An ai medical care and health management service system, method and storage medium thereof

Abstract

The application relates to an AI medical, nursing, health care and medicine management service system and a method thereof, and relates to the technical field of intelligent medical equipment and health management. The system comprises an intelligent terminal and a cloud data platform. The intelligent terminal is provided with a plurality of physically isolated and electrically independent intelligent medicine storage units. Each unit is provided with a lock control mechanism, a state indication component and a medicine taking sensing sensor. A master control unit sends an unlocking instruction at a preset medicine taking time and triggers an audible and light reminder. After an effective medicine taking signal is acquired through the medicine taking sensing sensor, behavior data containing a time stamp are generated and uploaded to the cloud, thereby forming a medicine taking behavior closed loop management mechanism. The application can realize forced verification and recording of medicine taking behavior, break the association link between health monitoring data and medicine taking records, and build a full-process family health management closed loop from reminding to replenishment.

Description

Technical Field

[0001] This application relates to the field of health management technology, and in particular to an AI-based medical care, rehabilitation, and drug management service system, method, and storage medium. Background Technology

[0002] Chronic diseases such as hypertension, diabetes, and cardiovascular and cerebrovascular diseases are characterized by long disease courses, the need for long-term medication, continuous monitoring, and comprehensive intervention. The key to controlling these diseases lies in continuous and standardized health management outside of hospitals—that is, in the home setting. However, current family-based chronic disease management still suffers from significant systemic deficiencies, leading to low patient medication adherence, frequent fluctuations in disease severity, and persistently high readmission rates, placing continuous pressure on individuals, families, and the healthcare system.

[0003] Specifically, current chronic disease management at the family level faces the following prominent issues: The inefficient approach to medication management is a primary bottleneck hindering the effective control of chronic diseases. Patients with chronic diseases often need to take multiple medications simultaneously, and the medication regimens are complex, with strict requirements on the timing, dosage, and order of administration. Currently, home medication management mainly relies on the patient's own memory and self-discipline, or the use of simple pill dispensers and other auxiliary tools. These methods are ill-suited to addressing medication inertia and memory lapses that can occur during long-term treatment, leading to frequent missed doses, incorrect doses, and duplicate medications, directly impacting treatment effectiveness and even causing adverse drug reactions. Simultaneously, family members and healthcare professionals cannot effectively supervise and objectively verify actual medication adherence, leaving the core aspect of chronic disease management—"medication compliance"—uncontrollable and unverifiable for extended periods, severely weakening the overall effectiveness of health management. Summary of the Invention

[0004] This application aims to address at least one of the technical problems existing in the prior art. To this end, this application proposes an AI-based medical, elderly care, rehabilitation, and medication management service system.

[0005] This application also proposes an AI-based medical care, rehabilitation, and drug management service method and a computer storage medium using the aforementioned AI-based medical care, rehabilitation, and drug management service system.

[0006] The AI-powered medical care, rehabilitation, and drug management service system according to a first aspect of this application includes: a smart terminal and a cloud data platform; The intelligent terminal includes a cabinet and a main control unit. The cabinet contains multiple intelligent medicine storage units, which are physically isolated from each other and electrically addressable and independently controllable. Each intelligent medicine storage unit is equipped with a locking mechanism, a status indicator component, and a medicine retrieval sensing sensor. The main control unit is configured to: send an unlock command to the corresponding target intelligent drug storage unit at a preset medication time and trigger an audio-visual reminder; after obtaining a valid medication retrieval signal through the medication retrieval sensing sensor, generate behavioral data containing a timestamp, target unit identifier, and medication retrieval confirmation information, and upload the behavioral data to the cloud data platform to form a closed-loop drug management mechanism from medication reminder, unlock control, medication retrieval confirmation to behavioral recording.

[0007] In one optional embodiment, the smart terminal further includes a health monitoring module; The health monitoring module is equipped with a data interface for connecting to external medical testing equipment to obtain physiological parameter measurement data generated by the equipment, and sending the physiological parameter measurement data to the main control unit. The main control unit is configured to automatically associate and match the received physiological parameter measurement data with the medication records within the corresponding time range, and generate and output an associated data packet containing medication information and physiological indicator information.

[0008] In an optional embodiment, a remote client communicating with the cloud data platform is also included; The remote client is configured to be accessible to healthcare professionals or caregivers to perform at least one of the following operations: View the corresponding user's health record; Receive early warning information from the system; Remote intervention operations are performed based on the health records or early warning information.

[0009] In one optional embodiment, the smart terminal further includes a remote communication unit; The remote communication unit is used to establish a communication connection between the smart terminal and the remote client to support remote consultation interaction based on the user's health record.

[0010] In one optional embodiment, the smart terminal further includes a temporary medication storage area; the temporary medication storage area is located inside the cabinet and is used to temporarily store medications to be replenished to each smart medication storage unit.

[0011] In an optional embodiment, the system further includes a predictive replenishment unit; The predictive replenishment unit is configured to build a drug consumption prediction model based on historical drug usage data and use the model to predict future stockout times. The predictive replenishment unit is also configured to automatically trigger a linkage with the cloud service platform when the drug inventory reaches a preset predictive alarm threshold, generating prescription renewal or replenishment suggestion information; the prescription renewal or replenishment suggestion information includes at least the medication plan currently being implemented by the user.

[0012] In one optional embodiment, the tonic storage area is equipped with a quantity detection unit; The main control unit is configured to receive the inventory signal sent by the inventory detection unit, and automatically generate and send a replenishment reminder to the user when it is determined that the current inventory is lower than a preset threshold. The system is configured to: after obtaining user authorization, generate an order with one click based on the replenishment reminder and medication plan information, and send the order information to the associated supply chain service platform to automatically start the drug delivery process.

