Regulation and control method and regulation and control device of electronic cigarette

Abstract

The embodiment of the invention discloses a regulation and control method and a regulation and control device for an electronic cigarette, which are used in the technical field of electronic cigarettes. The method comprises the steps of obtaining an intake prediction model trained based on training data detected when a user smokes the electronic cigarette, wherein the training data at least comprises first physiological response data and first tobacco tar atomization data, collected based on a multi-modal sensor, of the user; in the process that the user smokes the electronic cigarette every time, detection data detected when the user smokes the electronic cigarette are input into the intake prediction model, and the predicted nicotine intake output by the intake prediction model is obtained; the detection data at least comprises second physiological response data of the user and real-time atomization characteristic data; when the predicted nicotine intake meets a first intervention triggering condition, the nicotine release amount of the electronic cigarette is regulated and controlled; the user is effectively prevented from taking excessive nicotine, and the body health of the user is ensured.

Description

Technical Field

[0001] The embodiments of the present application relate to the technical field of electronic cigarettes, and in particular to a method and device for controlling an electronic cigarette. Background Art

[0002] E-cigarettes are electronic products that mimic cigarettes, with a similar appearance, smoke, taste, and feel. They are powered by a rechargeable lithium polymer battery that drives the atomizer, which heats the e-liquid in the tank and converts nicotine and other substances into vapor for the user to inhale.

[0003] Existing e-cigarette control systems, to tailor the volume of e-cigarette smoke to meet user needs, detect the current air pressure in the mouthpiece and adjust the atomization power based on the corresponding pressure signal. This system adjusts the atomization power based on the user's suction force, ensuring that the volume of smoke more closely matches the user's needs. However, because existing systems often rely on particle size or air pressure changes to control atomization power, they lack the ability to dynamically perceive the user's actual physiological reactions and ingested dose. This can easily lead to excessive nicotine intake when the user puffs excessively, posing a health risk. Summary of the Invention

[0004] The embodiments of the present application provide an electronic cigarette control method and control device, which can effectively avoid the risk of excessive nicotine intake by users and improve health and safety during use.

[0005] The present application provides an electronic cigarette control method, comprising:

[0006] Obtaining an intake prediction model trained based on training data detected when a user smokes an electronic cigarette, the training data including at least: first physiological response data and first e-liquid atomization data of the user collected based on a multimodal sensor;

[0007] During each puff of the electronic cigarette by the user, detection data detected by the user during the puff of the electronic cigarette is input into the intake prediction model to obtain a predicted nicotine intake output by the intake prediction model; the detection data includes at least: second physiological response data of the user and second e-liquid atomization data;

[0008] When the predicted nicotine intake meets the first intervention trigger condition, the nicotine release of the electronic cigarette is regulated.

[0009] Furthermore, obtaining an intake prediction model trained based on training data detected when a user smokes an electronic cigarette includes:

[0010] Normalizing the first physiological response data and the first e-liquid atomization data to generate a training feature vector;

[0011] Inputting the training feature vector into an initial intake prediction model to obtain a first predicted nicotine intake output by the initial intake prediction model;

[0012] constructing a first loss function based on the first predicted nicotine intake and the actual nicotine intake;

[0013] Until the first loss function converges, a trained intake prediction model is obtained.

[0014] Furthermore, it also includes:

[0015] A physiological response prediction model trained based on the training data and a second predicted nicotine intake output by the intake prediction model; the second predicted nicotine intake being the output of the intake prediction model after inputting the training data into the intake prediction model;

[0016] During each puff of the electronic cigarette by the user, the detection data and the predicted nicotine intake are input into the physiological response prediction model to obtain physiological response prediction data output by the physiological response prediction model after a preset delay; the physiological response prediction data includes at least: a predicted heart rate and / or a predicted blood oxygen saturation;

[0017] When the physiological response prediction data meets the second intervention trigger condition, the nicotine release amount of the electronic cigarette is regulated.

[0018] Furthermore, the method further includes: performing early intervention and regulation on the electronic cigarette based on the physiological response prediction data output by the physiological response prediction model.

[0019] Furthermore, the step of obtaining a physiological response prediction model based on the training data and the second nicotine intake prediction training output by the intake prediction model includes:

[0020] inputting the training data and the second predicted nicotine intake into an initial physiological response prediction model to obtain first physiological response prediction data output by the initial physiological response prediction model after a preset delay time;

[0021] constructing a second loss function based on a physiological response error between the first physiological response prediction data and the actual physiological response data;

[0022] Until the second loss function converges, a trained physiological response prediction model is obtained.

[0023] Furthermore, constructing a second loss function based on the physiological response error between the first physiological response prediction data and the actual physiological response data includes:

[0024] Obtaining an intake error between the second predicted nicotine intake and the actual nicotine intake;

[0025] The intake error and the physiological response error are weightedly combined to construct a second loss function.

[0026] Furthermore, it also includes:

[0027] When the error between the physiological response prediction data and the actual physiological response data satisfies a preset error condition, updating the model parameters of the intake prediction model so that the predicted nicotine intake output by the intake prediction model approaches the actual nicotine intake;

[0028] The physiological response prediction model is trained based on the third predicted nicotine intake output by the updated intake prediction model to obtain an updated physiological response prediction model; and the nicotine release of the electronic cigarette is regulated by using the physiological response prediction data output after a preset delay based on the updated physiological response prediction model.

[0029] Furthermore, the regulation of the nicotine release of the electronic cigarette includes:

[0030] determining a risk level of the user based on the predicted nicotine intake and the predicted physiological response data;

[0031] The heating power of the electronic cigarette and the air intake structure of the electronic cigarette are adjusted based on the risk level to regulate the nicotine release of the electronic cigarette.

[0032] Furthermore, it also includes:

[0033] After regulating the e-cigarette, target physiological response data from the user is obtained;

[0034] The first intervention trigger condition and the second intervention trigger condition, and / or a control mode of the electronic cigarette are adjusted based on the target physiological response data.

[0035] Furthermore, the training data also includes: first puffing behavior data of the user, first device state data of the electronic cigarette, and first external environment data of the electronic cigarette;

[0036] The detection data also includes: second puffing behavior data of the user, second device status data of the electronic cigarette, and second external environment data of the electronic cigarette.

