A mobile terminal use control system and a guide method based on a visual attention regulation mechanism

Abstract

The application discloses a mobile terminal use control system and guide method based on a visual attention regulation mechanism, relates to the technical field of human-computer interaction, digital health and mobile internet, and is based on online learning and content similarity control to perform hierarchical management and dynamic selection on reminding materials without changing the existing application ecology of a user, to inhibit reminding habituation, reduce night-time and long-time scrolling and other problems, and to upgrade to functional constraint when the reminding is invalid, so as to realize individualized, auditable and deployable mobile phone use inhibition capability.

Description

Technical Field

[0001] This invention relates to the fields of human-computer interaction, digital health, and mobile internet technology, and more specifically to a mobile terminal usage control system and guidance method based on a visual attention control mechanism. Background Technology

[0002] With the widespread application of mobile internet and content distribution algorithms, users' unplanned use of mobile devices has increased significantly. Applications continuously provide highly salient stimuli through push notifications, social feedback loops, and personalized recommendations, inducing a structural mismatch between immediate reinforcement and delayed loss. This leads to problems such as excessive nighttime use, distracted attention during study / work hours, and shortened sleep duration. Existing "digital health" functions mainly include fixed time quotas, usage statistics, targeted pop-ups, and parental controls. These solutions largely rely on static thresholds or timed rules, which can produce a reminder effect in the short term, but habituation and reminder fatigue are common in long-term interactions: users gradually desensitize to repetitive visual / textual stimuli, actively or passively bypassing reminders, thus rapidly diminishing the inhibitory effect.

[0003] Habituation is an adaptive process, often defined as a weakening of an individual's response when the same stimulus or task is repeatedly presented. Based on previous research, Massimo Turatto summarized nine characteristics of habituation: ① A specific stimulus elicits a response; repeated stimulation leads to a weakening of the response (habituation), and this reduction is usually a negative exponential function of the number of stimulus presentations; ② If the stimulus is inhibited, the response often recovers over time (spontaneous recovery); ③ With a series of repeated habituation training and spontaneous recovery, habituation becomes increasingly rapid (this can be called habituation reinforcement); ④ The weaker the stimulus, the faster the habituation; strong stimuli may not produce obvious habituation; ⑤ The faster the frequency of the stimulus, the faster the habituation; ⑥ Habituation training can also be implicit; ⑦ Habituation to a given stimulus exhibits generalization to other stimuli; ⑧ The appearance of another, stronger stimulus leads to a dehabituation response; ⑨ When dehabituation stimuli are repeatedly applied, the degree of dehabituation gradually decreases; this phenomenon can be called dehabituation habituation.

[0004] From a behavioral science perspective, habituation is a phenomenon where an individual's response amplitude to repetitive and predictable stimuli decreases, often associated with novelty seeking, stimulus specificity, and top-down attention allocation. For reminder-based interventions, continuous presentation of homogeneous stimuli, fixed timing, single modality, and lack of contextualization all accelerate the habituation process, manifesting as a decrease in exit rate, a shortened relapse time, and an increase in avoidance behavior. Some systems attempt to delay fatigue through frequency caps or cooling-off periods, but without systematic constraints on content similarity and cross-topic / cross-modal novelty, "homogeneous stacking" is still likely to occur, making it difficult to maintain significance in the long term.

[0005] At the algorithmic level, traditional offline A / B testing or manual rule tuning and update cycles are long and slow to respond to individual differences and non-stationary behavioral patterns. While recent reinforcement learning and online learning methods can balance exploration and exploitation in interactions, common problems in application to reminder scenarios include: coarse reward definitions (only counting clicks / dwells while ignoring behavioral indicators such as "reignition inhibition"), lack of similarity penalties and novelty budgeting mechanisms, and failure to organically integrate with tiered interventions and functional constraints. This results in algorithms that, while optimizing short-term interaction metrics, cannot effectively combat stimulus-specific habituation.

[0006] From the perspectives of interaction design and human factors engineering, the triggering mechanism of reminders is equally crucial. Reminders triggered upon unlocking or at fixed times offer strong predictability, allowing users to prepare mentally or automatically bypass them. Coupled with contextual variables such as conversation boundaries, prolonged scrolling, and circadian rhythms, and introducing jitter at specific times, reminders can improve salience and intervention fit without significantly increasing frequency. Furthermore, a single "light reminder" is often insufficient during high-load periods, requiring a tiered intervention path in conjunction with functional measures such as screen grayscale, scrolling speed limiting, and cooling lock: maintaining or downgrading if effective, and escalating if ineffective, to achieve gentle and sustainable suppression.

[0007] In terms of privacy and compliance, users are increasingly sensitive to the collection of behavioral and physiological data. Most countries / regions have differentiated restrictions on the content and interaction intensity for minors. Existing solutions often upload data to servers for centralized training, leaving room for improvement in terms of data minimization and local priority. How to achieve individual-level online learning and ethical classification while ensuring localized processing and anonymized aggregation is also an engineering challenge in this field.