[0013] In an optional embodiment, the AI-powered medical and elderly care and drug management service system according to claim 1 is characterized in that the intelligent terminal is equipped with a voice interaction unit; the voice interaction unit includes a microphone, a speaker and a local voice processing chip, and supports both online and offline working modes; The voice interaction unit is configured to: respond to voice wake-up, accept user commands through natural voice dialogue, and execute corresponding operations.

[0014] The AI-powered medical, elderly care, rehabilitation, and medication management service method according to the second aspect of this application, which uses the AI-powered medical, elderly care, rehabilitation, and medication management service system described above, includes the following steps: At the preset medication time, the main control unit of the smart terminal sends an unlock command to the target smart medication storage unit and activates an audio-visual reminder; The user's medication retrieval action is detected by the medication retrieval sensing sensor configured in the target intelligent medication storage unit; After a valid medication retrieval signal is detected, the main control unit generates behavioral data containing a timestamp, a corresponding unit identifier, and medication retrieval confirmation information. The behavioral data is uploaded to a cloud data platform to construct a medication behavior record; The health monitoring module configured on the smart terminal acquires physiological parameter measurement data collected by external medical testing equipment. The physiological parameter measurement data is automatically associated and matched with the medication records within the corresponding time period to form the correlation data between medication and physiological indicators; The user's health records and early warning information are received from the cloud data platform via a remote client, which can be viewed by medical staff or caregivers. Based on the health records or early warning information, remote intervention operations are performed through the remote client.

[0015] A computer-readable storage medium according to a third aspect embodiment of the present application stores a computer program thereon, which, when executed by a processor, implements the steps of the method described in the above embodiments.

[0016] The embodiments of this application have at least the following beneficial effects: This application provides an AI-powered medical, elderly care, rehabilitation, and medication management service system, method, and application. The solution utilizes multiple physically isolated and electrically addressable and independently controllable intelligent medication storage units to achieve precise control over medication use. A locking mechanism executes an unlocking command and triggers an audible and visual reminder at a preset medication time, ensuring the mandatory nature of medication reminders. After acquiring a valid medication retrieval signal using a medication retrieval sensor, the main control unit generates behavioral data containing a timestamp, target unit identifier, and medication retrieval confirmation information, thus upgrading medication use from passive reminders to active verification. This behavioral data is then uploaded to a cloud data platform, forming a closed-loop medication management mechanism from medication reminders, unlocking control, medication retrieval confirmation to behavioral recording, effectively solving the technical problem of unknown medication adherence. Based on this closed-loop management mechanism, the main control unit automatically correlates and matches physiological parameter measurement data obtained from the health monitoring module with medication records within the corresponding time range, generating a related data package containing medication information and physiological indicator information. This establishes a causal link between health data and treatment behavior, providing continuous data support for medical decision-making. By accessing the cloud data platform via a remote client, healthcare professionals can view health records, receive alerts, and perform remote interventions, achieving a deep integration of medical services and the home environment. Ultimately, through the independent control features of the intelligent medicine storage unit and the synergistic effect of the cloud data platform, the system constructs a complete closed loop for family health management, significantly improving the systematic and continuous nature of chronic disease management.

[0017] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0018] The accompanying drawings are used to provide a further understanding of the technical solutions disclosed in this application and form part of the specification. They are used together with the embodiments disclosed in this application to explain the technical solutions of this application and do not constitute a limitation on the technical solutions disclosed in this application.

[0019] Figure 1 This is a logic block diagram of an embodiment of this application; Figure 2 This is a schematic diagram of the intelligent drug storage unit according to an embodiment of this application; Figure 3 This is a structural diagram of the cabinet in this application. Detailed Implementation

[0020] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.

[0021] In the description of this application, the terms "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0022] Patients with chronic diseases often face complex medication regimens involving multiple drugs and fixed dosages at specific times when taking medication at home. Current technology relies primarily on patients' subjective memory or simple dispensing kits for management, lacking the ability to physically verify and digitally record actual medication dispensing, leading to frequent missed doses, incorrect doses, and duplicate doses. Furthermore, family members and healthcare professionals cannot monitor medication adherence in real time, leaving medication compliance in a state of "unknowable and unmanageable."

[0023] To solve the above problems, such as Figure 1 , Figure 2 , Figure 3 As shown in the figure, this application provides an AI-powered medical care, rehabilitation, and medication management service system, including: a smart terminal and a cloud data platform; The intelligent terminal includes a cabinet 100 and a main control unit. Multiple intelligent medicine storage units 110 are installed inside the cabinet. Each intelligent medicine storage unit 110 is physically isolated from each other and can be individually addressed and controlled electrically. Each intelligent medicine storage unit 110 is equipped with a locking mechanism 120, a status indicator component 130, and a medicine dispensing sensing sensor. The main control unit is configured to: send an unlock command to the corresponding target smart drug storage unit at the preset medication time and trigger an audio-visual reminder; after obtaining a valid drug retrieval signal through the drug retrieval sensing sensor, generate behavioral data containing a timestamp, target unit identifier and drug retrieval confirmation information, and upload the behavioral data to the cloud data platform to form a closed-loop drug management mechanism from medication reminder, unlock control, drug retrieval confirmation to behavior recording.

[0024] A smart terminal refers to a physical hardware device deployed in a user's home environment. Its main body is a vertical or desktop cabinet structure, and its overall size can be customized according to user needs. This application embodiment does not impose any special limitations on this.