[0037] The present application also provides an electronic cigarette control device, including:

[0038] an acquisition unit, configured to acquire an intake prediction model trained based on training data detected when a user smokes an electronic cigarette, the training data comprising at least first physiological response data and first e-liquid atomization data of the user collected based on a multimodal sensor;

[0039] an input unit, configured to input detection data detected by the user during each puff of the electronic cigarette into the intake prediction model after the user finishes puffing the electronic cigarette, to obtain a predicted nicotine intake output by the intake prediction model; the detection data including at least the user's second physiological response data and the second e-liquid atomization data;

[0040] The regulating unit is configured to regulate the nicotine release of the electronic cigarette when the predicted nicotine intake meets a first intervention triggering condition.

[0041] The present application also provides an electronic cigarette control device, including:

[0042] CPU, memory, input and output interfaces, wired or wireless network interfaces, power supply;

[0043] The memory is a transient storage memory or a persistent storage memory;

[0044] The central processing unit is configured to communicate with the memory and execute instructions in the memory on a control plane function entity to perform the above method.

[0045] An embodiment of the present application further provides a computer-readable storage medium, which includes instructions. When the instructions are executed on a computer, the computer executes the method described above.

[0046] It can be seen from the above technical solutions that the embodiments of the present application have the following advantages:

[0047] In an embodiment of the present application, an intake prediction model trained based on training data detected when a user puffs on an e-cigarette is obtained, where the training data includes at least first physiological response data of the user and first e-liquid atomization data collected by a multimodal sensor. During each puff of the e-cigarette by the user, the detection data detected when the user puffs on the e-cigarette is input into the intake prediction model to obtain a predicted nicotine intake output by the intake prediction model. The detection data includes at least second physiological response data of the user and second e-liquid atomization data. When the predicted nicotine intake meets a first intervention trigger condition, the nicotine release of the e-cigarette is regulated.

[0048] In the intake prediction model, the user's physiological response data and the e-liquid atomization data are used to predict and output the predicted nicotine intake. The predicted nicotine intake can accurately reflect the user's physiological response during the e-cigarette smoking process and the harmful substances inhaled by the user after atomization. By regulating the nicotine release of the e-cigarette through the predicted nicotine intake, the nicotine release of the e-cigarette can be timely regulated according to the user's physiological response and the harmful substances inhaled by the user, effectively preventing the user from consuming excessive nicotine and ensuring the user's health. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present application. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0050] Figure 1 This is a control flow chart of an electronic cigarette disclosed in an embodiment of the present application;

[0051] Figure 2 A training flow chart of an intake prediction model disclosed in an embodiment of the present application;

[0052] Figure 3 This is a control flow chart of another electronic cigarette disclosed in an embodiment of this application;

[0053] Figure 4 A training flow chart of a physiological response prediction model disclosed in an embodiment of the present application;

[0054] Figure 5 A flowchart of a model update disclosed in an embodiment of the present application;

[0055] Figure 6 A flowchart of feedback adjustment disclosed in an embodiment of the present application;

[0056] Figure 7 A diagram of a control device for an electronic cigarette disclosed in an embodiment of the present application;

[0057] Figure 8 This is a diagram of another electronic cigarette control device disclosed in an embodiment of the present application. DETAILED DESCRIPTION

[0058] In order to help those skilled in the art better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0059] In the description of the embodiments of the present application, it should be noted that the terms "center", "up", "down", "left", "right", "vertical", "horizontal", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings. They are only for the convenience of describing the embodiments of the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, they cannot be understood as limiting the embodiments of the present application.

[0060] In the description of the embodiments of the present application, it should be noted that, unless otherwise expressly specified or limited, the terms "installed," "connected," and "connected" should be understood in a broad sense. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to mechanical connections or electrical connections; they can refer to direct connections or indirect connections through an intermediate medium; and they can refer to internal connections between two components. Those skilled in the art will understand the specific meanings of the above terms in the embodiments of the present application based on the specific circumstances.

[0061] In the existing electronic cigarette control system, in order to make the electronic cigarette's smoke output more in line with the user's needs, the electronic cigarette control system detects the current air pressure of the electronic cigarette mouthpiece, and adjusts the electronic cigarette's atomization power based on the pressure electrical signal corresponding to the current air pressure, so as to adjust the electronic cigarette's atomization power based on the user's suction force on the electronic cigarette, so that the electronic cigarette's smoke output is more in line with the user's needs. However, during the control process, when the user inhales excessively, the electronic cigarette's smoke output also increases greatly, which can easily cause the user to ingest excessive nicotine and affect the user's health. Therefore, the embodiment of the present application provides an electronic cigarette control method that can effectively prevent users from ingesting excessive nicotine and ensure the user's health, such as Figure 1 As shown, the specific steps include:

[0062] 101. Obtain an intake prediction model trained based on training data detected when a user smokes an e-cigarette.

[0063] In an embodiment of the present application, an e-cigarette control device can obtain an intake prediction model trained based on training data detected during user e-cigarette puffs. This intake prediction model can be an LSTM (Long Short-Term Memory) model or an MLP (Multi-Layer Perceptron) regression model, the specifics of which are not limited herein. This intake prediction model is used to predict the amount of nicotine ingested by the user during each e-cigarette puff, i.e., to predict nicotine intake.

[0064] In an embodiment of the present application, a multimodal sensor can be integrated into the e-cigarette body. The multimodal sensor is connected to a processor (such as an MCU) in the e-cigarette's control device, for example, via I²C / SPI / ADC. The control device triggers the multimodal sensor based on a preset sampling frequency to detect relevant data from the user's e-cigarette puffing. This training data includes at least: first physiological response data and first e-liquid atomization data collected by the multimodal sensor. The first physiological response data represents the user's individual physiological state after puffing on the e-cigarette. The first physiological response data may include any one or more of heart rate (HR), heart rate variability (HRV), and blood oxygen saturation (SpO2), with the specific details not being limited herein. The first e-liquid atomization data represents the atomization state of the e-cigarette e-liquid, reflecting the atomized harmful substances inhaled by the user. The first e-liquid atomization data may include any one or more of conductivity change (dσ / dt) and aerosol density (AOD), with the specific details not being limited herein.

