WO2024002904A1 - Method and apparatus for efficient operation of an aerosol generation device - Google Patents
Method and apparatus for efficient operation of an aerosol generation device Download PDFInfo
- Publication number
- WO2024002904A1 WO2024002904A1 PCT/EP2023/067165 EP2023067165W WO2024002904A1 WO 2024002904 A1 WO2024002904 A1 WO 2024002904A1 EP 2023067165 W EP2023067165 W EP 2023067165W WO 2024002904 A1 WO2024002904 A1 WO 2024002904A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- agd
- time
- data
- resolved
- probability
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 61
- 239000000443 aerosol Substances 0.000 title claims abstract description 18
- 230000000977 initiatory effect Effects 0.000 claims abstract description 6
- 238000004590 computer program Methods 0.000 claims abstract description 4
- 238000013473 artificial intelligence Methods 0.000 claims description 33
- 238000012549 training Methods 0.000 claims description 19
- 239000007788 liquid Substances 0.000 claims description 5
- 230000033001 locomotion Effects 0.000 claims description 5
- 230000036772 blood pressure Effects 0.000 claims description 2
- 238000012545 processing Methods 0.000 description 18
- 238000012360 testing method Methods 0.000 description 13
- 238000004891 communication Methods 0.000 description 6
- 230000005540 biological transmission Effects 0.000 description 5
- 238000013480 data collection Methods 0.000 description 5
- 238000010801 machine learning Methods 0.000 description 5
- 238000007726 management method Methods 0.000 description 5
- 238000013459 approach Methods 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 238000007781 pre-processing Methods 0.000 description 3
- 230000000391 smoking effect Effects 0.000 description 3
- 241000208125 Nicotiana Species 0.000 description 2
- 235000002637 Nicotiana tabacum Nutrition 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 230000015556 catabolic process Effects 0.000 description 1
- 239000003795 chemical substances by application Substances 0.000 description 1
- 235000019504 cigarettes Nutrition 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000008020 evaporation Effects 0.000 description 1
- 238000001704 evaporation Methods 0.000 description 1
- 230000001815 facial effect Effects 0.000 description 1
- 238000010438 heat treatment Methods 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000013618 particulate matter Substances 0.000 description 1
- 230000001007 puffing effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000004043 responsiveness Effects 0.000 description 1
- 239000000758 substrate Substances 0.000 description 1
- 238000002759 z-score normalization Methods 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A24—TOBACCO; CIGARS; CIGARETTES; SIMULATED SMOKING DEVICES; SMOKERS' REQUISITES
- A24F—SMOKERS' REQUISITES; MATCH BOXES; SIMULATED SMOKING DEVICES
- A24F40/00—Electrically operated smoking devices; Component parts thereof; Manufacture thereof; Maintenance or testing thereof; Charging means specially adapted therefor
- A24F40/50—Control or monitoring
- A24F40/57—Temperature control
-
- A—HUMAN NECESSITIES
- A24—TOBACCO; CIGARS; CIGARETTES; SIMULATED SMOKING DEVICES; SMOKERS' REQUISITES
- A24F—SMOKERS' REQUISITES; MATCH BOXES; SIMULATED SMOKING DEVICES
- A24F40/00—Electrically operated smoking devices; Component parts thereof; Manufacture thereof; Maintenance or testing thereof; Charging means specially adapted therefor
- A24F40/50—Control or monitoring
-
- A—HUMAN NECESSITIES
- A24—TOBACCO; CIGARS; CIGARETTES; SIMULATED SMOKING DEVICES; SMOKERS' REQUISITES
- A24F—SMOKERS' REQUISITES; MATCH BOXES; SIMULATED SMOKING DEVICES
- A24F40/00—Electrically operated smoking devices; Component parts thereof; Manufacture thereof; Maintenance or testing thereof; Charging means specially adapted therefor
- A24F40/10—Devices using liquid inhalable precursors
Definitions
- the present disclosure relates to a method, apparatus, preferably an aerosol generation device (AGD), as well as a corresponding computer program for efficient operation of the device.
- ALD aerosol generation device
- RRPs risk reducing products
- HNB heat-not-burn
- International patent application WO 2021210900 proposes a system which tries to detect the distance of the device to the mouth of the user and to initiate a preheating stage based on that distance to ready the device for the first puff.
- this implementation lacks robustness (e.g., the mouth of a user may be covered if the user has a beard) as well as practicability due to the complexity introduced by the corresponding camera module and image analysis.
- European patent application EP 3869427 discloses a smoking substitute device configured to generate an aerosol from an aerosol forming substrate, the device comprising at least one processor configured to run a machine learning model, wherein the machine learning model is configured to receive usage data relating to usage of the smoking substitute device by a user and based on that usage data to determine one or more predicted parameters relating to future usage of the smoking substitute device by the user.
- this implementation also fails to provide a robust way for efficiently operating an aerosol generation device while enabling the user to experience a pleasant first puff.
- US patent application 2020337382 discloses an aerosol delivery includes sensor (s) to produce measurements of properties during use of the device, and a processing circuitry to record data for a plurality of uses of the device, for each use of which the data includes the measurements of the properties.
- the processing circuitry is configured to build a machine learning model to predict a target variable, using a machine learning algorithm, at least one feature selected from the properties, and a training set produced from the measurements of the properties.
- the processing circuitry is configured to then deploy the machine learning model to predict the target variable, and control at least one functional element of the device based thereon.
- the device may also include a digital camera to capture an image of a face of an attempted user to enable facial recognition to alter a locked state of the device.
- this implementation also fails to provide a robust way for efficiently operating an aerosol generation device while enabling the user to experience a pleasant first puff.
- AGD aerosol generation device
- a 1st embodiment of the invention is a computer-implemented method for operating an aerosol generation device (AGD, the method comprising the steps of: receiving time-resolved (i.e., without a time dimension) sensor data of the AGD representing a state of the AGD; estimating a future starting point in time of a usage of the AGD based on the time-resolved sensor data of the AGD; estimating a first probability which is a likelihood of correctness of the estimated future starting point in time of the usage of the AGD; and initiating a preheating of the AGD only if the estimated first probability is greater than or equal to a first threshold.
- AGD aerosol generation device
- Initiating the preheating in advance e.g., by at least a few tenths of a second
- a puff is imminent results in the AGD having sufficient time to generate the necessary heat to be able to deliver a pleasurable (first) puff with sufficient ACM/TPM to the user.
- the usage of the AGD is also improved as an active user intervention (e.g., pushing a button) is no longer necessary.
- the first probability is high enough (i.e., greater than or equal to the first threshold) an unnecessary or preheating is avoided reducing the drain on/of the battery of the AGD resulting in an improved power management of the AGD. Such improvement extends the lifetime of the battery.
- the method further comprises: determining, based on an amount of liquid inside the AGD and/or a position of the AGD, a second probability indicating whether the preheating of the AGD can be initiated safely; wherein the preheating of the AGD is initiated only if the second probability is equal to or greater than a second threshold and the estimated first probability is greater than the first threshold.
- the method further comprises: providing a safety issue notification to a user of the AGD at a second point in time if the second probability is smaller than the second threshold.
