A community pension service management system and method based on risk analysis

Abstract

The application discloses a community pension service management system and method based on risk analysis, and belongs to the technical field of pension service management. The application uses a YOLO algorithm to locate risk positions in a community, forms a community risk position set, constructs a community three-dimensional map, forms a risk area in the community three-dimensional map from the community risk position set, and labels risk information; exploratory data analysis is performed, a statistical model is constructed, and a risk area expansion factor is obtained; based on the risk area expansion factor and current weather data, the risk area of the community three-dimensional map and the corresponding risk information are adjusted; health data of a user is acquired, a scoring standard is set, and a health score is calculated; combined with individual conditions of the user, whether the user is currently suitable for going downstairs for exercise is evaluated based on the total health score, and a health evaluation result is obtained; when the health evaluation result of the user is suitable for exercise or cautious exercise, a planned exercise route of the user is adjusted, and risk information is prompted.

Description

Technical Field

[0001] This invention relates to the field of elderly care service management technology, specifically a community elderly care service management system and method based on risk analysis. Background Technology

[0002] Community-based elderly care has become an important model of elder care. In the community environment, the safety and health management of seniors' daily activities are key concerns for community-based elderly care services. Seniors face various potential risks from the community environment when engaging in activities such as exercise. At the same time, seniors have varying health conditions and different levels of exercise adaptability; therefore, how to rationally plan exercise routes based on seniors' health conditions and ensure their safety during exercise has become an urgent problem to be solved.

[0003] Current technologies often treat community risk assessment and exercise planning for the elderly as two separate components, failing to adequately consider the elderly's health status in conjunction with the actual risks in the community. This can lead to the elderly facing risks beyond their ability to cope during exercise. The community environment is dynamic, subject to changes such as construction site shifts, weather-related flooding, or fallen trees. However, current technologies lag in acquiring and updating this risk information, failing to provide the elderly with timely updates. Furthermore, there is a lack of differentiation based on the elderly's health assessment results; regardless of whether the elderly are in good or poor health, the same exercise route planning and risk warning methods are used, failing to meet the actual needs of different elderly individuals. Summary of the Invention

[0004] The purpose of this invention is to provide a community-based elderly care service management system and method based on risk analysis to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] In a first aspect, the present invention provides a community-based elderly care service management method based on risk analysis, comprising the following steps:

[0007] Using the YOLO algorithm, the locations of obstacles, building risks, fallen trees, and underground space risks within the community are identified to form a set of community risk locations. A 3D map of the community is then constructed, and the set of community risk locations is used to form risk areas within the 3D map, with risk information marked on the map.

[0008] By combining historical weather data with risk events that occurred in the community within the corresponding time period, exploratory data analysis was conducted to construct a statistical model and obtain a risk area expansion factor. Based on the risk area expansion factor and current weather data, the risk areas and corresponding risk information on the community's 3D map were adjusted.

[0009] Acquire users' health data, including heart rate, sleep quality, and exercise volume; set scoring criteria; calculate health scores; and, based on the user's individual circumstances and total health score, assess whether the user is currently suitable to go downstairs for exercise, thus obtaining a health assessment result.

[0010] When a user's health assessment result indicates that they are suitable for exercise or should exercise with caution, the system adjusts the user's planned exercise route based on the risk areas in the community's 3D map and provides the user with risk information alerts.

[0011] In conjunction with the first aspect, in the first implementation of the first aspect of this application, the use of the YOLO algorithm to locate the locations of obstacles, building risks, fallen trees, and underground spaces within the community to form a community risk location set is characterized by including:

[0012] Images captured by cameras along the main roads, public areas, and recreational spaces of the community were collected to obtain an obstacle identification dataset. Building exteriors within the community were photographed using drones along a planned flight path. Roof temperature distribution images were obtained via satellite remote sensing, and humidity data beneath the roofs was collected using ground-based humidity sensors to obtain a building risk identification dataset. Trees within the community were scanned using LiDAR to obtain tree parameters, including height, diameter, and crown morphology. The geographical location of each tree was determined using a geographic information system. Soil texture, groundwater level, and historical weather data were collected to obtain a tree fall risk identification dataset. Underground pipeline detection radar was used to detect pipelines along their routes, acquiring pipeline images. Pressure and flow within the pipelines were monitored in real time using sensors. Water levels, humidity, and drainage pump operation status were monitored at the entrances, exits, and low-lying areas of the underground parking lot to obtain an underground space risk identification dataset.

[0013] For datasets of obstacles to be identified, building risks to be identified, fallen trees risks to be identified, and underground space risks to be identified, data are labeled, YOLO algorithm models are trained, risk locations are located, and risk location sets are generated.

[0014] In conjunction with the first aspect, in the second implementation of the first aspect of this application, the step of labeling data, training a YOLO algorithm model, locating risk locations, and generating a risk location set for the obstacle identification dataset, building risk identification dataset, tree fall risk identification dataset, and underground space risk identification dataset includes:

[0015] The annotation of the obstacle identification dataset is specifically obstacle annotation; the annotation of the building risk identification dataset includes annotation of external wall detachment risk and roof leakage risk; the annotation of the tree fall risk identification dataset is specifically tree fall annotation; and the annotation of the underground space risk identification dataset includes annotation of underground pipe rupture risk and underground parking lot water accumulation risk.

[0016] For the obstacle identification dataset, for each obstacle in the image, draw a bounding box and indicate the obstacle category in the annotation information;

[0017] For the dataset of building risks to be identified, for images of building exterior walls taken by drones, an image segmentation and annotation tool is used to identify potential hazard areas in the images. The location of the hazard is marked by drawing polygons, and the hazard type is noted in the annotation information. The roof temperature distribution image is combined with humidity data. The temperature abnormality area is marked in the roof temperature distribution image, and the area with humidity higher than a set threshold is marked in the corresponding humidity data. For areas with roof leakage risk, the annotation information is associated and the risk level is noted.

[0018] For the dataset of trees to be identified for lodging risk, geographic information system software was used to integrate tree parameters with geographic location. For each tree, its lodging risk under severe weather conditions was assessed based on its tree parameters and the soil texture, groundwater level and historical weather data of its location. The trees were labeled as "high lodging risk", "medium lodging risk" and "low lodging risk", and the relevant assessment basis was recorded in the labeling information.

[0019] For the dataset of risks to be identified in underground space, pipeline images are combined with pressure and flow rates inside the pipelines. The location, material, and diameter of the pipelines are marked on the pipeline images. For areas where pressure and flow rates fluctuate abnormally, the locations of potential rupture hazards are marked on the images, and the type and level of the hazard are noted.