[0065] The corresponding multimodal sensors include: a heart rate sensor, a blood oxygen sensor, a conductivity sensor, and an optical density sensor. The heart rate sensor and blood oxygen sensor can utilize photoelectric or PPG modules and can be located in the e-cigarette's stem grip area to continuously monitor the user's heart rate, heart rate variability, and blood oxygen levels while the user is holding the device. The conductivity sensor, located in the e-liquid contact channel or atomizer core structure, can assist in determining the e-liquid's atomization efficiency by detecting changes in conductivity during heating. The optical density sensor, located in the e-cigarette's atomizer outlet, uses the principle of laser scattering to detect aerosol density (i.e., the particle size distribution and optical density of the particles in the aerosol). This aerosol density reflects the atomized concentration of the e-liquid.

[0066] In the embodiment of the present application, the intake prediction model is obtained by training the user's first physiological response data and the first e-liquid atomization data, so that the nicotine intake prediction output by the intake prediction model can better match the user's physiological response and the e-liquid atomization situation, and the nicotine intake prediction can be obtained more accurately.

[0067] 102. During each puff of the electronic cigarette by the user, the detection data detected by the user when puffing the electronic cigarette is input into the intake prediction model to obtain the predicted nicotine intake output by the intake prediction model.

[0068] In an embodiment of the present application, the electronic cigarette control device can input detection data detected by the user during each puff of the electronic cigarette into an intake prediction model to obtain a predicted nicotine intake output by the intake prediction model. The detection data includes at least: second physiological response data of the user and second e-liquid atomization data. The second physiological response data is similar to the first physiological response data described above, and the second e-liquid atomization data is similar to the first e-liquid atomization data described above. The details are not further described here.

[0069] In an embodiment of the present application, the control device of the electronic cigarette can detect the user's puffing behavior by using an airflow sensor (such as a MEMS airflow sensor) installed in the air inlet channel of the electronic cigarette. During the user's puffing process, the multimodal sensor is triggered to collect data and obtain corresponding detection data. The detection data is input into the intake prediction model to obtain the predicted nicotine intake output by the intake prediction model.

[0070] 103. When the predicted nicotine intake meets the first intervention trigger condition, the nicotine release of the e-cigarette is regulated.

[0071] When the predicted nicotine intake meets the first intervention trigger condition, the nicotine delivery of the electronic cigarette is regulated. The predicted nicotine intake can be compared with a safety threshold. If the predicted nicotine intake is greater than the safety threshold, it is determined that the predicted nicotine intake meets the first intervention trigger condition, and the nicotine delivery of the electronic cigarette is regulated, i.e., the nicotine delivery of the electronic cigarette is reduced. The safety threshold can be set voluntarily, such as 50mg or 60mg, and the specific value is not limited here.

[0072] Specifically, in the embodiment of the present application, the user will have multiple puffs in a day. After each puff, the intake prediction model will output a corresponding predicted nicotine intake Ypred. By accumulating the predicted nicotine intake Ypred within the day, the cumulative nicotine intake Y_day of the day can be obtained. When the cumulative nicotine intake Y_day is greater than the preset daily intake safety threshold θ_nic, it can be determined that the predicted nicotine intake meets the first intervention trigger condition.

[0073] In this embodiment of the present application, the electronic cigarette's control device automatically activates with each puff. Using the predicted nicotine intake output by the intake prediction model, it controls the nicotine delivery of the electronic cigarette in real time, rapidly responding to each puff to ensure that each inhalation does not exceed the safe nicotine intake threshold. The control response time can be controlled within 50 milliseconds, achieving instantaneous current limiting within milliseconds.

[0074] It can be seen that in the embodiment of the present application, an intake prediction model trained based on training data detected when the user smokes an e-cigarette is obtained, and the training data at least includes: first physiological response data of the user and first e-liquid atomization data collected based on a multimodal sensor; during each puff of the e-cigarette by the user, the detection data detected when the user smokes the e-cigarette is input into the intake prediction model to obtain a predicted nicotine intake output by the intake prediction model; the detection data at least includes: second physiological response data of the user and second e-liquid atomization data; when the predicted nicotine intake meets the first intervention trigger condition, the nicotine release of the e-cigarette is regulated.

[0075] In the intake prediction model, the user's physiological response data and the e-liquid atomization data are used to predict and output the predicted nicotine intake. The predicted nicotine intake can accurately reflect the user's physiological response during the e-cigarette smoking process and the harmful substances inhaled by the user after atomization. By regulating the nicotine release of the e-cigarette through the predicted nicotine intake, the nicotine release of the e-cigarette can be timely regulated according to the user's physiological response and the harmful substances inhaled by the user, effectively preventing the user from consuming excessive nicotine and ensuring the user's health.

[0076] Furthermore, in embodiments of the present application, during each puff of an e-cigarette, in addition to the predicted nicotine intake, the nicotine delivery of the e-cigarette can also be regulated by combining detected data: the user's second physiological response data and the second e-liquid atomization data, thereby achieving more precise control of nicotine delivery. Specifically, the user's second physiological response data can be analyzed to determine whether the user's physiological load is too high. For example, if the user's heart rate is significantly higher than the resting level or their blood oxygen saturation decreases significantly after puffing on the e-cigarette, it is determined that the user's physiological load is too high, and the e-cigarette needs to be intervened to regulate and reduce nicotine delivery. The second e-liquid atomization data can also be used to determine whether the user has inhaled an excessive amount of harmful substances. For example, if the aerosol density of a single inhalation is greater than a preset safety threshold θ_aod, the aerosol density of a single inhalation is too high, and the user's respiratory system is at risk of excessive particulate deposition, which may lead to acute respiratory irritation or addiction. In this case, the e-cigarette needs to be intervened to regulate and reduce nicotine delivery to prevent high concentrations of particulate matter from continuously entering the airway.

[0077] Furthermore, in an embodiment of the present application, in order to improve the accuracy of the intake prediction model, the training data of the intake prediction model also includes: the user's first puffing behavior data, the first device status data of the electronic cigarette, and the first external environment data of the electronic cigarette; the corresponding detection data of the intake prediction model also includes: the user's second puffing behavior data, the second device status data of the electronic cigarette, and the second external environment data of the electronic cigarette.