[0025] A cloud-based data platform refers to a data processing and storage system deployed on a remote server cluster. Its physical carrier can be a public cloud (such as Alibaba Cloud or Huawei Cloud), a private cloud, or a hybrid cloud architecture. The platform has user authentication, health record management, time-series storage of behavioral data, API interface services, and multi-terminal synchronization capabilities. It communicates bidirectionally with smart terminals via TLS-encrypted HTTPS or MQTT protocols to ensure the integrity and security of medication behavior data uploads.

[0026] Multiple intelligent drug storage units refer to several independent drug storage compartments arranged vertically in layers and horizontally in a modular fashion within a cabinet. Each drug storage unit is physically separated by partitions and is not interconnected, forming an independent, sealed space. In terms of electrical connection, each drug storage unit is equipped with a unique address code (such as an I²C bus address or RS485 node ID). The main control unit can send control commands or read status signals to any unit individually through address addressing.

[0027] In one embodiment, such as Figure 2 As shown, the cabinet is equipped with two independent 7×6 matrix medicine trays (100), each with 42 compartments, for a total of 84 independent medicine storage compartments, capable of storing 84 doses of dispensed medication. Each compartment is an independent intelligent medicine storage unit. By fully utilizing the two drawer layers, the system can be expanded to a maximum storage capacity of 168 compartments. Through three-dimensional vertical stacking and a modular matrix layout, it achieves high-density, zoned, and categorized storage of medications within a limited physical space, significantly improving space utilization efficiency.

[0028] The locking mechanism refers to the electromechanical integrated actuator installed at the door of each intelligent medicine storage unit. Its specific implementation can be, for example, a miniature electromagnetic lock (energized to engage, de-energized to release), or a miniature stepper motor driving a lead screw to raise and lower a shaft pin to achieve locking / unlocking. Its control signal is output by the main control unit through the drive circuit.

[0029] The status indicator component refers to the visual feedback unit integrated on the front of each smart medication storage unit. Its specific implementation can be, for example, a monochrome or tri-color LED light, or an OLED micro-display module. The LED light can display different colors (e.g., green for medication to be taken, blue for medication taken, and red for medication shortage alarm), different flashing frequencies (e.g., slow flashing for reminder, fast flashing for medication not taken within the time limit), and brightness levels (e.g., high brightness for emergency reminder). The OLED module can display text (e.g., "8:00 AM antihypertensive medication") or icons. All status indicators are uniformly scheduled by the main control unit and support dynamic updates according to the medication plan.

[0030] A medication dispensing sensor is a sensing device used to detect whether a user has completed the medication dispensing action. Specific implementation methods include, for example, a piezoelectric film sensor, attached to the inside of the compartment door, sensing changes in door opening pressure to detect whether medication has been dispensed; an infrared beam sensor, placed on both sides of the compartment door opening, detecting when a hand passes through the beam; or a miniature weighing sensor embedded in the bottom of the compartment, detecting the amount of weight reduction of the medication to detect whether medication has been dispensed. The sensor outputs an analog voltage or digital pulse signal, which is sampled by the main control unit's built-in ADC and judged by a threshold to identify a "valid medication dispensing signal." The judgment logic is: within a specified time window (e.g., 90 seconds) after unlocking, if the sensor output exceeds a preset trigger threshold and the duration is ≥200 ms, it is considered a valid medication dispensing event. This time window can be set within the range of 50–180 seconds according to the user's operating habits.

[0031] The main control unit refers to the embedded control core integrated inside the cabinet. Its hardware can be configured according to requirements. For example, it can be a main control unit including a quad-core ARM Cortex-A53 processor, 2 GB LPDDR4 memory, 16 GB eMMC storage, a real-time clock (RTC) chip, an audio codec, multiple GPIOs and serial port expansion interfaces, etc. It can be a main control unit running a lightweight Linux operating system with customized firmware, and has multi-task scheduling, local rule engine and edge computing capabilities. The main control unit communicates with each smart medicine storage unit through CAN bus or I²C bus, connects to external health monitoring devices through USB or Bluetooth, and connects to the cloud data platform through Wi-Fi or 4G modules.

[0032] The preset medication time refers to the set of time points set by users or medical staff during the initial system configuration phase. Its format is "HH:MM". It supports multiple times a day (such as 07:00, 12:00, 19:00) and periodic repetition (such as once every two days) to support some special medication needs. The time data is stored in the local database of the main control unit and synchronized with the cloud. The main control unit compares the current time with the preset time in real time through the RTC chip and triggers the matching action. The time error is controlled within ±10 seconds.

[0033] The sound and light reminder refers to the composite reminder signal output by the main control unit in conjunction with the voice module and the status indicator component. The "sound" part plays synthesized voice (such as "Please take your blood pressure medication") or prompt tone (such as a short double tone "beep-beep-") through the built-in speaker in the cabinet, and the volume can be adjusted within the range of 40–85 dB. The "light" part is the activation display of the status indicator component. The sound and light are activated synchronously and last for no less than 10 seconds. It supports manual shutdown by the user or automatic timeout termination.

[0034] A target intelligent drug storage unit refers to a specific drug storage unit associated with the current preset medication time. Each drug storage unit corresponds to a unit identifier, and the mapping relationship is defined by the medication plan configuration file. In a medication plan, each medication time point corresponds to one or more unit identifiers. The main control unit accurately locates and controls the target drug storage unit based on the mapping relationship. Multiple drug storage units can be associated at the same time point to support combined medication scenarios.

