Behavior problem analysis method and system based on 24-hour activity data

Abstract

The invention discloses a behavior problem analysis method and system based on 24-hour activity data, and mainly relates to the field of medical health. Comprising a multi-source data acquisition behavior problem modeling module, a wearable monitoring terminal and a cloud processing engine deployment XGBoost risk prediction sub-module. The method has the beneficial effects that medical resources are optimized; the family burden is reduced; the rehabilitation effect is improved; and the scientificity of intervention suggestions is ensured.

Description

[0001] The present invention relates to the field of medical health, and in particular to a method and system for analyzing behavioral problems in children and adolescents with autism spectrum disorders based on 24-hour activity monitoring. Background Art

[0002] Autism is a neurodevelopmental disorder. Data from the US CDC show that between 2000 and 2018, the prevalence of autism in eight-year-olds increased fourfold, from 1 in 166 to 1 in 44. China's "2021 Blue Book on Child Development Disorder Rehabilitation Industry" indicates that as of 2020, a conservative estimate of 3 million children and adolescents aged 0-18 with autism existed, with approximately 6 million adults caring for them. Approximately 9 million people are affected, and this number is expected to rise. Both the National 14th Five-Year Plan for Disabilities and the report to the 20th National Congress of the Communist Party of China emphasize protecting people's health. Focusing on the healthy growth of children and adolescents with autism is a national priority and requires social participation. Their physical and mental health is a crucial foundation for the "Healthy China" strategy and is of urgent importance.

[0003] The core characteristics of children and adolescents with autism are limited social interaction, communication deficits, narrow interests, and repetitive and stereotyped behaviors, often accompanied by other psychiatric comorbidities, with behavioral problems being particularly common. These symptoms affect their emotional, mental health, and personality development, weakening their ability to live independently and adapt to society. Recent studies have found that they are more likely to experience sensory problems such as anxiety and loneliness, as well as behavioral problems such as aggression and hyperactivity, the latter of which can aggravate symptoms and affect rehabilitation outcomes. Therefore, comprehensive attention should be paid to their behavioral problems, with timely screening and assessment to support rehabilitation.

[0004] Recent studies have further revealed that compared to their peers, children and adolescents with autism are more likely to experience emotional problems such as anxiety, withdrawal, and emotional instability. They are also more likely to exhibit behavioral issues such as aggressive behavior, attention deficit hyperactivity disorder (ADHD), and difficulty concentrating. These behavioral problems not only exacerbate the core symptoms of autism but may also hinder the effectiveness of educational interventions and rehabilitation training. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for analyzing behavioral problems in children and adolescents with autism spectrum disorders based on 24-hour activity monitoring. It uses high-density temporal feature extraction and causal inference models to achieve a closed-loop management of "full-cycle monitoring → precise attribution → spatiotemporal constraint instructions", thereby increasing the efficiency of behavioral problem intervention by 2.1 times.

[0006] To achieve the above-mentioned objectives, the present invention is implemented through the following technical solutions: a method and system for analyzing behavioral problems in children and adolescents with autism spectrum disorders based on 24-hour activity monitoring, comprising the following steps:

[0007] Multi-source data collection: Wearable sensors are used to continuously collect 24-hour activity time series data for at least 7 days, including motion acceleration, light intensity, and heart rate variability; log data entered by guardians and quantitative scores of clinical behavior scales are simultaneously obtained; wearable sensors (accelerometers / light sensors / HRV) and manual logs (emotional event markers) are integrated to solve the problem of single data source deviation.

[0008] Activity feature extraction: The time series data is segmented and the following core features are extracted:

[0009] Sleep quality indicators: sleep fragmentation index, deep sleep ratio;

[0010] Exercise intensity indicators: proportion of moderate to high intensity exercise (MVPA) time, average duration of sedentary behavior;

[0011] Rhythm stability indicators: variance of daytime light exposure fluctuations, activity-rest transition frequency;

[0012] Modeling behavioral problems:

[0013] Convert behavioral scale scores into numerical behavioral labels, including aggression intensity and frequency of emotional outbursts;

[0014] Establish a multivariate linear mixed model (LMM) with the activity characteristics as fixed effects and individual physiological parameters as random effects, and output the correlation coefficient matrix between activity characteristics and behavioral labels;

[0015] Personalized intervention generation: Generates structured action adjustment instructions when the model detects that pre-set risk conditions are met, including:

[0016] Sleep fragmentation index>0.5 and MVPA proportion<20%;

[0017] Sedentary behavior lasts for more than 3 hours at a time and the light exposure variance is less than 100 lux 2 . Achieve early warning of high-risk events.

[0018] The Granger causality test is used to verify the predictive significance of activity features on behavior labels (p<0.05), and only the features that pass the test are retained and input into the LMM model; thus solving the ambiguity problem of traditional suggestions.

[0019] Wearable monitoring terminal: Integrates a three-axis accelerometer, a light sensor, and a heart rate monitoring module to upload encrypted activity data in real time;

[0020] Cloud processing engine: activity feature extraction module, behavior association modeling module

[0021] User interaction terminal:

[0022] Guardian log input interface, supporting emotional event marking;

[0023] Visual report generation unit, outputting 24-hour activity-behavior heat map;

[0024] The instruction push unit, which triggers and adjusts instructions according to risk conditions, is a key design to improve execution rate.

