FR3035247A1 - METHOD AND SYSTEM FOR MONITORING A USER - Google Patents

Abstract

According to one aspect, the present invention relates to an intelligent system for the benevolence, detection and prevention of health risks for helping to keep people at risk of loss of autonomy at home. The present invention operates preferentially by detecting daily events, risk prevention and automatic home automation actions through a device communicating sensors and non-intrusive for the welfare of people without wearing a pendant, or remote video surveillance for total compliance with the rules. privacy and privacy of individuals.

Description

TECHNICAL FIELD OF THE INVENTION The present invention relates to the field of anticipating and monitoring health problems. More particularly, this invention relates to the sector of persons recovering from convalescence, loss of autonomy or suffering from cognitive disorders, and in particular the elderly sector. STATE OF THE ART The medical advances of recent decades have contributed to increasing the life expectancy of man in important ways. Today's society has to deal with the management and supervision of its elders. In addition, because of the growing longevity of humans, neurodegenerative diseases have emerged. These diseases result in loss of memory, spatio-temporal disorientation and motor malfunctions.

Along with this aging population, and requiring special attention on the part of society, interest in medical prediction is intensely strong. For example, the prevention of cardiovascular risks requires a statistical prediction: it is now possible to determine the probability that a person will suffer from cardiovascular disorders in his future, depending, for example, on data related to his activities and to his daily life. Thus, a strong will emerges in the medical world: Prevent rather than cure. With the recent technological advances concerning both the sensors but also the database analysis systems, it is now possible to express the probability that an individual will have to suffer or not, now or in the future. or less, of this or that health problem. Existing systems are mainly based on predictive algorithms with learning functions. The principle is simple: a typical profile is chosen for a user. Then an alert is triggered when the user's behavior differs from that of the profile beyond a threshold. This type of system has a significant disadvantage, namely the obtaining of a medical diagnosis sometimes too late, the premature care of some patients or the residence time too long in the clinic, the latter two disadvantages then leading to significant costs for medical and public services.

5 This basic profile is usually a typical profile. Indeed, the current systems have a database comprising some standard profiles theoretically established and / or resulting from the fusion of several real medical data. However, today's systems offer only a small number of profiles available to their users. This involves two major disadvantages. The first is the need for a system with highly efficient learning functions to learn as well as possible. The second drawback is that the time required for the proper functioning of the system is extremely long since it is necessary to create these type sheets before learning or after learning. As a result, current prediagnostic systems are more or less effective after a significant calibration time, leaving the user dependent on a system that is not yet operational. The present invention solves in part the disadvantages mentioned above. In addition, by design, the present invention improves the well-being of users at home since they feel monitored, not intrusively, but anonymously while identifying medical risks. The invention thus makes it possible to very greatly reduce the cost induced by clinical management.

SUMMARY OF THE INVENTION According to one aspect, the present invention relates to a method for monitoring at least one given user and detecting abnormal situations, characterized in that it comprises the following steps: A preliminary step comprising: - Selecting a user profile in at least one user profile database, said selected user profile 3035247 3 being an existing user profile made for another user; - Considering said selected existing user profile as a prior user profile of the given user; 5 - A plurality of steps performed repeatedly and comprising: - Collecting at least one data specific to the given user issued by at least one device; Analyzing at least one piece of data through at least one learning module comprising at least one learning rule; - Generate an updated user profile for the given user from said analysis step, and at least one prior user profile of the given user; - anonymously storing said updated user profile in the at least one database as an earlier user profile; - Make said previous user profile accessible as an existing user profile; - Confronting said updated user profile with at least one prior user profile of the given user so as to generate a comparative result; - Based on said comparative result generated in the confrontation step, generating at least one action; - Analyze the consequences on the given user of the action generation step based on data specific to the given user collected after the action generation step; - Modify the learning module according to the analysis step; Before or respectively after the step of generating at least one action, generating at least one prediction on the state of the given user according to the confrontation step or the generation step respectively Action. When performing said plurality of steps for the first time, the prior user profile of the given user is the selected existing user profile. Then, when performing said plurality of steps again for the nth time, the given user's previous user profile is the updated user profile that was generated when said plurality of steps were performed for the next time. N-1Ist time.

Based on the prior art, a person skilled in the art can easily devise solutions concerning the aid to diagnosis and medical forecasting. A first way of doing things is to use a database comprising only generic or standard profiles, that is to say, not very detailed and therefore remote from the user. These typical profiles will then have to be refined significantly during a more or less significant time of use of the system by the new user. This refinement step is essential for the system to work properly. This training time of the machine is therefore an important handicap for the rapid and safe implementation of this type of anticipated diagnostic system. A second way of proceeding is to generate a very large number of detailed user profiles so that the system is from the beginning of its operational use. However, this solution has a significant disadvantage, the time consumed to generate these profiles is very large, involving a significant monetary expenditure. Thus, each of these solutions have the same disadvantage, its operational time of implementation.