[0035] A valid drug retrieval signal refers to the original sensor signal output by the drug retrieval sensing sensor within a specified time window after the target unit is unlocked, satisfying both amplitude and duration conditions. The main control unit performs noise reduction filtering, threshold comparison, and edge detection on this signal, ultimately generating a Boolean confirmation flag. This flag, along with the current timestamp and the target unit identifier, is encapsulated into a structured behavioral data packet. Uploading behavioral data to the cloud data platform refers to the main control unit encrypting the generated behavioral data packets with AES-256 and publishing them to a designated topic in the cloud via the MQTT protocol. The platform message middleware receives, parses, verifies, and persists the data to the time-series database. The upload process supports offline caching and retransmission mechanisms. Each record generates a unique global ID in the cloud and is bound to the user's electronic health record.

[0036] Through the above technical solution, this application achieves the following: using physically isolated intelligent drug storage units as the execution basis, a locking mechanism to force the triggering of medication actions, a status indicator component to provide intuitive human-machine feedback, a drug retrieval sensing sensor to complete the objective verification of key behaviors, and a main control unit to uniformly schedule time, instructions, sensing and communication, and finally upload structured behavioral data to the cloud in a closed loop; the "reminder → unlock → retrieval → confirmation → upload" full-link constructed in this way ensures that every medication behavior has a traceable timestamp, a locatable unit identifier and verifiable action evidence, fundamentally solving the technical problem of unknowable and unmanageable medication adherence in home scenarios, and providing a real, continuous and reliable data source for subsequent health data analysis, remote medical intervention and supply chain linkage. Furthermore, by sending an unlock command to the corresponding target smart medication storage unit at the preset medication time and triggering an audio-visual reminder, it ensures that users can take specific medications at specific times, effectively avoiding medication errors and missed doses. Through the medication detection sensor, it can accurately record the user's medication time and promptly detect when the user fails to take the medication on time, and notify the user and caregiver through the cloud data platform, further ensuring that the user takes the medication on time and achieving better medication management results.

[0037] Among related technologies, home monitoring (such as blood pressure and blood sugar) is the cornerstone of chronic disease management. However, traditional home monitoring devices generate fragmented, discontinuous, and instantaneous data, recorded in scattered ways (such as paper records or different apps), completely disconnected from medication records and symptom changes. During follow-up visits, doctors cannot obtain accurate and coherent disease progress data, resulting in a lack of precise basis for medical decisions (such as adjusting medication), rendering home health management merely a formality, and severely delaying preventive interventions.

[0038] To address the aforementioned issues, in one optional embodiment of this application, the smart terminal further includes a health monitoring module; The health monitoring module is equipped with a data interface for connecting to external medical testing equipment to obtain physiological parameter measurement data generated by the equipment, and then sending the physiological parameter measurement data to the main control unit. The main control unit is configured to automatically associate and match the received physiological parameter measurement data with the medication records within the corresponding time range, and generate and output an associated data packet containing medication information and physiological indicator information.

[0039] The health monitoring module is located in a pre-reserved detection area on the front or top of the cabinet. This detection area is equipped with a standardized physical interface and a wireless communication protocol support unit. The data interface includes both wired and wireless interfaces: the wired interface is a USB Type-B or Micro-USB interface, used to connect to blood pressure monitors, electronic thermometers, or electrocardiogram (ECG) acquisition devices with USB output functionality; the wireless interface supports Bluetooth 5.0 or Bluetooth Low Energy protocols, used to establish pairing connections with external medical testing devices such as blood glucose meters, pulse oximeters, and smart scales that support Bluetooth communication. After the testing device is connected, the health monitoring module identifies the device type through a pre-installed device driver and automatically starts the data receiving process. Physiological parameter measurement data includes, but is not limited to, at least one of the following: systolic blood pressure, diastolic blood pressure, heart rate, fasting blood glucose level, postprandial blood glucose level, blood oxygen saturation, body temperature, weight, and raw ECG waveform data.

[0040] The health monitoring module and the main control unit transmit data via an internal serial bus or a high-speed SPI bus. The main control unit has a built-in time synchronization module to assign a unified timestamp to all access events. When the main control unit receives physiological parameter measurement data, it automatically extracts the embedded timestamp (if the device does not provide it, the local reception time is used as the timestamp) and compares it with the timestamp of the locally stored medication behavior data. The "corresponding time range" refers to a time window set by the user within a certain range before and after the detection time. For example, it can be a time window centered on the timestamp of the physiological parameter measurement data, extending forward by 15 minutes and backward by 60 minutes. Within this time window, if there is one or more medication behavior data (including timestamp, target unit identifier, and medication confirmation information), the physiological parameter measurement data is bound to the one or more medication behavior data.

[0041] After completing the association, the main control unit encapsulates the association data packet into JSON or Protocol Buffer format, and adds user ID, device serial number, and encrypted signature fields. It then uploads the data to the cloud data platform via the remote communication unit. After parsing, the cloud data platform stores the data in the "Medication-Vital Sign Association" sub-table of the user's health record and simultaneously generates visualization charts, including a line graph of blood pressure trend after medication, a heat map of blood glucose fluctuations, and a matrix of co-occurrence of medication adherence and abnormal indicators. The association data packet can also be used to trigger subsequent analysis logic. For example, if the systolic blood pressure does not decrease by ≥5 mmHg within 2 hours after 3 consecutive morning medications, and there are no missed doses during the same period, the system automatically generates a "Efficacy assessment pending manual review" flag and pushes it to the remote client.