[0020] YOLOv7 was selected, and a corresponding deep learning training environment was set up, including the PyTorch deep learning framework, its dependencies, and tools. The labeled dataset was divided into training, validation, and test sets according to a set ratio. The training set data was input into the YOLO algorithm model for training. The YOLO algorithm model learned the features of various risk locations by adjusting weights and biases. During training, the model's performance metrics on the validation set were monitored, including mean precision and recall. Training was stopped when the model's performance on the validation set no longer improved or reached the preset performance metrics. The trained model was evaluated on the test set and optimized.

[0021] The YOLO algorithm model is deployed on a data processing server. Real-time collected community image data and sensor-uploaded data are processed using the YOLO algorithm model. The risk locations predicted by the YOLO algorithm model are stored in the image coordinate system or sensor coordinate system and converted into geographic coordinates. By establishing the mapping relationship between image coordinates and geographic coordinates, as well as the correspondence between sensor locations and geographic coordinates, the conversion of risk locations is realized. The various community risk locations located by the YOLO algorithm model are integrated to form a community risk location set.

[0022] In conjunction with the first aspect, in the third implementation of the first aspect of this application, the construction of a community 3D map, which sets the community risk locations into risk areas on the community 3D map and marks the risk information, includes:

[0023] Collect terrain data, building data, road and public facility data, and use 3D map building software to construct a 3D map of the community; convert the community risk location set data into Shapefile format, import it into the 3D map building software, and mark each risk location in the community 3D map according to the geographic coordinate information of the risk location; for risk locations with a range, generate the corresponding risk area in the 3D map according to the impact range and characteristics of the risk;

[0024] Label the risk information, including the type of risk, description, risk level, discovery time, and recommended measures.

[0025] In conjunction with the first aspect, in the fourth implementation of the first aspect of this application, the step of combining historical weather data with risk events occurring within the community during the corresponding time period to conduct exploratory data analysis, construct a statistical model, and obtain a risk area expansion factor includes:

[0026] Historical weather data for the community area is collected, including weather conditions and weather variables such as rainfall, wind speed, temperature, humidity, and snowfall, with time granularity refined to the hourly level; historical risk event records are extracted from the community management system, including event type, occurrence time, specific location, scope of impact, and severity; weather data is matched with risk events by timestamp, and the geographical coordinates of risk events are mapped to the grid of the community's 3D map;

[0027] Calculate the Pearson correlation coefficient between weather variables and the frequency of various types of risk events; use Poisson regression to model the relationship between the number of risk events and weather variables; discretize the weather variables, divide them into training and test sets, use the AIC criterion to select the optimal model, and validate the model.

[0028] The coefficients of weather variables are extracted from the Poisson regression model to represent the impact of unit weather change on risk events, and the coefficients are normalized to the range of 0-1. For each risk area j and weather condition k, the expansion factor is calculated as follows:

[0029] Determine the standardized coefficients corresponding to different weather variables. For each weather variable, calculate the standardized coefficient and multiply it by the value of the weather variable to obtain the standardized value of the weather variable. Sum the standardized values ​​of all different weather variables and add 1 to the sum to obtain the risk area expansion factor.

[0030] In conjunction with the first aspect, in the fifth implementation of the first aspect of this application, the step of adjusting the risk areas and corresponding risk information of the community 3D map based on the risk area expansion factor and current weather data includes:

[0031] Based on the risk area expansion factor and current weather data, the risk areas that need to be adjusted are determined; the new risk area range is calculated using the following method:

[0032] When the risk area is a circular area, the original radius is multiplied by an expansion factor with the center of the original risk area as the center to obtain a new radius, thus determining the new circular risk area range. When the risk area is a polygonal area, the coordinates of each vertex of the polygon are scaled and offset accordingly. By multiplying the vertex coordinates with the expansion factor, new vertex coordinates are obtained. The vertices are then reconnected to form a new polygonal risk area. The calculated new risk area range is then updated in the community 3D map.

[0033] Based on the expansion of the risk area, reassess the risk level; supplement the risk information with descriptions related to weather conditions and the expansion of the risk area; and update the risk warning information for users in the 3D map according to the new risk area and risk level.

[0034] In conjunction with the first aspect, in the sixth implementation of the first aspect of this application, the step of obtaining the user's health data, including heart rate, sleep quality, and exercise volume, setting scoring criteria, and calculating a health score includes:

[0035] Equip users with smart wearable devices to monitor heart rate, sleep quality, and exercise volume in real time; set normal ranges for heart rate based on medical standards and the physical characteristics of users of different ages, and score the degree to which the heart rate deviates from the normal range; set sleep quality indicators based on sleep duration and the proportion of deep sleep stages; set exercise volume indicators based on the user's physical condition and exercise capacity, and score the actual exercise volume based on the comparison between the actual exercise volume and the target exercise volume.

[0036] Based on the importance of heart rate, sleep quality, and exercise volume to a user's health, corresponding weights are assigned; the scores for heart rate, sleep quality, and exercise volume are multiplied by their respective weights, and the results are added together to obtain the user's health score.

[0037] In conjunction with the first aspect, in the seventh implementation of the first aspect of this application, the step of assessing whether a user is currently suitable to go downstairs for exercise based on their total health score, taking into account the user's individual circumstances, to obtain a health assessment result includes:

[0038] The user's individual circumstances include health history, disease status, and exercise habits; based on the user's individual circumstances, the impact on the user's exercise is analyzed; by combining the user's individual circumstances and health score, a health assessment result is obtained, including assessment conclusions, explanations of the basis, exercise recommendations, and risk warnings.

[0039] In conjunction with the first aspect, in the eighth implementation of the first aspect of this application, when the user's health assessment result indicates suitable for exercise or cautious exercise, adjusting the user's planned exercise route based on the risk areas of the community's three-dimensional map and providing the user with risk information prompts includes:

[0040] When a user's health assessment result indicates that they are suitable for exercise or should exercise with caution, they are allowed to exercise; otherwise, they are advised to suspend exercise. When the result indicates that they should exercise with caution, restrictions are imposed.

[0041] The user's planned route is located on the community's 3D map, and key coordinate points on the route are extracted. The extracted route coordinate points are compared with risk areas in the 3D map to determine whether the route passes through risk areas. Using the spatial analysis function of the geographic information system, the distance between the route and the risk area is calculated. When the distance is less than the set safety threshold, it is determined that there is a risk. For road sections that are determined to be at risk, the corresponding risk information is associated.

[0042] For users whose health assessment results indicate they are suitable for exercise, the route will be adjusted based on the risk level and actual situation when there is a risk in the exercise route. Specifically, when passing through low-risk areas, the user will be prompted with risk information while keeping the original route unchanged. When passing through medium- or high-risk areas, the route will be replanned, and the map's path planning algorithm will be used to find an alternative route that avoids the risk area and is closest to the original route.