- Providing a safety issue notification (e.g., a vibration, a vibration pattern, a message sent to (e.g., a display of) the AGD, or another device connected to the AGD) to the user of the AGD may allow the user of the AGD to immediately recognize the potential unsafe situation of the AGD.
- the system safety is thus further increased as potential unsafe situations may be identified and solved/removed faster. If the AGD is in the user's pocket or bag, the user might overlook such safety issue notification performed on the AGD.
- performing the safety issue notification on another device e.g., smartphone, smartwatch, or earphone
- another device e.g., smartphone, smartwatch, or earphone
- the method further comprises: redetermining the second probability within a predefined third time interval starting from the second point in time.
- the safety assessment is again executed within a predefined third time interval. Should the safety assessment result in a conclusion that the situation is no longer unsafe for preheating, the preheating may be initiated without again determining the first probability.
- the removal of the unsafe situation may occur by the user changing the situation/position of the AGD (e.g., the AGD is no longer in user’s pocket).
- the removal of the unsafe situation may also occur if the user has given his approval (e.g.., by giving a corresponding response to the safety issue notification) in case the safety assessment was too strict (i.e., a situation/position of the AGD has been assessed as being unsafe while in reality it was safe).
- the safety assessment was too strict (i.e., a situation/position of the AGD has been assessed as being unsafe while in reality it was safe).
- the system safety as well as the power management of the AGD is improved as unnecessary executions of method steps (e.g., estimating the first probability) are avoided.
- the method further comprises: determining a first time interval between a first point in time and the estimated future starting point in time; wherein the preheating of the AGD is initiated only if the first time interval is smaller than or equal to a second, predetermined time interval; wherein the first point in time is a point in time of the receiving of the time-resolved sensor data, a point in time of estimating the first probability, or a point in time between the point in time of the receiving and the point in time of the estimating; and wherein the second time interval corresponds to an amount of time necessary for preheating the AGD.
- Preheating of the AGD is only initiated if the estimated future starting point in time of the usage of the AGD is close enough timewise. Should for example the first (and the second probability) be high enough, the preheating of the AGD would be initiated. However, if the AGD requires only a certain time for preheating (the second time interval) and the estimated future starting point of the usage of the AGD is further away (i.e., current point in time + second time interval ⁇ estimated future starting point of the usage), initiating the preheating may be unnecessary. This is because the AGD would stay in a heated mode for an unnecessarily long time. Staying in the heated mode for an unnecessarily long time may result in the degradation of aerosol taste.
- the AGD will repeat estimating the usage of the AGD. Therefore, preheating of the AGD will only be initiated if the future point in time is close enough (i.e., equal to or smaller than the second time interval). As a result, the AGD is preheated just-in time, which improves the power management of the AGD.
- the future starting point in time and the first probability are estimated by an artificial intelligence model.
- the artificial intelligence model has been trained according to a training method comprising the steps of: inputting training data to the artificial intelligence model to train the model, the training data including a plurality of data samples, each data sample comprising: time-resolved sensor data of a first AGD representing a state of the first AGD; and time-stamped usage data of the first AGD representing a starting point in time of the usage of the first AGD associated with the time-resolved data of the first AGD; and correlating the state of the first AGD and the starting point in time of the usage of the first AGD based on the plurality of data samples; wherein the model is configured to estimate the likelihood of correctness of the estimated future starting point in time of the usage of the AGD based on the correlation.
- Training the artificial intelligence model on a plurality of data samples allows the model to learn/train the correlation/relation between a state of the AGD represented by time- resolved sensor data and an associated starting point in time of a usage of the AGD. Accordingly, the trained model is able to predict a future starting point in time and a corresponding probability for an unknown input state (i.e., a state which was not comprised in the training data).
- Such a model is more accurate than conventional approaches (e.g., rule-based approaches), resulting in a more precise estimation of whether a usage of the AGD is imminent or not. Therefore, the operating of the AGD is improved.
- a further embodiment of the invention is a computer-implemented method for training an artificial intelligence for use in estimation of a future starting point in time of a usage of the AGD and/or a first probability according to the embodiments 6 and 7.
- the training data is processed before being used as input for training the mode; wherein the training data processing comprises at least one of the following steps: collecting a plurality of time-resolved sensor data and a plurality of time-stamped data representing starting points in time of the usage; generating the plurality of time resolved sensor data with a corresponding time-stamped usage data of the plurality of time-stamped data; and/or preprocessing the plurality of data samples according to a preprocessing configuration.
- Another embodiment of the invention is an artificial intelligence model for use in estimation of a first probability, which has been trained according to the computer-implemented method for training an artificial intelligence model for use in estimation of a future starting point in time of a usage of the AGD and/or a first probability according to some embodiments of the present invention.
- the preprocessing configuration comprises at least one of: a feature scaling, preferably at least one of a min-max normalization, a mean normalization, a z-score normalization and/or a scaling to unit length; an imputation configuration, preferably one of a mean imputation, a median imputation and/or a most-frequent imputation; and/or a training-validation-testing split configuration.
- the trained artificial intelligence model is employed on the AGD.
- Employing the trained artificial intelligence model on the AGD may reduce the latency/delay between the receiving of the time-resolved sensor data representing the state of the AGD and the executing/estimation. As a result, the method can be executed more frequently, which improves operation of the AGD regarding a just-in-time preheating.
- a potential future update of the model e.g., a retrained model on additional data
- the firmware of the AGD or of the above-mentioned thresholds and/or time intervals may be obtained via corresponding interfaces (e.g., via a wireless communication interface allowing over-the-air updates and/or via a traditional/wired communication interface like USB).
- the trained artificial intelligence model is employed on a location remote from the AGD; preferably a server or a cloud instance.
- the time resolved sensor data of the AGD representing the state of the AGD is transmitted to the remote location; and a command from the remote location to initiate the preheating of the AGD is received.
- a point in time of a conducted usage of the AGD corresponding to the transmitted time resolved sensor data may be transmitted to the remote location; a data sample may be created comprising the point in time of the conducted usage of the AGD and the corresponding time resolved sensor data; and the artificial intelligence model may be retrained using the created data sample.
- the artificial intelligence model can be adapted more flexibly and frequently, as the artificial intelligence model is continuously served with new usage data. Therefore, a model which is more customized to the user of the AGD can be created/trained, and which can better analyze the user’s patterns, resulting in an improved preheating. Furthermore, it may be possible to host several artificial intelligence models for one AGD in case the AGD is used by several users.
- a user of the plurality of users using the AGD at a certain moment may select his/her user profile (e.g., directly on the AGD or on another device connected to/with the AGD) which activates the corresponding artificial intelligence model.
- the AGD comprises a corresponding communication interface, which also allows over-the-air updates as in the above-described embodiment.
