Method and device for estimating a state of a user and controlling a device according to an estimated state

Abstract

The invention relates to a method relating in particular to an estimation of a current state of a user of one electronic device from among a plurality of electronic devices, the device being used to perform an activity comprising a cognitive component and the estimated state characterizing a risk for the user with regard to the activity, the method comprising: - obtaining (300) results of a test, the test comprising, for each of a plurality of iterations, presenting an indication to assist in choosing an instruction to continue or stop a process, wherein continuing the process involves a chance of additional gain and a risk of losing an accumulated gain; receiving, from the user, an instruction to continue or stop the process; executing the received instruction; - estimating (305) the current state of the user according to the results and an estimator comprising a behavioral model of the user.

Description

[0001] DESCRIPTION

[0002] TITLE: Method and device for estimating a user's state and controlling a device based on an estimated state

[0003] technical field

[0004] The invention relates to a method and device for estimating the state of a user, in particular a user of an electronic device used to perform an activity including a cognitive component, based on their behavior, by evaluating risk-taking in situations of uncertainty, using a Bayesian model. It finds particular applications in fields where the type of behavior of a user facing a risky situation must be determined, especially for managing that risk.

[0005] Previous techniques

[0006] There are many situations in which it is important to assess an individual's state, for example, to suggest, advise against, or even prohibit certain activities. This might involve granting access to a computer service, granting access while drawing the individual's attention to a particular risk, or denying access. Such a state can be assessed from among several predetermined point states, such as anxiety, fatigue, well-being, and so on. Other states might relate to how an individual reacts to specific stimuli, for example, based on their ability to cope with a risky situation—that is, a situation over which they have only partial control. Such states are generally relatively constant over time.For example, services whose access is suggested, whose access is subject to an alert message, or whose access is blocked may include online betting or gambling.

[0007] Solutions exist for estimating an individual's state from among several predetermined states. Commonly used solutions are based on analyzing the results of questionnaires administered to the individual. While such approaches allow for precisely targeting states to be estimated based on expected behaviors associated with those states, by adapting the questions to these behaviors (with questions posed directly or indirectly), the results obtained can be biased by the individual. The individual may answer spontaneously, but they may also be tempted to understand the purpose of the questions in order to provide the expected answers, according to their own expectations, or they may have a limited capacity for introspection.

[0008] While these solutions are often useful, they are difficult to implement because they are restrictive for users. Furthermore, they are not suitable for predicting an individual's potential vulnerability to a particular activity, such as the risk of gambling addiction, which would allow for early warning or even the restriction of access to such activities.

[0009] Therefore, there is a need for a simple solution to implement in order to assess an individual's condition and predict potential vulnerability to a particular activity.

[0010] Description of the invention

[0011] The aim of the invention is therefore to offer a solution enabling the estimation of the state of an individual confronted with a situation which he only partially controls in order to alert him to a risk linked to the pursuit of a particular activity, including a cognitive component.

[0012] According to one aspect, the invention relates to a computer method for estimating the current state of a user of at least one electronic device from among a plurality of predetermined states, the at least one device being used to perform an activity comprising at least one cognitive component, and the estimated state characterizing a risk to said user with regard to said activity, the method comprising:

[0013] - obtaining test results, the test comprising, for each of a plurality of iterations,

[0014] . presentation of at least one indication to aid in choosing an instruction to continue or stop a process whereby a continuation involves a chance of additional gain and a risk of loss of cumulative gain;

[0015] . receipt, from the user, of an instruction to continue or stop the process;

[0016] execution of said instruction received;

[0017] - estimation of the user's current state based on said results and an estimator including a behavioral model of the user.

[0018] The method according to the invention thus makes it possible to estimate an individual's state and identify a potential vulnerability of a user in the face of an activity including a cognitive component such as gambling.

[0019] Depending on a characteristic, the estimator also includes at least one classifier.

[0020] According to one aspect, the invention relates to a computer method for estimating the current state of a user of at least one electronic device from among a plurality of predetermined states, the at least one device being used to perform an activity comprising at least one cognitive component, and the estimated state characterizing a risk to said user with regard to said activity, the method comprising:

[0021] - obtaining test results, the test comprising, for each of a plurality of iterations,

[0022] . receipt, from the user, of an instruction to continue or stop a process whereby continuing involves a chance of additional gain and a risk of loss of cumulative gain;

[0023] execution of said instruction received;

[0024] - estimation of the user's current state based on said results and an estimator comprising a behavioral model of the user and a classifier.

[0025] The method according to the invention thus makes it possible to assess an individual's state and identify a user's potential vulnerability to an activity with a cognitive component, such as gambling. According to one feature, the method further includes presenting at least one indication to aid in choosing an instruction to continue or stop the process.

