alternatively, the machine learning model could implement other suitable machine learning or artificial intelligence algorithms, such as those based on supervised learning, unsupervised learning, deep learning, ensemble learning, dimensionality reduction, combinations thereof, and so on. In some instances, large language models or other foundation models (e.g., transformer-based models) may also be incorporated to analyze direct text inputs from subjects.

[0034] Accessing the trained machine learning model may include accessing model parameters that have been optimized or otherwise estimated by training the machine learning model on training data. In some instances, retrieving the machine learning model can alsoinclude retrieving, constructing, or otherwise accessing the particular machine learning model architecture to be implemented. For instance, data pertaining to the layers in a neural network architecture (e.g., number of layers, type of layers, ordering of layers, connections between layers, hyperparameters for layers) may be retrieved, selected, constructed, or otherwise accessed.

[0035] The subj ect health data are then input to the machine learning model, generating output as classified feature data, as indicated at step 106. As described above, the subject health data may include one or more of clinical test data, subject symptom data, and subject demographic data, amongst other types of subject health data or other relevant data. For example, the classified feature data may include a risk score. The risk score can provide physicians, other clinicians, and / or the subject with a recommendation to consider additional monitoring for subjects whose subject health data indicate the likelihood of the subject suffering from a migraine attack within a specified timeframe (e.g., the next 6-12 hours).

[0036] As another example, the classified feature data may indicate the probability for a particular classification; that is, the probability that the subject health data include patterns, features, or characteristics indicative of the subject having a migraine attack within a specified time frame (e.g., the next 6-12 hours). In some instances, a preset probability threshold may be used, such that the classified feature data indicate a binary outcome (e.g., yes the subject will have a migraine attack within the specified time frame, no the subject will not have a migraine attack within the specified time frame), or as a classifier to predict different migraine states / phases (e g., prodrome, migraine attacks, postdrome, non-migraine headache, interictal baseline) a subject might be in within a specific timeframe. In other instances, the classified feature data may include a probability score or percentage that the subject will have a migraine attack within the specified timeframe.

[0037] Additionally or alternatively, the classified feature data may classify the subject health data as indicating a particular medical condition. In these instances, the classified feature data can differentiate between different medical conditions. In still other embodiments, the classified feature data may indicate a severity of a medical condition. For example, the classified feature data may include a severity score that quantifies a severity of a medical condition.

[0038] The classified feature data generated by inputting the subject health data to the trained machine learning model(s) can then be displayed to a user, stored for later use or further processing, or both, as indicated at step 108.

[0039] In some instances, the machine learning model may be finetuned, retrained, or otherwise updated for a particular subject. As one example, the subject health data received from the subject and the corresponding classified feature data generated for the subject may be used to finetune the machine learning model parameters, or to otherwise retrained the machine learning model. For example, active learning, transfer learning, or the like, can be used to finetune, retrain, or otherwise update the machine learning model based on subject health data and classified feature data collected from the subject over multiple implementations. In this way, the machine learning model can be updated to provide more accurate results for a specific subject by learning which features of the subject health data are most relevant for predicting migraine attack onset and / or continuation for that particular subject.

[0040] In some instances, the updated machine learning model may require fewer inputs than the initial machine learning model. In these cases, the most relevant features of the subject health data may be identified for the updated model, such as by using a SHAP analysis, variable importance analysis, or the like. Based on one or more of these analyses, one or more relevant features of the subject health data are identified. For instance, relevant subject health data features can be selected as those with SHAP and / or variable importance values at or above a threshold, SHAP and / or variable importance values at or above a percentile rank of all subject health data features, or the like. Then, in future implementations the subject may be prompted to collect only the more limited subset of subject health data features. For example, the subject may only need to perform a single RAN test for use with an updated machine learning model. As another non-limiting example, the subject may only need to perform a single RAN test and record only one or a few symptoms for use with the updated machine learning model. In this way, the machine learning model can be personalized for the particular subject, which as one advantage can result in reducing the volume of subject health data that needs to be recorded by the subj ect.

[0041] Additionally or alternatively, the subject health data received from the subject and the corresponding classified feature data generated for the subject may also be used to reevaluate other machine learning model architecture, such that updating the machine learning model for the subject may include selecting a different machine learning model architecture to be used in future implementations for the subject. In another example, unsupervised machine learning frameworks can be applied to identify subject clusters and models with slightly different architectures can be then used with different individuals based on the clusters.

[0042] Referring now to FIG. 5. a flowchart is illustrated as setting forth the steps of an example method for training one or more machine learning models on training data, such that the one or more machine learning models are trained to receive subject health data as input data in order to generate classified feature data as output data, where the classified feature data are indicative of a prediction of migraine attack onset and / or continuation in a subject.