[0042] Through the above technical solutions, this application achieves deep coupling between the health monitoring module and the medication management module in four aspects: hardware access, time alignment, data binding, and cloud collaboration. Because the health monitoring module uses a data interface compatible with multiple communication methods, it allows plug-and-play use of home medical testing devices of different brands, models, and eras. Since the main control unit performs automatic association based on a high-precision local clock and dynamic time window strategy, it avoids subjective errors and operational burdens caused by manual annotation. Because the associated data packets carry complete context and verification mechanisms, it ensures the integrity and traceability of data during transmission and storage. Ultimately, without changing the user's original testing habits, a structured, time-aligned, and causally identifiable "medication-vital signs" joint data asset is naturally formed, providing a real, continuous, and reliable data foundation for efficacy evaluation, risk warning, and remote diagnosis and treatment.

[0043] In related technologies, patients with chronic diseases, especially the elderly or those with limited mobility, face the practical difficulties of frequently traveling to the hospital for check-ups and medication pickups. Although internet hospitals offer the possibility of online consultations, the lack of hardware terminals deeply integrated with the home environment renders remote diagnosis and treatment ineffective. Doctors cannot simultaneously access patients' recent medication records and vital sign trends during consultations, significantly reducing the effectiveness of so-called "remote consultations." The home remains the "periphery" of medical services rather than the "front line."

[0044] To address the aforementioned issues, in one optional embodiment of this application, a remote client that communicates with a cloud data platform is also provided. The remote client is configured to be accessible to healthcare professionals or caregivers to perform at least one of the following actions: View the corresponding user's health record; Receive early warning information from the system; Remote intervention operations are performed based on health records or early warning information.

[0045] The remote client is a software application running on a computer device or mobile terminal. It establishes a secure communication connection with the cloud data platform through standard network protocols (such as HTTPS and WebSocket), and accesses the corresponding user's exclusive data space after identity authentication and permission verification. Identity authentication methods include, but are not limited to, username / password, digital certificate, two-factor authentication, or integration with the unified identity authentication platform of the medical institution. Permission verification dynamically allocates the scope of data access and operation permissions according to the user's role (such as attending physician, community nurse, family caregiver). For example, attending physicians can view all historical health records and perform prescription adjustment interventions, while community nurses can view recent medication adherence and vital sign trends and initiate follow-up reminders.

[0046] A health record is a collection of personal health data that is structured and continuously updated by a cloud-based data platform. Its content includes, but is not limited to: time-series medication behavior records (including timestamps, target intelligent drug storage unit identifiers, and medication retrieval confirmation status), associated physiological parameter measurement data (including values ​​of indicators such as blood pressure, blood sugar, and heart rate, measurement time, device model, and calibration status), medication regimen version information, early warning event logs, remote consultation record summaries, and system-generated analysis reports. Health records support multi-dimensional retrieval and visualization by time axis, by medication regimen, and by physiological indicator type. For example, a line graph can be used to show the trend of the average systolic blood pressure during a certain medication cycle, and the nodes where missed doses occurred can be marked simultaneously.

[0047] The warning information is generated by the cloud data platform based on a preset rule engine or a lightweight AI model. The triggering conditions include, but are not limited to: N consecutive times without detecting a valid drug retrieval signal from the target smart drug storage unit; physiological parameter measurements exceeding the individual baseline threshold within the same time period and accompanied by medication interruption; abnormal patterns appearing in the associated data package where the medication record does not match the trend of improvement in vital signs; and drug inventory falling below the predicted alarm threshold without receiving confirmation of replenishment from the user. The warning information adopts a hierarchical mechanism, divided into Level 1 (requiring immediate response), Level 2 (recommended to be handled within 24 hours), and Level 3 (providing attention), and is sent to authorized remote clients through multiple channels such as pop-ups, SMS, APP push notifications, and voice outbound calls.

[0048] Remote intervention operations include, but are not limited to: issuing temporary medication instructions to smart terminals (such as adding an extra dose of antihypertensive medication), modifying subsequent medication plans (such as adjusting the daily frequency of a certain drug), initiating video consultation requests, generating and pushing personalized health guidance texts, submitting emergency medication replenishment instructions to the supply chain service platform, and accessing raw sensor data for specific time periods for clinical review. All remote intervention operations are fully recorded in the operation log of the health record, including the operator's identification, operation time, operation content, execution result feedback, and an audit trail code automatically generated by the system, ensuring that the entire process is traceable and auditable. At the same time, all remote intervention operations are simultaneously notified through the remote clients of the user and their family caregivers.

[0049] Through the above technical solutions, this application achieves the following: while maintaining the local closed-loop management capabilities of smart terminals, medication behavior and physiological parameter data generated in home scenarios are uniformly modeled and encapsulated on a cloud platform and then made available to professional caregivers in a controlled, hierarchical, and auditable manner; through structured access to health records via remote clients, medical staff can overcome time and space limitations to obtain continuous, authentic, and contextually complete evidence of the course of the disease, and can provide patients with consultation services based on detailed health records; through proactive push of early warning information and a multi-level response mechanism, the time window from anomaly identification to intervention initiation is significantly compressed; and through remote intervention operations supported by standardized interfaces, medical decisions can directly drive actions at the terminal execution layer, thereby truly constructing a full-chain medical and elderly care collaborative closed loop of "monitoring—analysis—early warning—intervention—feedback".

[0050] Furthermore, in an optional embodiment, the AI-powered medical care, rehabilitation, and medication management service system provided in this application further includes a remote communication unit in its smart terminal; The remote communication unit is used to establish a communication connection between the smart terminal and the remote client to support remote consultation interaction based on the user's health record.