[0026] Depending on a characteristic, at least one indication includes a probability of additional gain or loss of cumulative gain related to the instruction considered.

[0027] According to one characteristic, at least one indication includes a tendency for at least one other user to select a continue or stop instruction.

[0028] According to a characteristic, said results include at least one of the following elements: a total cumulative gain, a number of iterations, a number of iterations that led to a gain, a number of iterations that led to a loss, a number of instructions that led to obtaining a cumulative gain, a test execution time, and an average time between two iterations.

[0029] According to one characteristic, the process also includes the presentation of a warning, access to a service, personalization of a service or prohibition of access to a service based on said estimated current state.

[0030] According to one characteristic, the process includes a preliminary parameterization of said estimator, the results obtained being initial results, the parameterization comprising:

[0031] - creation of at least one model of said test;

[0032] - obtaining second results of said test using said at least one model;

[0033] - obtaining at least one expected state of a current user state corresponding to said second results; and

[0034] - determination of at least one parameter of the estimator according to the second results and the expected result.

[0035] Depending on one characteristic, the process includes:

[0036] - creation of a plurality of models of said test;

[0037] - obtaining third results of said test; and

[0038] - the selection of a plurality model based on the third results and the results obtained from each model of said plurality. The device according to the invention thus makes it possible to estimate an individual's state and identify potential vulnerability to an activity including a cognitive component, such as gambling.

[0039] A computer program, implementing all or part of the process described above, installed on pre-existing equipment, is in itself advantageous, since it allows us to estimate the state of an individual confronted with a situation that he only partially controls in order to alert him to a risk related to the pursuit of a particular activity, including a cognitive component.

[0040] Thus, the present invention also relates to a computer program comprising instructions for implementing the process described above, when this program is executed by a processor.

[0041] This program can use any programming language (e.g., an object-oriented language or other) and be in the form of interpretable source code, partially compiled code, or fully compiled code.

[0042] Another aspect concerns a non-transient storage medium for a computer-executable program, comprising a set of data representing one or more programs, said one or more programs comprising instructions for, when said one or more programs are executed by a computer comprising a processing unit operationally coupled to memory means and an input / output interface module, to execute all or part of the process described above.

[0043] Brief description of the drawings

[0044] Other objects, features and advantages of the invention will become apparent from the following description, given solely by way of non-limiting example, and made with reference to the accompanying drawings in which:

[0045] [Fig 1] schematically represents a graphical interface used to perform a modified BART type test to estimate an individual's state, according to embodiments of the invention; [Fig 2] illustrates an example of steps in a process according to embodiments to configure an estimator adapted to estimate an individual's state and identify a potential vulnerability to an activity including a cognitive component;

[0046] [Fig 3] illustrates an example of steps in a process according to embodiments to estimate an individual's state and identify potential vulnerability to an activity including a cognitive component;

[0047] [Fig 4] illustrates an example of the steps in a modified BART-type test according to embodiments; and

[0048] [Fig 5] illustrates an example of a device that can implement a process according to particular embodiments of the invention.

[0049] Detailed description of at least one embodiment

[0050] A detailed description of particular embodiments of the invention will be given below, with reference to the drawings in which the same references identify the same structural elements in each of the figures.

[0051] The inventors determined that the early detection of potential vulnerability to activities involving one or more cognitive components, such as gambling, could be achieved by assessing an individual's state based on their behavior and evaluating risk-taking under uncertainty. They also determined that behavior could be analyzed using results from modified BART (Balloon Analogue Risk Task) tests, which assess an individual's propensity to take risks (or risk aversion), specifically designed to enable the early detection of potential vulnerability to activities with a cognitive component, such as gambling.

[0052] It is important to remember that in a BART-type test, the participant must collect as many points as possible by inflating balloons and preventing them from popping. The larger the balloon, the more points the participant earns. However, a balloon can pop at any time, causing the participant to lose all the points accumulated for that balloon. Therefore, the participant can choose to save their points at any time by stopping inflating the balloon (before it pops) and moving on to the next balloon (if there are any left to inflate).

[0053] In summary, this test involves presenting a series of balloons to be inflated, for example, via a computer graphical interface. Points can be earned with each pump or inflation action (representing the virtual addition of a unit of air to the balloon, corresponding, for example, to pressing a specific key or clicking the mouse in a particular location on the interface). If the balloon bursts, the points associated with inflating it are lost. Conversely, if the individual stops inflating the balloon before it bursts, the accumulated points are considered permanently earned (they are added to any previously earned points). The number of balloons is predetermined.