[0043] In general, the machine learning model(s) can implement any number of different machine learning model architectures. For instance, the machine learning model(s) could implement gradient boosting machine (GBM) model, a distributed random forest (DRF) model or other random forest model architecture, a generalized linear model (GLM) model, an XGBoost model, a stacked ensemble model, or any suitable deep learning-based model, including a convolutional neural network, a residual neural network, or the like. Additionally or alternatively, models based on a large language model (LLM), other transformer-based models, or other foundation models may be used to directly process subject health data (e.g., text data stored in the subject health data, other data stored in the subject health data) input by the subject. Alternatively, the machine learning model could implement other suitable machine learning or artificial intelligence algorithms, such as those based on supervised learning, unsupervised learning, deep learning, ensemble learning, dimensionality reduction, combinations thereof, and so on.

[0044] The method includes accessing training data with a computer system, as indicated at step 502. Accessing the training data may include retrieving such data from a memory or other suitable data storage device or medium. Alternatively, accessing the training data may include acquiring such data from a subject and recording the data with the computer system, or otherwise transferring the data to the computer system.

[0045] In general, the training data can include sub] ect health. The method can include assembling training data from subject health data using a computer system. This step may include assembling the subject health data into an appropriate data structure on which the machine learning model can be trained. Assembling the training data may include assembling subject health data and other relevant data. For instance, assembling the training data may include generating labeled data and including the labeled data in the training data. Labeled data may include subject health or other relevant data that have been labeled as belonging to, or otherwise being associated with, one or more different classifications or categories. For instance, labeled data may include subject health data that have been labeled as being associated with one or more predictive outcomes.

[0046] In an example study, subjects filled out baseline demographics, headache characteristics, and the two migraine-specific questionnaires. A total of 4985 headache symptoms diary data and corresponding 4982 KDT data points from 30 subj ects were collected. The few missing data of KDT data points were imputed using the subject’s KDT scores at the closest timepoint. Each subject recorded 180 (median) diary data points.

[0047] During each recording in this example study, the subjects reported the migraine state (phase) that they were currently in. They were first asked whether they were experiencing a headache, and if yes whether it was a migraine attack or non-migraine headache. If they did not have a headache at the time of the recording, they were asked to choose from the following three options: feeling a migraine coming on, recovering from a migraine, or do not feel a migraine coming on. Therefore, a total of five possible migraine phases were recorded: migraine attack, non-migraine headache, prodrome, postdrome and interictal baseline. The migraine phase data collected at each timepoint can be used to create a binary7“outcome” for the prediction model to indicate whether the subject will have a migraine attack during the next recording. The outcome was marked “1” if the subject recorded “migraine attack” during the next recording, and “0” is the subject recorded any of the four phases (interictal baseline, non- migraine headache, prodrome, postdrome). A positive outcome at a timepoint indicates the subject would develop an impending migraine attack or would continue to be in a migraine attack phase during the next recording (after 6-12 hours).

[0048] After data preprocessing, the last recording of each subject was selected as the outcome, whether or not the subj ect had migraine at the next recording. A total of 294 variables, including 275 categorical and 19 numerical variables, and 4951 data points from 30 participants were used for model development.

[0049] Among the variables included, 136 were from baseline demographics including age, sex, race, ethnicity, headache characteristics, and questionnaire results; 4 variables were RAN test scores collected at each time point, including the RAN test score with zero errors, RAN test score of the first trial, the number of errors of the first trial, and the number of RAN test trials to zero errors; 154 variables were from headache symptom diary collected at each time points, including the state of health, sleep state, current migraine phase, and detailed symptoms the participants experienced at each time point.

[0050] The variables indicating whether the subj ect would be in a migraine attack phase during the next recording (6-12 hours) were used as an outcome during model development.

[0051] One or more machine learning models are trained on the training data, as indicated at step 504. In general, the machine learning model can be trained by optimizing model parameters based on minimizing a loss function. As one non-limiting example, the loss function may be a mean squared error loss function. The machine learning model may be any suitable machine learning model, including those based on supervised learning, unsupervised learning, ensemble learning, or other learning techniques. As a non-limiting example, the machine learning model may be a GBM model, a DRF model, a GLM model, an XGBoost model, or a stacked ensemble model.

[0052] In an example implementation, the training data were separated into training and test sets at the subject level using a 29: 1 ratio. A machine learning model was first trained using all data points from 29 subjects (except for subject 1), then the held out 161 data points from subject 1 were used as the test set to evaluate model performance. The process was repeated 30 times using each subject separately as the test set. Within each model training, all training data were randomly divided for 5 -fold cross validation, which also included automatic hyperparameter tuning and model selection among the top 20 models of different types. The cross-validation matrices were used to select the top-performing model, and the performance of the top model for each medication was then evaluated in the held-out test set.