- the time resolved sensor data of an AGD includes at least one of: time-resolved data from a gyroscope (preferably Micro Electro-Mechanical System (MEMS) based) comprised by the AGD; time-resolved data from a proximity sensor (preferably optical time-of-flight (ToF) or triangulation sensors operating in the near infra-red wavelength regime) comprised by the AGD; time-resolved data from a draw sensor comprised by the AGD; time-resolved data from a lip contact sensor (e.g., capacitive (touch) sensors or mechanical switches engaged by contact with the lips) comprised by the AGD (e.g., located on the mouth piece of the AGD); time-resolved data from a sensor comprised by the AGD and detecting a fill level of the AGD; time-resolved data from one or more temperature sensors measuring the temperature of a heater
- MEMS Micro Electro-Mechanical System
- the state of an AGD represents at least a motion of the AGD in space and/or a position of the AGD in space.
- the time-stamped usage data is data from a puff sensor, preferably a draw-activated puff sensor.
- the usage of an AGD is a puff of the AGD.
- a 14th embodiment of the invention is an aerosol generation device (AGD) comprising means for carrying out the method of any one of the preceding embodiments.
- ALD aerosol generation device
- a 15th embodiment of the invention is a computer program which when executed causes a computing device or a system to perform a method of any one of the above mentioned embodiments, preferably any one of the 1st to 13th embodiments.
- Fig. 1 illustrates an aerosol generation device (AGD) according to an embodiment of the invention.
- Fig. 2 illustrates a (computer-implemented) method for operating an AGD according to an embodiment of the invention.
- Fig. 3 illustrates a system for creating an AGD according to an embodiment of the invention.
- Fig. 4 illustrates a (computer-implemented) method for creating an AGD according to an embodiment of the invention.
- Fig. 1 shows an aerosol generation device (AGD) 110 according to an embodiment of the invention.
- the AGD 110 may be capable of/configured to perform(ing) the method 200 for operating an AGD 110.
- the AGD 110 may comprise sensors like a gyroscope, a proximity sensor (configured to measure the distance between the AGD 110 and the face of the user 120) and/or additional sensor(s) as described in corresponding embodiments.
- the AGD 110 may also comprise a mouthpiece, a data processing module and a heater. In general, the mouthpiece is positioned at the position closest to the face of the user 120. If the proximity sensor is installed on the mouthpiece, the proximity sensor may measure and provide a precise distance between the AGD 110 and the face of the user 120.
- the mouthpiece may be detachable from the AGD 110 for cleaning purposes. If the proximity sensor is installed on such a detachable part, the implementation may become more complex due to additional requirements on the electrical structure. Accordingly, in this embodiment, the proximity sensor is installed on the AGD 110 itself. This allows for a simpler implementation, while still being capable of measuring and providing a distance between the AGD 110 and the face of the user 120 which is precise enough. Alternatively, the proximity sensor may be installed on the mouthpiece, The AGD 110 may be used by a user 120. The sensor(s) of the AGD 110 may collect (time-resolved) sensor data which may represent a state of the AGD 110. The sensor data may be transmitted to/received by the data processing module of the AGD 110.
- the data processing module may perform the computer- implemented method for operating an AGD 110 according to some embodiments of the present invention (e.g., the method 200). Accordingly, if the data processing module determines (via estimation) that a usage/puff of the AGD 110 by the user 120 is imminent (i.e., the first probability is greater than or equal to the first threshold), the data processing module initiates preheating of the AGD 110, which causes the heater of the AGD 110 to heat up to a corresponding temperature (e.g., 250 to 300 degree Celsius).
- a corresponding temperature e.g. 250 to 300 degree Celsius
- Fig. 2 shows a computer-implemented method 200 for operating an AGD 110 according to an embodiment of the invention.
- the time-resolved sensor data representing a state of the AGD may be received (for example by the data processing module of the AGD 110).
- the data processing module may include a predictive model (e.g., conventionally programmed or an artificial intelligence model which may be trained according to some embodiments of the present invention) for estimating a future starting point in time of a usage of the AGD 110 based on the time-resolved sensor data of the AGD 110.
- a predictive model e.g., conventionally programmed or an artificial intelligence model which may be trained according to some embodiments of the present invention
- the predictive model may estimate a first probability (Pl) which may be a likelihood of correctness of the estimated future starting point in time of the usage (e.g., a puff) of the AGD 100, for example by the user 120.
- Pl a first probability
- "t" represents an estimated next puff (or draw) timing
- "x" represents the length of the time interval
- “y” represents a first threshold. If the first probability (Pl) value is large, the next puff (or draw) is most likely performed before the length of the time interval (x) is lapsed. On the other hand, if the first probability (Pl) value is small, the next puff (or draw) is most likely not performed before the length of the time interval (x) is lapsed.
- a large first probability (Pl) value may also mean that the estimated future starting point is close.
- a small first probability (Pl) may also mean that the estimated future starting point is far.
- the estimated first probability e.g., 80%
- the first threshold e.g. 60%
- preheating of the AGD 110 may be initiated in step 240 (e.g., via a preheating command of the data processing command to the heater of the AGD 110).
- the first probability is smaller (e.g., 50%) than the first threshold (e.g., 60%)
- preheating may not be initiated.
- the method 200 may then terminate and may be re-executed according to a configured frequency/period (e.g., pre-configured and fixed to execute each 0.5 s or a frequency/period which may adapt to for example the current power state of the AGD 110).
- a configured frequency/period e.g., pre-configured and fixed to execute each 0.5 s or a frequency/period which may adapt to for example the current power state of the AGD 110.
- the method may continue with step 230 after step 220 if the first probability is/was greater than or equal to the first threshold.
- a second probability P2 may be determined/estimated indicating whether the preheating of the AGD 110 can be initiated safely.
- Such a safety assessment may be based on for example an amount of liquid inside the AGD 110 and/or a position of the AGD 110.
- a second threshold z
- the preheating of the AGD 110 may be initiated in step 240. Otherwise, should the second probability be smaller than the second threshold, preheating of the AGD 110 may not be initiated.
- the method 200 may then terminate and may be re-executed according to a configured frequency/period.
- the AGD 110 can provide a safety issue notification (e.g., a vibration, a vibration pattern, a message sent to (e.g., a display of) the AGD, or another device connected to the AGD) to the user of the AGD 110. If the AGD is in the user's pocket or bag, the user might overlook such safety issue notification performed on the AGD. Thus, performing the safety issue notification on another device (e.g., smartphone, smartwatch, or earphone) may be advantageous in such situations.
- step 230 may be executed again within a predefined third time interval.
- step 230 If the result of step 230 shows that the situation is no longer unsafe for preheating (i.e., the re-calculated second probability is larger than the second threshold), the preheating may be initiated without again determining the first probability. As a result, the system safety as well as the power management of the AGD 110 is improved as unnecessary executions of method steps (e.g., estimating the first probability) are avoided.
- step 230 is performed before step 220. It may also be possible that some of the steps of method 200 are performed in a parallel manner to for example decrease the execution time of the method 200. It may also be possible that the method 200 comprises additional method steps as described with respect to some embodiments of the present invention.
- Fig. 3 shows a system 300 comprising a test AGD 310 for creating (i.e., setting up) an AGD 110 according to an embodiment of the invention.
- the test AGD 310 may comprise sensors like a gyroscope, proximity sensor (configured to measure the distance between the face of the test user) and/or additional sensor(s) as described in corresponding embodiments.