[0054] Thus, a BART-type test measures risk-taking behavior in situations of uncertainty among participants for whom, as in real-world situations, risk is rewarded up to a certain point, beyond which taking further risks leads to poorer results (participants are not informed of the balloons' bursting points; they must learn and adapt their behavior as they gain experience with the task's contingencies). The playful aspect of these tests ensures high acceptance, particularly among users who want to test themselves. Furthermore, these tests are reliable, valid, and have good predictive power. They help mitigate potential biases well-known in questionnaires (e.g., social desirability bias).This type of test is used, for example, to correlate behaviors with general risk-taking behaviors, including addictions such as the use of crack cocaine, cocaine, marijuana, alcohol, or tobacco. According to particular embodiments of the invention, a modified BART-type test is performed to obtain results representative of the behavior of a user confronted with a situation over which they have only partial control, combining the risk of loss and the chance of gain, by evaluating risk-taking under uncertainty.

[0055] In specific implementations, a Bayesian model, which may include a classifier, is used to analyze behavior and relevant parameters in order to associate a participant's state with test results. The estimated user state can then be used to alert them to a particular risk or to control a device, for example, to suggest, authorize, or deny access to a service. For instance, a user's state characterizing a potential vulnerability to gambling can be determined using a tool known as the Canadian Problem Gambling Index. This tool was developed by the Canadian Centre on Substance Abuse and Alcoholism. Each question in a multiple-choice questionnaire is assigned a score, and a state is determined based on the final score and predetermined thresholds.

[0056] The parameters of the modified BART test may include the number of balloons to be inflated, the resistance(s) of the balloons (there may be several types of balloons), the bursting point of the balloons, as well as contextual information (or feedback) such as the probability that the balloon being inflated will burst, the behavior of other participants in similar situations, etc.

[0057] Figure 1 schematically represents a graphical interface used to perform a modified BART test to assess an individual's state, according to embodiments of the invention. As illustrated, the graphical interface 100 includes a representation of a balloon 105 that can be inflated using a pump 110, here activated by a mouse and a clickable inflation area 115. According to the illustrated example, the number of points that can be earned per pump stroke is displayed on the pump. Here, by pumping, the user is likely to earn 3 points. The number of points that can be earned with each pump stroke can be constant or can vary depending on parameters such as the balloon's inflation level. The number of points accumulated for inflating the balloon is displayed here in area 120, corresponding to a one-time amount in a temporary bank.These points are earned if the user stops inflating the balloon and confirms the inflation, here using a mouse by clicking on the validation area 125. In certain specific embodiments, several pump strokes can be performed in quick succession. In this case, the user can, for example, enter the number of pump strokes to be performed beforehand and then confirm this number.

[0058] In some embodiments, assistance may be provided to the user in the form of feedback. For example, as shown in reference 130, this may include a risk (or probability) level of balloon bursting, for instance, expressed as a percentage, and / or a trend indicator showing the behavior of other users facing similar situations, indicating whether to continue the inflation process (arrow pointing upwards) or stop it (arrow pointing downwards). Again, for example, a warning such as "Caution, risk of bursting" may also be displayed. This may, for instance, be based on a balloon bursting risk level and thresholds. Other indicators may also be presented to the user.

[0059] If the balloon bursts, the points accumulated during its inflation are lost.

[0060] On the contrary, when the user stops inflating the balloon before it bursts, the points accumulated for its inflation are, where applicable, added to the points previously acquired and stored in a permanent bank, the amount of which is displayed here in zone 135.

[0061] As illustrated, interface 100 can include a 140 test progress gauge, the cursor of which evolves as the balloons are inflated, from the first balloon to the last balloon to be inflated (according to the example shown, the test includes 54 balloons).

[0062] The results are recorded as the test progresses. They may include, in addition to the number of points earned, the number of balloons inflated, the number of inflations (or pumps) per balloon before validation, the average number of inflations (or pumps) before validation, the number of inflations (or pumps) per balloon before bursting, the average number of inflations (or pumps) before bursting, the duration of the test, time deltas (i.e. for example the speed at which the participant clicks to inflate the balloon or the average time between two pumps) and / or the feedback information provided.

[0063] Figure 2 illustrates an example of steps in a process according to embodiments to configure an estimator suitable for estimating an individual's state and identifying potential vulnerability to an activity including a cognitive component.

[0064] According to the example shown in Figure 2, the estimator configuration includes the creation of modified BART-type test models.

[0065] As illustrated, a first step (step 200) aims to identify potentially relevant parameters for modeling modified BART-type tests and for estimating an individual's state and identifying potential vulnerability to an activity including a cognitive component.

[0066] Like the standard parameters of BART tests, the parameters of modified BART tests include the number of balloons an individual must inflate. This number must be high enough to achieve a certain level of accuracy in the results and low enough not to discourage the individual and encourage them to complete the test. Furthermore, the number of balloons must be sufficient so that the results are based on genuine risk-taking and not on uncertainty (participants discovering the balloons' resistance based on indicators such as the average number of pumps before bursting, or the average number of pumps before bursting related to balloon characteristics such as color, etc.). The number of balloons can range from 20 to 100, for example, between 45 and 65, and can be as high as 54.