[0043] For users whose health assessment results indicate that they should exercise with caution, when there are risks in the exercise route, they should prioritize avoiding the risk areas; when replanning the route, the constraints should be taken into account, and an alternative route that avoids the risk areas and is closest to the original route should be found.

[0044] Secondly, a community-based elderly care service management system based on risk analysis includes:

[0045] The 3D map construction module includes a community risk location unit and a 3D map construction unit. The community risk location unit uses the YOLO algorithm to locate the locations of obstacles, building risks, fallen trees, and underground spaces within the community, forming a set of community risk locations. The 3D map construction unit constructs a 3D map of the community, forming risk areas within the community risk location set and marking the risk information.

[0046] The 3D map adjustment module includes a risk area expansion factor calculation unit and a 3D map adjustment unit. The risk area expansion factor calculation unit combines historical weather data with risk events that occurred in the community within the corresponding time period to conduct exploratory data analysis, construct a statistical model, and obtain the risk area expansion factor. The 3D map adjustment unit adjusts the risk areas and corresponding risk information on the community's 3D map based on the risk area expansion factor and current weather data.

[0047] The health outcome assessment module consists of a health data acquisition unit, a health score calculation unit, and a health outcome assessment unit. The health data acquisition unit acquires the user's health data, including heart rate, sleep quality, and exercise volume. The health score calculation unit sets scoring criteria and calculates the health score. The health outcome assessment unit, based on the user's individual circumstances and total health score, assesses whether the user is currently suitable to go downstairs for exercise, thus obtaining the health assessment result.

[0048] Exercise route adjustment and prompt module: exercise route adjustment unit and risk information prompt unit; wherein, when the user's health assessment result is suitable for exercise or exercise with caution, the exercise route adjustment unit adjusts the user's planned exercise route based on the risk area of ​​the community 3D map, and the risk information prompt unit prompts the user with risk information.

[0049] Compared with the prior art, the beneficial effects of the present invention are:

[0050] 1. This invention obtains risk information from a three-dimensional map of the community in real time and combines it with the health assessment results of the elderly to make precise adjustments to the exercise route; it can effectively prevent the elderly from entering high-risk areas, reduce the probability of accidental injury, and provide strong protection for the elderly's exercise safety.

[0051] 2. This invention connects in real time with the community's 3D map database to ensure timely updates of risk information; it can also replan exercise routes for the elderly based on dynamic changes in the community environment, providing the latest risk alerts so that the elderly can always obtain accurate safety information.

[0052] 3. This invention develops different route adjustment strategies and risk warnings for elderly people with different health conditions, such as those who are "suitable for exercise" and those who are "cautious about exercise". It fully considers the individual differences of the elderly people's physical condition, exercise restrictions and other factors, and provides services that are more in line with the actual needs of the elderly. Attached Figure Description

[0053] Figure 1 This is a schematic diagram illustrating the steps of a community-based elderly care service management method based on risk analysis according to the present invention.

[0054] Figure 2 This is a system structure diagram of a community elderly care service management system based on risk analysis according to the present invention. Detailed Implementation

[0055] 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.

[0056] Example: Figures 1-2 As shown, the present invention provides a technical solution.

[0057] like Figure 1 The schematic diagram illustrates the steps of a risk analysis-based community-based elderly care service management method. This invention provides a risk analysis-based community-based elderly care service management method, comprising the following steps:

[0058] Step S100: Using the YOLO algorithm, locate the locations of obstacles, building risks, fallen trees, and underground spaces within the community to form a set of community risk locations; construct a 3D map of the community, and mark the risk information by placing the set of community risk locations in the 3D map.

[0059] Specifically, images captured by cameras along the community's main roads, public areas, and recreational spaces were collected to obtain an obstacle identification dataset; drones were used to photograph the exterior walls of buildings within the community along a planned flight path, and satellite remote sensing was used to obtain roof temperature distribution images, while ground-based humidity sensors collected humidity data below the roofs to obtain a building risk identification dataset; LiDAR was used to scan trees within the community to obtain tree parameters, including height, diameter, and crown morphology, and a geographic information system was used to determine the geographical location of each tree, while soil texture, groundwater level, and historical weather data were collected to obtain a tree fall risk identification dataset; underground pipeline detection radar was used to detect along the pipeline route, obtain pipeline images, and sensors were used to monitor the pressure and flow rate within the pipeline in real time, while water levels, humidity, and drainage pump operation status were monitored at the entrances, exits, and low-lying areas of the underground parking lot to obtain an underground space risk identification dataset;

[0060] For datasets of obstacles to be identified, building risks to be identified, fallen trees risks to be identified, and underground space risks to be identified, data are labeled, YOLO algorithm models are trained, risk locations are located, and risk location sets are generated.

[0061] Furthermore, the annotation of the obstacle identification dataset specifically includes obstacle annotation, the annotation of the building risk identification dataset includes annotation of external wall detachment risk and roof leakage risk, the annotation of the tree fall risk identification dataset specifically includes tree fall annotation, and the annotation of the underground space risk identification dataset includes annotation of underground pipe rupture risk and underground parking lot water accumulation risk.

[0062] YOLOv7 was selected, and a corresponding deep learning training environment was set up, including the PyTorch deep learning framework, its dependencies, and tools. The labeled dataset was divided into training, validation, and test sets according to a set ratio. The training set data was input into the YOLO algorithm model for training. The YOLO algorithm model learned the features of various risk locations by adjusting weights and biases. During training, the model's performance metrics on the validation set were monitored, including mean precision and recall. Training was stopped when the model's performance on the validation set no longer improved or reached the preset performance metrics. The trained model was evaluated on the test set and optimized.

[0063] The YOLO algorithm model is deployed on a data processing server. Real-time collected community image data and sensor-uploaded data are processed using the YOLO algorithm model. The risk locations predicted by the YOLO algorithm model are stored in the image coordinate system or sensor coordinate system and converted into geographic coordinates. By establishing the mapping relationship between image coordinates and geographic coordinates, as well as the correspondence between sensor locations and geographic coordinates, the conversion of risk locations is realized. The various community risk locations located by the YOLO algorithm model are integrated to form a community risk location set.

[0064] Furthermore, terrain data, building data, road and public facility data are collected, and a 3D map of the community is constructed using 3D map building software; the community risk location set data is converted into Shapefile format and imported into the 3D map building software, and each risk location is marked in the community 3D map according to the geographic coordinate information of the risk location; for risk locations with a range, the corresponding risk area is generated in the 3D map according to the impact range and characteristics of the risk.

[0065] Label the risk information, including the type of risk, description, risk level, discovery time, and recommended measures.