[0051] The remote communication unit is an embedded communication module integrated inside the smart terminal cabinet. Its hardware components include a baseband processor, an RF transceiver module, an antenna assembly, and a protocol stack firmware. It supports multiple network standards, including 4G LTE, 5G NR, and Wi-Fi 6 (IEEE 802.11ax), and can automatically select the optimal access link based on the deployment environment. Its communication protocol stack complies with RFC3261 (SIP), RFC 3984 (H.264 over RTP), and WebRTC standards, ensuring low-latency audio and video streaming transmission and end-to-end encryption capabilities.

[0052] The remote communication unit connects to the main control unit via a PCIe or USB 3.0 high-speed bus interface and is subject to unified scheduling by the main control unit. Its data channels are logically divided into control channels and media channels: the control channel is used to transmit consultation requests, identity authentication, consultation status synchronization, and health record access authorization commands; the media channel is used to carry real-time audio and video streams, screen sharing data, and incremental update frames of health record visualization charts; all media data is encrypted with AES-256 and encapsulated in the SRTP payload for transmission.

[0053] The remote communication unit supports dual-mode consultation initiation: one is the user-initiated mode, where the user initiates a consultation request via voice command (such as "Call Dr. Zhang") or by clicking the "One-Click Consultation" button on the touchscreen. The main control unit then calls the remote communication unit to send a SIP INVITE to remote clients in the preset whitelist. The other is the passive response mode, where when a remote client (medical staff) initiates a consultation invitation, the remote communication unit receives and parses the SIP INVITE message. After the main control unit completes local permission verification (including user biometric confirmation or PIN code input), it automatically wakes up the display screen, camera, and microphone, and simultaneously loads the current user's latest health record summary page (including a 7-day medication adherence heatmap, blood pressure / blood glucose trend line graph, the raw values ​​of the last three tests, and abnormal annotations) into the local cache for real-time access by the consultation interface.

[0054] The remote communication unit and the cloud data platform communicate via a bidirectional TLS 1.3 secure tunnel. During remote consultations, when a remote client requests access to a specific health record data item, the remote communication unit forwards the request to the main control unit. The main control unit then initiates an OAuth 2.0 authorization access request to the cloud data platform to obtain a temporary access token (JWT). The remote communication unit then embeds the token into the HTTP GET request header and retrieves the corresponding encrypted data block from the cloud API gateway. The retrieved data is decrypted and lightweight rendered by the main control unit and pushed to the remote client's browser or dedicated app in SVG vector graphics or JSON-LD structured format.

[0055] Through the above technical solutions, this application achieves collaborative cooperation between the remote communication unit, the intelligent terminal main control unit, the cloud data platform, and the remote client: Because the remote communication unit possesses multi-standard access capabilities and supports standardized audio and video protocols, it ensures the robustness and interoperability of remote consultation connections; because its communication channel strictly distinguishes between control and media streams and adopts end-to-end encryption and tokenized data access mechanisms, it supports real-time access to health records while ensuring the minimum necessary disclosure and secure transmission of patient privacy data; because its design combines local caching of summary pages with dynamic loading of raw data, it reduces cloud bandwidth pressure and avoids the risk of persistent storage of sensitive data on the terminal side; ultimately, without changing the existing internet hospital architecture, it achieves a reliable, efficient, and data-driven deep integration between home terminals and professional medical resources.

[0056] In related technologies, chronic disease management involves the long-term stable supply of materials such as medicines, test strips, and auxiliary materials. Under the current model, the purchase and replenishment of medicines are entirely initiated by the patient, often leading to treatment interruptions due to forgetfulness or inconvenience. The supply chain is not intelligently linked to the patient's actual medication progress or health status, making it impossible to achieve on-demand, proactive, and timely delivery services.

[0057] To address the aforementioned issues, in one optional embodiment of this application, the intelligent terminal of the AI-powered medical care, rehabilitation, and medication management service system further includes a medication storage area; the medication storage area is located within a cabinet and is used to temporarily store medications to be replenished to each intelligent medication storage unit.

[0058] The temporary storage area for replenished medicines is a separately defined physical space within the cabinet. It is located in the lower middle or back cavity of the cabinet and is integrated with multiple intelligent medicine storage units in the same cabinet layout. Structurally, this area is isolated from each intelligent medicine storage unit, but it is accessible to each intelligent medicine storage unit through internal channels or open access ports. It does not have a locking mechanism, status indicator components, or medicine retrieval sensors. It is only equipped with a basic support structure and partitions for classifying and temporarily storing medicines of different types, dosage forms, or batches to be replenished.

[0059] The volume of the temporary storage area for tonics can be set according to the actual frequency of medication use and the amount purchased at one time, ranging from 2L to 8L or 10L; its internal partitioning method can be set according to the actual situation, using adjustable sliding partitions, magnetic partition racks or modular slot structures to adapt to the storage needs of different forms of items such as tablets, capsules, external ointments, and test strips; this application embodiment does not make any special limitations in this regard.

[0060] The temporary medication storage area and each intelligent medication storage unit are equipped with clear operation path guidance signs, including silkscreen icons, LED directional indicator lights or dynamic guidance interfaces on the touch screen, to prompt users to allocate the temporarily stored medications to the corresponding target units according to the medication plan. This operation path supports manual replenishment and can also be used in subsequent extended embodiments with the help of a robotic arm or a medication pusher to achieve semi-automatic replenishment. However, in the current embodiment, manual operation is the default implementation method.

[0061] The medications stored in the temporary medication storage area are compliant drugs that have completed prescription review, label verification, and are within their expiration date. Their storage status does not participate in the medication reminder and behavior recording process of the main control unit, and they only exist as static transfer storage units. When a user takes out a medication from the temporary medication storage area and completes the loading action into the target intelligent medication storage unit, the status indicator of the target unit is updated to "medication prepared", and the main control unit synchronously updates the drug information database of the unit, including the drug name, specifications, expiration date, and initial stock.