[0078] The first user puffing behavior data is data related to the user's puffing behavior on the e-cigarette, including: puff flow rate (Q), puff duration (T), puff frequency (F), and puff volume (V). The corresponding multimodal sensor on the e-cigarette body also includes: a puff sensor and an airflow sensor. The puff sensor and the airflow sensor are arranged in the air inlet channel of the e-cigarette. The puff sensor is used to detect the puff flow rate, puff duration, and puff frequency, and the airflow sensor is used to detect the puff volume. The first device status data of the e-cigarette is the status data of the e-cigarette when the user puffs on the e-cigarette, including: atomization temperature peak (Tmax), atomization temperature average (Tavg), and heating power (P). The corresponding multimodal sensor on the e-cigarette body also includes: a temperature sensor and a voltage and current monitoring module. The temperature sensor and the voltage and current monitoring module are arranged in the atomization chamber of the e-cigarette. The temperature sensor is used to detect the temperature of the atomization chamber to obtain the atomization temperature peak and the atomization temperature average. The voltage and current monitoring module detects the voltage and current of the atomization chamber to obtain the heating power. The first external environment data of the electronic cigarette is the environmental data of the electronic cigarette when the user inhales the electronic cigarette, including: ambient temperature (Env_T) and ambient humidity (Env_H). The corresponding multimodal sensor on the electronic cigarette body also includes: a temperature and humidity sensor, which is arranged on the outer shell of the electronic cigarette and is used to detect the ambient temperature and ambient humidity.

[0079] In an embodiment of the present application, when training the intake prediction model, adding the user's first puffing behavior data, the first device status data of the electronic cigarette, and the first external environment data of the electronic cigarette to the training data can improve the accuracy of the intake prediction model, thereby improving the accuracy of the nicotine predicted intake output by the intake prediction model.

[0080] Furthermore, the following will be combined Figure 2 The training process of the intake prediction model in the embodiment of the present application is described, which specifically includes the following steps:

[0081] 201. Normalize the first physiological response data and the first e-liquid atomization data to generate a training feature vector.

[0082] In an embodiment of the present application, the control device of the electronic cigarette can normalize the first physiological response data and the first e-liquid atomization data to generate a training feature vector. When the training data also includes: the user's first puffing behavior data, the first device status data of the electronic cigarette, and the first external environment data of the electronic cigarette, the first physiological response data, the first e-liquid atomization data, the user's first puffing behavior data, the first device status data of the electronic cigarette, and the first external environment data of the electronic cigarette can be filtered and normalized to generate a training feature vector Xt, that is, Xt=[Q, T, F, V, Tmax, Tavg, P, dσ / dt, AOD, HR, HRV, SpO2, Env_T, Env_H, TimeSlot], where TimeSlot is a time segment identifier. The meaning of each feature dimension in the training feature vector Xt and the corresponding sensor are shown in Table 1 below:

[0083] Table 1

[0084] Characteristic dimension symbol meaning Sensor Source Q Suction flow rate (L / min) Airflow sensor T Single puff duration (s) Suction sensor F Suction frequency (times / min) Suction sensor V Aspirated volume (ml) Airflow sensor Tmax / Tavg Atomization temperature peak / average (℃) Temperature sensor P Heating power (W) Voltage and current sampling module dσ / dt E-liquid conductivity change rate Conductivity sensor AOD Aerosol optical density Laser scattering optical sensor HR / HRV Heart rate / heart rate variability PPG photoelectric sensor <![CDATA[SpO2]]> Blood oxygen saturation (%) Blood oxygen sensor Env_T / Env_H Ambient temperature / humidity Temperature and humidity sensors TimeSlot Identify the smoking time period (e.g. morning / night / before bedtime) System timestamp mapping

[0085] 202. Input the training feature vector into an initial intake prediction model to obtain a first predicted nicotine intake output by the initial intake prediction model.

[0086] The training feature vector is input into the initial intake prediction model to obtain the first nicotine intake prediction output by the initial intake prediction model. It can be understood that a mapping relationship can be constructed in the intake prediction model:

[0087] Ypred=f(Xt)+ε; where Ypred is the predicted nicotine intake output by the intake prediction model, f is the trained nonlinear function, which reflects the influence of the training feature vector Xt on the predicted nicotine intake Ypred, and learns the mapping relationship between the input training feature vector Xt and the predicted nicotine intake Ypred; ε is the residual error term, which is controlled within the preset tolerance range.

[0088] 203. Construct a first loss function based on the first predicted nicotine intake and the actual nicotine intake.

[0089] During the model training process, a supervised learning framework can be used to construct a first loss function based on the first predicted nicotine intake and the actual nicotine intake. The actual nicotine intake is recorded in the experimental dataset, and the actual nicotine intake = known e-liquid concentration × puff volume.

[0090] The first loss function is: Loss1=Ypred−Yt, where Yt is the actual nicotine intake.

[0091] 204. Until the first loss function converges, a trained intake prediction model is obtained.

[0092] Until the first loss function converges, a trained intake prediction model is obtained. It can be understood that in the training process of the embodiment of the present application, it is necessary to make the first nicotine predicted intake tend to the actual nicotine intake, that is, when the loss value corresponding to the first loss function is less than the preset threshold, it is determined that the first loss function converges. The preset threshold can be understood as, when the loss value is expressed using JS divergence, the preset threshold can be 0.4 or 0.5, which is not specifically limited here. After obtaining the trained intake prediction model, the trained intake prediction model can be used to obtain the predicted nicotine intake each time the user draws on the electronic cigarette.

[0093] In this implementation case, the structural selection strategies for the two core AI models are as follows:

[0094] Nicotine Intake Estimator: A multi-layer perceptron (MLP) or convolutional neural network (CNN) architecture is preferred. These are well-suited for processing normalized short-term temporal behavioral characteristics and physiological response characteristics, and offer efficient modeling capabilities. When sufficient test data is available, an LSTM network can also be introduced to model the temporal dependencies of inhalation behavior.