[0008] Furthermore, to avoid resistance caused by negative reminders and provide sustainable alternative behavioral paths, the system can not only prompt users to "stop," but also push local and offline high-value alternatives (e.g., "Stand up and look out the window for 20 seconds," "Take 3 deep breaths," "Drink a glass of water and look into the distance for 20 seconds") after detecting prolonged use, to divert attention and reset the context. Simultaneously, the system can output self-efficacy reinforcement information when milestones are achieved (e.g., "Your focus time increased by 20 minutes this week!" "You successfully resisted 5 unconscious screen-swiping attempts"). These alternatives, along with self-efficacy feedback, tiered intervention, cooling-off, and novelty mechanisms, work synergistically to reduce the risk of habituation and improve long-term adherence and the user's sense of control.

[0009] Therefore, there is an urgent need for a technical solution that can continuously, personally, and verifiably suppress excessive mobile device use while maintaining user experience and privacy protection, and that works in conjunction with alternative options and self-efficacy reinforcement mechanisms to address the shortcomings of existing technologies in terms of long-term effectiveness and user compliance. Summary of the Invention

[0010] In view of this, the present invention provides a mobile terminal usage control system and guidance method based on a visual attention regulation mechanism, which solves the problems existing in the background technology.

[0011] To achieve the above objectives, the present invention provides the following technical solution:

[0012] A mobile terminal usage control system based on a visual attention regulation mechanism, employing a closed-loop collaborative architecture, specifically includes:

[0013] The data acquisition unit is used to acquire user behavior data and context information on mobile terminals in real time, generate time-series behavior logs and output them to the habituation evaluation unit, the dehabituation scheduling unit and the learning and updating unit.

[0014] The habituation assessment unit receives historical reminder response records from the data acquisition unit, combines them with stimulus similarity data provided by the stimulus material library and the grading unit, and calculates a habituation score in the [0,1] interval using a time decay model. It also outputs multidimensional habituation scores based on themes or modalities;

[0015] The stimulus material library and hierarchical unit are used to store multi-dimensional reminder materials and classify and manage them according to intensity level, theme and modality, and maintain metadata for each reminder material;

[0016] De-habituation scheduling unit, used to receive habituated scores Metadata and usage load indicators output by the data acquisition unit are used to select target alert materials in the candidate material set through online exploration and utilization strategies. The intensity level of the material or the combination of stimulus schemes are dynamically determined using load indicators and output to the triggering and presentation unit;

[0017] The triggering and presentation unit is used to monitor triggering conditions in real time and present the target reminder material selected by the de-habituation scheduling unit in various forms when the conditions are met. It also executes various functional constraints in conjunction with the intensity level of the material and synchronously feeds back the execution timestamp of the reminder action to the feedback collection unit.

[0018] The feedback acquisition unit is used to record user behavior feedback and optional physiological signals within a preset observation window after the triggering and presentation unit outputs a reminder, generate feedback parameters and calculate inhibition rewards;

[0019] The learning and updating unit is used to calculate key indicators based on the suppression reward data and the cumulative usage data of the data collection unit within the rolling evaluation window, to update habitual scores, scheduling parameters and material weights online, and to demote or replace inefficient materials to achieve iterative optimization of the strategy.

[0020] Optionally, usage behavior data includes: app foreground dwell time, unlock events and session boundaries, continuous scrolling time and scrolling speed, nighttime usage markers, and exit, re-ignite, and bypass behaviors after the reminder is displayed.

[0021] Optionally, stimulus similarity data is determined by the cosine similarity of the material embedding vectors; similarity threshold. This is used to determine homogeneous stimuli; when the similarity is ≥ θ, the stimulus material library and the hierarchical unit set a cooldown period for the current material. .

[0022] Optionally, the habituation assessment unit performs a weighted summation of the similarity between the ineffective response and the current candidate stimulus based on the time decay coefficient, and then maps it to the [0,1] interval through a compression function to obtain the habituation score. .

[0023] Optionally, the default action constraints for intensity levels are as follows: Level A corresponds to reminder materials in the form of lightweight text or icons, with no additional action constraints by default; Level B corresponds to reminder materials in the form of images combined with short sentences, with a reading wait constraint set by default; Level C corresponds to reminder materials in the form of images combined with short sentences, with an action constraint that triggers screen grayscale or scrolling speed limit of the application interface by default; Level D corresponds to reminder materials in the form of combined stimuli, with an action constraint that triggers cooldown lock by default.

[0024] Optional metadata includes: unique identifier, intensity level, topic tag, modal tag, embedding vector, ethical level, cooling parameters, exposure count, reward statistics, and the last performance evaluation time.