The present invention makes it possible to considerably increase the learning speed of the medical diagnostic aid and monitoring system since any user starts with a previously tested profile corresponding to another user. The fact that each profile is saved in a secure database makes it possible in a very short time to have a large number of precise profiles that can thus easily correspond to any new user. The learning and therefore the quality of the predictions of the present invention are optimized by this sharing of typical user profiles. Sharing profiles anonymously allows any new user to benefit from hundreds of hours, for example, perfecting a profile very close to his own so that the invention can reach an optimized prediction level. as quickly as possible without having to start with a typical profile. The database grows very quickly allowing any new user to benefit from an initial profile as refined as possible to its own characteristics and at the same time already proven and ready to evolve again. Each user in addition to being followed by the system of the present invention participates in the new user's watch by sharing his profile. In addition, by design, the present invention improves the welfare of users at home, delay their clinical management or advance their return home in case of return hospitalization. This ability to save money is possible through constant and innovative monitoring of the health status of users. Thus a significant money saving can be achieved at the level of medical services but also public.

According to another aspect, the present invention also relates to a computer program product comprising instructions which, when executed by at least one processor, performs the steps of the method described above.

Finally, in another aspect, the present invention also relates to a system for assisting in the medical diagnosis and monitoring of at least one given user, characterized in that it comprises: 3035247 6 - At least one database user profiles; - At least one device transmitting at least one data specific to the given user; - A learning module and at least one central unit 5 configured to analyze the at least one data through the learning module; - At least one central server; and that it is configured to perform the steps of the method as described above.

10 This system makes it possible to predict in a probabilistic manner the state of health of the user, to communicate automatically without user action and thus to take the corrective or preventive measures or emergencies necessary for his well-being .

BRIEF DESCRIPTION OF THE FIGURES The objects, objects, as well as the features and advantages of the invention will become more apparent from the detailed description of an embodiment thereof which is illustrated by the following accompanying drawings in which: FIGURE 1 shows the complete architecture, according to one embodiment, of the present invention. FIG. 2 represents a predictive method, according to an embodiment, of the present invention. FIG. 3 represents the multi-scale operation of the present invention, according to one embodiment. The drawings are given by way of examples and are not limiting of the invention. They are schematic representations of principle intended to facilitate the understanding of the invention and are not necessarily at the scale of practical applications.

DETAILED DESCRIPTION OF THE INVENTION Before going into the details of preferred embodiments, in particular with reference to the figures, various options are given below, which may preferably, but not exclusively, be presented in the invention, these options being possible. implemented, either separately or in any combination between them: Advantageously, each updated user profile is saved as an earlier user profile in said at least one database after the step of generating the updated user profile so as to generate an evolution history of the user profile. The backup of the updated user profile makes it possible at any time to have an increasing number of user profiles since at each step of comparing profiles, if there is any difference between the updated user profile and the previous user profile, the Updated user profile is saved while preserving the previous user profile. So that it is possible to follow the history of the evolution of the user profile and to go back if necessary, just as borrowed from any of the previous profiles as a user profile for a new user. As a result, for each user, there is a constantly growing number of previous user profiles feeding the user profile database. Advantageously, each updated user profile is saved as an earlier user profile in said at least one database after the step of generating the updated user profile. Advantageously, the method generates a history of evolution of the user profile. Advantageously, when said plurality of steps is performed the first time, the prior user profile of the given user is the existing user profile selected. Advantageously, when said plurality of steps is performed for the nth time, the prior user profile of the given user is the updated user profile that was generated when performing said plurality of steps for the N- 1st time. Advantageously, each saved previous user profile is made accessible to a new given user. The accessibility of these previous user profiles makes it possible to enrich the database very quickly since at each user profile update, it is usable as a first profile for any new user having characteristics similar to this prior user profile. . Advantageously, a step of modifying the learning module 20 is performed according to the analysis step. This step of modifying the learning module makes it possible to perfect said module according to the variations observed between the analysis of the system and the behavior of the user. This step thus makes it possible to significantly improve the accuracy of the forecasts. Advantageously, the action generation step is performed as a function of the evolution of at least one parameter with respect to the updated profile and at least one prior profile. Advantageously, the action generation step is performed as a function of the evolution of at least one parameter with respect to the updated profile and a plurality of previous profiles selected over a predetermined period. Advantageously, the action generation step is performed as a function of the evolution of at least one combination of parameters with respect to the updated profile and at least one prior profile. Advantageously, the action generation step is performed as a function of the evolution of at least one combination of parameters with respect to the updated profile and a plurality of previous profiles selected over a predetermined period. Advantageously, any user profile at any time is saved in said at least one database so as to generate a history of evolution of the user profile. This history makes it possible to follow the state of the user and optimize the learning system by comparing the modifications made and the possibility of reloading an earlier user profile if necessary. Advantageously, a self-diagnosis module of said learning module analyzes said history of evolution of the user profile and allows the optimization of said learning module.

This history of previous user profiles makes it possible to monitor the state of the user and optimize the learning system by comparing the modifications made and the possibility of reloading an earlier user profile if necessary. Advantageously, the step of selecting an existing user profile is performed automatically according to at least one medical or physiological characteristic of the given user.

