Systems and methods for the detection of eating disorder behaviors using wearable sensors

Abstract

An exemplary system (180) and method (300) are disclosed for detecting and preventing a future eating disorder behavior for a user. A user is provided a wearable device (105) that includes multiple physiological sensors (106) (305). At regular intervals, the wearable device collects physiological data(102) of the user including heart rate, electrodermal activity, and skin temperature (310). Features (108) are generated from the physiological data and used as input to a machine learning model (112) that is trained to predict a future eating disorder behavior based on features of physiological data (315; 320). If the model predicts a future eating disorder behavior, one or more actions (116) to prevent the eating disorder behavior are performed including alerting the user through the wearable device or a smart phone associated with the user, or alerting a caregiver, parent, or medical practitioner associated with the user (325).

Description

SYSTEMS AND METHODS FOR THE DETECTION OF EATING DISORDER BEHAVIORS USING WEARABLE SENSORSCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to U.S. Provisional Patent Application Serial No. 63 / 590,290, filed on October 13, 2023, and entitled "USING MACHINE LEARNING WITH PASSIVE WEARABLE SENSORS TO PILOT THE DETECTION OF EATING DISORDER BEHAVIORS IN EVERYDAY LIFE." The contents of which are hereby incorporated by reference in their entirety.STATEMENT OF GOVERNMENT SUPPORT

[0002] This invention was made with government support under Grant No. R15 MH121445 awarded by National Institute of Mental Health, National Institutes of Health (NIH). CRN is also funded by grant P20GM103436-20 (KY-INBRE) from the National Institute of General Medical Sciences. CEC is funded by the National Science Foundation (NSF) Graduate Research Fellowship Program under grant No. 2021320143. The government has certain rights in the invention.BACKGROUND

[0003] Eating disorders are serious psychiatric illnesses, taking a life every 52 minutes. Only ~50% of adults with eating disorders respond to existing evidence-based treatments. Relapse is common which contributes to the "revolving door phenomenon" where individuals cyclically admit and discharge from eating disorder treatment. Given the high mortality rate associated with eating disorders and the financial burden of accessing treatment, improved treatment that is effective for more individuals seeking recovery from eating disorders is critical. However, there are currently no support or effective intervention therapeutics for individuals with eating disorders in their everyday life.SUMMARY

[0004] An exemplary system and method are disclosed for detecting and preventing a future eating disorder behavior for a user. A user is provided a wearable device that includes multiple physiological sensors. At regular intervals, the wearable device collects physiological data of the user including heart rate, electrodermal activity, and skintemperature. Features are generated from the physiological data and used as input to a machine learning model that is trained to predict a future eating disorder behavior based on features of physiological data. The eating disorder behaviors may include self-induced vomiting, laxative or diuretic use, binge eating, restricting food intake, and excessive exercise. If the model predicts a future eating disorder behavior, one or more actions to prevent the eating disorder behavior are performed including alerting the user through the wearable device or a smart phone associated with the user, or alerting a caregiver, parent, or medical practitioner associated with the user.

[0005] In some aspects, the techniques described herein relate to a computer- implemented method for predicting an eating disorder behavior including: providing a wearable device to a user; while the user is wearing the wearable device, receiving a set of physiological data of the user taken by the wearable device by a computing device; and based on at least some of the received set of physiological data and a model, determining that the user is likely to have an eating disorder behavior at a future time by the computing device; and in response to the determination, taking one or more actions by the computing device.

[0006] In some aspects, the techniques described herein relate to a method, wherein the wearable device includes a smart watch.

[0007] In some aspects, the techniques described herein relate to a method, wherein the wearable device includes an Empatica™ wristband.

[0008] In some aspects, the techniques described herein relate to a method, wherein the eating disorder behavior includes one or more of self-induced vomiting, laxative or diuretic use, binge eating, restricting food intake, and excessive exercise.

[0009] In some aspects, the techniques described herein relate to a method, wherein the set of physiological data of the user includes one or more of heart rate, electrodermal activity, and skin temperature.

[0010] In some aspects, the techniques described herein relate to a method, wherein the model includes a machine learning model.

[0011] In some aspects, the techniques described herein relate to a method, wherein the one or more actions include generating an alert for the user, notifying amedical practitioner associated with the user, recording an indication of the determination, and providing a digital intervention.

[0012] In some aspects, the techniques described herein relate to a method, wherein the digital intervention includes a just-in-time adaptive intervention.