[0025] Optional load metrics include: continuous rolling time threshold, nighttime cumulative time threshold, and recent resurgence delay quantile; when habitual scoring... Alternatively, when the load index reaches a preset condition, the intensity can be increased or a combination of stimuli can be triggered.

[0026] Optionally, the construction of the candidate material set must meet all of the following conditions: ethical level and user age / region compliance, homogeneous material cooling-off period, topic cooling-off pool constraints and similarity constraints.

[0027] Optionally, the de-habituation scheduling unit adopts an online exploration-exploitation strategy as a multi-armed gambling machine strategy, with the arm space corresponding to candidate materials.

[0028] Optionally, the objective function of the de-habituation scheduling unit introduces a similarity penalty and a novelty budget, which reduces the score of highly similar materials while giving novelty rewards to cross-topic and cross-modal materials.

[0029] Optionally, triggering conditions include: session boundary triggering, unlock triggering, long scrolling triggering, and circadian rhythm triggering; presentation methods include: lock screen overlay, system notification, or floating window; the activation condition for functional constraints is: continuous q The reminder was invalid or The constraint duration is not lower than the threshold H_max, and the constraint duration increases according to a stepwise strategy.

[0030] Optionally, the suppression reward can be binary or continuous: 1 if the user does not return to high intensity within the observation window after the alert, and 0 otherwise; or, normalized to a continuous reward based on the re-ignition delay, exit delay, and the magnitude of the decrease in the rolling rate.

[0031] Optionally, the learning and updating unit calculates key indicators and makes a compliance judgment within a rolling evaluation window of 7 days, 14 days, or 28 days. The judgment conditions include: the total nighttime usage does not exceed the target value, the median re-ignition delay is not lower than the minimum value, and the high-load re-ignition ratio after the reminder is not higher than the upper limit value. When the standard is met, the existing intensity and cooling strategy are maintained and a low-probability exploration is retained. When the standard is not met, at least one of the following is adjusted according to preset rules: intensity, combination, similarity penalty parameters, cooling parameters, and triggering strategy.

[0032] This embodiment also proposes a mobile terminal usage guidance method based on a visual attention control mechanism, applied to a mobile terminal usage control system based on a visual attention control mechanism as described above, including the following steps:

[0033] S1. Real-time acquisition of user behavior data and context information on mobile terminals to generate time-series behavior logs;

[0034] S2. Calculate the habituation score based on historical reminder response records, stimulus similarity data, and a time decay model. ;

[0035] S3. Construct a candidate material set from the graded stimulus material library based on ethical level, cooling, similarity threshold and topic cooling pool;

[0036] S4. Employ online exploration – utilize strategies to select target reminder materials from the candidate material set, combined with… Dynamically determine the intensity level of materials or combined stimulus schemes using load indicators;

[0037] S5. When the triggering conditions are met, present target reminder materials in multiple forms and execute multiple functional constraints in conjunction with the intensity level of the materials; within the preset observation window, record user behavior feedback and optional physiological signals, generate feedback parameters and calculate inhibition rewards;

[0038] S6. Within the rolling evaluation window, determine the compliance of key indicators based on the suppression reward data. If the compliance is met, maintain the existing strategy. If the compliance is not met, adjust at least one of the following according to the preset rules: intensity, combination, similarity penalty parameters, cooling parameters, and triggering strategy, and return to S4 for iteration execution.

[0039] Optionally, in S3, candidate materials with a similarity of not less than the threshold θ for the most recently used material are only allowed to enter the candidate material set if the cooling-off period expires and the novelty budget remains, and the novelty budget is deducted after selection.

[0040] Optionally, it also includes: when the habituation rating reaches a combined threshold or is continuous q When the inhibition reward is 0, cross-modal combined stimulation is used and functional constraints are enabled.

[0041] As can be seen from the above technical solution, compared with the prior art, the present invention discloses a mobile terminal usage control system and guidance method based on a visual attention control mechanism, which has the following beneficial effects:

[0042] This invention utilizes H(t) to quantify habituation, similarity penalty and novelty budget constraint selection, and multi-armed gambling machine to achieve online exploration and exploitation. It also links with scenario-based triggering, jitter strategy, and hierarchical / functional constraints to systematically combat reminder failure caused by homogeneous stimuli. While ensuring privacy and user experience, it continuously and individually suppresses excessive usage behaviors such as nighttime and long-term scrolling. It has the advantages of simple implementation, calibrable parameters, auditability and verification, and easy deployment. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0044] Figure 1 This is a schematic diagram of the overall system architecture of the present invention;

[0045] Figure 2 This is a flowchart of the method of the present invention. Detailed Implementation

[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] Example 1

[0048] To continuously suppress problematic mobile phone usage without increasing the burden on users, embodiments of this invention disclose a mobile terminal usage control system based on a visual attention control mechanism, such as... Figure 1 As shown, a closed-loop collaborative architecture is adopted, specifically including:

[0049] The data acquisition unit is used to acquire user behavior data and context information on mobile terminals in real time, generate time-series behavior logs and output them to the habituation evaluation unit, the dehabituation scheduling unit and the learning and updating unit.