This automatic selection increases the reliability of the present invention and saves time in a user profile search among the database. In addition, the automatic selection is done by simply recording certain characteristics specific to the given user serving as comparison markers for the determination of the most compatible prior profile. Advantageously, the step of selecting an existing user profile is performed manually according to at least one medical or physiological characteristic of the given user. This allows an experienced medical service, professional caregivers or certified social welfare services to directly choose the existing user profile for the given user based solely on accurate and medical data of preference. Advantageously, the step of collecting at least one datum transmitted by at least one device is performed in real time. This allows a real-time analysis of the user's situation and health condition. This real-time process allows a determination of possible accidents and / or health problems very quickly. Advantageously, said at least one data item is a data item taken from at least one physiological datum, an emotional datum, an environmental datum, a position datum, a consumption datum of a resource, a behavior datum.

This makes it possible to take into account various parameters in establishing the pre-diagnosis of the state of health of the user. The system is configured to cross all of these data and to allow a precise deduction of possible future health problems. Advantageously, said at least one device is a sensor selected from at least: a motion sensor, a sensor of at least one physical environmental parameter, an emotion sensor, a flow meter, an event sensor, a device for alert, a switch. This allows you to collect a lot of information about the user, his actions and his environment. Advantageously, the collection step comprises unidirectional communication between said at least one device and said at least one central unit. This makes it possible to use sensors that work well and are easy to use. Advantageously, the collection step comprises bidirectional communication between said at least one device and said at least one central unit.

This allows the use of latest generation sensors with increased functionality to maximize the collected data. Advantageously, the analysis step comprises the use of a method taken from at least: an artificial intelligence, a neural network, a learning algorithm.

This makes it possible to aggregate all the various data collected so as to be able to create a synergy between all of these data in order to extract a user-specific behavior while allowing the system to learn and to perfect oneself. Advantageously, the analysis step is performed at the level of at least one central unit. This makes it possible to perform the analysis step locally without resorting to a communication network external to the installation zone of the invention. Advantageously, the analysis step is performed at the level of at least one central server.

This allows the analysis step to be outsourced in order to have a larger computing resource and thus process more data and faster. Advantageously, the step of generating the current user profile is performed at said at least one central unit. This makes it possible to work locally without resorting to a communication network external to the installation area of the invention. Advantageously, the step of generating the current user profile is performed at the level of said at least one central server. This makes it possible to work off-site in order to have a larger computing resource and thus process more data and more quickly. Advantageously, the step of comparing user profiles is performed at said at least one central unit. This makes it possible to work locally without resorting to a communication network external to the installation area of the invention. Advantageously, the step of comparing user profiles is performed at the level of said at least one central server.

This allows for off-site work to have a larger computing resource to process more data and faster. Advantageously, said at least one action is an action taken among at least 15: displaying a visual message, sending a text message, sending an audio message. This allows the invention to communicate in a simple and versatile manner with the user and any other chosen and authorized person. Advantageously, the analysis step is performed at the level of said at least one central unit. This makes it possible to work locally without resorting to a communication network external to the installation area of the invention. Advantageously, the analysis step is performed at the level of said at least one central server.

This allows delocalized work to have a larger computing resource to process more data and faster. Advantageously, the analysis step is performed at said at least one central unit and / or at the level of said at least one central server. Advantageously, the step of modifying the learning module is performed at said at least one central unit. This makes it possible to work locally without resorting to a communication network external to the installation area of the invention. Advantageously, the step of modifying the learning module is performed at the level of said at least one central server.

This allows delocalized work in order to have a larger computing resource and thus process more data and faster. The present invention relates to a method of anticipating health problems, as well as monitoring the health status of a user. The invention applies in particular, but not restrictively, to the elderly, for example. This method is performed by a system comprising at least one central unit 1000, sensors (100, 200, 300, 400, 500), but also a central server 2000 and alerting devices 600.

The present invention includes at least one database 2300 having a plurality of user profiles. Each new user, called user B for example, is assigned an existing user profile. This existing user profile is intended to be modified as user B utilizes the present invention. Indeed, this existing user profile serves as a starting user profile for the creation of a profile specific to the user B, called updated user profile. This existing user profile is therefore used in order to configure the present invention a first time. This existing user profile comes from the database 2300 of existing user profiles of the present invention. According to a preferred embodiment, this existing user profile is the updated and / or previous user profile of another user, called user A. In fact, in order to allow the machine to progress and learn as quickly as possible a user profile, derived from a first learning process, from a user A having features common to the new user B is used.

According to one embodiment, which is also preferential, the database 2300 of the existing user profiles is configured so that the user profiles are stored in an anonymous form. This makes it possible to preserve the identity of the users. For example, database 2300 may have computer encryption to ensure privatization and controlled access to and content thereof. We will now describe in a non-limiting way the place of life of the user B of the present invention. We consider the place of life of the user B as being a house type dwelling comprising an entrance, a living room, a bedroom, a bathroom and a kitchen for example non-limiting. A set of sensors is arranged throughout the house. For example, an input / output sensor 500 is located at the input gate so that any input and output can be detected. Similarly motion detectors 100 are positioned at strategic points throughout the house so that they can easily know the movements of the user B but also his position. Regarding the chamber, a pressure sensor can be disposed at the bed so as to measure the rest time of the user B for example.