[0066] In one specific embodiment, 20 high-definition cameras were installed within the community, continuously collecting image data for a week, resulting in 5,000 valid images, forming an obstacle identification dataset. Drones were used to photograph building exteriors, flying 20 times and acquiring 300 images; simultaneously, 20 images of roof temperature distribution were acquired via satellite remote sensing, combined with ground humidity sensor data, to form a building risk identification dataset. LiDAR was used to scan 300 trees within the community, acquiring tree parameters; combined with a geographic information system to determine tree locations, and collecting soil texture, groundwater level, and historical weather data from the past 5 years, a tree fall risk identification dataset was obtained. Underground pipeline detection radar equipment was used to detect pipelines, acquiring 100 pipeline images; sensors were used to monitor pipeline pressure and flow in real time, and water level, humidity, and drainage pump operation status were monitored at key locations in the underground parking lot, resulting in an underground space risk identification dataset.

[0067] Five professional annotators were organized to annotate images in the obstacle identification dataset, identifying a total of 2000 obstacles (such as vehicles, debris, construction equipment, etc.). For the building risk identification dataset, exterior wall images were annotated, identifying 50 locations with potential for exterior wall collapse. Combining roof temperature distribution images and humidity data, 30 locations with potential for roof leakage were identified. Based on tree parameters and environmental data, 80 locations with potential for tree fall were identified. Underground pipe images and sensor data were annotated, identifying 20 locations with potential for underground pipe rupture. Based on underground parking lot monitoring data, 40 locations with potential for water accumulation were identified. The YOLOv7 algorithm was selected, and a deep learning training environment was built on a server equipped with an NVIDIA RTX 5090 GPU, installing PyTorch 1.10.0 and related dependencies and tools. The annotated datasets were divided into training, validation, and test sets at a ratio of 70%, 15%, and 15%, respectively.

[0068] The training set data was input into the YOLOv7 algorithm model for training, with an initial learning rate of 0.001, 500 iterations, and a batch size of 16. During training, the model continuously adjusted weights and biases to learn the features of various risk locations. The model's performance metrics on the validation set, including mean precision (mAP) and recall, were monitored in real time during training. At the 300th iteration, mAP reached 0.85 and recall reached 0.80. Training was stopped when the model's performance on the validation set no longer improved (no significant change after 50 iterations). The trained model was evaluated on the test set, revealing poor performance in identifying small obstacles and hidden pipe rupture hazards. Optimization was achieved by increasing the number of small-sized samples and adjusting the model's anchor box parameters. After optimization, the model's mAP on the test set improved to 0.90, and recall improved to 0.85.

[0069] The optimized YOLOv7 algorithm model was deployed to a data processing server to process real-time community image data and sensor-uploaded data. During testing, 1000 real-time image data and 500 sets of sensor data were processed. The risk locations predicted by the model were initially stored in the image coordinate system or sensor coordinate system. Through the established mapping and correspondence relationships, over 95% of the risk locations were successfully and accurately converted to geographic coordinates. For example, in one real-time monitoring session, the model accurately identified the location of a potential building exterior wall collapse hazard and converted it from image coordinates to geographic coordinates, with a positioning error within 1 meter.

[0070] Collect terrain data, building data, road and public facility data of the community, and use ArcGIS software to build a 3D map of the community.

[0071] The community risk location data was converted to Shapefile format and imported into ArcGIS. Based on the geographic coordinates of the risk locations, each risk location was marked on the community's 3D map, and corresponding risk areas were generated. A total of 500 risk locations of various types were marked on the 3D map, generating 300 risk areas. Each risk location and area was labeled in detail, including the risk type, description, risk level, discovery time, and recommended measures. For example, for a tree fall risk area, the labeling information is: "Risk type: Tree fall; Description: Trees located next to the community's central square are at risk of falling due to aging root systems and recent strong winds; Risk level: High; Discovery time: October 15, 2024; Recommended measures: Immediately arrange for personnel to reinforce or cut down the trees."

[0072] Step S200: Combine historical weather data with risk events that occurred in the community within the corresponding time period to conduct exploratory data analysis, construct a statistical model, and obtain the risk area expansion factor; based on the risk area expansion factor and current weather data, adjust the risk areas and corresponding risk information on the community's 3D map;

[0073] Specifically, historical weather data of the community area is collected, including weather conditions and weather variables, such as rainfall, wind speed, temperature, humidity, and snowfall, with time granularity refined to the hourly level; historical risk event records are extracted from the community management system, including event type, occurrence time, specific location, scope of impact, and severity; weather data is matched with risk events by timestamp, and the geographical coordinates of risk events are mapped to the grid of the community's 3D map;

[0074] Calculate the Pearson correlation coefficient between weather variables and the frequency of various types of risk events; use Poisson regression to model the relationship between the number of risk events and weather variables; discretize the weather variables, divide them into training and test sets, use the AIC criterion to select the optimal model, and validate the model.

[0075] The coefficients of weather variables are extracted from the Poisson regression model to represent the impact of unit weather change on risk events, and the coefficients are normalized to the range of 0-1. For each risk area j and weather condition k, the expansion factor is calculated as follows:

[0076] Determine the standardized coefficients corresponding to different weather variables. For each weather variable, calculate the standardized coefficient and multiply it by the value of the weather variable to obtain the standardized value of the weather variable. Sum the standardized values ​​of all different weather variables and add 1 to the sum to obtain the risk area expansion factor.

[0077] Furthermore, based on the risk area expansion factor and current weather data, the risk areas that need to be adjusted are determined; the new risk area range is calculated using the following method:

[0078] When the risk area is a circular area, the original radius is multiplied by an expansion factor with the center of the original risk area as the center to obtain a new radius, thus determining the new circular risk area range. When the risk area is a polygonal area, the coordinates of each vertex of the polygon are scaled and offset accordingly. By multiplying the vertex coordinates with the expansion factor, new vertex coordinates are obtained. The vertices are then reconnected to form a new polygonal risk area. The calculated new risk area range is then updated in the community 3D map.

[0079] Based on the expansion of the risk area, reassess the risk level; supplement the risk information with descriptions related to weather conditions and the expansion of the risk area; and update the risk warning information for users in the 3D map according to the new risk area and risk level.