[0013] In some aspects, the techniques described herein relate to a method, wherein generating the alert for the user includes causing a notification to be sent to the user at the wearable device or at a smart phone associated with the user.

[0014] In some aspects, the techniques described herein relate to a method, wherein determining that the user is likely to have the eating disorder behavior at the future time includes: generating a set of features from the set of physiological data of the user; and determining that the user is likely to have the eating disorder behavior using the model and the set of features.

[0015] In some aspects, the techniques described herein relate to a system for predicting an eating disorder behavior including: a computing device; and a wearable device adapted to be worn by a user, to collect a set of physiological data of the user, and to provide the set of physiological data to the computing device, and wherein the computing device is adapted to: receive the set of physiological data; based on at least some of the received set of physiological data and a model, determine that the user is likely to have an eating disorder behavior at a future time; and in response to the determination, take one or more actions.

[0016] In some aspects, the techniques described herein relate to a system, wherein the wearable device includes a smart watch.

[0017] In some aspects, the techniques described herein relate to a system, wherein the wearable device includes an Empatica™ wristband.

[0018] In some aspects, the techniques described herein relate to a system, wherein the eating disorder behavior includes one or more of self-induced vomiting, laxative or diuretic use, binge eating, restricting food intake, and excessive exercise.

[0019] In some aspects, the techniques described herein relate to a system, wherein the set of physiological data of the user includes one or more of heart rate, electrodermal activity, and skin temperature.

[0020] In some aspects, the techniques described herein relate to a system, wherein the model includes a machine learning model.

[0021] In some aspects, the techniques described herein relate to a system, wherein the one or more actions include generating an alert for the user, notifying a medical practitioner associated with the user, recording an indication of the determination, and providing a digital intervention.

[0022] In some aspects, the techniques described herein relate to a system, wherein the digital intervention includes a just-in-time adaptive intervention.

[0023] In some aspects, the techniques described herein relate to a system, wherein generating the alert for the user includes causing a notification to be sent to the user at the wearable device or at a smart phone associated with the user.

[0024] In some aspects, the techniques described herein relate to a system, wherein determining that the user is likely to have the eating disorder behavior at the future time includes: generating a set of features from the set of physiological data of the user; and determining that the user is likely to have the eating disorder behavior using the model and the set of features.

[0025] The systems and methods described herein provide the following advantages. As described above, there is currently no effective intervention therapeutics for individuals with eating disorders in their everyday life. The present invention solves this problem by providing a non-invasive wearable device that predicts a future eating disorder with enough time for the user or an associated medical professional to intervene. In addition, because the wearable device uses physiological data of the user that is passively collected by sensors of the wearable device to make the prediction, there is no onus on the user to try to anticipate a future eating disorder.BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The accompanying figures, which are incorporated herein and form part of the specification, illustrate an image transmission system and method. Together with the description, the figures further serve to explain the principles of the system and method described herein and thereby enable a person skilled in the pertinent art to make and use the system and method.

[0027] FIG. 1 is an example environment for predicting and preventing future eating disorder behaviors;

[0028] FIG. 2 is an illustration of an example method for training a model to predict future eating disorder behaviors;

[0029] FIG. 3 is an illustration of an example method for predicting and preventing future eating disorder behaviors; and

[0030] FIG. 4 shows an exemplary computing environment in which example embodiments and aspects may be implemented.DETAILED DESCRIPTION

[0031] To facilitate an understanding of the principles and features of various embodiments of the present invention, they are explained hereinafter with reference to their implementation in illustrative embodiments.

[0032] FIG. 1 is an example environment for predicting and preventing future eating disorder behaviors. As shown, the environment 100 includes a wearable device 104 in communication with a server 180 through a network. The network may include a combination of private and public networks (e.g., the internet). The wearable device 104 and the server 180 may each be implemented by one or more general purpose computing devices such as the computing device 400 illustrated with respect to FIG. 4. Note, while shown as separate from the wearable device 104, in some embodiments some or all of the functionality performed by the server 180 may be performed by the wearable device 104.

[0033] The wearable device 104 may include one or more physiological sensors 106. The physiological sensors 106 may measure physiological data 102 associated with a user or wearer of the wearable device 104. The physiological data 102 may include one or more of heart rate, electrodermal activity, and skin temperature of the user. Other physiological data 102 may be measured depending on the physiological sensors 106 associated with the wearable device 104. An example wearable device 104 is a general-purpose smartwatch or smart ring. The wearable device 104 may include the Empatica™ wristband. Other wearable devices 105 may be supported.