According to a very advantageous embodiment, the detector positioned at the bed also makes it possible to know the position of the user B in his bed but also his movements during sleep so as to provide enough information to the unit. central 1000 and the central server 2000 3035247 16 to extract information related to its sleep cycle and thus to know if the user B sleeps sufficiently or not. This sleep quality parameter may be an important consideration in establishing a diagnosis for predicting the health of user B. For example, a series of bad nights alone may be sufficient to explain some loss of memory and / or fatigue that could be experienced by the user B. Conversely, a series of bad nights can also prevent possible future falls due to the lack of rest of the user B. Light intensity sensors 300 can also be used to detect whether the person sleeps in the dark or not, thereby signaling anxious behavior for example in the case where the user B sleeps with the light on a regular basis. In the chamber may also be sound and / or visual warning devices 600 such as loudspeakers, and / or billboards or television screen for example non-limiting. These devices 600 can be used to alert and / or more generally to communicate with the user B and to inform him of various things, such as the need to get up in the morning, to eat, to wash or simply to date, time, and / or daily or future events.

In the living room sensors pressure can be arranged on all surfaces that can accommodate the user B as carpets, chairs, armchairs and sofa for example. As in the room, audible and / or visual warning devices 600 and / or information systems can be arranged in the living room so that user B can have easy access to them. The kitchen may have sensors 500 at the refrigerator and various closets so as to know whether the user B regularly opens or not the food storage spaces, and therefore feeds. A flowmeter 400 may also be positioned at water circuits connected to the kitchen so as to know whether the user B regularly or not his dishes for example.

Visual sensors 200 may also be arranged throughout the house so as to identify for example the facial expression of the user B and thus extract information related to his emotional state. The whole of this dwelling description is a non-limitative example, the present invention is partly configured to extract as optimally as possible information from various real and / or virtual sources to deduce a present state of health. and to extrapolate a future state of health of user B.

The present invention may, for example, relate to convalescent persons, but also to the elderly. Indeed, in a non-limiting manner, any person whose state of health can evolve quickly and / or over a longer time can be a user of the present invention. The sharing of user profiles anonymously allows any new user to benefit from hundreds of hours, for example, of refining an existing user profile very close to his own so that the invention can reach a level of optimized prediction as quickly as possible without having to start with a typical, generic or neutral user profile.

This feature is an advantageous element of the present invention and is based in part on at least one database 2300 gathering all the existing user profiles real-time updating, for example non-limiting. This database 2300 therefore grows very rapidly, allowing any new user to benefit from an existing user profile that is as refined as possible to its own characteristics and at the same time already proven and ready to evolve further. Each user in addition to being careful by the present invention participates in the watch of new users by sharing their existing and / or previous user profile.

According to one embodiment, each user profile at each modification is recorded in the database 2300 without erasing previous versions of the same user profile. So the 2300 database of profiles is growing very fast.

The database 2300 preferably has a restricted access and the data is preferentially encrypted so as to preserve the anonymity of the users.

According to a preferred embodiment, each updated user profile is saved regularly so as to generate a history illustrating the evolution of the user profile according to relevant parameters, for example, derived from the activities of the user. This history then has the advantage of allowing a restoration of the updated user profile to an earlier date, but also of optimizing the learning system by studying the choices made by it. Thus, according to one embodiment, the present invention comprises a self-diagnosis module of its learning system for its optimization.

Advantageously, the first prior profile used in the present invention corresponds to the existing user profile selected from an existing user. Then, as soon as a new user's own updated user profile is generated, his previous user profile is saved and can be considered as an earlier user profile. Figure 1 shows, in one embodiment, the architecture of the present invention. A series of sensors and / or detectors (100, 200, 300, 400, 500) are in communication, preferably bilateral, with at least one central unit 1000. According to one embodiment, the communication between the central unit and the least one sensor (100, 200, 300, 400, 500, 600) is bilateral. According to one embodiment, the communication between the central unit and all the sensors (100, 200, 300, 400, 500) is bilateral.

The two-way communication between the central unit 1000 and at least one sensor (100, 200, 300, 400, 500) allows the exchange of data such as measurements of the sensors (100, 200, 300, 400, 500). to the central unit 1000 and the sensor setting parameters (100, 200, 3035247, 300, 400, 500, 600) sent by the central unit 1000 to said sensors (100, 200, 300, 400, 500). The use of intelligent sensors (100, 200, 300, 400, 500) and high technology, latest generation, is possible.

According to one embodiment, the communication between the central unit and all the sensors (100, 200, 300, 400, 500) is unilateral. Unilateral communication between the central unit 1000 and at least some sensors (100, 200, 300, 400, 500) allows the use of simple and inexpensive sensors.