[0080] In one specific embodiment, historical weather data for the past five years was collected from the local meteorological department for the community area. Weather conditions included sunny, cloudy, overcast, light rain, heavy rain, torrential rain, strong winds, high temperatures, and low temperatures; weather variables included rainfall, wind speed, temperature, humidity, and snowfall. The time granularity was refined to the hourly level, resulting in approximately 43,800 data records. Historical risk event records for the past five years were extracted from the community management system, yielding 800 records. Event types included building facade collapses, fallen trees, underground pipe ruptures, flooding in underground parking lots, and accidents caused by road obstructions; record content included the time of occurrence, specific location (accurate to latitude and longitude), affected area (in square meters or meters), and severity (divided into minor, moderate, and severe levels). Weather data was matched with risk events by timestamp, resulting in 750 successful matches. Simultaneously, the geographical coordinates of the risk events were mapped onto a grid of the community's 3D map, with each grid measuring 10 meters × 10 meters, facilitating subsequent analysis.

[0081] The Pearson correlation coefficients between weather variables and the frequency of various types of risk events were calculated. For example, the calculations showed that the Pearson correlation coefficient between rainfall and the frequency of underground parking lot flooding events was 0.75, indicating a strong positive correlation between the two; the Pearson correlation coefficient between wind speed and the frequency of tree fall events was 0.68, also showing a strong positive correlation.

[0082] Poisson regression was used to model the relationship between the frequency of risk events and weather variables. Weather variables were discretized, such as dividing rainfall into intervals of 0-10 mm, 10-50 mm, 50-100 mm, and over 100 mm. The dataset was divided into training and test sets at a ratio of 70% and 30%, respectively, and the optimal model was selected using the AIC criterion. After multiple trials and adjustments, the final optimal model showed that variables such as rainfall, wind speed, and temperature had a significant impact on the frequency of risk events. The model was validated on the test set, and the mean absolute error between the model's predicted and actual frequency of risk events was 5.6%, with a relative error within 15%, indicating that the model has good predictive ability.

[0083] The coefficients of weather variables are extracted from the Poisson regression model to represent the impact of a unit of weather change on risk events. For example, the coefficient for rainfall is 0.05, meaning that for every 1 mm increase in rainfall, the average number of risk events increases by 0.05. The coefficients are normalized to the range of 0-1; for example, the normalized coefficient for rainfall is 0.8.

[0084] Taking a flood-prone area in an underground parking lot as an example, under current weather conditions, the rainfall is 50 mm, the wind speed is 10 m / s, the temperature is 25℃, and the humidity is 70%. According to standardized coefficients, the standardized coefficient for rainfall is 0.8, for wind speed it is 0.6, for temperature it is 0.3, and for humidity it is 0.2. The standardized values ​​for each weather variable are calculated as follows: Rainfall standardized value = 0.8 × 50 = 40, Wind speed standardized value = 0.6 × 10 = 6, Temperature standardized value = 0.3 × 25 = 7.5, Humidity standardized value = 0.2 × 70 = 14. Adding all the standardized values ​​of the weather variables together and adding 1, we obtain the risk area expansion factor = 40 + 6 + 7.5 + 14 + 1 = 68.5.

[0085] The water accumulation risk area in the underground parking lot was originally a circular area with a radius of 10 meters. Based on the calculated expansion factor of 68.5, the new radius = 10 × 68.5 = 685 meters, thus determining the new circular risk area.

[0086] Taking a polygonal risk area formed by the collapse of a building's exterior wall as an example, the original vertex coordinates of this polygon are [(100,100),(150,100),(150,150),(100,150)]. Scaling and offset calculations are performed on each vertex coordinate. For example, for vertex (100,100), the new coordinates are (100×68.5,100×68.5)=(6850,6850). These new vertices are then reconnected to form a new polygonal risk area.

[0087] The calculated new risk area ranges were updated in the community 3D map, with a total of 50 risk areas updated, including 20 circular risk areas and 30 polygonal risk areas.

[0088] Based on the expansion of risk areas, risk levels were reassessed. For example, an underground parking lot flooding risk area, originally classified as medium risk, was reassessed as high risk due to the expansion of the risk area. In this experiment, the risk levels of 30 risk areas were reassessed, with 10 risk areas upgraded and 5 downgraded. Descriptions related to weather conditions and risk area expansion were added to the risk information. For example, for a risk area where trees were uprooted due to strong winds, the description was added: "Current wind speed is 15 m / s, the risk area of ​​uprooted trees has expanded, please take precautions." A total of 80 risk descriptions were added. Based on the new risk areas and risk levels, the risk warning information for users on the 3D map was updated. Updated risk warnings were pushed to users through the community management platform, with a total of 500 warning messages pushed, covering 80% of the residents in the community.

[0089] Step S300: Obtain the user's health data, including heart rate, sleep quality, and exercise volume; set scoring criteria and calculate a health score; based on the user's individual situation and the total health score, assess whether the user is currently suitable to go downstairs for exercise and obtain a health assessment result.

[0090] Specifically, users are equipped with smart wearable devices to monitor heart rate, sleep quality, and exercise volume in real time; a normal range for heart rate is set according to medical standards and the physical characteristics of users of different ages, and a score is given based on the degree to which the heart rate deviates from the normal range; sleep quality indicators are set, and a score is given based on sleep duration and the proportion of deep sleep stages; and exercise volume indicators are set according to the user's physical condition and exercise capacity, and a score is given based on the comparison between the actual exercise volume and the target exercise volume.

[0091] Based on the importance of heart rate, sleep quality, and exercise volume to a user's health, corresponding weights are assigned; the scores for heart rate, sleep quality, and exercise volume are multiplied by their respective weights, and the results are added together to obtain the user's health score.

[0092] Furthermore, the user's individual circumstances include health history, disease status, and exercise habits; based on the user's individual circumstances, the impact on the user's exercise is analyzed; by combining the user's individual circumstances and health score, a health assessment result is obtained, including assessment conclusions, explanations of the basis, exercise recommendations, and risk warnings.

[0093] In one specific embodiment, 50 users of different ages within the community were selected to participate in the experiment. Each user was equipped with a smart bracelet as a wearable device for real-time monitoring of heart rate, sleep quality, and activity levels. The smart bracelet is equipped with a high-precision sensor, capable of accurately recording relevant data and transmitting it in real-time to the community health management platform via Bluetooth. The data collection period was set at one month, during which users were required to wear the smart bracelet regularly to ensure data continuity and integrity. Simultaneously, individual information such as users' health history, disease conditions, and exercise habits was collected through community health records and face-to-face interviews.

[0094] According to medical standards, normal heart rate ranges are set for different age groups. For example, for users aged 60-70, the normal resting heart rate range is set at 55-90 beats per minute; for users aged 70-80, the normal range is set at 50-95 beats per minute. The scoring rules are as follows: 100 points are awarded for a heart rate within the normal range; for every 5 beats per minute above the upper limit of the normal range, 10 points are deducted; for every 5 beats per minute below the lower limit of the normal range, 10 points are deducted.