[0053] In some implementations, a SHAP analysis can be performed on the top performing models to analyze the variables that were most important in model prediction. These results can be stored as feedback data and used to guide the selection of subject health data variables to be input to the trained machine learning model. In a non-limiting example, a SHAP analysis determined that relevant subject health data features included variables for the models were: current migraine state, time from last migraine attack (minutes), RAN test score (first attempt), RAN test score (without making any errors), sleep state, hours of sleep the night before the recording, headache location, and headache severity.

[0054] Optionally, the machine learning model may be a pretrained machine learning model (e g., a stored, pretrained machine learning model or a machine learning model trained in step 504) that may be retrained, fine-tuned, or otherwise updated on additional training data that are received by the computer system. In these cases, the pretrained machine learning model may be retrained, fine-tuned, or otherwise updated on these additional training data before storing the updated machine learning model for later use.

[0055] The one or more trained machine learning models are then stored for later use, as indicated at step 506. Storing the machine learning model(s) may include storing modelparameters, which have been computed or otherwise estimated by training the machine learning model(s) on the training data. Storing the trained machine learning model(s) may also include storing the particular machine learning model architecture to be implemented. For instance, data pertaining to the layers in a neural network architecture (e.g., number of layers, type of layers, ordering of layers, connections between layers, hyperparameters for layers) may be stored.

[0056] In an example study, the systems and methods described in the present disclosure were implemented to predict migraine attack onset or continuation in a group of 30 adult subjects. The subjects suffered from migraine with or without aura, according to International Classification of Headache Disorders 3rd(ICHD-3) edition, for at least 8 migraine attacks and 2 months (up to 4 months). The inclusion criteria included: 18 years of age or older, being diagnosed with migraine with or without aura per ICHD-3, experiencing 4-10 migraine attacks per month, having fewer than 15 headache days per month (average over the prior 3 months), and willing to provide RAN tests and headache diary entries 3 times per day. Subjects with the following conditions were excluded: having greater than 10 migraine attacks per month, 15 or greater headache days per month (average over the prior 3 months), pre-existing ocular conditions, or other neurologic diseases, history of intracranial surgery, and history of concussion within past 2 years.

[0057] At the time of enrollment, detailed demographics (age, gender, race, ethnicity, handedness), baseline headache characteristics (headache frequency, duration, location, quality, intensity; associated symptoms; aura; number of years with migraine; current preventive and acute therapies), and answers to the Migraine Scog and Utah Photophobia Symptom Impact Scale (UPSIS-17) questionnaires were recorded for each participant.

[0058] Following enrollment, participants recorded a minimum of three King-Devick test and symptom diary entries per day; in the morning (~8 AM), mid-day (~ 2 PM), in the evening (~8PM), and additionally if there w as impending migraine attack or migraine attacks outside of these windows. Each data entry, recorded multiple times a day on a daily bases included: (1) RAN test scores including RAN test time with zero errors, RAN test time of first trial, number of errors of the first trial, number of trials to zero errors, and (2) Headache Symptom Diary, included state of health- current state of health, menstruation, sleep related information such as quality of sleep, bed time, wake time, hours of sleep, and whether the participant experienced headache in evening or overnight prior to that data entry; headache state- including the presence of headache or migraine, and detailed symptom questionnaire ateach of the migraine phases- non-migraine headache phase, migraine atack phase, prodrome and postdrome phase. The study protocol used in this example is illustrated in FIG. 2.

[0059] The recorded subject health data were extracted from a database. The RAN test scores, headache symptoms diary data, and baseline demographics data were mapped according to the participant and the time data was recorded. All participants filled out the baseline demographics, headache characteristics, and the Mig-Scog and UPSIS-17 questionnaires. A total of 4985 headache symptoms diary data and corresponding 4982 RAN test data points (e.g., KDT data points) from 30 participants were collected. The few missing data of RAN test data points were imputed using the participant’s RAN test scores at the closest timepoint. On average, each participant recorded 166 (range 11-227) headache symptoms diary and RAN test data entries.

[0060] As the baseline RAN test scores for each participant were different, all RAN test scores were individually scaled according to a participant’s minimum and maximum RAN test scores across multiple entries using a min-max scaler (e.g., MinMaxScaler in the skleam.preprocessing package). All other numerical and categorical variables were scaled at the group level.