[0095] Physiological Response Forecaster: Optimally, LSTM, Bi-LSTM, or Transformer architectures can be used to model the dynamic causal relationship between nicotine intake and delayed physiological responses (such as heart rate and blood oxygen saturation). For scenarios requiring longer prediction horizons and more precise responses, the Transformer architecture combined with positional encoding to optimize sequence learning is recommended.

[0096] The system can flexibly switch model structures based on device computing resources, supporting more complex structures when deployed in the cloud and using lightweight structures such as the TinyML model for local inference.

[0097] Furthermore, when a user ingests nicotine, the user's body absorbs nicotine with a certain delay. After nicotine ingestion, there is a certain lag (generally about 30 seconds or 40 seconds) in the process of nicotine propagation in the blood and the nervous system. Therefore, in the embodiment of the present application, in order to obtain a more realistic physiological response after nicotine ingestion, a physiological response prediction model is used. The physiological response prediction model models the causal relationship between nicotine intake and delayed physiological response, and the physiological response prediction data after nicotine ingestion is obtained through the physiological response prediction model, such as Figure 3 As shown, the specific steps include:

[0098] 301. Obtain a second predicted nicotine intake based on the training data and the output of the intake prediction model, and a trained physiological response prediction model.

[0099] In an embodiment of the present application, a second predicted nicotine intake based on training data and the output of the intake prediction model can be obtained, and a trained physiological response prediction model can be obtained. The training data includes at least: the user's first physiological response data and the first e-liquid atomization data. The second predicted nicotine intake is the output of the intake prediction model after the training data is input into the intake prediction model; that is, the second predicted nicotine intake output by the intake prediction model after the training data is input into the trained intake prediction model. The physiological response prediction model is used to predict the physiological response after a certain delay after the user finishes smoking the electronic cigarette. The certain time can be 30 seconds or 40 seconds, and the specific time is not limited here. It is understandable that the physiological response prediction model can predict the physiological response within a preset time range after a certain delay, such as predicting the physiological response within 90 seconds after a delay of 30 seconds, that is, predicting the physiological response within the next 30 seconds to 120 seconds.

[0100] The physiological response prediction model is trained using the second predicted nicotine intake output by the trained intake prediction model. The physiological response prediction data output by the trained physiological response prediction model can be matched with the predicted nicotine intake predicted by the intake prediction model to accurately obtain the physiological response after a certain delay.

[0101] 302. During each puff of the electronic cigarette by the user, the detection data and the predicted nicotine intake are input into the physiological response prediction model to obtain the physiological response prediction data after the preset delay time output by the physiological response prediction model.

[0102] In an embodiment of the present application, after obtaining a trained physiological response prediction model, the user inputs the test data and predicted nicotine intake into the physiological response prediction model each time they puff on an e-cigarette, obtaining physiological response prediction data output by the physiological response prediction model after a preset delay. The test data includes at least the user's second physiological response data and second e-liquid atomization data; the predicted nicotine intake is the predicted nicotine intake output by the trained intake prediction model, obtained by inputting the feature vector corresponding to the test data into the trained intake prediction model. The physiological response prediction data includes at least a predicted heart rate and / or a predicted blood oxygen saturation.

[0103] 303. When the physiological response prediction data meets the second intervention trigger condition, the nicotine release amount of the electronic cigarette is regulated.

[0104] In the embodiment of the present application, when the physiological response prediction data meets the second intervention trigger condition, the nicotine release of the electronic cigarette is regulated, that is, the nicotine release of the electronic cigarette can be reduced in advance through the future physiological response prediction data, and the nicotine release can be regulated in advance through a more realistic physiological response after nicotine intake, thereby further ensuring the health of the user. Among them, the physiological response prediction data can be the predicted heart rate If the predicted heart rate is higher than the preset heart rate threshold, it is determined that the physiological response prediction data meets the second intervention trigger condition, wherein the preset heart rate threshold is the normal maximum heart rate of the human body, which can be 140 beats / minute or 150 beats / minute, which is not limited here. The physiological response prediction data can also be the predicted blood oxygen saturation If the predicted blood oxygen saturation is lower than the preset blood oxygen threshold, it is determined that the physiological response prediction data meets the second intervention trigger condition, wherein the preset blood oxygen threshold is the normal minimum blood oxygen saturation of the human body, which can be 90% or 92%, and the specific details are not limited here.

[0105] It is understandable that after obtaining the physiological response prediction data, the actual physiological response data after a preset delay can be detected. If the actual physiological response data continues to be high-risk and deviates from the physiological response prediction data, it is necessary to immediately reduce the nicotine release amount and control the electronic cigarette to stop working.

[0106] It can be seen that in the embodiments of the present application, the physiological response prediction data after nicotine intake is obtained through the physiological response prediction model, and the nicotine release amount of the electronic cigarette is further regulated in advance by the physiological response prediction data after the preset delay time output by the physiological response prediction model, thereby effectively avoiding health risks to the user's body.

[0107] Furthermore, in an embodiment of the present application, after obtaining the physiological response prediction data output by the physiological response prediction model, the e-cigarette can be intervened and regulated in advance based on the physiological response prediction data output by the physiological response prediction model. It is understandable that the physiological response after nicotine intake has a certain time delay in the human body, which is mainly affected by factors such as blood circulation and nerve conduction, and the lag time is usually 30 to 40 seconds. This embodiment introduces a "hysteresis prediction mechanism" to perceive potential risk trends in advance.

[0108] The specific hysteresis prediction mechanism is as follows: after the user completes a puff, the control system uses the nicotine intake prediction model to output the nicotine intake prediction Ypred, and links the physiological response prediction model to predict the physiological response prediction data (physiological indicators) after a delay of Δt (for example, 40 seconds), such as the predicted heart rate. Predicted blood oxygen saturation If the prediction result exceeds the set threshold (such as >140 bpm or <92%), the e-cigarette intervention mechanism is triggered in advance, including dynamically reducing the heating power and shrinking the channel of the air intake structure.

[0109] This hysteresis prediction mechanism offers proactive, proactive response, significantly different from traditional passive control methods based on suction / temperature feedback. This mechanism ensures that the control system doesn't base its decisions on the current state, but rather on the prediction of future high-risk situations. This creates an intelligent closed-loop "inhalation-prediction-intervention" system, further ensuring user health.