[0067] These standard parameters for BART-type tests may also include a balloon burst distribution law. This could be, for example, a normal or random distribution, using a uniform or geometric distribution.

[0068] In certain embodiments, standard parameter values ​​for BART tests are defined by default and can be modified by an administrator. Also in certain embodiments, the standard parameters for BART tests remain constant over time (i.e., they are not changed between different tests), so that test results are comparable to one another.

[0069] To assess an individual's state and identify potential vulnerability in an activity with a cognitive component, additional parameters are needed. For example, parameters potentially relevant for modeling modified BART-type tests might include providing guidance to help a user decide whether to continue inflating a balloon or stop. This could involve indicating the probability of the balloon bursting or the reactions of other participants faced with similar situations. Furthermore, this removes an element of uncertainty that complicates the analysis of the results.As a further illustration, the parameters used to assess an individual's state and identify potential vulnerability to an activity with a cognitive component may include at least some of the results obtained from modified BART-type tests, such as the number of points earned, the number of balloons inflated, the number of balloons bursting, the average number of inflations (or pumps) before validation, the average number of inflations (or pumps) before bursting, the test duration, and / or time deltas. In a subsequent step, one or more models of the modified BART-type tests are created and validated. These test models may be based on similar or different mathematical approaches and on identical or different parameters. Some of these models may allow for the direct estimation of a user's state (step 205).Other models can be combined with other mathematical tools, for example classifiers, to estimate a user state (steps 205' and 215). As an illustration, eight test models can be created, for example with the following characteristics, based on the mathematical models presented in the appendix and called Model A, Model B, Model C and Model D:.

[0070] Test model 1:

[0071] - Mathematical basis: Model A

[0072] Number of samples used to parameterize the model: 160 Number of balloons: 54

[0073] Balloon resistance: 3 possible resistances

[0074] Balloon bursting point: random probability of bursting with a uniform distribution

[0075] Additional parameters: none,

[0076] Test model 2:

[0077] - Mathematical basis: Model B and Model D

[0078] Number of samples used to parameterize the model: 160 Number of balloons: 54

[0079] Balloon resistance: 3 possible resistances

[0080] Balloon bursting point: random probability of bursting with a uniform distribution

[0081] Additional parameters: none,

[0082] Test model 3:

[0083] - Mathematical basis: Model C

[0084] Number of samples used to parameterize the model: 160 Number of balloons: 54

[0085] Balloon resistance: 3 possible resistances

[0086] Balloon bursting point: random bursting probability with a uniform distribution; additional parameters: display of the balloon bursting probability for each inflation.

[0087] Test model 4:

[0088] - Mathematical basis: Model A

[0089] Number of samples used to parameterize the model: 160 Number of balloons: 54

[0090] Balloon resistance: 3 possible resistances

[0091] Balloon burst point: predetermined burst point; additional parameters: none.

[0092] Test model 5:

[0093] - Mathematical basis: Model B and Model D

[0094] Number of samples used to parameterize the model: 160 Number of balloons: 54

[0095] Balloon resistance: 3 possible resistances

[0096] Balloon burst point: predetermined burst point; additional parameters: none.

[0097] Test model 6:

[0098] - Mathematical basis: Model A

[0099] Number of samples used to parameterize the model: 160 Number of balloons: 54

[0100] Balloon resistance: 3 possible resistances

[0101] Balloon bursting point: predetermined bursting point; additional parameters: display of the number of points that could be earned for each balloon before moving on to the next balloon.

[0102] Test model 7:

[0103] - Mathematical basis: Model B and Model D

[0104] Number of samples used to parameterize the model: 160 Number of balloons: 54

[0105] Balloon resistance: 3 possible resistances

[0106] Balloon bursting point: predetermined bursting point. Additional parameters: display of the number of points that could be earned for each balloon before moving on to the next balloon, and Test Model 8:

[0107] - Mathematical basis: Model C

[0108] Number of samples used to parameterize the model: 160 Number of balloons: 54

[0109] Balloon resistance: 3 possible resistances

[0110] Balloon bursting point: random bursting probability with a geometric distribution

[0111] Additional parameters: display of the probability of balloons bursting for each inflation.

[0112] These test models can be parameterized based on previously obtained results from modified BART tests performed by participants to create a training database, using test conditions similar to those of the models to be parameterized. These results are, for example, stored in a database. The results used to parameterize the models are chosen according to the test conditions and the characteristics of the test models, for example, to account for the presentation of help prompts to assist a user in deciding whether to continue inflating a balloon or to stop. Validation of a test model can also be performed using previously obtained results; for example, a test model is validated if the mean error is below a given threshold.