- the test AGD 310 may also comprise a mouthpiece, a transmission module, and a draw detector.
- the test AGD 310 may be used by a test user 320.
- the system 300 may further comprise a data collection device 340 which may at least comprise a reception module.
- the test AGD 310 may collect data of the usage of the test AGD 310 by the test user 320.
- the collected data may be transmitted to the data collection device 340.
- the data may then be used to develop for example the data processing module of the AGD 110 which is capable of/configured to perform(ing) the method 200 for operating the AGD 110 according to some embodiments of the present invention.
- the transmission module may be omitted from the AGD 110. Otherwise, such transmission module may be mounted on the AGD 110.
- the deployed artificial intelligence model can be optimized by communication with data collection device 340. In other words, the data collection device 340 can optimize or re-train the deployed artificial intelligence model by using collected data from sensors mounted on the AGD 110. Since such data is collected from actual user of the AGD 110, the deployed artificial intelligence model can be optimized and thus the prediction quality.
- the data may also be used to train/develop an artificial intelligence model according to some embodiments of the present invention, which may then be deployed on the AGD 110 as part of the data processing module of the AGD 110.
- the artificial intelligence model may be deployed in a remote location as explained with respect to some embodiments. Accordingly, the data processing module of the AGD 110 may then be configured to communicate with the remote location via a communication interface.
- Fig. 4 shows a computer-implemented method 400 for creating (i.e., setting up) an AGD 110 according to an embodiment of the invention.
- Method 400 may be performed by the system 300.
- the transmission module of a test AGD 310 may collect usage data from several sensor sources of the test AGD 310 being used by a test user 320 and may transmit the data to a data collection device 340 comprising at least a reception module.
- a predictive model may be developed in step 440.
- the collected data may also be used for developing and training an artificial intelligence model according to some embodiments of the present invention in step 430, which may be used as the predictive model.
- the predictive model may then be deployed on the AGD 110 as part of the data processing module or at a remote location, in which case the data processing module of the AGD 110 may be configured to communicate with the remote location via a communication interface.
Abstract
A computer-implemented method for operating an aerosol generation device (AGD) as well as a corresponding computer program and a corresponding AGD is disclosed. The method comprises the steps of receiving time-resolved sensor data of the AGD representing a state of the AGD, estimating a future starting point in time of a usage of the AGD based on the time-resolved sensor data of the AGD, estimating a first probability which is a likelihood of correctness of the estimated future starting point in time of the usage of the AGD and initiating a preheating of the AGD only if the estimated first probability is greater than or equal to a first threshold.
Description
Method and Apparatus for efficient Operation of an Aerosol Generation Device
Field of the Invention
[0001] The present disclosure relates to a method, apparatus, preferably an aerosol generation device (AGD), as well as a corresponding computer program for efficient operation of the device.
Background of the Invention
[0002] Poor responsiveness of risk reducing products (RRPs) may be perceived as a nuisance by consumers who are used to light a cigarette and immediately puff away.
[0003] Currently, commercial heat-not-burn (HNB) tobacco devices, which may be a kind of RRP, require pre-heating periods of around 20 - 40 seconds to bring the tobacco to the right temperature for a pleasurable first puff. In open tank e-vapor systems, consumers typically start the heat-up phase (i.e., the preheating) by the push of a button prior to puffing. Even e-vapor pod devices such as the Logic 1.1 require several puffs to heat up sufficiently and provide a satisfactory experience. Replacing the direct e-liquid mechanism of a wick and coil system with a heater-in- device (HID) approach further increases the heat-up time to achieve sufficient evaporation rates of e-liquid, because the heat has to be conducted through an interface material from the heater into the pod. This will diminish the first puff aerosol collected mass (ACM)/total particulate matter (TPM) delivered by draw-activated devices.
[0004] International patent application WO 2021210900 proposes a system which tries to detect the distance of the device to the mouth of the user and to initiate a preheating stage based on that distance to ready the device for the first puff. However, this implementation lacks robustness (e.g., the mouth of a user may be covered if the user has a beard) as well as practicability due to the complexity introduced by the corresponding camera module and image analysis.
[0005] European patent application EP 3869427 discloses a smoking substitute device configured to generate an aerosol from an aerosol forming substrate, the device comprising at least one processor configured to run a machine learning model, wherein the machine learning model is configured to receive usage data relating to usage of the smoking substitute device by a user and based on that usage data to determine one or more predicted parameters relating to future usage of the smoking substitute device by the user. However, this implementation also fails to provide a robust way for efficiently operating an aerosol generation device while enabling the user to experience a pleasant first puff.
[0006] US patent application 2020337382 discloses an aerosol delivery includes sensor (s) to produce measurements of properties during use of the device, and a processing circuitry to record data for a plurality of uses of the device, for each use of which the data includes the measurements of the properties. The processing circuitry is configured to build a machine learning model to predict a target variable, using a machine learning algorithm, at least one feature selected from the properties, and a training set produced from the measurements of the properties. The processing circuitry is configured to then deploy the machine learning model to predict the target variable, and control at least one functional element of the device based thereon. The device may also include a digital camera to capture an image of a face of an attempted user to enable facial recognition to alter a locked state of the device. However, this implementation also fails to provide a robust way for efficiently operating an aerosol generation device while enabling the user to experience a pleasant first puff.
Summary of the Invention
[0007] It is thus an object of the present invention to increase the efficiency of operating an aerosol generation device (AGD). This may inter alia be achieved by efficient preheating of the AGD. Due to such an efficient preheating of the AGD a user of the AGD may experience a more pleasant use of the AGD. Furthermore, the efficient preheating may result in a more efficient power management of the AGD. It is a further object of the invention to increase the safety when operating the AGD by enabling the preheating only in situations when this is safe.
[0008] One or more of these objects are achieved by the subject-matter of the independent claims. Preferred embodiments are subject of the dependent claims.
[0009] A 1st embodiment of the invention is a computer-implemented method for operating an aerosol generation device (AGD, the method comprising the steps of: receiving time-resolved (i.e., without a time dimension) sensor data of the AGD representing a state of the AGD; estimating a future starting point in time of a usage of the AGD based on the time-resolved sensor data of the AGD; estimating a first probability which is a likelihood of correctness of the estimated future starting point in time of the usage of the AGD; and initiating a preheating of the AGD only if the estimated first probability is greater than or equal to a first threshold.
[0010] Initiating the preheating in advance (e.g., by at least a few tenths of a second) if a puff is imminent results in the AGD having sufficient time to generate the necessary heat to be able to deliver a pleasurable (first) puff with sufficient ACM/TPM to the user. The usage of the AGD is also improved as an active user intervention (e.g., pushing a button) is no longer necessary. Furthermore, as preheating is only initiated if the first probability is high enough (i.e., greater than or equal to the first threshold) an unnecessary or preheating is avoided reducing the drain on/of the battery of the AGD resulting in an improved power management of the AGD. Such improvement extends the lifetime of the battery.