[0061] During each data entry, the participant reported the migraine state that they were currently in. They were first asked whether they were experiencing a headache, and if yes, whether it was a migraine atack or non-migraine headache. If they did not have a headache at the time of the recording, they were asked to choose from the following three options: feeling a migraine coming on, recovering from a migraine, or do not feel a migraine coming on. Therefore, a total of five possible migraine phases were recorded: migraine atack, non- migraine headache, prodrome, postdrome and interictal baseline. We used the migraine phase data collected to create a binary outcome. The outcome was marked “1” if the participant recorded '‘migraine atack” during the next recording, and “0” if the participant recorded any of the four phases (interictal baseline, non-migraine headache, prodrome, postdrome). The outcome was used to train the machine learning model to predict whether the participant will be in a migraine atack phase at the next recording. A positive outcome at a timepoint indicates the subject would develop an impending migraine atack, or would continue to be in a migraine attack during the next recording (after 6 hours). After data preprocessing, we removed the last data entry of each subject, as the outcome, whether the subject had migraine at the next recording, was unknown.

[0062] A total of 294 variables, including 275 categorical and 19 numerical variables, and 4951 data points from 30 participants were used for model development. Among the variables included, 136 were from baseline demographics including age, sex, race, ethnicity, headache characteristics, Mig-Scog and UPSIS-17 questionnaire results; 4 variables were RAN test scores collected at each time point, including the RAN test score with zero errors, RAN test score of the first trial, the number of errors of the first trial, and the number of RAN test trials to zero errors; 154 variables were from headache symptom diary collected at each time point, including the state of health, sleep state, current migraine phase, and detailed symptoms the participants experienced at each time point. The outcome variable, indicating whether the subject would be in a migraine attack phase during the next recording (after approximately 6 hours) were removed from model development.

[0063] The dataset was separated into training and test sets at the participant level using a 29: 1 ratio. We first trained a model using all data entries from 29 participants (except for participant 1), then evaluated the model performance on all 161 data points from participant 1 as the test set, unseen by the model during training. The process was repeated 30 times using each participant separately as the test set. Within each model training, all training data were randomly divided for 5 -fold cross validation, which also included automatic hyperparameter tuning and model selection among the top 20 models of different types, including GBM, DRF, GLM, XGBoost, or stacked ensemble models. The cross-validation metrics were used to select the top-performing model, and the performance of the top model for each participant was then evaluated in the held-out test set. As the number of binary outcomes were imbalanced (approximately 10.4% positive), class balancing was employed in the model building.

[0064] The area under receiver operating characteristics curve (AUROC or AUC), precision, recall, accuracy, and Fl scores were measured for each model. Additionally, the average performance metrics across all models were reported using 1,000 bootstrapped samples. The model output was the probability of whether a subject will be in a migraine attack phase during the next recording. The prediction threshold was set such that could achieve the maximal Fl score so that the model reported a binary outcome.

[0065] Variable importance analysis along with SHAPley analysis was performed among the five best performing models to extract the top 20 most important variables for the model development in training and making predictions.

[0066] Among the 30 participants enrolled, the average age was 44.6 (23.0-65.0). 29 (96.7%) were female, 27 (90%) were White. 2 (6.7%) were Asian, 1 (3.3%) was AmericanIndian; 4 (13.3%) were Hispanic, and 26 (86.7%) were non-hispanic. 15 participants had migraine with aura, and 15 had migraine without aura. The mean duration of having migraine was 25.5 years, and the median days of headache and migraine days per month was 10 and 7 days, respectively.

[0067] After data preprocessing, 4951 data entries from 30 subjects, average 166 (range 10 to 226 per subject) were used to construct the machine learning models. An example of RAN test scores for individuals are shown in FIG. 6. Overall, there were 516, 288, 180, 288 and 3678 during migraine attacks, non-migraine headache, prodrome, post-drome and interictal phase, respectively.

[0068] Thirty different machine learning models were constructed, leaving one participant out as the held-out test set for each model, and the model performance in the test set was evaluated using standard metrics. When evaluating the performances for all 30 models, the average performance in the held-out test sets were: AUC 0.66 (0.61-0.72), precision 0.39 (0.28-0.45), recall 0.53 (0.48-0.63), accuracy 0.81 (0.74-0.85), and Fl score 0.39 (0.31 to 0.44). The performances of all models are listed in Table 3.Table 3: Performance Metrics from Example Study

[0069] The best performing model was a gradient boosting machine (GBM) model, that achieved an AUC of 0.95, precision 0.67, recall 0.88, accuracy 0.95, and Fl score 0.76 in the held-out test set. The second and third best performing models with the highest AUCs. 0.93 and 0.87. respectively, achieved the following: precision 0.67. 0.71; recall 0.62. 0.83; accuracy 0.93, 0.91; and Fl score 0.64 and 0.76, respectively.