[0113] Mathematical models are, for example, based on models known as BSR models (Bayesian Sequential Risk-Taking in Anglo-Saxon terminology) or on the Lejuez model. BSR models are notably described in the following documents: Wallsten et al, Wallsten T. S, Lejuez C. W, and Pleskac T. J (2005) Modeling Behavior in a Clinically Diagnostic Sequential Risk-Taking Task, Zhou et al, Zhou Ran, Myung JL, and Pitt MA (2021) Revisiting learning in the balloon analogue risk task, and Coon and Lee, Coon J and Lee M D. (2021) A Bayesian Method for Measuring Risk Propensity in the Balloon Analogue Risk Task. The Lejuez model is described in particular in the document Lejuez et al, Lejuez C. W, Richards J. B, Read J. P, Kahler C. W, Ramsey S. E, Stuart G. L, Strong D. R and Brown RA (2002) Evaluation of a Behavioral Measure of Risk Taking: The Balloon Analogue Risk Task.

[0114] Of course, these test models are provided for illustrative purposes only; many other test models can be created and evaluated, for example, based on other mathematical models, by varying the number of balloons, etc.

[0115] As indicated by the dotted arrow, the creation and validation of a test model is an iterative process that allows for fine-tuning the parameters and converging towards an error level below a desired threshold.

[0116] When a model needs to be combined with another mathematical tool, such as a classifier, to estimate a user's state, such a classifier is created and validated (step 215) for each relevant model previously created and validated. The classifier analyzes the previously created and validated models to create classes (or user profiles) and establish links between results of modified BART tests and the states expected for those results. For example, results of modified BART tests can be obtained for given participants who have also been analyzed, for instance, via a form such as the Canadian Problem Gambling Index. As illustrated, the used results of modified BART tests and the corresponding states can be stored in database 210. In other embodiments, they are stored in a separate database.Again, as indicated by the dotted arrow, the creation and validation of a classifier is an iterative process.

[0117] As an illustration, the classifier can be of the Bayesian mixture model type and defined by Model D in the appendix.

[0118] In a subsequent step, a test model, possibly combined with a classifier (called an estimator), is selected (step 220). This might be, for example, the test model or the test model combined with a classifier that provides the best results and whose error rate is below a given threshold. As indicated by the dashed arrows, the step of creating and validating a test model, the step of creating and validating a classifier, or the steps of creating and validating both a model and a classifier can be repeated, for example, if the classifier's error rate is greater than or equal to a given threshold.

[0119] Figure 3 illustrates an example of steps in a process according to embodiments to estimate an individual's state and identify potential vulnerability to an activity including a cognitive component.

[0120] As illustrated, the first step involves executing a test (step 300), for example, a modified BART test. This could be the test described with reference to Figure 1. This test is offered, for example, to a user wishing to subscribe to or access a service, such as an online gambling offer. It can be offered on the device used by the user wishing to subscribe or access the service, for example, as a web interface, with the test steps executed remotely. Alternatively, these steps can be executed on the user's device, for example, as a plugin.

[0121] According to particular embodiments, the execution conditions of this test are similar to those of the test model selected as estimator, possibly combined with a classifier.

[0122] The results obtained are then analyzed to estimate a user state (step 305), for example using the test model selected in step 220 of Figure 2 as an estimator, possibly combined with a classifier. In specific embodiments, the state is either a state characterizing a potential vulnerability to an activity with a cognitive component, such as online gambling, or a state characterizing an absence of potential vulnerability to such an activity.

[0123] A test can then be performed (step 310) to determine whether the estimated state characterizes a potential vulnerability to an activity involving a cognitive component. If so, specific measures can be taken (step 315), for example, to alert the user to a particular risk, control a device, for example, to allow, suggest, or deny access to a service, or personalize a service or a service environment.

[0124] Conversely, if the estimated state indicates an absence of potential vulnerability to an activity with a cognitive component, information can be provided to the user (step 320), for example, in the form of a displayed message. Such information can be used for educational purposes, such as providing explanations about the meaning of the test the user has completed.

[0125] Figure 4 illustrates an example of steps in a modified BART test according to embodiments.

[0126] As illustrated, a first step (step 400) is for the initialization of the indexes i and j, the index i representing an index on the balloons to be inflated and the index j representing an index on the pump strokes during the inflation of balloon i, and of the total account cpt tot and temporary account cmp tmp, the total account cpt tot representing the total gain of a participant and the temporary account cpt tmp representing a potential gain related to the balloon being inflated.