[0011] According to a 2nd embodiment of the invention, in the 1st embodiment, the method further comprises: determining, based on an amount of liquid inside the AGD and/or a position of the AGD, a second probability indicating whether the preheating of the AGD can be initiated safely; wherein the preheating of the AGD is initiated only if the second probability is equal to or greater than a second threshold and the estimated first probability is greater than the first threshold.
[0012] Initiating the preheating of the device based on a further probability assessing whether the preheating can be initiated safely further increases the efficiency of the AG operating in terms of increased system safety (i.e., the safety of the AGD as well as of the user). Accordingly, a situation can be avoided where a preheating of an AGD is automatically initiated even though the AGD is in a situation (e.g., the user’s pocket or bag) which may potentially be dangerous for the user. By employing both probabilities, not only the preheating, but also the safety is improved at the same time.
[0013] According to a 3rd embodiment of the invention, in the 2nd embodiment, the method further comprises: providing a safety issue notification to a user of the AGD at a second point in time if the second probability is smaller than the second threshold.
[0014] Providing a safety issue notification (e.g., a vibration, a vibration pattern, a message sent to (e.g., a display of) the AGD, or another device connected to the AGD) to the user of the AGD may allow the user of the AGD to immediately recognize the potential unsafe situation of the AGD. The system safety is thus further increased as potential unsafe situations may be identified and solved/removed faster. If the AGD is in the user's pocket or bag, the user might overlook such safety issue notification performed on the AGD. Thus, performing the safety issue notification on another device (e.g., smartphone, smartwatch, or earphone) may be advantageous in such situations.
[0015] According to a 4th embodiment of the invention, in the 2nd or 3rd embodiments, the method further comprises: redetermining the second probability within a predefined third time interval starting from the second point in time.
[0016] In a situation where it was determined that preheating cannot be initiated safely and maybe even a safety issue notification was provided to the user, but the first probability was greater than or equal to the first threshold (i.e., an imminent puff was detected/predicted), the safety assessment is again executed within a predefined third time interval. Should the safety assessment result in a conclusion that the situation is no longer unsafe for preheating, the preheating may be initiated without again determining the first probability. The removal of the unsafe situation may occur by the user changing the situation/position of the AGD (e.g., the AGD is no longer in user’s pocket). The removal of the unsafe situation may also occur if the user has given his approval (e.g.., by giving a corresponding response to the safety issue notification) in case the safety assessment was too strict (i.e., a situation/position of the AGD has been assessed as being unsafe while in reality it was safe). As a result, the system safety as well as the power management of the AGD is improved as unnecessary executions of method steps (e.g., estimating the first probability) are avoided.
[0017] According to a 5th embodiment of the invention, in any one of the preceding embodiments, the method further comprises: determining a first time interval between a first point
in time and the estimated future starting point in time; wherein the preheating of the AGD is initiated only if the first time interval is smaller than or equal to a second, predetermined time interval; wherein the first point in time is a point in time of the receiving of the time-resolved sensor data, a point in time of estimating the first probability, or a point in time between the point in time of the receiving and the point in time of the estimating; and wherein the second time interval corresponds to an amount of time necessary for preheating the AGD.
[0018] Preheating of the AGD is only initiated if the estimated future starting point in time of the usage of the AGD is close enough timewise. Should for example the first (and the second probability) be high enough, the preheating of the AGD would be initiated. However, if the AGD requires only a certain time for preheating (the second time interval) and the estimated future starting point of the usage of the AGD is further away (i.e., current point in time + second time interval < estimated future starting point of the usage), initiating the preheating may be unnecessary. This is because the AGD would stay in a heated mode for an unnecessarily long time. Staying in the heated mode for an unnecessarily long time may result in the degradation of aerosol taste. As the computer-implemented method for operating the AGD according to the present invention may be periodically executed, the AGD will repeat estimating the usage of the AGD. Therefore, preheating of the AGD will only be initiated if the future point in time is close enough (i.e., equal to or smaller than the second time interval). As a result, the AGD is preheated just-in time, which improves the power management of the AGD.
[0019] According to a 6th embodiment of the invention, in any one of the preceding embodiments, the future starting point in time and the first probability are estimated by an artificial intelligence model.
[0020] Using an artificial intelligence model increases the accuracy of the estimation resulting in an improved preheating operation of the AGD.
[0021] According to a 7th embodiment of the invention, in the 6th embodiment, the artificial intelligence model has been trained according to a training method comprising the steps of: inputting training data to the artificial intelligence model to train the model, the training data including a plurality of data samples, each data sample comprising: time-resolved sensor data of a first AGD representing a state of the first AGD; and time-stamped usage data of the first AGD
representing a starting point in time of the usage of the first AGD associated with the time-resolved data of the first AGD; and correlating the state of the first AGD and the starting point in time of the usage of the first AGD based on the plurality of data samples; wherein the model is configured to estimate the likelihood of correctness of the estimated future starting point in time of the usage of the AGD based on the correlation.
[0022] Training the artificial intelligence model on a plurality of data samples allows the model to learn/train the correlation/relation between a state of the AGD represented by time- resolved sensor data and an associated starting point in time of a usage of the AGD. Accordingly, the trained model is able to predict a future starting point in time and a corresponding probability for an unknown input state (i.e., a state which was not comprised in the training data). Such a model is more accurate than conventional approaches (e.g., rule-based approaches), resulting in a more precise estimation of whether a usage of the AGD is imminent or not. Therefore, the operating of the AGD is improved.
[0023] A further embodiment of the invention is a computer-implemented method for training an artificial intelligence for use in estimation of a future starting point in time of a usage of the AGD and/or a first probability according to the embodiments 6 and 7. In this further embodiment, the training data is processed before being used as input for training the mode; wherein the training data processing comprises at least one of the following steps: collecting a plurality of time-resolved sensor data and a plurality of time-stamped data representing starting points in time of the usage; generating the plurality of time resolved sensor data with a corresponding time-stamped usage data of the plurality of time-stamped data; and/or preprocessing the plurality of data samples according to a preprocessing configuration.
[0024] Another embodiment of the invention is an artificial intelligence model for use in estimation of a first probability, which has been trained according to the computer-implemented method for training an artificial intelligence model for use in estimation of a future starting point in time of a usage of the AGD and/or a first probability according to some embodiments of the present invention.
[0025] According to another embodiment of the invention, the preprocessing configuration comprises at least one of: a feature scaling, preferably at least one of a min-max normalization, a
mean normalization, a z-score normalization and/or a scaling to unit length; an imputation configuration, preferably one of a mean imputation, a median imputation and/or a most-frequent imputation; and/or a training-validation-testing split configuration.
[0026] Processing the training data accordingly before training the artificial intelligence model results in an optimized training. Thus, the prediction accuracy of the artificial intelligence mode is increased, which again results in a further improvement of the preheating procedure of an AGD.
[0027] According to an 8th embodiment of the invention, in any one of the 6ths to 7th embodiments, the trained artificial intelligence model is employed on the AGD.
[0028] Employing the trained artificial intelligence model on the AGD may reduce the latency/delay between the receiving of the time-resolved sensor data representing the state of the AGD and the executing/estimation. As a result, the method can be executed more frequently, which improves operation of the AGD regarding a just-in-time preheating. A potential future update of the model (e.g., a retrained model on additional data) or of the firmware of the AGD or of the above-mentioned thresholds and/or time intervals may be obtained via corresponding interfaces (e.g., via a wireless communication interface allowing over-the-air updates and/or via a traditional/wired communication interface like USB).