[0034] The wearable device 104 may provide the physiological data 102 to the server 180 through the network. Depending on the embodiment, the wearable device 104may provide the physiological data 102 using a cellular, Bluetooth, or Wi-Fi modem associated with the wearable device 104. Other methods for providing data may be used.

[0035] The wearable device 104 may stream the physiological data 102 to the server 180 as the physiological data 102 is generated by the physiological sensors 106. Each piece of data may be associated with a time stamp for the time that the data was generated by a physiological sensor 106. Alternatively, rather than provide the physiological data 102 as it is generated, the wearable device 104 may batch the data 102 and may periodically send the batched physiological data 102 to the server 180. For example, the physiological sensor 106 may send batches of data 102 every five seconds or ten seconds. The frequency with which the wearable device 104 transmits the physiological data 102 to the server 180 may be set by the user or an associated administrator.

[0036] The server 180 includes several components including, but not limited to, a data collection engine 105, a training engine 110, and an action engine 115. More or fewer components may be supported. Each of the components may be implemented together or separately using one or more general purpose computing devices 400. In addition, some or all of the components of the server 180 may be implemented on the wearable device 104. For example, where the wearable device 104 is a smartwatch with processing and storage capability, the data collection engine 105 and action engine 115 may be implemented by an "app" or other software application executed by the wearable device 104.

[0037] The data collection engine 105 may receive the time stamped physiological data 102 from the wearable device 104 and may organize the received physiological data 102 into windows based on the time stamps. Depending on the embodiment, each window may be approximately 5 seconds. Other sized windows may be used.

[0038] The data collection engine 105, for each window of physiological data 102, may generate a set of features 108 for the window. The features 108 generated for the physiological data 102 may include, for each measured value from a physiological sensor 106, a mean value, median value, minimum value, maximum value, standard deviation, root mean square, mean absolute deviation, and mean absolute values for the 25thand 75thpercentiles. Other features 108 may be supported.

[0039] As will be described further below, the server 180 may use the features 108 to predict whether or not the user associated with the wearable device 104 will have what isreferred to herein as an eating disorder behavior. Eating disorder behaviors may include any behavior associated with eating disorders or disordered eating and may include selfinduced vomiting, laxative or diuretic use, binge eating, restricting food intake, and excessive exercise. Other behaviors may be included.

[0040] To facilitate the prediction of the eating disorder behavior based on the features 108, the training engine 110 may generate a model 112. The model 112 may be a machine-learning model 112 and may have been trained by the training engine 110 to receive a set of features 108, and to generate a probability that the user associated with the features 108 will experience an eating disorder behavior at some future time. Depending on the embodiment, the future time may be twenty minutes from the time stamps associated with the features 108. Other future times may be used depending on the embodiment.

[0041] In some embodiments, a single model 112 may be used to predict multiple eating disorder behaviors. For example, a single model 112 may be trained to predict selfinduced vomiting, binge eating, and laxative use. Alternatively, multiple models 112 may be used and each model 112 may be used to predict a different eating disorder behavior. For example, a first model 112 may be used to predict binge eating, and a second model 112 may be used to predict self-induced vomiting.

[0042] The training engine 110 may train the model 112 using training data 111. In some embodiments, the training data 111 may include a plurality of sets of features 108. Each set of features 108 may correspond to a window of physiological data 102 collected for a variety of subjects. The set of features 108 may include a label that indicates whether or not set of features 108 was extracted from physiological data 102 of a subject who experienced an eating disorder behavior within some period of time after the physiological data 102 was collected from the user. The period of time may include a variety of time periods such as 5 minutes, ten minutes, twenty minutes, etc. The period of time may be selected by a user or administrator. Any method for training a machine learning model may be used.

[0043] The training data 111 used by the training engine 110 to train the model 112 may be generated using a variety of methods. In one embodiment, a plurality of test subjects is selected. Each test subject wears a wearable device 104, and physiological data 102 is collected from each subject. While the physiological data 102 is collected for eachsubject, the subjects are asked to record each time that they experience an eating disorder behavior. For example, each test subject may be instructed to activate a button on the wearable device 104 each time they have a disordered eating behavior. The behaviors may be recorded along with time stamps as eating disorder behavior data 103. In some embodiments, each subject may be later asked to indicate the type of disorder eating behavior that they experienced.