[0070] Variable importance analysis was performed along with SHAPley analysis among the five best performing models to extract the top 20 most important variables for the model development in trainings and making predictions in test sets. Among all migraine phases, self-reported current migraine phase, time from last migraine attack (minutes), RAN test score at the first trial, RAN test score without making any error, the duration and quality of sleep the night before, if currently experiencing a headache, the severity and location of the headache, and current state of health were consistently the most important variables. Duringprodrome phase, experiencing phonophobia and photophobia, and subject belief that the current symptoms indicate a migraine is coming on were important variables to predict impending migraine attacks. During migraine attacks, endorsing photophobia and fatigue were important to predict continuation of migraine. Furthermore, data collected regarding baseline demographics and headache characteristics including age. headache frequency, years lived with migraine, the percentage of time subjects report experiencing prodrome and postdrome symptoms, and certain UPSIS-17 questionnaire elements were among the most important features in model development.

[0071] The variable importance plot of the top performing model is shown in FIG. 7.

[0072] The example study demonstrated that by leveraging detailed headache symptoms data and RAN test scores collected multiple times a day over a four-month period, robust machine learning models were constructed for accurately forecasting impending migraine attack, or continuation of an ongoing migraine attack over the next 6-12 hours for an individual. Additionally, several clinical features that were most important when forecasting migraine attacks were identified, including current migraine phase. RAN test scores, sleep state, photophobia, phobophobia, and fatigue.

[0073] This example study demonstrates the feasibility to use a prediction model incorporating detailed symptoms and an objective measure (e.g., the RAN test score, including different aspects as previously described) to forecast the probability of whether an individual will have an impending migraine attack, or whether an ongoing migraine attack would continue or subside over the next 6-12 hours, or another specified timeframe. Using the disclosed systems and methods, early intervention or treatment for possible impending migraine attacks can be enabled for subjects. The systems and methods can also be used to help forecasting whether a current migraine attack is probable to continue, enabling individuals to plan their activities accordingly.

[0074] It is an advantage of the disclosed systems and methods to leverage the realtime, self-administered headache symptoms and RAN test entries, recorded several times a day and daily over an extended period of time, to construct prediction models that are trained to predict migraine attack onset and / or continuation over a specified timeframe (e.g., the next 6- 12 hours).

[0075] FIG. 8 illustrates an example of a system 800 for predicting migraine attack onset and / or continuation in accordance with some embodiments of the systems and methods described in the present disclosure. As shown in FIG. 8, a computing device 850 can receiveone or more types of data (e.g., subject health data) from data source 802. In some embodiments, computing device 850 can execute at least a portion of a migraine attack onset and / or continuation prediction system 804 to predict the onset and / or continuation of a migraine attack in a subject from data received from the data source 802.

[0076] Additionally or alternatively, in some embodiments, the computing device 850 can communicate information about data received from the data source 802 to a server 852 over a communication network 854, which can execute at least a portion of the migraine attack onset and / or continuation prediction system 804. In such embodiments, the server 852 can return information to the computing device 850 (and / or any other suitable computing device) indicative of an output of the migraine attack onset and / or continuation prediction system 804.

[0077] In some embodiments, computing device 850 and / or server 852 can be any suitable computing device or combination of devices, such as a desktop computer, a laptop computer, a smartphone, a tablet computer, a wearable computer, a server computer, a virtual machine being executed by a physical computing device, and so on. The computing device 850 and / or server 852 can be generally referred to as a computer system, which can include a desktop computer, a laptop computer, a smartphone, a tablet computer, a wearable computer, a server computer, a virtual machine being executed by a physical computing device, or the like.

[0078] As a non-limiting example, a subject can use the computing device 850 to record one or more types of subject health data. For example, the computing device 850 can be used by the subject to perform a RAN test or other clinical test, such that the test scores are recorded as clinical test data. Additionally or alternatively, the computing device 850 can be used to generate and display a questionnaire to the subject, such that the subject’s responses to the questionnaire are recorded as subject symptom data.

[0079] In some embodiments, data source 802 can be any suitable source of data (e.g., subject health data, processed subject health data, other data extracted or derived from subject health data), another computing device (e.g., a server storing subject health data, processed subject health data, other data extracted or derived from subject health data), and so on. In some embodiments, data source 802 can be local to computing device 850. For example, data source 802 can be incorporated with computing device 850 (e.g., computing device 850 can be configured as part of a device for measuring, recording, estimating, acquiring, or otherwise collecting or storing data). As another example, data source 802 can be connected to computing device 850 by a cable, a direct wireless link, and so on. Additionally or alternatively, in someembodiments, data source 802 can be located locally and / or remotely from computing device 850, and can communicate data to computing device 850 (and / or server 852) via a communication network (e.g., communication network 854).