[0029] According to a 9th embodiment of the invention, in any one of the 6ths to 7th embodiments, the trained artificial intelligence model is employed on a location remote from the AGD; preferably a server or a cloud instance. Even more preferably, the time resolved sensor data of the AGD representing the state of the AGD is transmitted to the remote location; and a command from the remote location to initiate the preheating of the AGD is received. Advantageously, a point in time of a conducted usage of the AGD corresponding to the transmitted time resolved sensor data may be transmitted to the remote location; a data sample may be created comprising the point in time of the conducted usage of the AGD and the corresponding time resolved sensor data; and the artificial intelligence model may be retrained using the created data sample.
[0030] Employing the trained artificial intelligence model on a remote location increases the flexibility and scalability of the proposed system. Accordingly, the artificial intelligence model
can be adapted more flexibly and frequently, as the artificial intelligence model is continuously served with new usage data. Therefore, a model which is more customized to the user of the AGD can be created/trained, and which can better analyze the user’s patterns, resulting in an improved preheating. Furthermore, it may be possible to host several artificial intelligence models for one AGD in case the AGD is used by several users. In such a case, a user of the plurality of users using the AGD at a certain moment may select his/her user profile (e.g., directly on the AGD or on another device connected to/with the AGD) which activates the corresponding artificial intelligence model. In case the artificial intelligence model is employed on the remote location, the AGD comprises a corresponding communication interface, which also allows over-the-air updates as in the above-described embodiment.
[0031] According to a 10th embodiment of the invention, in any one of the preceding embodiments, the time resolved sensor data of an AGD includes at least one of: time-resolved data from a gyroscope (preferably Micro Electro-Mechanical System (MEMS) based) comprised by the AGD; time-resolved data from a proximity sensor (preferably optical time-of-flight (ToF) or triangulation sensors operating in the near infra-red wavelength regime) comprised by the AGD; time-resolved data from a draw sensor comprised by the AGD; time-resolved data from a lip contact sensor (e.g., capacitive (touch) sensors or mechanical switches engaged by contact with the lips) comprised by the AGD (e.g., located on the mouth piece of the AGD); time-resolved data from a sensor comprised by the AGD and detecting a fill level of the AGD; time-resolved data from one or more temperature sensors measuring the temperature of a heater comprised by the AGD, the environment of the AGD, a battery comprised by the AGD and/or electronics comprised by the AGD; time-resolved data from a draw pressure and/or air flow sensor comprised by the AGD; and/or time-resolved data from at least one bio sensor comprised by the AGD and measuring the heart rate of a user of the AGD, a movement of the user of the AGD and/or the blood pressure of the user of the AGD. In some embodiments, the state of the AGD may be represented as a feature vector based on/comprising the time resolved sensor data.
[0032] Collecting sensor data from a plurality of different sensor sources increases the accuracy of describing the state of an AGD, resulting in an improved assessment of the state, which allows for more precise estimations.
[0033] According to an 11th embodiment of the invention, in any one of the preceding embodiments, the state of an AGD represents at least a motion of the AGD in space and/or a position of the AGD in space.
[0034] According to a 12th embodiment of the invention, in any one of the preceding embodiments, the time-stamped usage data is data from a puff sensor, preferably a draw-activated puff sensor.
[0035] According to a 13th embodiment of the invention, in any one of the preceding embodiments, the usage of an AGD is a puff of the AGD.
[0036] Estimating the occurrence of a puff by a puff sensor increases the accuracy of preheating, as the point in time of the usage/puff can be precisely measured. Therefore, the correlation between a motion of the AGD and/or the position of the AGD with the actual point in time of the usage/puff can be more precisely determined.
[0037] A 14th embodiment of the invention is an aerosol generation device (AGD) comprising means for carrying out the method of any one of the preceding embodiments.
[0038] A 15th embodiment of the invention is a computer program which when executed causes a computing device or a system to perform a method of any one of the above mentioned embodiments, preferably any one of the 1st to 13th embodiments.
Brief Description of the Drawings
[0039] Various embodiments of the present invention are described in more detail in the following by reference to the accompanying figures, without the present invention being limited to the embodiments of these figures.
Fig. 1 illustrates an aerosol generation device (AGD) according to an embodiment of the invention.
Fig. 2 illustrates a (computer-implemented) method for operating an AGD according to an embodiment of the invention.
io
Fig. 3 illustrates a system for creating an AGD according to an embodiment of the invention.
Fig. 4 illustrates a (computer-implemented) method for creating an AGD according to an embodiment of the invention.
[0040] Throughout the present drawings and specification, the same reference numerals refer to the same elements. In the drawings, reference signs are illustrated exemplarily without limiting the embodiments of the drawings to merely comprising the illustrated reference signs.
Detailed Description of Preferred Embodiments
[0041] In the following, exemplary embodiments of the present invention are described in more detail.
[0042] Fig. 1 shows an aerosol generation device (AGD) 110 according to an embodiment of the invention. The AGD 110 may be capable of/configured to perform(ing) the method 200 for operating an AGD 110. The AGD 110 may comprise sensors like a gyroscope, a proximity sensor (configured to measure the distance between the AGD 110 and the face of the user 120) and/or additional sensor(s) as described in corresponding embodiments. The AGD 110 may also comprise a mouthpiece, a data processing module and a heater. In general, the mouthpiece is positioned at the position closest to the face of the user 120. If the proximity sensor is installed on the mouthpiece, the proximity sensor may measure and provide a precise distance between the AGD 110 and the face of the user 120. The mouthpiece may be detachable from the AGD 110 for cleaning purposes. If the proximity sensor is installed on such a detachable part, the implementation may become more complex due to additional requirements on the electrical structure. Accordingly, in this embodiment, the proximity sensor is installed on the AGD 110 itself. This allows for a simpler implementation, while still being capable of measuring and providing a distance between the AGD 110 and the face of the user 120 which is precise enough. Alternatively, the proximity sensor may be installed on the mouthpiece, The AGD 110 may be used by a user 120. The sensor(s) of the AGD 110 may collect (time-resolved) sensor data which may represent a state of the AGD 110. The sensor data may be transmitted to/received by the data processing module of the AGD 110. The data processing module may perform the computer- implemented method for operating an AGD 110 according to some embodiments of the present invention (e.g., the method 200). Accordingly, if the data processing module determines (via
estimation) that a usage/puff of the AGD 110 by the user 120 is imminent (i.e., the first probability is greater than or equal to the first threshold), the data processing module initiates preheating of the AGD 110, which causes the heater of the AGD 110 to heat up to a corresponding temperature (e.g., 250 to 300 degree Celsius).