According to one embodiment, the devices 600 are, for example, warning devices. These may be visual and / or audible alerts to the user and / or persons other than the user. According to one embodiment, the alerting devices 600 include, for example, a display panel for sending messages 1440 to the user. These 1440 messages can be alerts but also reminders, such as "it's time for lunch" or "think about taking your medicine." Advantageously, warning messages 1440 can also be displayed, for example: "the air is charged with carbon monoxide, you are asked to evacuate the room" or "the temperature is abnormally low, heater ". According to one embodiment, the alerts may also be voice 1420, a prerecorded voice and / or a speech synthesizer may be used to inform the user of the alert and / or callback messages. Advantageously, the system can be configured so that the voice used corresponds to a voice known to the user, so that it is not worried about an unknown voice. According to one embodiment, the alert messages can also be sent by texto 1410, electronic messages 1410 and / or notification on a television 1440. According to one embodiment, confirmation of receipt of the alert messages and / or reminders can be requested from the user. The devices 600 then comprise confirmation means such as sound receiver, push button or simply the detection of the action performed. According to one embodiment, a timer can be taken into account for the recall of alert and callback messages in the case where the state of the situation at the origin of the alert or callback has not evolved in a satisfactory way for the system. According to one embodiment, a visual reminder 1440 may be followed by a sound alert 1430 in the case where the reminder would not have been sufficient for the user to act. Similarly, in the case of user inaction, the message may be relayed to one and / or persons registered in the system as to be contacted in the event of such an alert. Indeed, each alert may require the intervention of different actors. For example, in the case of a food or medical neglect, a family member of the user can be warned for it to pass and make sure to eat the user for example. For example, if the alert concerns the high level of carbon monoxide and the user is not acting, more specific services such as medical emergency services can be contacted first.

According to one embodiment, the sensors 100 are, for example, motion sensors. Advantageously, some of these motion sensors 100 comprise at least one thermal image sensor and / or infrared emitting diodes and / or infrared radiation detecting photodiodes.

Advantageously, this type of sensor 100 allows the detection of movement, fall, light, opening / closing of elements and / or the operation of electrical equipment. A certain advantage in the use of the sensors 100 is that the anonymity of the detected individuals is guaranteed. This anonymity may sometimes be desired by the user. According to one embodiment, the sensors 200 are, for example, emotion sensors. These emotion sensors 200 include, for example, facial analysis of at least one individual. This facial analysis then allows the determination of the emotional state of the individual. According to one embodiment, these emotion sensors 200 require a calibration phase during which the individual will express before the camera 5 a series of facial expression translating a certain number of emotions. This calibration then allows the system driving the present facility to detect the emotional state of the individual by simple facial analysis. Advantageously, the number of emotions that can be detected is between 1 and 20, preferably between 1 and 15, and advantageously equal to 13.

Advantageously, these emotion sensors 200 make it possible, for example, to detect presence, to recognize the person, to detect hands, to estimate the direction of the gaze, to estimate the age, to estimate feminine or masculine, the estimation of emotions of satisfaction, dissatisfaction, joy, surprise, anger, sadness or neutral emotion.

According to one embodiment, these emotion sensors 200 may include a voice analysis of the individual so as to extract emotional information from intonation, pitch and pitch based again on a calibration. prior.

According to one embodiment, the sensors 300 are environmental sensors. For example, these sensors may include sensors for temperature, humidity, air quality analysis, the amount of oxygen, the amount of carbon dioxide, the amount of carbon monoxide, atmospheric pressure, the amount of suspended particles and all other environmental variables. These environmental sensors 300 have the function of measuring the environment of the user. These measures then make it possible to alert the user, for example, and / or to control certain environmental systems in order to adjust this or that parameter.

For example, if the temperature of a room is too low compared to an established reference specific to the user and / or his condition, an alert can be signaled to the user via an electronic message, a text message, a text message, a visual alert such as a billboard, a television overprint and / or a voice alert. This alert indicates to the user that the temperature of the room is too low, for example. According to various possible embodiments, the system simply alerts the user, if it does nothing, the system can alert people registered in its database for this type of situation, and / or the system can directly control the environmental system to correct the problem, such as the radiator for example. In the case of air quality, a system of filtration or renewal of the air can be triggered.

According to one embodiment, the sensors 400 are flow meters for example. These flowmeters can be positioned at various strategic points of the pipes. The interest of flowmeters 400 is to be able to measure whether or not the user uses the washbasin of the bathroom, the toilets, the shower, the kitchen sink, the washing machine, the dishwasher, and / or any other system using water or a liquid circulating in a pipe. These sensors 400 may be flow meters indicating precisely the amount of liquid flowing in the pipe or simply a detector 20 to know that liquid flows in the pipe. This information is essential to prevent neurodegenerative disorders for example in the user. For example, forgetting to wash or incontinence problems is important information for a preventive diagnosis.

According to one embodiment, the sensors 500 are, for example, event sensors, so-called event sensors 500. These sensors 500 comprise, for example, pressure, contact, opening / closing, touch and / or or any other interaction or lack of physical interaction. For example, an open / close sensor can be installed at the refrigerator to determine if the user is feeding for example. Likewise, a sensor at the sanitary level makes it possible to determine in combination with the flow meter sensors 400 if the user uses the toilets regularly. A correlation between the opening of the refrigerator door and the use of sanitary can be for example important information for the medical diagnosis of the user.

Similarly, advantageously, a pressure sensor 500 can be installed at the level of the bed and / or the sofa and / or chairs in order to be able to measure the presence or absence of the user but also the time of presence / absence and the position of the user. According to one embodiment, the agitation during the sleep of the user 10 can also be measured by the event sensors 500. Advantageously, this information is important for the medical prediagnosis of the user. Indeed, the time spent lying or sitting or in bed is an indication of its physical condition but also psychic.