[0050] The habituation assessment unit receives historical reminder response records from the data acquisition unit, combines them with stimulus similarity data provided by the stimulus material library and the grading unit, and calculates a habituation score in the [0,1] interval using a time decay model. This reflects the degree of desensitization of users to a particular topic or modality of stimulation, and outputs a multidimensional habituation score based on the topic or modality. In this embodiment, the habituation score is based on the response intensity of the last k reminders. r Similarity to the current candidate stimulus s The reaction intensity is obtained by normalizing indicators such as exit delay, reignition delay, and rolling rate decrease, as well as the time decay factor calculation.

[0051] The stimulus material library and hierarchical unit are used to store multi-dimensional reminder materials and classify and manage them according to intensity level, theme and modality, and maintain metadata for each reminder material;

[0052] De-habituation scheduling unit, used to receive habituated scores Metadata and usage load indicators output by the data acquisition unit are used to select target alert materials in the candidate material set through online exploration and utilization strategies. The intensity level of the material or the combination of stimulus schemes are dynamically determined using load indicators and output to the triggering and presentation unit;

[0053] The triggering and presentation unit is used to monitor the triggering conditions in real time and present the target reminder material selected by the de-habituation scheduling unit in various forms when the conditions are met. It also executes various functional constraints in conjunction with the intensity level of the material and synchronously feeds back the execution timestamp of the reminder action to the feedback acquisition unit. In this embodiment, the triggering and presentation unit uses a jitter strategy to randomize the triggering time, with the jitter amplitude being 10% to 25% of the target time, in order to reduce the predictability of the reminder.

[0054] The feedback acquisition unit is used to record user behavior feedback and optional physiological signals within a preset observation window after the triggering and presentation unit outputs a reminder, generate feedback parameters and calculate inhibition rewards;

[0055] The learning and updating unit is used to calculate key indicators based on the suppression reward data and the cumulative usage data of the data collection unit within the rolling evaluation window, to update habitual scores, scheduling parameters and material weights online, and to demote or replace inefficient materials to achieve iterative optimization of the strategy.

[0056] See Figure 1 The various units in this embodiment constitute a closed loop of "collection-evaluation-scheduling-presentation-feedback-learning". Through the technical link of "quantification of habituation-diversified scheduling-tiered upgrade-closed-loop learning", it can effectively delay or block the habituation of reminders, maintain the salience of stimuli, and achieve quantifiable improvement on total time, relapse delay and nighttime time. It has beneficial effects such as individualization, scalability and privacy localization.

[0057] Furthermore, usage behavior data includes: application foreground dwell time, unlock events and session boundaries (unlock / lock screen, foreground / background switching), continuous scrolling time and scrolling speed, nighttime usage markers, and exit, re-ignition, and bypass behaviors after the reminder is displayed; this data can be written to a local time-series log with timestamps, with a period of 1 to 5 seconds.

[0058] Furthermore, the stimulus similarity data is determined by the cosine similarity of the material embedding vectors, ranging from [0,1], with larger values ​​indicating more pronounced habituation; the similarity threshold... This is used to determine homogeneous stimuli; when the similarity is ≥ θ, the stimulus material library and the hierarchical unit set a cooldown period for the current material. To avoid repeated deployments in the short term.

[0059] Furthermore, the habituation assessment unit performs a weighted summation of the similarity between the ineffective response and the current candidate stimulus based on the time decay coefficient, and then maps it to the [0,1] interval through a compression function to obtain the habituation score. When the similarity or the proportion of invalid responses increases, It increases accordingly.

[0060] Furthermore, the default action constraints for each intensity level are as follows: Level A corresponds to lightweight text or icon-based reminder materials, with no additional action constraints by default; Level B corresponds to reminder materials combining images and short sentences, with a default reading wait constraint; Level C corresponds to reminder materials combining images and short sentences, triggering screen grayscale or application interface scrolling speed limits by default; Level D corresponds to reminder materials with combined stimuli, triggering a cooldown lock by default. In addition, the grading can be based on... Adaptive to load adjustment; cooling parameters include the cooling period of homogeneous materials. With topic-level cooldown times u .

[0061] Furthermore, the metadata includes: unique identifier, intensity level (A / B / C / D), topic tags, modal tags (image / text / audio / statistical visualization), embedding vector (dimension ≥128), ethical level, cooling-off parameters, exposure count, return statistics, and the last performance evaluation time.

[0062] Furthermore, load metrics used include: continuous rolling time threshold, nighttime cumulative time threshold, and recent resurgence delay quantile; when habitual scoring... Alternatively, when the load index reaches a preset condition, the intensity can be increased or a combination of stimuli can be triggered.