[0095] Sleep indicators are set, with a recommended sleep duration of 7-9 hours for the elderly, and deep sleep accounting for about 20%-30% of the total sleep time. Sleep duration within the reasonable range earns a base score of 60 points, with 10 points deducted for each hour of deviation; the proportion of deep sleep within the normal range earns 40 points, with 5 points deducted for each 5% deviation from the normal range.

[0096] The exercise volume targets are set based on the user's physical condition and exercise capacity. For users in good physical condition, the daily step count target is set at 6,000-8,000 steps, and the calorie expenditure target is set at 300-500 kcal; for users in average physical condition, the corresponding targets are appropriately lowered. Reaching the target exercise volume earns 100 points, and for every 10% below the target, 10 points are deducted.

[0097] Based on the relative importance of heart rate, sleep quality, and exercise volume to a user's health, heart rate is weighted at 40%, sleep quality at 30%, and exercise volume at 30%. For example, a 65-year-old user has an average heart rate of 80 beats per minute, within the normal range, scoring 100 points; sleep duration of 8 hours with a deep sleep phase of 25%, resulting in a sleep quality score of 60 + 40 = 100 points; and daily steps of 7000 steps, resulting in an exercise volume score of 100 points. Therefore, the user's health score is: 100 × 40% + 100 × 30% + 100 × 30% = 40 + 30 + 30 = 100 points.

[0098] Among the 50 users, 10 had a history of hypertension, 5 had a history of diabetes, and 20 consistently exercised. For example, for a user with a history of hypertension and recently unstable blood pressure control, analysis showed that their blood pressure might fluctuate during exercise, posing a significant risk; for users who consistently exercised, their exercise capacity and endurance were considered to be relatively good.

[0099] The following is an example of a health assessment result:

[0100] Assessment conclusion: For the above-mentioned users with a history of hypertension and unstable blood pressure control, the assessment conclusion is that they are not suitable to go downstairs for exercise.

[0101] Explanation: The basis is the user's history of hypertension and the current unstable blood pressure control, as well as the health score (the user's health score is 80 points, and the exercise assessment is affected by blood pressure issues).

[0102] Exercise recommendations: After blood pressure stabilizes, users are advised to gradually increase their exercise volume under the guidance of a doctor, starting with simple indoor activities.

[0103] Risk warning: Exercise may cause a sharp rise in blood pressure, triggering cardiovascular and cerebrovascular diseases.

[0104] Step S400: When the user's health assessment result indicates that the user is suitable for exercise or should exercise with caution, the user's planned exercise route is adjusted based on the risk areas of the community 3D map, and the user is given risk information prompts.

[0105] Specifically, when a user's health assessment result indicates that they are suitable for exercise or should exercise with caution, they are allowed to exercise; otherwise, they are advised to suspend exercise. When the result indicates that they should exercise with caution, restrictions are imposed.

[0106] The user's planned route is located on the community's 3D map, and key coordinate points on the route are extracted. The extracted route coordinate points are compared with risk areas in the 3D map to determine whether the route passes through risk areas. Using the spatial analysis function of the geographic information system, the distance between the route and the risk area is calculated. When the distance is less than the set safety threshold, it is determined that there is a risk. For road sections that are determined to be at risk, the corresponding risk information is associated.

[0107] For users whose health assessment results indicate they are suitable for exercise, the route will be adjusted based on the risk level and actual situation when there is a risk in the exercise route. Specifically, when passing through low-risk areas, the user will be prompted with risk information while keeping the original route unchanged. When passing through medium- or high-risk areas, the route will be replanned, and the map's path planning algorithm will be used to find an alternative route that avoids the risk area and is closest to the original route.

[0108] For users whose health assessment results indicate that they should exercise with caution, when there are risks in the exercise route, they should prioritize avoiding the risk areas; when replanning the route, the constraints should be taken into account, and an alternative route that avoids the risk areas and is closest to the original route should be found.

[0109] In one specific embodiment, from the 50 users who participated in the previous health assessment experiment, 30 users whose health assessment results indicated they were suitable for exercise and 10 users who indicated they should exercise cautiously were selected as the subjects of this experiment. A previously constructed 3D community map was used, with various risk areas marked on the map, such as construction areas, waterlogged sections, areas at risk of fallen trees, and areas with potential for building facade collapse. Risk areas were categorized into low-risk, medium-risk, and high-risk based on their risk level.

[0110] The community health management platform's user-end application collected information on 50 experimental users, including their planned exercise routes, exercise times, and exercise methods (such as walking, jogging, cycling, etc.). For example, User A planned to jog for 30 minutes within the community, starting from their own doorstep, passing through the community's central square, circling the community garden, and returning.

[0111] The health assessment module retrieves the user's health assessment results to determine whether the user is "suitable for exercise" or "exercises with caution." For users who exercise with caution, the restrictions are also retrieved. For example, if user B is judged to exercise with caution, the restrictions are that the exercise intensity should not be too high and that someone should accompany them.

[0112] The user's planned route is located on a community 3D map, and key coordinate points are extracted. For complex routes, a coordinate point is extracted every 50 meters. For example, 20 key coordinate points were extracted for user A's route. These extracted coordinate points are compared with risk areas on the 3D map, and the spatial analysis function of a Geographic Information System (GIS) is used to calculate the distance between the route and the risk areas. A safety threshold of 10 meters is set; when the distance is less than 10 meters, a risk is identified. In the analysis of user A's route, it was found that 5 coordinate points were less than 10 meters from a construction area (medium risk), thus identifying this section as risky. For sections identified as risky, relevant risk information is associated. For example, risky sections in user A's route are associated with risk information about the construction area, including the construction start time, estimated end time, construction content, and risk level.

[0113] Among 30 users suitable for exercise, 5 users' routes passed through low-risk areas, such as sections of road with slight water accumulation. While keeping the original routes largely unchanged, these users were given risk warnings, such as "Slight water accumulation ahead, please proceed with caution." Ten users suitable for exercise passed through medium- to high-risk areas, such as construction zones or areas with fallen trees. Using map-based path planning algorithms, their routes were replanned. For example, user A's route passed through a construction zone; the replanned route bypassed the construction zone, taking an adjacent road, and the new route was 200 meters longer than the original.

[0114] Of the 10 users who exercised cautiously, 8 had routes that posed a risk. For these users, priority was given to avoiding the risk areas. For example, if User B's route passed through an area with a high risk of falling building facades, when replanning the route, the restrictions (the exercise intensity should not be too high, and someone should accompany them) were taken into account, and an alternative route that avoided the risk area and was closest to the original route was found. The new route avoided the risk area and also chose a relatively flat road to meet User B's exercise intensity restrictions.