[0025] The wearable monitoring terminal adopts a tear-resistant wristband structure, and the sensor module and the wristband are detachably connected by magnetic attraction.

[0026] The cloud-based processing engine deploys an XGBoost risk prediction submodule, which inputs activity features to predict the risk level (high / medium / low) of behavioral problems within the next 48 hours with an accuracy rate of ≥85%.

[0027] Compared with the prior art, the beneficial effects of the present invention are:

[0028] 1. Decompose the 24-hour activity rhythm into a continuous dynamic system of sleep-movement-restlessness. By extracting temporal features (such as sleep fragmentation index and activity transition frequency) and using a mixed-effects model (LMM), we reveal the nonlinear cumulative effects of activity patterns on behavioral problems. Granger causality testing is used to screen for predictive features (e.g., "afternoon restlessness → aggression") to ensure the scientific validity of intervention recommendations.

[0029] 2. Optimize medical resources, reduce 60% of ineffective outpatient follow-up; reduce family burden, save medical expenses for patients; improve rehabilitation effects. DETAILED DESCRIPTION

[0030] The present invention provides a method and system for analyzing behavioral problems in children and adolescents with autism spectrum disorders based on 24-hour activity monitoring. The main structure includes multi-source data collection, behavioral problem modeling, a wearable monitoring terminal, and a cloud-based processing engine deploying an XGBoost risk prediction submodule.

[0031] Detailed instructions for use:

[0032] Wear the tear-resistant magnetic wristband on the non-dominant wrist of the ASD child, ensure that the light sensor is exposed to ambient light, press and hold the terminal button for 3 seconds to start, and Bluetooth will automatically connect to the guardian's mobile phone APP.

[0033] When an emotional outburst / aggressive behavior occurs during use, click the "Event Record" button; complete the ABC scale every Sunday; for example, when a child engages in self-harm behavior in the afternoon, the guardian immediately clicks "Aggressive Event" in the APP and adds the description "patting the head for 2 minutes."

[0034] The guardian uses the APP to receive instructions → click "Start Execution" → the system starts a countdown reminder and evaluates the degree of behavior improvement in the APP.

Claims

1. A method for analyzing behavioral problems in children and adolescents with autism spectrum disorders based on 24-hour activity monitoring, characterized by: The following steps are involved: Multi-source data collection: Wearable sensors continuously collect 24-hour activity time series data for at least 7 days, including motion acceleration, light intensity, and heart rate variability. Synchronously obtain log data entered by guardians and quantitative scores of clinical behavior scales; Activity feature extraction: The time series data is segmented and the following core features are extracted: Sleep quality indicators: sleep fragmentation index, deep sleep ratio; Exercise intensity indicators: proportion of medium- and high-intensity exercise time, average duration of sedentary behavior; Rhythm stability indicators: variance of daytime light exposure fluctuations, activity-rest transition frequency; Modeling behavioral problems: Convert behavioral scale scores into numerical behavioral labels, including aggression intensity and frequency of emotional outbursts; Establish a multivariate linear mixed model with the activity characteristics as fixed effects and individual physiological parameters as random effects, and output the correlation coefficient matrix between the activity characteristics and the behavior labels; Personalized intervention generation: Generates structured action adjustment instructions when the model detects that pre-set risk conditions are met, including: Sleep fragmentation index>0.5 and MVPA proportion<20%; Sedentary behavior lasts for more than 3 hours at a time and the light exposure variance is less than 100 lux 2 .

2. The method for analyzing behavioral problems in children and adolescents with autism spectrum disorders based on 24-hour activity monitoring according to claim 1, characterized in that: The Granger causality test is used to verify the predictive significance of activity features for behavior labels, and only features that pass the test are retained and entered into the LMM model.

3. The method for analyzing behavioral problems in children and adolescents with autism spectrum disorders based on 24-hour activity monitoring according to claim 1, characterized in that: The structured activity adjustment instructions include time constraints: The time window associated with the "Increase daytime outdoor exercise" command is 10:00-15:00; The time window associated with the "Establish a Sleep Rituals" command is 60 minutes before bedtime.

4. A system for analyzing behavioral problems in children and adolescents with autism spectrum disorders based on 24-hour activity monitoring according to any one of claims 1 to 4, characterized in that: Wearable monitoring terminal: Integrates a three-axis accelerometer, a light sensor, and a heart rate monitoring module to upload encrypted activity data in real time; Cloud processing engine: activity feature extraction module, behavior association modeling module User interaction terminal: Guardian log input interface, supporting emotional event marking; Visual report generation unit, outputting 24-hour activity-behavior heat map; The instruction push unit triggers adjustment instructions based on risk conditions.

5. The behavioral problem analysis system for children and adolescents with autism spectrum disorders based on 24-hour activity monitoring according to claim 5 is characterized by: The wearable monitoring terminal adopts a tear-resistant wristband structure, and the sensor module and the wristband are detachably connected by magnetic attraction.

6. The behavioral problem analysis system for children and adolescents with autism spectrum disorders based on 24-hour activity monitoring according to claim 5, characterized in that: The cloud-based processing engine deploys an XGBoost risk prediction submodule, which inputs activity features to predict the risk level of behavioral problems within the next 48 hours with an accuracy rate of ≥85%.