[0063] Furthermore, the construction of the candidate material set must meet all of the following conditions: ethical level and user age / region compliance, expiration of the cooling-off period for homogeneous materials, topic cooling-off pool constraints, and similarity constraints. In this embodiment, the construction of the candidate material set includes a topic cooling-off pool: the most recent u The next campaign will have a theme-based cooldown period and will be subject to mandatory cross-theme rotation. u The value ranges from 1 to 5.

[0064] Furthermore, the de-habituation scheduling unit employs an online explore-exploitation strategy of a multi-armed gambler strategy, with the arm space corresponding to candidate materials. In this embodiment, the multi-armed gambler can be Thompson Sampling, with each material arm maintaining... The posterior distribution is used to update parameters based on observed returns. Multi-armed gambling machines can also be UCB or UCB-Tuned, with the arm selection determined by the sum of the average return and the upper bound of the uncertainty. Based on the above design, this embodiment supports contextual gambling machine extensions, using time period, scene type, recent usage intensity, and individual characteristics as contextual features, and estimating the returns of each material arm using linear UCB or Bayesian linear regression.

[0065] Furthermore, the objective function of the de-habituation scheduling unit introduces a similarity penalty and a novelty budget, which reduces the score of highly similar materials while giving novelty rewards to cross-topic and cross-modal materials.

[0066] Furthermore, the triggering conditions include: session boundary triggering, unlock triggering, long scrolling triggering, and circadian rhythm triggering; the presentation methods include: lock screen overlay, system notification, or floating window; the activation condition for functional constraints is: continuous q The reminder was invalid or The constraint duration is not lower than the threshold H_max, and the constraint duration increases according to a stepwise strategy.

[0067] Furthermore, the suppression reward is a binary or continuous value: within the observation window after the reminder, it is recorded as 1 if it does not return to high intensity use, otherwise it is 0; or, it is normalized to a continuous reward based on the re-ignition delay, exit delay, and the reduction in rolling rate.

[0068] Furthermore, the learning and updating unit calculates key indicators and makes a compliance judgment within a rolling evaluation window of 7 days, 14 days, or 28 days. The judgment conditions include: the total nighttime usage does not exceed the target value, the median re-ignition delay is not lower than the minimum value, and the high-load re-ignition ratio after the reminder is not higher than the upper limit value. When the target is met, the existing intensity and cooling strategy are maintained and a low-probability exploration is retained. When the target is not met, at least one of the following is adjusted according to preset rules: intensity, combination, similarity penalty parameters, cooling parameters, and triggering strategy.

[0069] In this embodiment, the system can employ exponential forgetting or sliding window statistics on historical samples and reset or pull back the posterior when inefficiency occurs over a long period. The system stores logs and model parameters locally by default, only uploading anonymous aggregate metrics, and provides options for cleaning and differential privacy noise injection. The underage mode restricts the ethical and intensity levels of candidate materials, blocks Category D and sensitive topics, and automatically reduces contrast and mutes at night. Furthermore, the system is deployed as an application-layer SDK, mobile device management component, or system service, and interacts with mobile operating system notifications, overlays, and accessibility interfaces to implement presentation and functional constraints.

[0070] In addition, this embodiment also proposes a mobile terminal usage guidance method based on a visual attention control mechanism, which is applied to any of the above-described mobile terminal usage control systems based on a visual attention control mechanism, such as... Figure 2 As shown, it includes the following steps:

[0071] S1. Data Acquisition: Real-time acquisition of user behavior data and context information on mobile terminals, generating time-series behavior logs; recording feedback signals such as exit / re-launch, and optional acquisition of physiological / kinematic signals such as heart rate and acceleration;

[0072] S2. Habituation Assessment: Based on historical reminder response records, stimulus similarity data, and a time decay model, a habituation score is calculated. ;

[0073] S3. Candidate set construction: Construct a candidate set from the graded stimulus material library based on ethical level, cooling, similarity threshold and topic cooling pool;

[0074] S4. Online Scheduling: Employing online exploration—using strategies to select target reminder materials from the candidate material set, combined with... Dynamically determine the intensity level of materials or combined stimulus schemes using load indicators;

[0075] S5, Triggering and Presentation, Feedback and Reward: When the triggering condition is met, target reminder materials are presented in multiple forms, and multiple functional constraints are executed in conjunction with the intensity level of the materials; within the preset observation window, user behavior feedback and optional physiological signals are recorded, feedback parameters are generated and suppression rewards are calculated;

[0076] S6, Learning Update: Within the rolling evaluation window, key indicators are judged based on the suppression reward data. If the target is met, the existing strategy is maintained. If the target is not met, at least one of the following is adjusted according to preset rules: intensity, combination, similarity penalty parameters, cooling parameters, and triggering strategy, and then returned to S4 for iteration execution.