[0044] After collecting physiological data 102 and eating disorder behavior data 103 from a subject, the collected physiological data 102 is divided into windows, and a set of features 108 is extracted for each window as described above. The eating disorder behavior data 103 is then processed to identify times associated with eating disorder behaviors for the subject. Using these identified times and the times associated with each window of the physiological data 102, windows that proceed an eating disorder behavior are determined. Depending on the embodiment, a window is determined to proceed an eating disorder behavior if it is within 20 minutes of the time associated with the disordered eating behavior. The sets of features 108 of windows that are determined to proceed any eating disorder behavior are labeled as predictive of eating disorder behavior. The sets of features 108 of windows not determined to proceed any eating disorder behavior are labeled as not predictive of eating disorder behaviors. Depending on the embodiment, each label may further indicate the particular eating disorder that the window is predictive of. The labeled sets of features 108 for each subject are the training data 111.

[0045] The data collection engine 105 may use the trained model 112 to predict whether or not a user of the wearable device 104 will experience an eating disorder behavior based on received physiological data 102. While the user wears the wearable device 104, the physiological data 102 collected by the physiological sensors 106 is provided to the data collection engine 105. The action engine 115 may, for a window of the physiological data 102, generate a set of features 108 for the window, and may use the set of features 108 as input to the model 112. In response, the model 112 may output an indication as to whether the user is likely to experience an eating disorder behavior within the next 20 minutes. The indication may be a binary value or may be a probability that the eating disorder behavior will occur. Where multiple models 112 are used, the data collection engine 105 may provide the set of features 108 to each of the models 112.

[0046] When the model 112 predicts that the user may have a future disordered eating behavior, the action engine 115 may perform one or more actions 116 with respect to the user. In some embodiments, the actions may include one or more digital interventions or just-in-time adaptive interventions. The actions may include actions such as notifying the user by causing the wearable device 104 to vibrate or make a sound, or by sending an electronic notification to a computer or smart phone associated with the user. The user may then take one or more preventive measures to reduce the likelihood of the behavior such as meditating or removing themselves from whatever stresses or situation may have been triggering the behavior.

[0047] The actions 116 may further include actions such as notifying one or more contacts associated with the user. For example, the user may have indicated that they would like a relative, a parent, a spouse, or a friend to be notified of a possible eating disorder behavior. The actions 116 may further include notifying a medical professional associated with the user, and / or logging the predicted eating disorder behavior in a file or data structure associated with the user. The medical professional may then review the file at a later time and may ask the user if the predicted behavior did or did not occur.

[0048] FIG. 2 is an illustration of an example method for training a model to predict future eating disorder behaviors. The method 200 may be implemented by one or more wearable devices 105 and the server 180.

[0049] At 205, physiological data collected for a set of subjects is received. The physiological data 102 may be received by the training engine 110 of the server 180 from a wearable device 104 worn by each subject of the set of subjects. The physiological data collected from a subject may include a variety of time stamped values of physiological data 102 including, but not limited to heart rate, skin temperature, and electrodermal activity. The values may be generated by one or more physiological sensors 106 of the wearable devices 105.

[0050] At 210, eating disorder behavior data collected for the set of subjects is received. The eating disordered behavior data 103 may be received by the training engine 110 of the server 180 from each subject of the set of subjects. In some embodiments, each subject may indicate that they are currently experiencing an eating disorder behavior by pressing a button or their wearable device 104, or by using a smartphone or othercomputing device associated with the subject. Each piece of eating disorder behavior data may indicate the type of eating disorder behavior that is or was experienced, and a time stamp that indicates approximately when the eating disorder behavior occurred. Depending on the embodiment, the eating disorder behaviors may include self-induced vomiting, laxative or diuretic use, binge eating, restricting food intake, and excessive exercise. Other behaviors may be included.

[0051] At 215, the physiological data is divided into a plurality of windows. The physiological data for each subject may be divided into windows by the training engine 110. Depending on the embodiment, each window may be approximately 5 seconds. Other sized windows may be used. The windows may be overlapping or non-overlapping depending on the embodiment.

[0052] At 220, based on the eating disorder data collected for the set of subjects, each window is labeled. Each window may be labeled by the training engine 110. In some embodiments, the training engine 110 may label each window based on the eating disorder behavior data 103 associated with the same subject. In particular, the training engine 110 may label windows that have time stamps that proceed the time stamp of an eating disorder behavior by more than a threshold time as being associated with an eating disorder behavior. The time may be 20 minutes. Other times may be used. Any other windows may be labeled as not associated with an eating disordered behavior. Depending on the embodiment, where the types of eating disorder behaviors are known, the window label may indicate the associated type of each eating disorder behavior.