[0127] In a subsequent step, and according to the illustrated embodiment, a decision aid is estimated (step 405) to help a user decide whether to continue inflating a balloon or confirm the balloon's inflation to acquire the accumulated points. For example, such a decision aid could represent a probability pburstQU un rj Sq ueof balloon bursting during inflation, for example expressed as a percentage, related to k ieme pump strokes of the l ieme inflatable balloon. This risk can, for example, be estimated as follows:

[0128] pb.urst. _ _ k x 100

[0129] n

[0130] where k is the number of pump strokes required to inflate the balloon in question (i.e., balloon 1) and n is the maximum possible number of pump strokes on the balloon (n can vary depending on the balloon's resistance; for example, a value of n can be associated with each level of balloon resistance). For example, the value of n could be 15 for the first balloon resistance, 20 for the second, and 25 for the third.

[0131] The user is then provided with a selection aid (step 410), for example, as a display on a graphical interface, and the number of points that can be earned with an additional pump is determined. In some embodiments, this number is displayed. For example, it may be constant and predetermined or vary, for instance, according to the number of pumps already given for the balloon in question.

[0132] The user is then prompted to enter their choice.

[0133] Following receipt of the user's command or instruction (step 415), for example via a keyboard keystroke or by clicking on a specific area, a test is performed (step 420) to determine whether the command is a continue command to give an additional pump stroke or a stop (validation) command to cease inflating the balloon and acquire the points accumulated for inflating that balloon. If the received command is a continue command to give an additional pump stroke, a virtual pump stroke is given to determine whether the balloon bursts or not (step 425). In certain embodiments, the balloon burst points are pre-generated (i.e., a number of pump strokes corresponding to the burst point is associated with each balloon) according to a distribution law and the maximum possible number of pump strokes on the balloon (e.g., related to its resistance).Thus, to know whether the balloon bursts or not after this additional pump, it is enough to compare the number of pumps made on the balloon in question with its bursting point.

[0134] If, after the additional pump, the balloon has not burst, the temporary count is updated (step 430) by adding the gain from the pump (referred to as gain in Figure 4) to the previous value of the temporary count (cpt tmp). The index on the number of pumps is incremented by one. The algorithm then loops back to step 405 to propose another pump. If, after the additional pump, the balloon bursts, the temporary count is reset to zero and the index on the number of pumps is reset to one (step 435). A test is then performed to determine if all the balloons have been inflated (step 440), that is, if the index i is less than the number of balloons to be inflated (denoted nb tot). If all the balloons have been inflated, the test is terminated.On the contrary, if there are still balloons to inflate, the index i is incremented by one (step 445) and the algorithm loops back to step 405 to inflate the next balloon.

[0135] If the received command is a stop (or validation) command to cease inflating the balloon and to accrue points for inflating that balloon, the total count is updated (step 450) by adding the value of the temporary count to the previous value of the total count. The temporary count is then reset to zero, and the index on the number of pump strokes is reset to one. The algorithm then loops back to step 440 to determine if all the balloons have been inflated.

[0136] It is noted here that other implementations can be considered. In particular, other information can be presented to the user to guide their choices. Similarly, other calculation formulas can be used.

[0137] Figure 5 illustrates an example of a data processing device that can be used to implement, at least partially, embodiments of the invention, in particular the steps described with reference to Figures 2, 3 and 4.

[0138] The 500 device preferably includes a 502 communication bus to which the following are connected:

[0139] - a central processing unit or microprocessor 504 (CPU, acronym for Central Processing Unit in Anglo-Saxon terminology);

[0140] - a read-only memory (ROM) 506, which may contain the operating system and programs such as "prog", "progl", and "prog2"; - a random access memory or cache (RAM) 508, which contains registers adapted to store variables and parameters created and modified during the execution of the aforementioned programs; and

[0141] - a 510 communication interface connected to a 512 distributed communication network, for example a wireless communication network and / or a local communication network, the interface being capable of transmitting and receiving data, including to and from other devices.

[0142] Optionally, the 500 device may also include the following features:

[0143] - a 518 hard drive capable of containing the aforementioned "prog", "progl" and "prog2" programs and data processed or to be processed according to the invention;

[0144] - an input / output interface 520 to which can be connected a keyboard 522, a mouse 524 and / or any other pointing device such as a light pen, a touch screen, a voice recognition device, a gesture recognition device, a camera, a microphone or a remote control allowing the user to interact with the programs according to the invention;

[0145] - a graphics card and a sound card or an audio / video card 514 connected to a monitor and speakers 516; and / or

[0146] - a 530 removable storage media reader 532 such as a memory card.

[0147] The communication bus enables communication and interoperability between the various elements included in or connected to the 500 device. The bus representation is not exhaustive; in particular, the central processing unit can communicate instructions to any element of the 500 device directly or through another element of the 500 device.