[0077] Furthermore, in S3, candidate materials whose similarity to the most recently used material is not lower than the threshold θ are only allowed to enter the candidate material set if the cooling-off period expires and the novelty budget remains, and the novelty budget is deducted after selection.

[0078] Furthermore, the method also includes: when the habituation score reaches a combined threshold or is continuous q When the inhibition reward is 0, cross-modal combined stimulation is used and functional constraints are enabled.

[0079] To ensure the feasibility and reproducibility of this embodiment, the implementation range of the parameters and thresholds is given below:

[0080] (1) Similarity threshold 0.6~0.9; Homogeneous cooling : 6h~72h; number of theme cooldowns u : 1~5;

[0081] (2) Observation window Evaluation window: 5~30min; W : 7 / 14 / 28 days;

[0082] (3) Trigger jitter amplitude: ±10%~±25%;

[0083] (4) Continuous scrolling duration threshold: 60~600s; Nighttime cumulative time target: 10~30min;

[0084] (5) Functional constraint ladder: benchmark locking Increasing coefficient ;

[0085] (6) Continuous invalid threshold q : 2~5; Upper limit of habituation : 0.6~0.9;

[0086] (7) Novelty budget B The number of evaluations is 3 to 10 per evaluation window, and increases linearly with the increase of H(t).

[0087] This application provides a mobile terminal usage control system and method based on a de-habituation mechanism, applied to a management device in an augmented reality system. The augmented reality system also includes: an augmented reality device, an olfactory stimulation device, a tactile stimulation device, a physiological signal monitoring device, a motion capture device, and a printer. The general operation steps are as follows:

[0088] First, the user selects a scenario or uploads scenario information. They can choose a target scenario from a preset list, such as "watching short videos at night / long scrolling / distraction at work," or upload a custom scenario (specifying time period, application category / package name, and trigger threshold). The client generates a scenario description, including the time period, application class, and session threshold, and persists it locally.

[0089] Secondly, users upload persistent information, including: goals, daily routines, and time limits. This includes their usual sleep schedule [23:00, 07:00] and the total nighttime time spent on their daily goals (T). night =30min, maximum single session T sess =10min, minor mode on / off, sensitive topic blocking list, etc.

[0090] The user selects a reminder target, either by checking a template or by creating a custom one: meeting G for 14 consecutive days. Based on this, the system generates a compliance window W7 / 14 / 28 and a threshold (T). night Δ min p max Generate target scene models, such as session boundaries and circadian rhythms. Real-time detection of unlocking, screen locking, and foreground / background switching on the device side identifies session boundaries; scene weights are assigned according to circadian rhythms, such as a nighttime weight. =1.3, daytime =1.0, used for subsequent triggering and reward weighting. Generate a target grading model, including intensity, similarity, cooldown, etc., and establish a four-level AD system. A represents lightweight text notifications, such as: no obstruction, no speed limit, etc. B represents text + image, with a selectable reading wait time (t). read =2-4s. C represents the combination of text, image, screen grayscale, or scrolling speed limit, with a speed limit coefficient. ∈[0.5,0.8]. D represents a combination stimulus + cooling lock, temporarily locking the target using t.lock =30-120s. Configure similarity threshold. (Default 0.75) and theme cooling pool (homogeneous themes require cooling) =2-6h).

[0091] Control the presentation device and output the target object. When the trigger condition is met, call the system notification / overlay API to output the selected material item and synchronously record the timestamp and session context. Determine the visual / copy parameters corresponding to the target scene. Set the following for the "watching short videos at night" scene: Night-themed copy template (including independent variable slots, such as "This round has been continuously scrolling") min), font size / contrast accessibility parameters, and reading wait time (t). read And the specific values ​​of the grayscale curve / speed limit threshold.

[0092] Determine the intensity of the negative reminder and the parameters of the combined stimulus. Select A / B / C / D based on the current H(t) and the progress towards the target; if H(t) ≥ 0.7 or continuous... q If the return is the lowest possible, upgrade to the next level or activate a combination of stimuli. Distribute and record the immediate suppression effect. Distribute materials to the front end while simultaneously calculating the immediate suppression return. r If in the observation window W If the system does not return to "high load" within 10 minutes (e.g., continuous scrolling > 60 seconds or switching to a similar application), then... r =1; otherwise, normalize according to the reignition delay Δ. r =min(Δ / W,1), and record the exit delay, number of bypasses, and exposure count.

[0093] A similarity penalty and a novelty budget are set to form a candidate set. Using the most recent T deployments H as history, a candidate set C = {x | x is compliant and its cooling-off period has expired, and max s(x,y) < θ, y ∈ H} is constructed. If too small, use the "novelty budget" for cross-topic / cross-modal completion (budget percentage). =10%-20%).

[0094] Collect physiological / behavioral signals to generate feedback parameters. Mandatory edge-side sampling: exit delay, re-ignition delay, scroll rate decrease percentage, and bypass path. Optional sampling: heart rate / accelerometer / grip posture (if linked to wearable devices). All signals are only anonymized locally for statistical analysis.