[0043] Fig. 2 shows a computer-implemented method 200 for operating an AGD 110 according to an embodiment of the invention. In step 210, the time-resolved sensor data representing a state of the AGD may be received (for example by the data processing module of the AGD 110). The data processing module may include a predictive model (e.g., conventionally programmed or an artificial intelligence model which may be trained according to some embodiments of the present invention) for estimating a future starting point in time of a usage of the AGD 110 based on the time-resolved sensor data of the AGD 110. In step 220, the predictive model may estimate a first probability (Pl) which may be a likelihood of correctness of the estimated future starting point in time of the usage (e.g., a puff) of the AGD 100, for example by the user 120. In step 220, "t" represents an estimated next puff (or draw) timing, "x" represents the length of the time interval, and “y” represents a first threshold. If the first probability (Pl) value is large, the next puff (or draw) is most likely performed before the length of the time interval (x) is lapsed. On the other hand, if the first probability (Pl) value is small, the next puff (or draw) is most likely not performed before the length of the time interval (x) is lapsed. A large first probability (Pl) value may also mean that the estimated future starting point is close. A small first probability (Pl) may also mean that the estimated future starting point is far. In some cases, if the estimated first probability (e.g., 80%) is greater than or equal to the first threshold (e.g., 60%), preheating of the AGD 110 may be initiated in step 240 (e.g., via a preheating command of the data processing command to the heater of the AGD 110). In contrast, if the first probability is smaller (e.g., 50%) than the first threshold (e.g., 60%), preheating may not be initiated. In such cases, the method 200 may then terminate and may be re-executed according to a configured frequency/period (e.g., pre-configured and fixed to execute each 0.5 s or a frequency/period which may adapt to for example the current power state of the AGD 110).
[0044] However, in some other cases, the method may continue with step 230 after step 220 if the first probability is/was greater than or equal to the first threshold. In step 230, a second probability (P2) may be determined/estimated indicating whether the preheating of the AGD 110 can be initiated safely. Such a safety assessment may be based on for example an amount of liquid
inside the AGD 110 and/or a position of the AGD 110. Should the second probability be greater than or equal to a second threshold (z), the preheating of the AGD 110 may be initiated in step 240. Otherwise, should the second probability be smaller than the second threshold, preheating of the AGD 110 may not be initiated. In such cases, the method 200 may then terminate and may be re-executed according to a configured frequency/period. In such case, the AGD 110 can provide a safety issue notification (e.g., a vibration, a vibration pattern, a message sent to (e.g., a display of) the AGD, or another device connected to the AGD) to the user of the AGD 110. If the AGD is in the user's pocket or bag, the user might overlook such safety issue notification performed on the AGD. Thus, performing the safety issue notification on another device (e.g., smartphone, smartwatch, or earphone) may be advantageous in such situations. Alternatively, step 230 may be executed again within a predefined third time interval. If the result of step 230 shows that the situation is no longer unsafe for preheating (i.e., the re-calculated second probability is larger than the second threshold), the preheating may be initiated without again determining the first probability. As a result, the system safety as well as the power management of the AGD 110 is improved as unnecessary executions of method steps (e.g., estimating the first probability) are avoided.
[0045] It is to be understood that the above-described sequence of steps of the method 200 is only an example and can be reordered accordingly. For example, it may be possible that step 230 is performed before step 220. It may also be possible that some of the steps of method 200 are performed in a parallel manner to for example decrease the execution time of the method 200. It may also be possible that the method 200 comprises additional method steps as described with respect to some embodiments of the present invention.
[0046] Fig. 3 shows a system 300 comprising a test AGD 310 for creating (i.e., setting up) an AGD 110 according to an embodiment of the invention. The test AGD 310 may comprise sensors like a gyroscope, proximity sensor (configured to measure the distance between the face of the test user) and/or additional sensor(s) as described in corresponding embodiments. The test AGD 310 may also comprise a mouthpiece, a transmission module, and a draw detector. The test AGD 310 may be used by a test user 320. The system 300 may further comprise a data collection device 340 which may at least comprise a reception module. The test AGD 310 may collect data of the usage of the test AGD 310 by the test user 320. The collected data may be transmitted to the data collection device 340. The data may then be used to develop for example the data processing
module of the AGD 110 which is capable of/configured to perform(ing) the method 200 for operating the AGD 110 according to some embodiments of the present invention. If an artificial intelligence model is deployed on the AGD 110, the transmission module may be omitted from the AGD 110. Otherwise, such transmission module may be mounted on the AGD 110. If the AGD 110 has such transmission module, the deployed artificial intelligence model can be optimized by communication with data collection device 340. In other words, the data collection device 340 can optimize or re-train the deployed artificial intelligence model by using collected data from sensors mounted on the AGD 110. Since such data is collected from actual user of the AGD 110, the deployed artificial intelligence model can be optimized and thus the prediction quality.
[0047] In some cases, the data may also be used to train/develop an artificial intelligence model according to some embodiments of the present invention, which may then be deployed on the AGD 110 as part of the data processing module of the AGD 110. In some cases, the artificial intelligence model may be deployed in a remote location as explained with respect to some embodiments. Accordingly, the data processing module of the AGD 110 may then be configured to communicate with the remote location via a communication interface.
[0048] Fig. 4 shows a computer-implemented method 400 for creating (i.e., setting up) an AGD 110 according to an embodiment of the invention. Method 400 may be performed by the system 300. In step 410 of the method 400, the transmission module of a test AGD 310 may collect usage data from several sensor sources of the test AGD 310 being used by a test user 320 and may transmit the data to a data collection device 340 comprising at least a reception module. Based on the collected usage data, a predictive model may be developed in step 440. In some cases, the collected data may also be used for developing and training an artificial intelligence model according to some embodiments of the present invention in step 430, which may be used as the predictive model. In step 450, the predictive model may then be deployed on the AGD 110 as part of the data processing module or at a remote location, in which case the data processing module of the AGD 110 may be configured to communicate with the remote location via a communication interface.
Claims
1. A computer-implemented method (200) for operating an aerosol generation device, AGD, (110) the method comprising the steps of: receiving (210) time-resolved sensor data of the AGD representing a state of the AGD; estimating a future starting point in time of a usage of the AGD based on the time- resolved sensor data of the AGD; estimating (220) a first probability which is a likelihood of correctness of the estimated future starting point in time of the usage of the AGD; and initiating (240) a preheating of the AGD only if the estimated first probability is greater than or equal to a first threshold.
2. The method of the claim 1, wherein the method further comprises: determining (230), based on an amount of liquid inside the AGD and/or a position of the AGD, a second probability indicating whether the preheating of the AGD can be initiated safely; wherein preheating of the AGD is initiated only if the second probability is equal to or greater than a second threshold and the estimated first probability is greater than or equal to the first threshold.
3. The method of the claim 2, wherein the method comprises: providing a safety issue notification to a user of the AGD at a second point in time if the second probability is smaller than the second threshold.
4. The method of any one of claims 2 to 3, wherein the method further comprises: redetermining the second probability within a predefined third time interval starting from the second point in time.
5. The method of any one of the preceding claims, wherein the method comprises: determining a first time interval between a first point in time and the estimated future starting point in time; wherein the preheating of the AGD is initiated only if the first time interval is smaller than or equal to a second, predetermined time interval; wherein the first point in time is a point in time of the receiving of the time- resolved sensor data, a point in time of estimating the first probability, or a point in time between the point in time of the receiving and the point in time of the estimating; and wherein the second time interval corresponds to an amount of time necessary for preheating the AGD.