[0110] Furthermore, in the embodiment of the present application, the training process of the physiological response prediction model is as follows: Figure 4 The specific steps are as follows:

[0111] 401. Input the training data and the second predicted nicotine intake into an initial physiological response prediction model to obtain first physiological response prediction data output by the initial physiological response prediction model after a preset delay time.

[0112] In the embodiment of the present application, the control device of the electronic cigarette can input the training data and the second predicted nicotine intake into the initial physiological response prediction model to obtain the first physiological response prediction data output by the initial physiological response prediction model after the preset delay time. That is, the training feature vector Xt after the training data is normalized can be input into the trained intake prediction model to obtain the second predicted nicotine intake Ypred2 output by the intake prediction model; the training feature vector Xt and the second predicted nicotine intake Ypred2 are input into the initial physiological response prediction model to obtain the first physiological response prediction data output by the initial physiological response prediction model after the preset delay time. For example, the first physiological response prediction data obtained is the first predicted heart rate. .

[0113] 402. Construct a second loss function based on a physiological response error between the first physiological response prediction data and the actual physiological response data.

[0114] After obtaining the first physiological response prediction data, a second loss function can be constructed based on the physiological response error between the first physiological response prediction data and the true physiological response data. Constructing the second loss function based on the physiological response error between the first physiological response prediction data and the true physiological response data can make the first physiological response prediction data tend to the true physiological response data, that is, the physiological response prediction data output by the trained physiological response prediction model is more consistent with the true physiological response data. In this embodiment of the present application, the physiological response error between the first physiological response prediction data and the true physiological response data can be directly used as the second loss function.

[0115] Preferably, the intake error between the second predicted nicotine intake and the actual nicotine intake can be obtained, and the intake error and the physiological response error are weighted and combined to construct a second loss function. , the corresponding second loss function is: Loss2=‖Ypred2−Yt‖+λ‖ − ‖² , where λ is the risk weight and ‖Ypred2−Yt‖ is the intake error, which is used to minimize the deviation between the predicted nicotine intake output by the intake prediction model and the actual nicotine intake. is the actual heart rate after the preset delay, − ‖ is the delayed heart rate error.

[0116] 403. Until the second loss function converges, a trained physiological response prediction model is obtained.

[0117] Until the second loss function converges, a trained intake prediction model is obtained. It can be understood that in the training process of the embodiment of the present application, it is necessary to make the first physiological response prediction data tend to the real physiological response data, that is, when the loss value corresponding to the second loss function is less than the preset threshold, it is determined that the second loss function converges. The preset threshold can be understood as, when the loss value is expressed using JS divergence, the preset threshold can be 0.5 or 0.6, which is not limited here. After obtaining the trained physiological response prediction model, each time the user draws on the electronic cigarette, the trained physiological response prediction model can be used to obtain the physiological response prediction data after a preset delay.

[0118] Furthermore, in the embodiment of the present application, the nicotine release of the electronic cigarette can be regulated by adjusting the heating power of the electronic cigarette and the air intake structure of the electronic cigarette, wherein a first dynamic inhibition factor can be configured for the heating power of the electronic cigarette. , configure the second dynamic suppression factor for the heating power of the electronic cigarette ; Among them, the first dynamic inhibitory factor If it is greater than 0 and less than or equal to 1, the corresponding heating power is adjusted to: By reducing the current or frequency of the heating wire, the heating power can be reduced to reduce the evaporation rate of the e-liquid, thereby reducing the amount of nicotine released. The second dynamic inhibition factor If it is greater than 0 and less than or equal to 1, the corresponding intake structure is adjusted as follows: , the aerosol density can be diluted by reducing the opening of the air inlet, increasing the inhalation resistance or extending the airflow path, thereby reducing the amount of nicotine released.

[0119] Among them, different intervention and control methods are used to control the nicotine release of the e-cigarette based on different abnormal conditions obtained through detection data, intake prediction model, and physiological response prediction model. Specifically, when the aerosol density AOD in the second e-liquid atomization data is greater than the preset safety threshold θ_aod, the nicotine release of that time needs to be immediately reduced, which can be set as: 0.6, achieving a sudden drop in heating power, 0.5, to achieve immediate tightening of the progress structure; when the user's physiological load is determined to be too high based on the second physiological response data, mandatory intervention needs to be initiated immediately, which can be set to: 0.5, to achieve heating power limitation, 0.4, to achieve airflow restriction by the intake structure; when the physiological response prediction data meets the second intervention trigger condition, such as the predicted heart rate is higher than the preset heart rate threshold, prospective intervention can be performed. Initially lowered to 0.75, if the user continues to puff, Decrease; when the predicted nicotine intake meets the first intervention trigger condition, such as the cumulative nicotine intake Y_day obtained by the predicted nicotine intake is greater than the preset daily intake safety threshold θ_nic, the flexible intervention can be initiated and can be gradually decreased as well as , such as 0.9→0.8→0.7, to achieve a gradual reduction in heating power and a gradual tightening of the air intake structure.

[0120] Furthermore, in an embodiment of the present application, the control device of the electronic cigarette can determine the user's risk level based on the predicted nicotine intake and the physiological response prediction data; specifically, the predicted nicotine intake for the day can be accumulated to obtain the cumulative nicotine intake Y_day. If the cumulative nicotine intake Y_day is less than 70% of the preset daily intake safety threshold and the physiological response prediction data shows that the physiological indicators are normal, the user's risk level can be determined to be low risk; if the cumulative nicotine intake Y_day is less than 90% of the preset daily intake safety threshold and greater than 70% of the preset daily intake safety threshold, and the physiological response prediction data shows that the physiological indicators are slightly abnormal, such as a slightly increased heart rate, the user's risk level can be determined to be medium risk; if the cumulative nicotine intake Y_day is greater than 90% of the preset daily intake safety threshold and the physiological response prediction data shows that the physiological indicators are significantly abnormal, such as a sharply increased heart rate, the user's risk level can be determined to be high risk.