[0095] Update the habituation rating H(t). (Based on recent...) k The rewards and similarity of subsequent topic-related reminders are updated with time decay, obtained through cold-start empirical values ​​or online learning. A higher H(t) indicates a need for stronger or more novel stimuli. A multi-armed gambling machine is run, and the posterior / upper confidence bounds of each arm are updated. For the candidate set... C Maintenance of each arm .

[0096] Determine whether the target inhibition effect has been achieved. Evaluate the conjunction rule within a 7 / 14 / 28-day rolling window: total nighttime duration ≤ T. night The median reignition delay is ≥ Δ min (e.g., 6 minutes); High-load reignition ratio ≤ p max (e.g., ≤20%). If all conditions are met, it is considered "compliant"; otherwise, it is considered "non-compliant". If "yes", maintain the current intensity and cooling strategy, proceed to the next round; maintain the current level and topic distribution, only fine-tune the reading wait or trigger jitter ratio (10%–25%), and extend the cooling time for homogeneous topics to prevent regression. If "no", adjust the intensity / combination / cooling; enable functional constraints if necessary. Priority order: increase intensity level or enable combined stimuli (B→C, or B+C). Increase similarity penalty. With threshold Strengthen de-homogenization. Shorten the trigger interval and increase jitter; trigger functional constraints (grayscale / scrolling speed limit / cooling lock level D), and increase the duration according to a step strategy (e.g., 30s→60s→120s), and return to the loop after the strategy is updated.

[0097] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. The methods disclosed in the embodiments are described simply because they correspond to the systems disclosed in the embodiments; relevant details can be found in the method section.

[0098] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A mobile terminal usage control system based on a visual attention control mechanism, characterized in that, A closed-loop collaborative architecture is adopted, specifically including: The data acquisition unit is used to acquire user behavior data and context information on mobile terminals in real time, generate time-series behavior logs and output them to the habituation evaluation unit, the dehabituation scheduling unit and the learning and updating unit. The habituation assessment unit receives historical reminder response records from the data acquisition unit, combines them with stimulus similarity data provided by the stimulus material library and the grading unit, and calculates a habituation score in the [0,1] interval using a time decay model. It also outputs multidimensional habituation scores based on themes or modalities; The stimulus material library and hierarchical unit are used to store multi-dimensional reminder materials and classify and manage them according to intensity level, theme and modality, and maintain metadata for each reminder material; De-habituation scheduling unit, used to receive habituated scores Metadata and usage load indicators output by the data acquisition unit are used to select target alert materials in the candidate material set through online exploration and utilization strategies. The intensity level of the material or the combination of stimulus schemes are dynamically determined using load indicators and output to the triggering and presentation unit; The triggering and presentation unit is used to monitor triggering conditions in real time and present the target reminder material selected by the de-habituation scheduling unit in various forms when the conditions are met. It also executes various functional constraints in conjunction with the intensity level of the material and synchronously feeds back the execution timestamp of the reminder action to the feedback collection unit. The feedback acquisition unit is used to record user behavior feedback and optional physiological signals within a preset observation window after the triggering and presentation unit outputs a reminder, generate feedback parameters and calculate inhibition rewards; The learning and updating unit is used to calculate key indicators based on the suppression reward data and the cumulative usage data of the data collection unit within the rolling evaluation window, to update habitual scores, scheduling parameters and material weights online, and to demote or replace inefficient materials to achieve iterative optimization of the strategy.

2. The mobile terminal usage control system based on a visual attention control mechanism according to claim 1, characterized in that, User behavior data includes: app foreground dwell time, unlock events and session boundaries, continuous scrolling time and scrolling speed, nighttime usage markers, and exit, re-ignition, and bypass behaviors after the reminder is displayed.

3. A mobile terminal usage control system based on a visual attention control mechanism according to claim 1, characterized in that, Stimulus similarity data is determined by the cosine similarity of the material embedding vectors; similarity threshold. This is used to determine homogeneous stimuli; when the similarity is ≥ θ, the stimulus material library and the hierarchical unit set a cooldown period for the current material. .

4. A mobile terminal usage control system based on a visual attention control mechanism according to claim 1, characterized in that, The habituation assessment unit performs a weighted summation of the similarity between the ineffective response and the current candidate stimulus based on the time decay coefficient, and then maps it to the [0,1] interval through a compression function to obtain the habituation score. .

5. A mobile terminal usage control system based on a visual attention control mechanism according to claim 1, characterized in that, The default action constraints for each intensity level are as follows: Level A corresponds to reminder materials in the form of lightweight text or icons, with no additional action constraints by default; Level B corresponds to reminder materials in the form of images combined with short sentences, with a reading wait constraint set by default; Level C corresponds to reminder materials in the form of images combined with short sentences, with an action constraint that triggers screen grayscale or scrolling speed limit of the application interface by default; Level D corresponds to reminder materials in the form of combined stimuli, with an action constraint that triggers cooldown lock by default.