6. The method of any one of the claims 1 to 5, wherein the future starting point in time and the first probability are estimated by an artificial intelligence model.
7. The method of claim 6, wherein the artificial intelligence model has been trained according to a training method, the training method comprising the steps of: inputting training data to the artificial intelligence model to train the model, the training data including a plurality of data samples, each data sample comprising: time-resolved sensor data of a first AGD representing a state of the first AGD; and time-stamped usage data of the first AGD representing a starting point in time of the usage of the first AGD associated with the time-resolved data of the first AGD; and correlating the state of the first AGD and the starting point in time of the usage of the first AGD based on the plurality of data samples; wherein the model is configured to estimate the likelihood of correctness of the estimated future starting point in time of the usage of the AGD based on the correlation.
8. The method of any one of the claims 6 to 7, wherein the trained artificial intelligence model is employed on the AGD.
9. The method of any one of the claims 6 to 7, wherein the trained artificial intelligence model is employed on a location remote from the AGD, preferably a server or a cloud instance.
10. The method of any one of the preceding claims, wherein the time-resolved sensor data of an AGD includes at least one of: time-resolved data from a gyroscope comprised by the AGD; time-resolved data from a proximity sensor comprised by the AGD; time-resolved data from a draw sensor comprised by the AGD; time-resolved data from a lip contact sensor comprised by the AGD; time-resolved data from a sensor comprised by the AGD and detecting a fill level of the AGD; time-resolved data from one or more temperature sensors measuring the temperature of a heater comprised by the AGD, the environment of the AGD, a battery comprised by the AGD and/or electronics comprised by the AGD; time-resolved data from a draw pressure and/ or air flow sensor comprised by the AGD; and/or time-resolved data from at least one bio sensor comprised by the AGD and measuring the heart rate of a user of the AGD, a movement of the user of the AGD and/ or the blood pressure of the user of the AGD.
11. The method of any one of the preceding claims, wherein the state of an AGD represents at least a motion of the AGD in space and/or a position of the AGD in space.
12. The method of any one of the preceding claims, wherein the time-stamped usage data is data from a puff sensor, preferably a draw-activated puff sensor.
13. The method of any one of the preceding claims, wherein the usage of an AGD is a puff of the AGD.
14. An aerosol generation device, AGD, (110) comprising means for carrying out the method according to any one of preceding claims.
15. A computer program having instructions which when executed cause a computing device or a system to perform a method according to any one of the claims 1 to 14.
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP22181953.5 | 2022-06-29 | ||
EP22181953 | 2022-06-29 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2024002904A1 true WO2024002904A1 (en) | 2024-01-04 |
Family
ID=82492884
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/EP2023/067165 WO2024002904A1 (en) | 2022-06-29 | 2023-06-23 | Method and apparatus for efficient operation of an aerosol generation device |
Country Status (1)
Country | Link |
---|---|
WO (1) | WO2024002904A1 (en) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20200154772A1 (en) * | 2017-04-11 | 2020-05-21 | Kt&G Corporation | Aerosol generation system of preheating heater |
US20200337382A1 (en) | 2019-04-25 | 2020-10-29 | Rai Strategic Holdings, Inc. | Artificial intelligence in an aerosol delivery device |
EP3869427A1 (en) | 2020-02-19 | 2021-08-25 | Nerudia Limited | Smoking substitute device |
WO2021210900A1 (en) | 2020-04-16 | 2021-10-21 | Kt&G Corporation | Aerosol generating device |
EP4011226A1 (en) * | 2019-10-18 | 2022-06-15 | Shenzhen Happy Vaping Technology Limited | Electronic cigarette capable of automatically preheating after pre-determination of smoking action and control method for electronic cigarette |
EP3932225B1 (en) * | 2020-06-30 | 2022-06-15 | Philip Morris Products S.A. | Aerosol-generating device providing gesture initiation of a status query |
-
2023
- 2023-06-23 WO PCT/EP2023/067165 patent/WO2024002904A1/en unknown
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20200154772A1 (en) * | 2017-04-11 | 2020-05-21 | Kt&G Corporation | Aerosol generation system of preheating heater |
US20200337382A1 (en) | 2019-04-25 | 2020-10-29 | Rai Strategic Holdings, Inc. | Artificial intelligence in an aerosol delivery device |
EP4011226A1 (en) * | 2019-10-18 | 2022-06-15 | Shenzhen Happy Vaping Technology Limited | Electronic cigarette capable of automatically preheating after pre-determination of smoking action and control method for electronic cigarette |
EP3869427A1 (en) | 2020-02-19 | 2021-08-25 | Nerudia Limited | Smoking substitute device |
WO2021210900A1 (en) | 2020-04-16 | 2021-10-21 | Kt&G Corporation | Aerosol generating device |
EP3932225B1 (en) * | 2020-06-30 | 2022-06-15 | Philip Morris Products S.A. | Aerosol-generating device providing gesture initiation of a status query |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP3382696B1 (en) | Method for operating speech recognition service and electronic device supporting the same | |
EP3076865B1 (en) | Heart rate monitor device | |
JP6670418B2 (en) | State prediction apparatus and state prediction control method | |
EP3513927B2 (en) | Shaving appliance including a notification circuit for communicating cumulative shave event information | |
US10417901B2 (en) | Electronic device and control method thereof | |
CN105792740B (en) | The detection and calculating that heart rate in non-clinical restores | |
EP2688311B1 (en) | Method for processing sensor input, in particular for adjusting mobile device state based on user intentions and/or identity | |
EP3513926B1 (en) | Shaving appliance including a notification circuit for communicating shave stroke direction information | |
US20160364002A1 (en) | Systems and methods for determining emotions based on user gestures | |
EP3513925B2 (en) | Networked shaving appliance system | |
EP3292817B1 (en) | Authentication system and authentication method for detecting breath alcohol | |
JP2014117328A (en) | Sensor device, measuring system, and measuring program | |
TaheriNejad et al. | Comprehensive Observation and its Role in Self-Awareness; An Emotion Recognition System Example. | |
JP5372487B2 (en) | Action record input support system and server | |
JP6436663B2 (en) | Blood pressure estimation device | |
TWI621999B (en) | Method for face detection | |
CN105844130B (en) | Unlocked by fingerprint method under standby mode | |
US10426382B2 (en) | Detection device and detection method | |
EP3767182A1 (en) | Method and heating apparatus for estimating status of heated object | |
CN108351715A (en) | Resident sensor device applied to human body touch | |
JP2020531995A5 (en) | ||
Mikos et al. | Real-time patient adaptivity for freezing of gait classification through semi-supervised neural networks | |
US11930142B2 (en) | Identification apparatus, processing apparatus, processing method, and storage medium | |
WO2024002904A1 (en) | Method and apparatus for efficient operation of an aerosol generation device | |
EP3372155B1 (en) | Meal time estimation method, meal time estimation program, and meal time estimation device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 23734287 Country of ref document: EP Kind code of ref document: A1 |