[0115] For users who exercise cautiously, in addition to providing route adjustments and risk warnings, the frequency and detail of the warnings are increased. For example, user B will receive a risk warning every 5 minutes during exercise, which includes the location and type of risk area and precautions.

[0116] like Figure 2 The system structure diagram of a community-based elderly care service management system based on risk analysis is shown in this invention. The community-based elderly care service management system based on risk analysis includes:

[0117] The 3D map construction module includes a community risk location unit and a 3D map construction unit. The community risk location unit uses the YOLO algorithm to locate the locations of obstacles, building risks, fallen trees, and underground spaces within the community, forming a set of community risk locations. The 3D map construction unit constructs a 3D map of the community, forming risk areas within the community risk location set and marking the risk information.

[0118] The 3D map adjustment module includes a risk area expansion factor calculation unit and a 3D map adjustment unit. The risk area expansion factor calculation unit combines historical weather data with risk events that occurred in the community within the corresponding time period to conduct exploratory data analysis, construct a statistical model, and obtain the risk area expansion factor. The 3D map adjustment unit adjusts the risk areas and corresponding risk information on the community's 3D map based on the risk area expansion factor and current weather data.

[0119] The health outcome assessment module consists of a health data acquisition unit, a health score calculation unit, and a health outcome assessment unit. The health data acquisition unit acquires the user's health data, including heart rate, sleep quality, and exercise volume. The health score calculation unit sets scoring criteria and calculates the health score. The health outcome assessment unit, based on the user's individual circumstances and total health score, assesses whether the user is currently suitable to go downstairs for exercise, thus obtaining the health assessment result.

[0120] Exercise route adjustment and prompt module: exercise route adjustment unit and risk information prompt unit; wherein, when the user's health assessment result is suitable for exercise or exercise with caution, the exercise route adjustment unit adjusts the user's planned exercise route based on the risk area of ​​the community 3D map, and the risk information prompt unit prompts the user with risk information.

[0121] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A community elder care service management method based on risk analysis, characterized in that, The method comprises the following steps: Using the YOLO algorithm, the positions of obstacles, building risks, tree collapse risks and underground space risks in the community are located to form a community risk position set; A three-dimensional map of the community is constructed, the community risk position set is formed into a risk area in the three-dimensional map of the community, and risk information is labeled; Historical weather data is combined with risk events occurring in the community in the corresponding time period to perform exploratory data analysis, a statistical model is constructed, and a risk area expansion factor is obtained; Based on the risk area expansion factor and the current weather data, the risk area and the corresponding risk information of the three-dimensional map of the community are adjusted; The method of combining historical weather data with risk events occurring in the community in the corresponding time period to perform exploratory data analysis, constructing a statistical model, and obtaining a risk area expansion factor comprises: Collect historical weather data of the area where the community is located, including weather conditions and weather variables, the weather variables including rainfall, wind speed, temperature, humidity and snowfall, and the time granularity is refined to the hour level; extract historical risk event records from the community management system, including event type, occurrence time, specific location, influence range and severity; match weather data and risk events by timestamp, and map the geographic coordinates of risk events to the grid of the three-dimensional map of the community; Calculate the Pearson correlation coefficient of weather variables and the occurrence frequency of each type of risk event; use Poisson regression to model the relationship between the number of risk events and weather variables; discretize the weather variables, divide the training set and the test set, and use the AIC criterion to select the optimal model for model validation; Extract the coefficients of weather variables from the Poisson regression model, which represent the influence of unit weather change on risk events, and normalize the coefficients to the range of 0-1; for each risk area j and weather condition k, calculate the expansion factor as follows: Determine the standardized coefficients corresponding to different weather variables, for each weather variable, calculate the standardized coefficient multiplied by the value of the weather variable to obtain the standardized value of the weather variable; add all the standardized values of different weather variables corresponding to the weather variables to obtain the sum, and add 1 to the sum to obtain the risk area expansion factor; Obtain the user's health data, including heart rate, sleep quality and exercise amount, set the scoring standard, calculate the health score, and combine the user's individual situation to assess whether the user is currently suitable for going downstairs for exercise based on the total health score to obtain a health assessment result; When the user's health assessment result is suitable for exercise or cautious exercise, adjust the user's planned exercise route based on the risk area of the three-dimensional map of the community, and prompt the user with risk information.

2. The community pension service management method based on risk analysis according to claim 1, wherein the YOLO algorithm is used to locate the obstacle position, building risk position, tree collapse risk position and underground space risk position in the community to form a community risk position set. It comprises: The camera images collected from the main roads, public areas and leisure places of the community are used to obtain the obstacle to-be-identified dataset; the unmanned aerial vehicle is used to shoot the outer walls of the buildings in the community along the planned flight route, the satellite remote sensing is used to obtain the roof temperature distribution image, and the humidity sensor on the ground is used to collect the humidity data below the roof to obtain the building risk to-be-identified dataset; the laser radar is used to scan the trees in the community to obtain the tree parameters, including the height, diameter and crown shape of the trees, the geographic information system is used to determine the geographic position of each tree, and the soil texture, underground water level and historical weather data of the community are collected to obtain the tree collapse risk to-be-identified dataset; the underground pipeline detection radar equipment is used to detect along the pipeline direction to obtain the pipeline image, the sensors are used to monitor the pressure and flow in the pipeline in real time, and the water level, humidity and drainage pump operation state are monitored at the entrance, exit and low-lying sections of the underground parking lot to obtain the underground space risk to-be-identified dataset; The data labeling is performed on the obstacle to-be-identified dataset, the building risk to-be-identified dataset, the tree collapse risk to-be-identified dataset and the underground space risk to-be-identified dataset, the YOLO algorithm model is trained, the risk position is located, and the risk position set is generated.

3. The community pension service management method based on risk analysis according to claim 2, wherein the data labeling is performed on the obstacle to-be-identified dataset, the building risk to-be-identified dataset, the tree collapse risk to-be-identified dataset and the underground space risk to-be-identified dataset, the YOLO algorithm model is trained, the risk position is located, and the risk position set is generated, including: The obstacle to-be-identified dataset labeling is obstacle labeling, the building risk to-be-identified dataset labeling includes outer wall falling hidden danger labeling and roof leakage risk labeling, the tree collapse risk to-be-identified dataset labeling is tree collapse labeling, and the underground space risk to-be-identified dataset labeling includes underground pipeline rupture hidden danger labeling and underground parking lot waterlogging risk labeling; YOLOv7 is selected, and a corresponding deep learning training environment is built; the labeled dataset is divided into a training set, a validation set and a test set according to a proportion; the training set data is input into the YOLO algorithm model for training, and the YOLO algorithm model learns the characteristics of various risk positions by adjusting the weights and biases; During the training process, the performance indicators of the model on the validation set are monitored, including the average precision mean and the recall rate; when the performance of the model on the validation set no longer improves or reaches the preset performance indicators, the training is stopped; the trained model is evaluated on the test set for optimization. The YOLO algorithm model is deployed on a data processing server, and the real-time collected community image data and the data uploaded by the sensor are processed through the YOLO algorithm model; the risk position predicted by the YOLO algorithm model is saved in an image coordinate system or a sensor coordinate system and is converted into a geographic coordinate; the mapping relationship between the image coordinate and the geographic coordinate and the corresponding relationship between the sensor position and the geographic coordinate are established to realize the conversion of the risk position; and the various community risk positions positioned by the YOLO algorithm model are integrated to form a community risk position set.