[0148] The executable code of each program enabling the programmable device to implement the processes according to the invention can be stored, for example, in the hard drive 518 or in read-only memory 506. According to an alternative, the executable code of the programs can be received via the communication network 512, through the interface 510, to be stored in the same way as described above.

[0149] More generally, the program(s) can be loaded into one of the storage means of device 500 before being executed.

[0150] The central processing unit 504 will command and direct the execution of the instructions or portions of software code of the program(s) according to the invention, instructions which are stored in the hard drive 518 or in the read-only memory 506 or in the other aforementioned storage elements. Upon power-up, the program(s) stored in non-volatile memory, for example the hard drive 518 or the read-only memory 506, are transferred to the random access memory 508, which then contains the executable code of the program(s) according to the invention, as well as registers for storing the variables and parameters necessary for the implementation of the invention.

[0151] Of course, the present invention is not limited to the embodiments described above by way of example. It extends to other variants.

[0152] Depending on the chosen embodiment, certain acts, actions, events, or functions of each of the methods described in this document may be performed or occur in a different order than described, or may be added, merged, or omitted, as appropriate. Furthermore, in some embodiments, certain acts, actions, or events are performed or occur concurrently rather than sequentially. Additionally, while the examples provided are based on BART-type tests, other similar tests may be used.

[0153] Although described through a number of detailed embodiments, the proposed device, system, and method include various variants, modifications, and improvements that will be obvious to those skilled in the art, it being understood that these various variants, modifications, and improvements form part of the scope of the invention, as defined by the following claims. Furthermore, different aspects and features described above may be implemented together, separately, or substituted for one another, and all the different combinations and subcombinations of aspects and features form part of the scope of the invention. In addition, some systems and equipment described above may not incorporate all the modules and functions described for the preferred embodiments. APPENDIX

[0154] Mathematical model A (probability of a balloon bursting)

[0155] According to this model, the probability, denoted p k belief that the k ieme The balloon bursting with each pump stroke is expressed by the following relationship:

[0156] p k belief = 1 -

[0157]

[0158] belief > - 1-° 1 (0 < a < [1])

[0159] “ K .. i yfc-1 pumps v

[0160] ”r Zji=o

[0161] OR

[0162] (1 — α / μ) represents the a priori belief about the probability of the balloon bursting,

[0163] ^=0 TLi success represents the number of pumps on the K-1 balloons that did not burst and

[0164]

[0165] Σ k-1 i=0 n i pumps represents the total number of inflations on the k-1 balloons.

[0166] The other components of the model are as follows:

[0167] wk = - γ + / log (1 - p k belief )

[0168] p kl pump = 1 / (1 + exp[β(l - w k )])

[0169] p(D|θ) = Ϡ N k=1 Pi n_k l=1 p kl pump (1-p k,n_k+1 pump ) d_k

[0170] Or

[0171] Wk represents the optimal number of pump strokes at k ieme ball,

[0172] y + represents the propensity for risk,

[0173] P represents behavioral consistency,

[0174] p kl pump represents the probability of inflating the k ième balloon the l ième times (1 pump stroke),

[0175] dk is an indication of whether or not k has burst ieme ball and

[0176] nk represents the number of pump strokes performed for the k iemeballoon. The mathematical model A allows for the direct estimation of a user's state without the need to present an indication to help the choice of continuing or stopping the balloon inflation.

[0177] Mathematical model B (probability of a balloon bursting)

[0178] This model allows us to estimate the number of pump strokes that the i ieme participant intended to perform for the k ieme balloon according to the following relationship:

[0179] y' ik ~Gaussian + (ρ i , β i 2 )

[0180] = yi'k if yi'k < b k

[0181] yik ib k if y k > b k

[0182] Or

[0183] yik represents the number of pump strokes actually performed by the i ieme participant to inflate the k ieme ball,

[0184] y i ' krepresents the number of pump strokes that the i ieme participant intended to perform to inflate the k ieme ball and bk is the bursting point of k ieme ball.

[0185] Key parameters of the model here are risk propensity p and behavioral consistency p.

[0186] Mathematical model C (displaying geometric and uniform probability)

[0187] This model is based on prospect theory, assuming that the probability P k f rst bursting of the l ieme balloon at the k ieme The "coup de pompe" (pump) is defined by the following relationship:

[0188] P kl burst = 100 k / n

[0189] n

[0190] where n represents the number of possible pump strokes for the l ieme ball.