[0080] In some embodiments, communication network 854 can be any suitable communication network or combination of communication networks. For example, communication network 854 can include a Wi-Fi network (which can include one or more wireless routers, one or more switches, etc.), a peer-to-peer network (e.g., a Bluetooth network), a cellular network (e.g., a 3G network, a 4G network, etc., complying with any suitable standard, such as CDMA, GSM, LTE, LTE Advanced, WiMAX, etc.), other ty pes of wireless network, a wired network, and so on. In some embodiments, communication network 854 can be a local area network, a wide area network, a public network (e.g., the Internet), a private or semi-private network (e.g., a corporate or university intranet), any other suitable type of netw ork, or any suitable combination of networks. Communications links show n in FIG. 8 can each be any suitable communications link or combination of communications links, such as wired links, fiber optic links, Wi-Fi links. Bluetooth links, cellular links, and so on.

[0081] Referring now to FIG. 9, an example of hardware 900 that can be used to implement data source 802, computing device 850, and server 852 in accordance with some embodiments of the systems and methods described in the present disclosure is shown.

[0082] As shown in FIG. 9, in some embodiments, computing device 850 can include a processor 902. a display 904, one or more inputs 906, one or more communication systems 908, and / or memory 910. In some embodiments, processor 902 can be any suitable hardware processor or combination of processors, such as a central processing unit (“CPU”), a graphics processing unit (“GPU”), and so on. In some embodiments, display 904 can include any suitable display devices, such as a liquid crystal display (“LCD”) screen, a light-emitting diode (“LED”) display, an organic LED (“OLED”) display, an electrophoretic display (e.g., an “e- ink” display), a computer monitor, a touchscreen, a television, and so on. In some embodiments, inputs 906 can include any suitable input devices and / or sensors that can be used to receive user input, such as a keyboard, a mouse, a touchscreen, a microphone, and so on.

[0083] In some embodiments, communications systems 908 can include any suitable hardware, firmware, and / or software for communicating information over communication network 854 and / or any other suitable communication netw orks. For example, communications systems 908 can include one or more transceivers, one or more communication chips and / or chip sets, and so on. In a more particular example, communications systems 908 can includehardware, firmware, and / or software that can be used to establish a Wi-Fi connection, a Bluetooth connection, a cellular connection, an Ethernet connection, and so on.

[0084] In some embodiments, memory 910 can include any suitable storage device or devices that can be used to store instructions, values, data, or the like, that can be used, for example, by processor 902 to present content using display 904, to communicate with server 852 via communications system(s) 908, and so on. Memory 910 can include any suitable volatile memory, non-volatile memory, storage, or any suitable combination thereof. For example, memory 910 can include random-access memory (“RAM”), read-only memory (“ROM”), electrically programmable ROM (“EPROM”), electrically erasable ROM (“EEPROM"), other forms of volatile memory, other forms of non-volatile memory, one or more forms of semi-volatile memory, one or more flash drives, one or more hard disks, one or more solid state drives, one or more optical drives, and so on. In some embodiments, memory 910 can have encoded thereon, or otherwise stored therein, a computer program for controlling operation of computing device 850. In such embodiments, processor 902 can execute at least a portion of the computer program to present content (e.g., images, user interfaces, graphics, tables), receive content from server 852, transmit information to server 852, and so on. For example, the processor 902 and the memory 910 can be configured to perform the methods described herein (e.g., the method of FIG. 1, the method of FIG. 5).

[0085] In some embodiments, server 852 can include a processor 912. a display 914, one or more inputs 916. one or more communications systems 918. and / or memory 920. In some embodiments, processor 912 can be any suitable hardware processor or combination of processors, such as a CPU, a GPU, and so on. In some embodiments, display 914 can include any suitable display devices, such as an LCD screen, LED display, OLED display, electrophoretic display, a computer monitor, a touchscreen, a television, and so on. In some embodiments, inputs 916 can include any suitable input devices and / or sensors that can be used to receive user input, such as a keyboard, a mouse, a touchscreen, a microphone, and so on.

[0086] In some embodiments, communications systems 918 can include any suitable hardware, firmware, and / or software for communicating information over communication network 854 and / or any other suitable communication networks. For example, communications systems 918 can include one or more transceivers, one or more communication chips and / or chip sets, and so on. In a more particular example, communications systems 918 can include hardware, firmware, and / or software that can be used to establish a Wi-Fi connection, a Bluetooth connection, a cellular connection, an Ethernet connection, and so on.

[0087] In some embodiments, memory' 920 can include any suitable storage device or devices that can be used to store instructions, values, data, or the like, that can be used, for example, by processor 912 to present content using display 914, to communicate with one or more computing devices 850, and so on. Memory 920 can include any suitable volatile memory', non-volatile memory', storage, or any suitable combination thereof. For example, memory 920 can include RAM. ROM, EPROM, EEPROM, other ty pes of volatile memory, other ty pes of non-volatile memory, one or more types of semi-volatile memory, one or more flash drives, one or more hard disks, one or more solid state drives, one or more optical drives, and so on. In some embodiments, memory 920 can have encoded thereon a server program for controlling operation of server 852. In such embodiments, processor 912 can execute at least a portion of the server program to transmit information and / or content (e.g., data, images, a user interface) to one or more computing devices 850, receive information and / or content from one or more computing devices 850, receive instructions from one or more devices (e.g., a personal computer, a laptop computer, a tablet computer, a smartphone), and so on.