Preferably, an event / event sensor 500 for opening / closing is arranged at each room and / or dwelling outlet so as to be able to determine the output frequency for example of the user. According to one embodiment the sensors 500 have the function of detecting whether the pillbox of the user is open each day or not and at what time, for example. Advantageously, these sensors 500 make it possible to detect the forgetfulness of the taking of medication on the part of the user and thus to warn him and / or any other person registered for this function. Likewise, alerts similar to the alerts of the system of the present invention may be used. According to one embodiment, the central unit 1000 comprises wired and / or non-wired communication means with at least some sensors (100, 200, 300, 400, 500) and / or warning devices 600.

Advantageously, the central unit 1000 is configured to process and analyze all the information sent by the sensors. According to one embodiment a central unit 1000 is unique to a single user.

According to one embodiment, the central unit 1000 is specific to a plurality of users. According to one embodiment, the central unit is unique to a single system.

According to one embodiment, the central unit 1000 comprises a neural network for analyzing the information transmitted by the sensors (100, 200, 300, 400, 500) and / or the alerts devices 600. realization, the central unit 1000 comprises an artificial intelligence analyzing the information emitted by the sensors (100, 200, 300, 400, 500) and / or the alerts devices 600. According to one embodiment, the central unit 1000 is configured to generate a pre-diagnosis of the state of health of the user. To do this the system considers the profile of the user. This user profile is then reworked regularly by the CPU 1000 as the user uses the present invention. The user profile is the beginning of the existing user profile of another user with very similarities to the current user. This existing user profile serves as the basis for the central unit to analyze and prevent the health problems of the user. As time passes and user-specific data is collected and analyzed by CPU 1000, the user profile changes and becomes unique. This updated and unique user profile may, if necessary, also serve as a basis for establishing the user profile of another user. To this end, the set of updated user profiles are stored anonymously and securely in a database 2300. This database 2300 is contained in a data center configured to communicate with all the central units. 1000 deployed. Access to this database is restrictive and regulated. The central units 1000 access this database through at least one central server 2000.

According to one embodiment, a central server 2000 is common to a single central unit 1000. According to one embodiment, a central server 2000 is common to at least two central units 1000.

According to one embodiment, the central server 2000 comprises wired and / or non-wired communication means with the central unit 1000 and / or with a communication network such as the Internet, allowing the use and transmission of the data. access to the central server 2000 via a web interface 2100, and all types of communication system 2400 such as telephone, smartphone, tablet and / or computer. According to one embodiment, the central server 2000 is specific to a single central unit 1000.

According to one embodiment, the central server 2000 is specific to at least two central units 1000. Advantageously, the central server 2000 is connected to at least one data center 2300 comprising at least one database. Preferably, the central server 2000 is in communication with the actors 2200 corresponding to the persons registered and to be contacted in case of alerts, but also having access to at least some data from the central server 2000 and / or the unit 1000. Advantageously, these actors can be close to the user intended to be contacted if the user does not eat enough, for example.

According to one embodiment, these actors may be members of a medical emergency service, and / or the doctors treating the user. These actors may be, for example, engineers performing the maintenance of the present invention, as well as private and / or public agents using a portion of the data generated by the CPU 1000 for statistical purposes, for example. FIG. 2 represents a predictive method carried out by the central unit 1000. This predictive algorithm corresponds to a method which comprises at least the following steps: Collecting 1100 of the data transmitted by the sensors (100, 200, 300, 400, 500 and / or the alerting devices 600. - Analyzing 1200 said data through at least one learning module comprising at least one learning rule (1210a, 1210b, 1210c, 1210d). - Generate 1230 an updated user profile. 5 - Confront 1300 the updated user profile thus generated with at least one previous user profile. - Generate 1400 a series of actions specific to the result of the previous confrontation stage. Analyze 1500 the consequences of the preceding action generation step based on data collected after the action generation step 1400. - Modify 1510 the learning module according to the result of the previous analysis step 1500. Before or after the generation step 1400 of the series of actions, generating 1600 at least one prediction on the state of the user. According to one embodiment, the data collected in step 1100 is transmitted to an integrated services platform 1150. Advantageously, this platform is configured to have the possibility of directly transmitting raw data and / or the data collected at step 1100 to external data processing centers such as approved public bodies to allow real-time access to the entire system whether locally by the user and / or remotely by the persons authorized for such access , as close to the user and / or technicians and / or doctors of the user. According to one embodiment, step 1200 comprises analyzing the data extracted in step 1100. This analysis can be done, for example, by a neuronal system and / or an artificial intelligence and / or any suitable predictive algorithm. collecting a plurality of data and processing them all together and not one by one.

The present invention includes at least one algorithm configured to learn and predict the state of health of the user based on various information sources. Learning allows the algorithm to evolve and thus improve its predictive capabilities.

Step 1230 then consists in generating a new profile, the updated user profile of the user. This generation of the updated user profile is derived from the modification of the previous user profile based on the data scanned by the present invention. For example, if the user has broken his leg, this characteristic must be taken into account and significantly modify his user profile as it was before the accident. Indeed, the absence of physical activity should not, in this case, be taken as a health problem index but as a consequence of this accident. According to one embodiment, a significant modification of the characteristics of a user may result in the choice of a new existing user profile. Indeed, following certain events, the profile of the updated user may no longer correspond to the user, and the choice in the database 2300 of a new existing user profile is necessary. Updating the user profile is an advantageous element of the present invention. Indeed, each update is preferentially saved in the database, enriching it and allowing a faster and optimized learning for the systems specific to future users.