6. A mobile terminal usage control system based on a visual attention control mechanism according to claim 1, characterized in that, Metadata includes: Unique identifier, intensity level, topic tag, modal tag, embedding vector, ethical rating, cooling parameters, exposure count, reward statistics, and the last performance evaluation time.

7. A mobile terminal usage control system based on a visual attention control mechanism according to claim 1, characterized in that, The load metrics used include: continuous rolling time threshold, nighttime cumulative time threshold, and recent resurgence delay quantile; when habitual scoring Alternatively, when the load index reaches a preset condition, the intensity can be increased or a combination of stimuli can be triggered.

8. A mobile terminal usage control system based on a visual attention control mechanism according to claim 1, characterized in that, The construction of the candidate material set must meet all of the following conditions: ethical level and user age / region compliance, homogeneous material cooling-off period, theme cooling-off pool constraints and similarity constraints.

9. A mobile terminal usage control system based on a visual attention control mechanism according to claim 1, characterized in that, The de-habituation scheduling unit adopts an online exploration-exploitation strategy similar to a multi-armed gambling machine strategy, with the arm space corresponding to candidate materials.

10. A mobile terminal usage control system based on a visual attention control mechanism according to claim 1, characterized in that, The objective function of the de-habituation scheduling unit introduces a similarity penalty and a novelty budget, which reduces the score of highly similar materials while rewarding novelty materials across topics and modalities.

11. A mobile terminal usage control system based on a visual attention control mechanism according to claim 1, characterized in that, Triggering conditions include: session boundary triggering, unlocking triggering, long scrolling triggering, and circadian rhythm triggering; presentation methods include: lock screen overlay, system notification, or floating window; the activation condition for functional constraints is: continuous q The reminder was invalid or The constraint duration is not lower than the threshold H_max, and the constraint duration increases according to a stepwise strategy.

12. A mobile terminal usage control system based on a visual attention control mechanism according to claim 1, characterized in that, Suppression rewards are binary or continuous values: within the observation window after the alert, a value of 1 is recorded if the user does not return to high-intensity use, otherwise a value of 0; or, the value is normalized to a continuous reward based on the re-ignition delay, exit delay, and the decrease in scroll rate.

13. A mobile terminal usage control system based on a visual attention control mechanism according to claim 1, characterized in that, The learning and updating unit calculates key indicators and makes a compliance judgment within a rolling evaluation window of 7 days, 14 days, or 28 days. The judgment conditions include: the total nighttime usage does not exceed the target value, the median re-ignition delay is not lower than the minimum value, and the high-load re-ignition ratio after the reminder is not higher than the upper limit value. When the standard is met, the existing intensity and cooling strategy are maintained and a low-probability exploration is retained. When the standard is not met, at least one of the following is adjusted according to the preset rules: intensity, combination, similarity penalty parameters, cooling parameters, and triggering strategy.

14. A method for guiding the use of mobile terminals based on a visual attention control mechanism, characterized in that, The mobile terminal usage control system based on a visual attention control mechanism as described in any one of claims 1-13 includes the following steps: S1. Real-time acquisition of user behavior data and context information on mobile terminals to generate time-series behavior logs; S2. Calculate the habituation score based on historical reminder response records, stimulus similarity data, and a time decay model. ; S3. Construct a candidate material set from the graded stimulus material library based on ethical level, cooling, similarity threshold and topic cooling pool; S4. Employ online exploration – utilize strategies to select target reminder materials from the candidate material set, combined with… Dynamically determine the intensity level of materials or combined stimulus schemes using load indicators; S5. When the triggering conditions are met, present target reminder materials in multiple forms and execute multiple functional constraints in conjunction with the intensity level of the materials; within the preset observation window, record user behavior feedback and optional physiological signals, generate feedback parameters and calculate inhibition rewards; S6. Within the rolling evaluation window, determine the compliance of key indicators based on the suppression reward data. If the compliance is met, maintain the existing strategy. If the compliance is not met, adjust at least one of the following according to the preset rules: intensity, combination, similarity penalty parameters, cooling parameters, and triggering strategy, and return to S4 for iteration execution.

15. A mobile terminal usage guidance method based on a visual attention control mechanism according to claim 14, characterized in that, In S3, candidate materials with a similarity of not less than the threshold θ to the most recently used material are only allowed to enter the candidate material set if the cooling-off period has expired and the novelty budget remains, and the novelty budget is deducted after selection.

16. A mobile terminal usage guidance method based on a visual attention control mechanism according to claim 14, characterized in that, Also includes: When habitual ratings reach the combined threshold or are continuous q When the inhibition reward is 0, cross-modal combined stimulation is used and functional constraints are enabled.

Images