[0088] In some embodiments, the server 852 is configured to perform the methods described in the present disclosure. For example, the processor 912 and memory 920 can be configured to perform the methods described herein (e.g., the method of FIG. 1, the method of FIG. 5).

[0089] In some embodiments, data source 802 can include a processor 922, one or more data acquisition systems 924, one or more communications systems 926. and / or memory 928. In some embodiments, processor 922 can be any suitable hardware processor or combination of processors, such as a CPU, a GPU, and so on. In some embodiments, the one or more data acquisition systems 924 are generally configured to acquire or otherwise receive subject health data, and can include databases; smartphones, tablet computers, or other mobile devices; computer systems; smartwatches or other wearable devices; and the like. Additionally or alternatively, in some embodiments, the one or more data acquisition systems 924 can include any suitable hardware, firmware, and / or software for coupling to and / or controlling operations of databases; smartphones, tablet computers, or other mobile devices; computer systems; smartwatches or other wearable devices; and the like. In some embodiments, one or more portions of the data acquisition system(s) 924 can be removable and / or replaceable.

[0090] Note that, although not shown, data source 802 can include any suitable inputs and / or outputs. For example, data source 802 can include input devices and / or sensors that can be used to receive user input, such as a keyboard, a mouse, a touchscreen, a microphone, atrackpad, a trackball, and so on. As another example, data source 802 can include any suitable display devices, such as an LCD screen, an LED display, an OLED display, an electrophoretic display, a computer monitor, a touchscreen, a television, etc., one or more speakers, and so on.

[0091] In some embodiments, communications systems 926 can include any suitable hardware, firmware, and / or software for communicating information to computing device 850 (and, in some embodiments, over communication network 854 and / or any other suitable communication networks). For example, communications systems 926 can include one or more transceivers, one or more communication chips and / or chip sets, and so on. In a more particular example, communications systems 926 can include hardware, firmware, and / or software that can be used to establish a wired connection using any suitable port and / or communication standard (e.g.. VGA. DVI video. USB, RS-232, etc.). Wi-Fi connection, a Bluetooth connection, a cellular connection, an Ethernet connection, and so on.

[0092] In some embodiments, memory 928 can include any suitable storage device or devices that can be used to store instructions, values, data, or the like, that can be used, for example, by processor 922 to control the one or more data acquisition systems 924. and / or receive data from the one or more data acquisition systems 924; to generate images from data; present content (e.g., data, images, a user interface) using a display; communicate with one or more computing devices 850; and so on. Memory 928 can include any suitable volatile memory, non-volatile memory, storage, or any suitable combination thereof. For example, memory 928 can include RAM. ROM, EPROM. EEPROM, other types of volatile memory, other types of non-volatile memory, one or more types of semi-volatile memory, one or more flash drives, one or more hard disks, one or more solid state drives, one or more optical drives, and so on. In some embodiments, memory 928 can have encoded thereon, or otherwise stored therein, a program for controlling operation of data source 802. In such embodiments, processor 922 can execute at least a portion of the program to generate images, transmit information and / or content (e.g., data, images, a user interface) to one or more computing devices 850, receive information and / or content from one or more computing devices 850, receive instructions from one or more devices (e.g., a personal computer, a laptop computer, a tablet computer, a smartphone, etc.), and so on.

[0093] In some embodiments, any suitable computer-readable media can be used for storing instructions for performing the functions and / or processes described herein. For example, in some embodiments, computer-readable media can be transitory or non-transitory. For example, non-transitory computer-readable media can include media such as magneticmedia (e.g., hard disks, floppy disks), optical media (e.g., compact discs, digital video discs, Blu-ray discs), semiconductor media (e.g.. RAM, flash memory, EPROM. EEPROM), any suitable media that is not fleeting or devoid of any semblance of permanence during transmission, and / or any suitable tangible media. As another example, transitory' computer- readable media can include signals on networks, in wires, conductors, optical fibers, circuits, or any suitable media that is fleeting and devoid of any semblance of permanence during transmission, and / or any suitable intangible media.

[0094] As used herein in the context of computer implementation, unless otherwise specified or limited, the terms “component,” “system,” “module,” “framework,” and the like are intended to encompass part or all of computer-related systems that include hardware, software, a combination of hardware and software, or software in execution. For example, a component may be, but is not limited to being, a processor device, a process being executed (or executable) by a processor device, an object, an executable, a thread of execution, a computer program, or a computer. By way of illustration, both an application running on a computer and the computer can be a component. One or more components (or system, module, and so on) may reside within a process or thread of execution, may be localized on one computer, may be distributed between two or more computers or other processor devices, or may be included within another component (or system, module, and so on).