[0121] Based on the risk level, the heating power and air intake structure of the electronic cigarette are adjusted to control the nicotine release of the electronic cigarette. 0.95, 0.9, medium risk, 0.8, 0.7, high risk, 0.6, 0.5. The higher the risk level, the less nicotine is released; the lower the risk level, the more nicotine is released. By simultaneously adjusting the e-cigarette's heating power and air intake structure, nicotine release can be more precisely controlled. Furthermore, during the regulation process, the e-cigarette does not require additional structural changes, effectively reducing regulation costs.

[0122] Furthermore, when using the trained physiological response prediction model to predict physiological response prediction data of different users, since different users have different physiological responses to the ingested nicotine, it is easy to cause the physiological response prediction data predicted by the physiological response prediction model to be different. Therefore, in the embodiment of the present application, it is necessary to optimize the intake prediction model and the physiological response prediction model so that the nicotine release prediction output by the intake prediction model and the physiological response prediction model are more compatible with the human bodies of different users. Figure 5 The specific steps are as follows:

[0123] 501. When the error between the physiological response prediction data and the actual physiological response data meets a preset error condition, the model parameters of the intake prediction model are updated.

[0124] In an embodiment of the present application, when the error between the physiological response prediction data and the actual physiological response data meets a preset error condition, the model parameters of the intake prediction model are updated so that the predicted nicotine intake output by the intake prediction model approaches the actual nicotine intake. Specifically, the actual physiological response data can be detected in real time to obtain a true physiological curve. The true physiological curve and the corresponding physiological response prediction data are subjected to an error evaluation at regular intervals (e.g., 10 minutes). If the error between the physiological response prediction data and the actual physiological response data is greater than a preset error threshold and persists for a certain period of time, it is determined that the error between the physiological response prediction data and the actual physiological response data meets the preset error condition. The preset error threshold can be 10% or 15% of the actual physiological response data, which is not specifically limited here; the certain period of time can be 10 seconds or 15 seconds, which is not specifically limited here. At this point, the residual error term in the mapping relationship of the intake prediction model can be adjusted so that the predicted nicotine intake output by the intake prediction model approaches the actual nicotine intake, thereby optimizing and updating the intake prediction model.

[0125] 502. Train a physiological response prediction model based on the third predicted nicotine intake output by the updated intake prediction model to obtain an updated physiological response prediction model.

[0126] Next, the physiological response prediction model can be trained based on the third predicted nicotine intake output by the updated intake prediction model to obtain an updated physiological response prediction model; and the nicotine release of the electronic cigarette can be regulated by using the physiological response prediction data output after a preset delay based on the updated physiological response prediction model.

[0127] Specifically, the training data is input into the updated intake prediction model to obtain a third predicted nicotine intake output. The physiological response prediction model is then trained based on the training data and the third predicted nicotine intake output. Updating the intake prediction model first and then updating the physiological response prediction model based on the updated intake prediction model effectively avoids error propagation.

[0128] Furthermore, in an embodiment of the present application, a confidence interval is added to the second predicted nicotine intake used to train the physiological response prediction model. When the second predicted nicotine intake exceeds the confidence interval, the model parameters of the intake prediction model are updated to avoid the uncertainty of the output physiological response prediction data caused by the intake error during the learning process of the physiological response prediction model.

[0129] Furthermore, in the embodiment of the present application, the corresponding control process can also be adjusted according to the target physiological response data fed back by the user, so that the control of the electronic cigarette is more in line with the individual state of the user, such as Figure 6 As shown, the specific steps are as follows;

[0130] 601. After regulating the electronic cigarette, obtain target physiological response data fed back by the user.

[0131] In the embodiment of the present application, after the electronic cigarette is regulated, target physiological response data fed back by the user can be obtained. For example, the user's heart rate, blood oxygen saturation, etc. can be detected after the electronic cigarette is regulated.

[0132] 602. Adjust the first intervention trigger condition and the second intervention trigger condition, and / or the regulation mode of the electronic cigarette based on the target physiological response data.

[0133] Then, the first intervention trigger condition and the second intervention trigger condition, and / or the control mode of the electronic cigarette can be adjusted based on the target physiological response data. Specifically, the relevant thresholds in the first intervention trigger condition and the second intervention trigger condition, such as the preset daily intake safety threshold, the preset heart rate threshold, etc., can be adjusted based on the target physiological response data. For example, when the target physiological response data is the target heart rate, after the electronic cigarette is regulated, the target heart rate is still at a high state, then the preset daily intake safety threshold and the preset heart rate threshold can be reduced. The dynamic inhibition factor ( 、 ), for example, when the target physiological response data is the target heart rate, after the electronic cigarette is regulated, the target heart rate is still at a high state, the dynamic inhibition factor can be reduced ( 、 ).

[0134] Furthermore, in embodiments of the present application, the e-cigarette's control device can also provide feedback on relevant risk information to the user and upload relevant control information to the system. For example, device feedback can be provided: by changing the color of an LED light (e.g., green → yellow → red) to indicate different levels of health risk, alerting the user to the current warning or intervention state. Vibration feedback can also be provided: when the system identifies a high-risk state (e.g., abnormal physiological indicators, excessive intake), the e-cigarette will vibrate briefly to enhance user perception. Furthermore, various test data, risk levels, intake prediction models, and predicted nicotine intake and physiological response data output by the physiological response prediction model can also be uploaded to the system for subsequent system processing.

[0135] The present application also provides an electronic cigarette control device, such as Figure 7 Shown, including:

[0136] An acquisition unit 701 is configured to acquire an intake prediction model trained based on training data detected when a user smokes an electronic cigarette, wherein the training data includes at least first physiological response data and first e-liquid atomization data of the user collected by a multimodal sensor;

[0137] An input unit 702 is configured to input detection data detected by the user during each puff of the electronic cigarette into the intake prediction model after the user finishes puffing the electronic cigarette, to obtain a predicted nicotine intake output by the intake prediction model; the detection data includes at least the user's second physiological response data and the second e-liquid atomization data;

[0138] The regulating unit 703 is configured to regulate the nicotine release of the electronic cigarette when the predicted nicotine intake meets the first intervention triggering condition.