[0191] The value function used here is as follows:

[0192] v(r) = { r γ+for gains relative to a status quo r > 0

[0193]

[0194] (— Â|r| y for losses relative to a status quo r < 0 r represents the gain,

[0195] v(r) represents a value function,

[0196] γ + = γ - represents the propensity for risk and

[0197] X represents risk aversion

[0198] It is observed that since the probability of a balloon bursting is known each time a user can choose to perform an additional pump, evaluating the options of inflating or not inflating the balloon before giving or not giving an additional pump can be used. Thus, it is possible to define the expected inflation and deflation values ​​on the k ieme ball at the l ieme occasion, as follows:

[0199] C”” = (1 - p^ rst )r r * - - Dr)''- K t = 0

[0200]

[0201] U kl pump represents the subjective value of inflating and

[0202] U^ op represents the subjective value of stopping swelling.

[0203] A probabilistic choice rule, representing the probability of performing an additional pump to inflate the balloon and the probability of stopping, such as the one given below, can be used:

[0204] P kl (pump,stop) = e βU /

[0205]

[0206] e βU + e βU

[0207] Considering that β (which represents behavioral consistency) is greater than zero and that U kl stop is equal to zero, as stated above, the probability of performing an additional pump rather than stopping inflation of the balloon can be defined as follows:

[0208] "pump

[0209] eP kl

[0210] ppump _ _

[0211] “kl ~ RnP um P

[0212] e PU kl + 1

[0213] P kl pump = 1 / (1 + e -βU )

[0214]

[0215] 1 + e -βU Mathematical model D (classification)

[0216] According to this example, the model's assumptions are as follows:

[0217] π~DDir(1,1')

[0218] (z i |π)~categorical(π1, π2)

[0219] (

[0220]

[0221] (ρ k , β k )~informativePriors

[0222] for example, as described previously, with reference to model B:

[0223] y' ik ~Gaussian + (ρ i , β i 2 )

[0224] = [yi'k if yi'k < b k

[0225]

[0226] b k if y' ik ≥ b k

Claims

DEMANDS 1. A computer method for estimating a current state of a user of at least one electronic device from among a plurality of predetermined states, the at least one device being used to perform an activity comprising at least one cognitive component, and the estimated state characterizing a risk to said user with regard to said activity, the method comprising: - obtaining (300) results from a test, the test comprising, for each of a plurality of iterations, . presentation (410) of at least one indication to aid in choosing an instruction to continue or stop a process whereby a continuation involves a chance of additional gain and a risk of loss of cumulative gain; . receipt (415), from the user, of an instruction to continue or stop the process; . execution (420) of said instruction received; - estimation (305) of the user's current state based on said results and an estimator comprising a behavioral model of the user.

2. A method according to claim 1, wherein the estimator further comprises at least one classifier.

3. A computer method for estimating a current state of a user of at least one electronic device from among a plurality of predetermined states, the at least one device being used to perform an activity comprising at least one cognitive component, and the estimated state characterizing a risk to said user with regard to said activity, the method comprising: - obtaining (300) results from a test, the test comprising, for each of a plurality of iterations, . receipt (415), from the user, of an instruction to continue or stop a process whereby a continuation involves a chance of additional gain and a risk of loss of cumulative gain; . execution (420) of said instruction received; - estimation (305) of the user's current state based on said results and an estimator comprising a behavioral model of the user and a classifier.

4. Method according to claim 3, further comprising a presentation (410) of at least one indication to aid in choosing an instruction to continue or stop the process.

5. A method according to claim 1, claim 2 or claim 4, wherein at least one indication includes a probability of additional gain or loss of cumulative gain related to the instruction considered.

6. A method according to claim 1, claim 2, claim 4 or claim 5, wherein at least one indication includes a tendency for at least one other user to select a continue or stop instruction.

7. A method according to any one of claims 1 to 6, wherein said results comprise at least one of the following: a total cumulative gain, a number of iterations, a number of iterations that resulted in a gain, a number of iterations that resulted in a loss, a number of instructions that resulted in a cumulative gain, a test execution time, and an average time between two iterations.

8. A method according to any one of claims 1 to 7, further comprising the presentation (315) of a warning, access to a service, personalization of a service or prohibition of access to a service based on said estimated current state.

9. A method according to any one of claims 1 to 8, comprising a preliminary parameterization of said estimator, the results obtained being initial results, the parameterization comprising: - creation (205, 205') of at least one model of said test; - obtaining second results of said test using said at least one model; - obtaining at least one expected state of a current user state corresponding to said second results; and - determination of at least one parameter of the estimator according to the second results and the expected result.

10. A method according to claim 9, comprising: - creation of a plurality of models of said test; - obtaining third results of said test; and - the selection (220) of a plurality model according to the third results and the results obtained from each model of said plurality.

11. Device comprising at least one data processing unit and one memory unit, the device being configured to implement each of the steps of the process according to any one of claims 1 to 10.

12. Computer program comprising instructions for carrying out each of the steps of the process according to any one of claims 1 to 10 when the program is executed by a computer.

Images