[0095] In some implementations, devices or systems disclosed herein can be utilized or installed using methods embodying aspects of the disclosure. Correspondingly, description herein of particular features, capabilities, or intended purposes of a device or system is generally intended to inherently include disclosure of a method of using such features for the intended purposes, a method of implementing such capabilities, and a method of installing disclosed (or otherwise known) components to support these purposes or capabilities. Similarly, unless otherwise indicated or limited, discussion herein of any method of manufacturing or using a particular device or system, including installing the device or system, is intended to inherently include disclosure, as embodiments of the disclosure, of the utilized features and implemented capabilities of such device or system.

[0096] The present disclosure has described one or more preferred embodiments, and it should be appreciated that many equivalents, alternatives, variations, and modifications, aside from those expressly stated, are possible and within the scope of the invention.

Claims

CLAIMS1. A method for predicting a migraine attack in a subject, the method comprising:(a) accessing subject health data with a computer system, wherein the subject health data comprise at least one of clinical test or measurement data received from the subject or subject symptom data received from the subject:(b) accessing a trained machine learning model with the computer system, wherein the trained machine learning model has been trained on training data to predict a likelihood of migraine attack occurring or continuing within a specified timeframe based on features in subject health data;(c) inputting the subject health data to the trained machine learning model using the computer system, generating classified feature data as an output, wherein the classified feature data indicate a likelihood of the subject having a migraine attack within the specified timeframe; and(d) outputting the classified feature data using the computer system.

2. The method of claim 1, wherein the subject health data comprise both clinical test or measurement data and subject symptom data.

3. The method of claim 2. wherein the subject health data further comprise subject demographic data.

4. The method of claim 1, wherein the subject health data comprise the clinical test or measurement data comprising rapid automatized naming (RAN) test score data.

5. The method of claim 4, wherein the RAN test score data comprise at least one of RAN test scores recorded at a time with zero errors, RAN test scores recorded at a time of a first trial for the subject, errors of the first trial for the subject, or number of trials to reach zero errors.

6. The method of claim 4, wherein the RAN test score data comprise rapid number naming test score data.

7. The method of claim 1. wherein the subject health data are recoded multiple times a day on a daily basis by the subject.

8. The method of claim 1, wherein the subject health data comprise the subject symptom data, which include subject responses to a headache symptom questionnaire.

9. The method of claim 8, wherein the subject symptom data comprise at least one of headache frequency, headache duration, headache location, headache quality, headache intensity, associated symptoms, aura, number of years with migraine, or current preventive and acute therapies.

10. The method of claim 9, wherein the subject symptom data comprise the associated symptoms, which include at least one of photophobia, phobophobia. or fatigue.

11. The method of claim 8, wherein the subject symptom data comprise subject symptom data recorded during a current migraine phase reported by the subject.

12. The method of claim 1. further comprising updating the machine learning model using the subject health data and the classified feature data, thereby generating a subject-specific machine learning model.

13. The method of claim 1. wherein the machine learning model is one of a gradient boosting machine (GBM) model, a distributed random forest (DRF) model, a generalized linear model (GLM) model, an XGBoost model, or a stacked ensemble model.

14. The method of claim 1, wherein the training data comprise clinical test or measurement data, subject symptom data, and subject demographic data collected from a group of test subjects.

15. The method of claim 14, wherein the clinical test or measurement data include rapid automatized naming (RAN) test score data.

16. The method of claim 15, wherein the RAN test score data comprise rapid number naming test score data.

17. The method of claim 14, wherein the subject demographic data comprise at least one of subject age, subject gender, subject race, subject ethnicity, or subject handedness.

18. The method of claim 1, wherein the specified timeframe is between 6 and 12 hours.

19. A method for predicting a migraine attack in a subject, the method comprising:(a) accessing subject health data with a computer system, wherein the subject health data comprise rapid automatized naming (RAN) test score data received from the subject and subject symptom data received from the subject;(b) accessing a trained machine learning model with the computer system, wherein the trained machine learning model has been trained on training data to predict a likelihood of migraine attack occurring or continuing within a specified timeframe based on features in subject health data;(c) inputting the subject health data to the trained machine learning model using the computer system, generating classified feature data as an output, wherein the classified feature data indicate a likelihood of the subject having a migraine attack within the specified timeframe; and(d) outputting the classified feature data using the computer system.

20. The method of claim 19, wherein the RAN test score data comprise at least one of RAN test scores recorded at a time with zero errors, RAN test scores recorded at a time of a first trial for the subject, errors of the first trial for the subject, or number of trials to reach zero errors.

Images