[0139] The present application also provides an electronic cigarette control device 800, such as Figure 8 As shown, the control device 800 of the embodiment of the present application may include one or more central processing units (CPUs) 801 and a memory 802 , in which one or more application programs or data are stored.

[0140] Memory 802 may be volatile or persistent storage. The program stored in memory 802 may include one or more modules, each of which may include a series of instruction operations on the electronic device. Furthermore, the central processing unit 801 may be configured to communicate with memory 802, and execute the series of instruction operations in memory 802 on the control device 800.

[0141] The control device 800 may also include one or more power supplies 805, one or more wired or wireless network interfaces 804, one or more input and output interfaces 803, and / or one or more operating systems, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0142] The central processing unit 801 can execute the operations performed by the aforementioned first aspect or any specific method embodiment of the first aspect, and the details will not be repeated here.

[0143] An embodiment of the present application further provides a computer-readable storage medium, which includes instructions. When the instructions are executed on a computer, the computer executes the method described above.

[0144] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0145] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0146] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0147] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0148] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, read-only memory), random access memory (RAM, random access memory), disk or optical disk, and other media that can store program code.

Claims

1. A method for regulating an electronic cigarette, characterized in that: include: Obtaining an intake prediction model trained based on training data detected when a user smokes an electronic cigarette, the training data including at least: first physiological response data and first e-liquid atomization data of the user collected based on a multimodal sensor; During each puff of the electronic cigarette by the user, detection data detected by the user during the puff of the electronic cigarette is input into the intake prediction model to obtain a predicted nicotine intake output by the intake prediction model; the detection data includes at least: second physiological response data of the user and second e-liquid atomization data; When the predicted nicotine intake meets the first intervention trigger condition, the nicotine release of the electronic cigarette is regulated.

2. The control method according to claim 1, wherein The step of obtaining an intake prediction model trained based on training data detected when a user smokes an electronic cigarette comprises: Normalizing the first physiological response data and the first e-liquid atomization data to generate a training feature vector; Inputting the training feature vector into an initial intake prediction model to obtain a first predicted nicotine intake output by the initial intake prediction model; constructing a first loss function based on the first predicted nicotine intake and the actual nicotine intake; Until the first loss function converges, a trained intake prediction model is obtained.

3. The control method according to claim 1, characterized in that Also includes: Obtaining a second predicted nicotine intake based on the training data and the output of the intake prediction model, and a trained physiological response prediction model; The second predicted nicotine intake is output by the intake prediction model after the training data is input into the intake prediction model; During each puff of the electronic cigarette by the user, the detection data and the predicted nicotine intake are input into the physiological response prediction model to obtain physiological response prediction data output by the physiological response prediction model after a preset delay time; The physiological response prediction data includes at least: predicted heart rate and / or predicted blood oxygen saturation; When the physiological response prediction data meets the second intervention trigger condition, the nicotine release amount of the electronic cigarette is regulated.

4. The control method according to claim 3, characterized in that Also includes: Based on the physiological response prediction data output by the physiological response prediction model, electronic cigarettes are intervened and regulated in advance.

5. The control method according to claim 3, characterized in that: The step of obtaining a physiological response prediction model based on the training data and the second nicotine intake prediction training output by the intake prediction model includes: inputting the training data and the second predicted nicotine intake into an initial physiological response prediction model to obtain first physiological response prediction data output by the initial physiological response prediction model after a preset delay time; constructing a second loss function based on a physiological response error between the first physiological response prediction data and the actual physiological response data; Until the second loss function converges, a trained physiological response prediction model is obtained.

6. The control method according to claim 5, characterized in that: The constructing a second loss function based on the physiological response error between the first physiological response prediction data and the actual physiological response data includes: Obtaining an intake error between the second predicted nicotine intake and the actual nicotine intake; The intake error and the physiological response error are weightedly combined to construct a second loss function.

7. The control method according to claim 3, characterized in that: Also includes: When the error between the physiological response prediction data and the actual physiological response data satisfies a preset error condition, updating the model parameters of the intake prediction model so that the predicted nicotine intake output by the intake prediction model approaches the actual nicotine intake; The physiological response prediction model is trained based on the third predicted nicotine intake output by the updated intake prediction model to obtain an updated physiological response prediction model; and the nicotine release of the electronic cigarette is regulated by using the physiological response prediction data output after a preset delay based on the updated physiological response prediction model.

8. The control method according to claim 3, characterized in that: The regulating the nicotine release of the electronic cigarette includes: determining a risk level of the user based on the predicted nicotine intake and the predicted physiological response data; The heating power of the electronic cigarette and the air intake structure of the electronic cigarette are adjusted based on the risk level to regulate the nicotine release of the electronic cigarette.

9. The control method according to claim 3, characterized in that: Also includes: After regulating the e-cigarette, target physiological response data from the user is obtained; The first intervention trigger condition and the second intervention trigger condition, and / or a control mode of the electronic cigarette are adjusted based on the target physiological response data.

10. The control method according to claim 1, characterized in that: The training data also includes: first puffing behavior data of the user, first device state data of the electronic cigarette, and first external environment data of the electronic cigarette; The detection data also includes: second puffing behavior data of the user, second device status data of the electronic cigarette, and second external environment data of the electronic cigarette.

11. An electronic cigarette control device, characterized in that: include: an acquisition unit, configured to acquire an intake prediction model trained based on training data detected when a user smokes an electronic cigarette, the training data comprising at least first physiological response data and first e-liquid atomization data of the user collected based on a multimodal sensor; an input unit, configured to input detection data detected by the user during each puff of the electronic cigarette into the intake prediction model after the user finishes puffing the electronic cigarette, to obtain a predicted nicotine intake output by the intake prediction model; the detection data including at least the user's second physiological response data and the second e-liquid atomization data; The regulating unit is configured to regulate the nicotine release of the electronic cigarette when the predicted nicotine intake meets a first intervention triggering condition.

12. An electronic cigarette control device, characterized in that: include: CPU, memory, input and output interfaces, wired or wireless network interfaces, power supply; The memory is a transient storage memory or a persistent storage memory; The central processing unit is configured to communicate with the memory and execute instructions in the memory on a control plane function entity to perform the method according to any one of claims 1 to 10.

Images