[0062] Through the above technical solutions, this application achieves spatial centralization and process pre-positioning of medication replenishment operations: By setting up a dedicated medication replenishment storage area within the cabinet, users can complete the pre-filling preparation of multiple days' worth of medication at once during non-medication periods, avoiding the need for temporary unpacking, repackaging, and identification of medications before each dose. Furthermore, family caregivers can replenish the corresponding medications, preventing elderly users from taking the wrong medication when faced with complex medication needs. Because the medication replenishment storage area and the intelligent medication storage unit are located in the same cabinet and have a compact layout, the user's operational path is significantly shortened, reducing the burden on elderly or mobility-impaired users. As the storage area serves as a medication transfer node, the medication replenishment behavior is decoupled from the medication execution behavior, ensuring the certainty and closed-loop nature of the medication administration process while improving the system's long-term flexibility in medication scheduling and its autonomous management capabilities.

[0063] Furthermore, in an optional embodiment, this application also provides a predictive replenishment unit; The predictive replenishment unit is configured to build a drug consumption prediction model based on historical drug usage data and use the model to predict future stockout times. The predictive replenishment unit is also configured to automatically trigger a linkage with the cloud service platform when the drug inventory reaches a preset predictive alarm threshold, generating prescription or replenishment suggestions; the prescription or replenishment suggestions shall include at least the medication plan currently being implemented by the user.

[0064] The prediction supply unit is an embedded computing module. Its hardware carrier can be a coprocessor inside the main control unit or an edge computing board independently deployed in the cabinet. It communicates with the main control unit through a CAN bus or SPI interface. The unit has a built-in lightweight machine learning inference engine, which supports model loading, feature extraction and time series prediction calculations to be completed locally without continuously relying on cloud computing power, so as to ensure the real-time performance of prediction response and offline availability.

[0065] Historical medication data originates from user medication behavior records synchronized locally from a cloud data platform, including but not limited to: timestamps of each smart drug storage unit being unlocked, corresponding unit identifiers, the time of generation of the drug retrieval confirmation signal, the duration of a single drug retrieval action, and the cumulative frequency of drug retrieval per unit time. After preprocessing by the main control unit, the data is categorized and aggregated according to drug type, specifications, and dosage cycle (e.g., once daily, twice daily, or once every other day) to form a structured time series dataset. This dataset is used as training samples and input into the drug consumption prediction model deployed by the prediction replenishment unit.

[0066] The drug consumption prediction model is a regression prediction model built based on time series analysis. Its optional implementation forms include: exponential smoothing model (Holt-Winters), ARIMA model, LSTM neural network model, or XGBoost ensemble learning model. The model input variables include at least: historical medication frequency, single dose setting value, medication interval stability coefficient, and holiday / seasonal fluctuation factors. The model output is an estimate of the remaining days of use for each type of drug and its corresponding confidence interval. The model can be incrementally updated online based on newly generated medication behavior data to adapt to dynamic scenarios such as changes in user medication adherence, adjustments to medical orders, or temporary discontinuation of medication. The model parameters and structure can be tailored to the actual deployment resource constraints, balancing accuracy and computational cost. For example, a simplified version of LSTM can be used on resource-constrained terminals, while a multivariate fusion XGBoost model can be deployed in versions with edge computing power.

[0067] The future stockout time point refers to the calendar time point corresponding to the remaining days of medication as output by the model being less than or equal to zero. This time point is calculated based on the system's local clock, combined with the current date and the predicted number of days. For example, if the current date is April 10, 2025, and the model predicts that a certain antihypertensive drug has 12.3 days of remaining medication, then the predicted stockout time point is April 22, 2025. This time point can be further superimposed with a safety buffer period (e.g., 3 days in advance) to generate the "predicted alarm time point" that will actually trigger an alarm.

[0068] The preset prediction alarm threshold is a configurable parameter, and its value is set comprehensively based on the drug type, storage stability, delivery cycle and user geographical location. For example, for tablets stored at room temperature, the threshold can be set to "remaining days of use ≤ 7 days"; for biological agents that require refrigeration, the threshold can be set to "remaining days of use ≤ 5 days". This threshold is stored in the non-volatile memory of the main control unit and can be modified via remote client or local voice command. When the prediction replenishment unit determines that the remaining days of use for any drug fall within the threshold range, the alarm process is triggered.

[0069] Automatic triggering of linkage with the cloud service platform refers to the prediction and replenishment unit initiating an HTTPS protocol request to the cloud service platform through the remote communication unit, carrying an alarm event packet in standard JSON format. This event packet includes at least: a unique user identifier, the generic name of the drug (including ATC code), specifications, current remaining quantity, predicted shortage time, and the currently executed medication regimen ID. The medication regimen ID is associated with complete structured prescription information stored in the cloud, including drug name, dosage, frequency, start and end times of treatment, contraindications, and physician signature information. This linkage process supports a network outage caching mechanism—if communication fails, the event packet is temporarily stored in the local Flash storage area and automatically retransmitted after the network is restored.

[0070] Refill or restocking recommendations are generated by the cloud service platform and returned to the smart terminal. This information includes at least: a list of recommended refill drugs (including name, specifications, quantity, and dosage), a summary of the treatment plan, pharmacist review status, a list of available supply chain service providers, and the estimated delivery time. This information can be pushed to remote clients for review by medical staff via the main control unit, or displayed on a local screen in a split-screen format, and supports voice broadcast of key fields. Users can confirm acceptance of the recommendations via touch or voice commands, and the authorized system will automatically call preset payment methods and logistics interfaces to complete the order.