The confrontation 1300 of the updated user profile with at least one prior user profile makes it possible to take measurements and actions, the step 1400. Indeed, depending on this comparison, it may be necessary to indicate to the user various things like eating time or having to exercise. In addition, in some cases it is not only the user who is alerted but also specific individuals, who may be the family or health professionals. The alert may be broadcast via an electronic message 1410, a text message 1410, a voice message 3035247 28 1420 and 1430, a visual message 1440 and / or any other form of communication. For example, one of the actions may be a backup of certain data, such as the updated user profile of the user, in the central server 2000. In a particularly advantageous embodiment, the central server 2000 is accessible via various communication means such as: for example a mobile phone 2100, a computer 2300, another server 2200 and / or any other communication system 2400. Preferably, the central server 2000 is protected so as to allow access to the data only to authorized persons . According to a preferred embodiment of the present invention, when the method described in FIG. 2 is carried out for the first time, in the confrontation step 1300, the prior user profile of the given user is the previously selected existing user profile and belonging to another existing user. Then, when performing this process at least a second time, from the generation at step 1230 of an updated user profile of the given user, the prior user profile used in the confrontation step 1300 can be the user profile refreshed the first time the process was run.

More generally, and according to one embodiment, when performing the method of FIG. 2 for the nth time, N being different from 1, the prior user profile of the given user is the updated user profile which was generated when the process of Figure 2 was performed for the N-th time.

An analysis step 1500 of the results of the actions initiated in step 1400 also makes it possible to improve the user's updated user profile and to resort to new actions if necessary. For example if the user does not react or reacts. Any actions following step 1400 must also be analyzed to modify the updated user profile and learning rules, step 1510.

Finally, step 1600, in parallel with the generation of actions, makes it possible to establish a prediagnosis of the state of the user. This pre-diagnosis can also lead to a series of actions if necessary, such as the medical care of the user for example.

According to one embodiment, the analysis and information processing steps are performed at the level of the central unit. According to one embodiment, the analysis and information processing steps are performed at the central server.

Figure 3 illustrates, in accordance with an embodiment of the present invention, the implementation of the present invention in an environment. Each living area 3100a, 3100b and 3100c comprises, for example, a central unit 1000 connected to a central server 2000 according to the present invention, and is connected to at least one departmental server 4100a, 4100b, 4100c. The central server 2000 is advantageously configured to process mainly information in real time and in connection with the present moment, such as a fall, an alert, an incident for example. In addition this central server also deals with short-term prediction such as the probability of future falls, medical relapses, future health problems. These analyzes preferably relate to the user who is in the place of life connected to this central server. The departmental servers have a medium-term function and 25 for a larger user population. For example departmental servers can deal with statistical questions relating to the number of falls in relation to the age of users, the need to promulgate more or less care, a better classification of people at risk for example for a particular health problem. These analyzes concern a larger number of users and on a larger time scale as well.

Each of these departmental servers 4100a, 4100b, 4100c is connected to at least one regional server 5000. The regional server 5000 is connected to various public services. For example, the service 6100a may correspond to the urban planning service and may concern the construction of houses, adapted residences for certain populations, and spatial planning. The 6100b service may correspond to the training service for public service employees, for example, to train them to take care of certain populations. Finally, the service 6100c can correspond to the service of the jobs and the creation of post in the public sector in connection with the monitoring, the assistance and the accompaniment of certain populations for example. The regional server here again has a role of larger scale. For example, the regional server processes data for an even larger number of users and for issues whose operational kinetics are over an even longer time. It is a matter here of dealing with questions relating to the demographic evolution, the typology of the populations, the displacement of the population, ... All these domains are treated at the level of the regional server so as to be able to allocate more or fewer resources in both virtual and real places. For example, it may be possible to increase the number of staff trained to care for the elderly in the hospitals of a particular city, because the analyzes show an increase of 10% per year in the number of elderly people admitted to the geriatric services of this city. This increase can take effect by increasing recruitment quotas, but also by mobilizing resources for the training of personnel in this sector. According to one embodiment, the present invention relates to an intelligent system for the benevolence, detection and prevention of health risks for helping to keep people at risk of loss of autonomy at home. The present invention preferably operates by detecting daily events, risk prevention and automatic home actions 3035247 31 thanks to a device communicating sensors and non-intrusive for the welfare of people without wearing a pendant or remote video surveillance for full compliance privacy and privacy of individuals. The system works advantageously through an evolutionary algorithm 5 allowing learning as its use and based on the pace of life of the person. It adapts to any profile, any habitat because its installation is non-destructive, interacts with the user and communicates with the outside in an intuitive and automatic way. The system preferably allows the large-scale, real-time collection of anonymous demographic data, user profile trends according to specific geographic areas, regional, national and international, for macroeconomic utilities of general interest such as training. specialized personnel, for example, the allocation of specific resources or the construction of specialized establishments. The invention is not limited to the previously described embodiments and extends to all the embodiments covered by the claims. 20

Claims (20)