4. The community pension service management method based on risk analysis according to claim 1, wherein the community three-dimensional map is constructed, the community risk position set is formed into a risk area in the community three-dimensional map, and risk information is labeled, comprising: collecting terrain data, building data, road data and public facility data, and constructing a community three-dimensional map by using three-dimensional map construction software; converting the community risk position set data into a Shapefil format and importing it into the three-dimensional map construction software, marking each risk position in the community three-dimensional map according to the geographic coordinate information of the risk position; for a risk position with a range, generating a corresponding risk area in the three-dimensional map according to the influence range and characteristics of the risk; labeling risk information, including the type, description, risk level, discovery time and recommended measures of the risk.

5. The community pension service management method based on risk analysis according to claim 1, wherein the risk area expansion factor and the current weather data are used to adjust the risk area of the community three-dimensional map and the corresponding risk information, comprising: determining the risk area that needs to be adjusted based on the risk area expansion factor and the current weather data; calculating the new risk area range, specifically as follows: when the risk area is a circular area, taking the center of the original risk area as the center, multiplying the original radius by the expansion factor to obtain a new radius, and thus determining the range of the new circular risk area; when the risk area is a polygonal area, performing corresponding scaling and offset calculation on each vertex coordinate of the polygon, obtaining new vertex coordinates by multiplying the vertex coordinates by the expansion factor, reconnecting the vertices to form a new polygonal risk area; updating the calculated new risk area range to the community three-dimensional map; reassessing the risk level according to the expansion of the risk area; supplementing the description related to the weather condition and the risk area expansion in the risk information; updating the risk prompt information for the user in the three-dimensional map according to the new risk area and the risk level.

6. The community pension service management method based on risk analysis according to claim 1, wherein the health data of the user, including heart rate, sleep quality and exercise amount, are obtained, and a scoring standard is set to calculate a health score, comprising: equipping the user with an intelligent wearable device to monitor the heart rate, sleep quality and exercise amount in real time; setting the normal range of the heart rate according to the medical standard and the physical characteristics of users in different age groups, and scoring according to the degree of deviation of the heart rate from the normal range. Set the sleep index of sleep quality, score according to sleep duration and deep sleep stage proportion; Set the exercise amount index according to the user's physical condition and exercise ability, score according to the comparison between actual exercise amount and target exercise amount; According to the importance of heart rate, sleep quality and exercise amount to user health, set the corresponding weight; Multiply the scores of heart rate, sleep quality and exercise amount by their respective weights, and add the results to get the user's health score.

7. The community pension service management method based on risk analysis according to claim 1, wherein the user's individual situation includes health history, disease condition and exercise habit; the influence of the user's individual situation on the user's exercise is analyzed; and the health assessment result is obtained by comprehensively considering the user's individual situation and health score, including assessment conclusion, basis explanation, exercise suggestion and risk prompt.

8. The community pension service management method based on risk analysis according to claim 1, wherein when the user's health assessment result is suitable for exercise or cautious exercise, the user's planned exercise route is adjusted based on the risk area of the community three-dimensional map, and the user is prompted with risk information, including: when the user's health assessment result is suitable for exercise or cautious exercise, the user is allowed to carry out exercise, otherwise the user is suggested to suspend exercise; when it is cautious exercise, the limiting condition is obtained; the user's planned exercise route is located on the community three-dimensional map, and the key coordinate points on the route are extracted; the extracted route coordinate points are compared with the risk area in the three-dimensional map to determine whether the route passes through the risk area, and the distance between the route and the risk area is calculated by using the spatial analysis function of the geographic information system, when the distance is less than the set safety threshold, it is determined that there is risk; for the road section judged to have risk, the corresponding risk information is associated; for the user whose health assessment result is suitable for exercise, when the exercise route has risk, the route is adjusted according to the risk level and actual situation; when passing through low-risk area, the user is prompted with risk information without changing the original route; when passing through high-risk area, a new route is planned to find a replacement route that avoids the risk area and is closest to the original route; for the user whose health assessment result is cautious exercise, when the exercise route has risk, the risk area is avoided first; when planning a new route, the limiting condition is used as a premise to find a replacement route that avoids the risk area and is closest to the original route. The three-dimensional map construction module includes: a community risk position positioning unit and a three-dimensional map construction unit; wherein the community risk position positioning unit uses YOLO algorithm to locate the positions of obstacles, building risks, tree collapse risks and underground space risks in the community to form a community risk position set; the three-dimensional map construction unit constructs a community three-dimensional map, forms a risk area in the community three-dimensional map by using the community risk position set, and labels risk information. ​ ​ ​ ​ ​ ​ ​ 9. A risk analysis-based community elder care service management system using the risk analysis-based community elder care service management method according to any one of claims 1 to 8, characterized by, ​ ​ The three-dimensional map adjustment module comprises a risk area expansion factor calculation unit and a three-dimensional map adjustment unit; the risk area expansion factor calculation unit combines historical weather data with risk events occurring in the community in a corresponding time period to perform exploratory data analysis, constructs a statistical model, and obtains a risk area expansion factor; the three-dimensional map adjustment unit adjusts the risk area and corresponding risk information of the three-dimensional map of the community based on the risk area expansion factor and current weather data; The health result evaluation module comprises a health data acquisition unit, a health score calculation unit, and a health result evaluation unit; the health data acquisition unit acquires health data of the user, including heart rate, sleep quality, and exercise amount; the health score calculation unit sets a scoring standard and calculates a health score; the health result evaluation unit combines the individual situation of the user, evaluates whether the user is suitable for going downstairs for exercise based on the health total score, and obtains a health evaluation result; The exercise route adjustment and prompt module comprises an exercise route adjustment unit and a risk information prompt unit; when the health evaluation result of the user is suitable for exercise or cautious exercise, the exercise route adjustment unit adjusts the planned exercise route of the user based on the risk area of the three-dimensional map of the community; and the risk information prompt unit prompts the user of the risk information.

Images