[0053] At 225, for each window, features are extracted from the physiological data associated with the window. The features may be extracted by the training engine 110. The extracted features from each window and the corresponding label is the training data 111.

[0054] At 230, a model is trained using the extracted features and label associated with each window. The model 112 may be trained by the training engine 110 using the training data 111.

[0055] FIG. 3 is an illustration of an example method 300 for predicting and preventing future eating disorder behaviors. The method 300 may be implemented by one or more wearable devices 105 and the server 180.

[0056] At 305, a wearable device is provided to a user. The wearable device 104 may be provided to the user by a medical professional. The wearable device 104 may include a plurality of physiological sensors 106 that may measure a variety of physiological data signals including heart rate, electrodermal activity, and skin temperature.

[0057] At 310, while user is wearing the wearable device, a set of physiological data is received. The set of physiological data 102 may be received by the data collection engine 105 of the server 180.

[0058] At 315, features are generated from the physiological data. The features 108 may be generated by the data collection engine 105 from the set of physiological data 102.

[0059] At 320, based on the features and a model, determine that the user is likely to have an eating disorder behavior at a future time. The determination may be made by the data collection engine 105 from the features 108 using a model 112. The model 112 may output a probability or likelihood that the user will have an eating disorder behavior at some future time. Depending on the embodiment, the data collection engine 105 may make the determination when the probability is greater than a threshold probability.

[0060] At 325, one or more actions are taken based on the determination. The one or more actions 116 may be taken by the action engine 115 of the server 180. Depending on the embodiment, the actions 116 may include one or more of causing the wearable device 104 to vibrate or make a sound, sending a message to a smartphone or computing device associated with the user, altering the medical professional that provided the wearable device 104 to the user, or performing a just-in-time adaptive intervention.

[0061] FIG. 4 shows an exemplary computing environment in which example embodiments and aspects may be implemented. The computing device environment is only one example of a suitable computing environment and is not intended to suggest any limitation as to the scope of use or functionality.

[0062] Numerous other general purpose or special purpose computing devices environments or configurations may be used. Examples of well-known computing devices, environments, and / or configurations that may be suitable for use include, but are not limited to, personal computers, server computers, handheld or laptop devices, multiprocessor systems, microprocessor-based systems, network personal computers (PCs),minicomputers, mainframe computers, embedded systems, distributed computing environments that include any of the above systems or devices, and the like.

[0063] Computer-executable instructions, such as program modules, being executed by a computer may be used. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. Distributed computing environments may be used where tasks are performed by remote processing devices that are linked through a communications network or other data transmission medium. In a distributed computing environment, program modules and other data may be located in both local and remote computer storage media including memory storage devices.

[0064] With reference to FIG. 4, an exemplary system for implementing aspects described herein includes a computing device, such as computing device 400. In its most basic configuration, computing device 400 typically includes at least one processing unit 402 and memory 404. Depending on the exact configuration and type of computing device, memory 404 may be volatile (such as random access memory (RAM)), non-volatile (such as read-only memory (ROM), flash memory, etc.), or some combination of the two. This most basic configuration is illustrated in FIG. 4 by dashed line 406.

[0065] Computing device 400 may have additional features / functionality. For example, computing device 400 may include additional storage (removable and / or nonremovable) including, but not limited to, magnetic or optical disks or tape. Such additional storage is illustrated in FIG. 4 by removable storage 408 and non-removable storage 410.

[0066] Computing device 400 typically includes a variety of computer readable media. Computer readable media can be any available media that can be accessed by the device 400 and includes both volatile and non-volatile media, removable and nonremovable media.

[0067] Computer storage media include volatile and non-volatile, and removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Memory 404, removable storage 408, and non-removable storage 410 are all examples of computer storage media. Computer storage media include, but are not limited to, RAM, ROM, electrically erasable program read-only memory (EEPROM), flash memory orother memory technology, CD-ROM, digital versatile disks (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can be accessed by computing device 700. Any such computer storage media may be part of computing device 400.

[0068] Computing device 400 may contain communication connection(s) 412 that allow the device to communicate with other devices. Computing device 400 may also have input device(s) 414 such as a keyboard, mouse, pen, voice input device, touch input device, etc. Output device(s) 416 such as a display, speakers, printer, etc. may also be included. All these devices are well known in the art and need not be discussed at length here.