REVENDICATIONS1. A method of monitoring at least one given user, said method being executed by a system comprising at least one database (2300), at least one central server (2000) and at least one central unit (1000) in communication with the at least one sensor (100, 200, 300, 400, 500, 600), said method being characterized in that it comprises the following steps: - A preliminary step comprising: selecting a user profile in the at least one database ( 2300) of user profiles, said selected user profile being an existing user profile made for another user; - Considering said selected existing user profile as a prior user profile of the given user; A plurality of steps performed repeatedly and comprising: collecting (1100) at least one data specific to the given user transmitted by the at least one sensor (100, 200, 300, 400, 500, 600); said at least one data item being data taken from at least: physiological data, emotional data, environmental data, position data, consumption data of a resource, behavioral data; Analyzing (1200) the at least one data through at least one learning module (1200) having at least one learning rule (1210a, 1210b, 1210c, 1210d); Generating (1230) an updated user profile for the given user from said analyzing step (1200), and at least one prior user profile of the given user; Confronting (1300) said updated scheduler with at least one prior user profile of the given user (1310a, 1310b, 1310c, 1310d) so as to generate a comparative result; According to said comparative result generated in the confrontation step (1300), generating (1400) at least one action (1410, 1420, 1430, 1440, 2000); Analyzing (1500) the consequences for the given user of the action generating step (1400) (1410, 1420, 1430, 1440, 2000) based on data specific to the given user collected after the generation step (1400) of action; Before or respectively after the step of generating (1400) at least one action, generating (1600) at least one prediction 10 on the state of the given user according to the confrontation step (1300) or respectively the step of generating (1400) action; and wherein the step of analyzing (1200) the at least one data through at least one learning module, the step of generating (1230) an updated user profile, the step of of confrontation (1300), the analysis step (1500) of the consequences on the given user of the generation step (1400) of action (1410, 1420, 1430, 1440, 2000) according to own data to the given user collected after the generating step (1400) of action are performed at the at least one the central unit (1000) or at the at least one central server (2000) . The method of the preceding claim wherein each updated user profile is saved as an earlier user profile in said at least one database (2300) after the step of generating (1230) the updated user profile so as to generate a history of evolution of the user profile. 3. Method according to the preceding claim wherein each saved previous user profile is made accessible to a new user 30. 3035247 34 The method of any preceding claim wherein when said plurality of steps is performed the first time, the prior user profile of the given user is said selected existing user profile. 5 A method according to any one of the preceding claims wherein when said plurality of steps is performed for the nth time, the prior user profile of the given user is the updated user profile that was generated when performed said plurality of steps for the N-1 th time. 10 The method of any of the preceding claims wherein a step of modifying (1510) the learning module is performed according to the analyzing step (1500). 15 The method of any one of the preceding claims wherein the step of generating (1400) action (1410, 1420, 1430, 1440, 2000) is performed according to the evolution of at least one parameter by to the updated profile and at least one previous profile. 20 The method according to any one of the preceding claims wherein the step of generating (1400) action (1410, 1420, 1430, 1440, 2000) is performed according to the evolution of at least one combination of parameters with respect to the updated profile and at least one previous profile. 25 9. A method according to any one of the preceding claims wherein the step of selecting an existing user profile is performed automatically based on at least one medical or physiological characteristic of the given user. 30 The method of any one of claims 1 to 8 wherein the step of selecting an existing user profile is performed manually based on at least one medical or physiological characteristic of the given user. 3035247 11. Method according to any one of the preceding claims wherein the step of collecting (1100) at least one data transmitted by the at least one sensor (100, 200, 300, 400, 500, 600) is performed in real time. The method according to any one of the preceding claims wherein said at least one sensor (100, 200, 300, 400, 500, 600) is a sensor selected from at least one motion sensor, a sensor of at least a physical environmental parameter, an emotion sensor, a flow meter, an event sensor, an alarm sensor, a switch. 10 The method of any of the preceding claims wherein the collecting step (1100) comprises unidirectional communication between said at least one sensor and said at least one CPU (1000). 15 The method of any one of claims 1 to 12 wherein the collecting step (1100) comprises bidirectional communication between said at least one sensor and said at least one CPU (1000). 15. The method of any one of the preceding claims wherein the analyzing step (1200) comprises using a method taken from at least one: an artificial intelligence, a neural network, a learning algorithm. 16. A method according to any one of the preceding claims wherein said at least one action (1410, 1420, 1430, 1440, 2000) is one of at least one of: displaying a visual message, sending a message text, sending an audio message. 17. A method according to any one of the preceding claims wherein the step of modifying (1510) the learning module is performed at said at least one CPU (1000). 3035247 36 18. A method according to any one of claims 1 to 16 wherein the step of modifying (1510) the learning module is performed at said at least one central server (2000). 5 A computer program product comprising instructions, which when executed by at least one processor, performs the steps of the method according to any one of the preceding claims. 20. A system for monitoring at least one given user, characterized in that it comprises: at least one database (2300) of user profiles; At least one sensor (100, 200, 300, 400, 500, 600) emitting at least one data specific to the given user; A learning module and at least one central unit (1000) configured to analyze the at least one data item through the learning module (1200); At least one central server (2000); and configured to perform the steps of the method of any one of claims 1 to 18.