[0069] It should be understood that the various techniques described herein may be implemented in connection with hardware components or software components or, where appropriate, with a combination of both. Illustrative types of hardware components that can be used include Field-programmable Gate Arrays (FPGAs), Application-specific Integrated Circuits (ASICs), Application-specific Standard Products (ASSPs), System-on-a-chip systems (SOCs), Complex Programmable Logic Devices (CPLDs), etc. The methods and apparatus of the presently disclosed subject matter, or certain aspects or portions thereof, may take the form of program code (i.e., instructions) embodied in tangible media, such as floppy diskettes, CD-ROMs, hard drives, or any other machine-readable storage medium where, when the program code is loaded into and executed by a machine, such as a computer, the machine becomes an apparatus for practicing the presently disclosed subject matter.

[0070] Although exemplary implementations may refer to utilizing aspects of the presently disclosed subject matter in the context of one or more stand-alone computer systems, the subject matter is not so limited, but rather may be implemented in connection with any computing environment, such as a network or distributed computing environment. Still further, aspects of the presently disclosed subject matter may be implemented in or across a plurality of processing chips or devices, and storage may similarly be effected across a plurality of devices. Such devices might include personal computers, network servers, and handheld devices, for example.

[0071] Although the subject matter has been described in language specific to structural features and / or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims.

Claims

What is claimed is:

1. A computer-implemented method for predicting an eating disorder behavior comprising: providing a wearable device to a user; while the user is wearing the wearable device, receiving a set of physiological data of the user taken by the wearable device by a computing device; based on at least some of the received set of physiological data and a model, determining that the user is likely to have an eating disorder behavior at a future time by the computing device; and in response to the determination, taking one or more actions by the computing device.

2. The method of claim 1, wherein the wearable device comprises a smart watch.

3. The method of claim 1, wherein the wearable device comprises an Empatica wristband.

4. The method of claim 1, wherein the eating disorder behavior comprises one or more of self-induced vomiting, laxative or diuretic use, binge eating, restricting food intake, and excessive exercise.

5. The method of claim 1, wherein the set of physiological data of the user comprises one or more of heart rate, electrodermal activity, and skin temperature.

6. The method of claim 1, wherein the model comprises a machine learning model.

7. The method of claim 1, wherein the one or more actions comprise generating an alert for the user, notifying a medical practitioner associated with the user, recording an indication of the determination, and providing a digital intervention.

8. The method of claim 7, wherein the digital intervention comprises a just-in- time adaptive intervention.

9. The method of claim 7, wherein generating the alert for the user comprises causing a notification to be sent to the user at the wearable device or at a smart phone associated with the user.

10. The method of claim 1, wherein determining that the user is likely to have the eating disorder behavior at the future time comprises: generating a set of features from the set of physiological data of the user; and determining that the user is likely to have the eating disorder behavior using the model and the set of features.

11. A system for predicting a disordered eating event comprising: a computing device; and a wearable device adapted to be worn by a user, to collect a set of physiological data of the user, and to provide the set of physiological data to the computing device, and wherein the computing device is adapted to: receive the set of physiological data; based on at least some of the received set of physiological data and a model, determine that the user is likely to have an eating disorder behavior at a future time; and in response to the determination, take one or more actions.

12. The system of claim 11, wherein the wearable device comprises a smart watch.

13. The system of claim 11, wherein the wearable device comprises an Empatica wristband.

14. The system of claim 11, wherein the eating disorder behavior comprises one or more of self-induced vomiting, laxative or diuretic use, binge eating, restricting food intake, and excessive exercise.

15. The system of claim 11, wherein the set of physiological data of the user comprises one or more of heart rate, electrodermal activity, and skin temperature.

16. The system of claim 11, wherein the model comprises a machine learning model.

17. The system of claim 11, wherein the one or more actions comprise generating an alert for the user, notifying a medical practitioner associated with the user, recording an indication of the determination, and providing a digital intervention.

18. The system of claim 17, wherein the digital intervention comprises a just-in- time adaptive intervention.

19. The system of claim 17, wherein generating the alert for the user comprises causing a notification to be sent to the user at the wearable device or at a smart phone associated with the user.

20. The system of claim 11, wherein determining that the user is likely to have the eating disorder behavior at the future time comprises: generating a set of features from the set of physiological data of the user; and determining that the user is likely to have the eating disorder behavior using the model and the set of features.

Images