[0071] The user's current medication regimen is managed uniformly by the cloud data platform. Its data structure includes: regimen version number, effective time, termination conditions, associated disease diagnosis code (ICD-11), and an array of drug entries (each containing drug ID, dosage, frequency, route of administration, and treatment cycle). This regimen can be remotely updated by doctors and synchronously distributed to the terminal. The predictive replenishment unit only references entries in the regimen that are directly related to the currently monitored drug, without parsing or executing regimen logic, and is only used for context binding. The regimen content is presented in the suggestion information in read-only summary form to ensure that the replenishment content is strictly consistent with the clinical treatment intention.

[0072] Through the above technical solutions, this application achieves the following: drug consumption modeling driven by real medication behavior data, enabling stockout prediction to have individualized and dynamic characteristics; by strongly binding the prediction results with the medication plan and embedding them into the standardized prescription information flow, ensuring that prescription renewal recommendations comply with clinical norms; through a hybrid architecture of local prediction + cloud collaboration, completing the automated service leap from "stock alert" to "prescription generation" while ensuring privacy and real-time performance; and ultimately achieving a fundamental upgrade in drug supply from passive response to proactive prediction and from manual intervention to process autonomy, significantly improving the continuity of chronic disease management and treatment stability.

[0073] In one optional embodiment, this application also provides an AI-powered medical care, rehabilitation, and medication management service system; The temporary storage area for tonics is equipped with a stock detection unit; The main control unit is configured to receive the inventory signal from the inventory detection unit, and automatically generate and send a replenishment reminder to the user when it is determined that the current inventory is lower than the preset threshold; The system is configured to automatically initiate the drug delivery process by generating an order with one click based on replenishment reminders and medication plan information after obtaining user authorization, and sending the order information to the associated supply chain service platform.

[0074] The temporary storage area for replenishing medicines is located inside the cabinet and is a physical space independent of each intelligent medicine storage unit. Its structure can be a pull-out tray, a sliding storage compartment, or a detachable modular medicine box. Its internal volume can be set according to the actual medicine packaging specifications, such as adapting to standard aluminum-plastic sheet packaging (e.g., 10 tablets / sheet), bottle packaging (e.g., 30 tablets / bottle), or bagged medicines. The surface of this area is flat and unobstructed, which facilitates stable sensing by the inventory detection unit. Its material can be opaque plastic or metal to avoid interference from ambient light with optical detection, or it can be a semi-transparent material used in conjunction with a light shield. This application embodiment does not make any special limitations on this.

[0075] The inventory detection unit is used to monitor the current inventory of drugs in the replenishment storage area in real time. Its specific implementation methods include, but are not limited to, any one or more combinations of the following: Weight sensing method: A high-precision pressure sensor or weighing module (range range, for example, 0–2 kg, resolution 1 g) is integrated at the bottom of the replenishment storage area. The inventory is calculated by measuring the overall weight change. When the initial total weight of the replenished drugs is W0, and the weight decreases by ΔW after each subsequent replenishment to each intelligent drug storage unit, the current remaining weight is W0-ΣΔW; Infrared beam sensing method: Infrared emitting and receiving tubes are installed at corresponding positions on both sides of the replenishment storage area to form a horizontal array detection line. When the height of the drug stack covers a certain infrared line, the receiving signal is blocked, thereby determining whether the inventory is higher than a preset height threshold; Ultrasonic ranging method: An ultrasonic probe is fixed at the top of the replenishment storage area, emitting ultrasonic waves vertically downwards and receiving the reflected echoes. The distance h from the top of the drug stack to the probe is obtained by calculating the flight time. Combined with the known empty chamber height H, the remaining height H-h can be obtained, which can then be converted into a volume percentage or equivalent number of tablets. Image recognition method: A miniature camera is set above the temporary storage area for medicines. In conjunction with the edge computing module, the acquired images are used to perform target detection and stacking counting to identify the number of medicine plates or the liquid level in bottles. This method can be redundantly deployed with infrared or weight methods to improve detection robustness.

[0076] All of the above detection methods can output analog voltage signals, digital switch signals, or serial communication data (such as I). 2 C. UART protocol format), which is provided for the main control unit to read in real time; the installation location, quantity and detection frequency of the inventory detection unit can be set according to actual needs, such as sampling once every 10 seconds, or triggering a single detection only before or after the user opens the cabinet door and performs the refill operation. This application embodiment does not make any special limitations on this.

[0077] The main control unit is configured to receive the inventory signal from the inventory detection unit and automatically generate and send a replenishment reminder to the user when the current inventory is determined to be lower than a preset threshold. The preset threshold is a configurable parameter, and its value can be set comprehensively based on the average daily consumption of medicines, the replenishment cycle, and the safety redundancy. For example, it can be set as "remaining quantity ≤ 3 days' supply" or "remaining weight ≤ 150 g". This threshold can be set by the system's factory default or can be personalized by the user through the smart terminal interface, remote client, or voice command. The replenishment reminder can take the form of, but is not limited to: local sound and light prompts on the cabinet (such as a buzzer beeping 3 times + a red indicator light on the replenishment area), pop-up prompts on the smart terminal screen, message notifications pushed to the bound mobile APP, and synthesized voice broadcasts (such as "The medicines in the replenishment temporary storage area are insufficient, please replenish them in time"). The reminder content can include the name of the medicine to be replenished, the suggested replenishment quantity, the current remaining quantity, and the recommended replenishment time, so that the user can respond quickly.

[0078] The system is configured to: upon obtaining user authorization, generate an order with one click based on replenishment reminders and medication plan information, and send the order information to the associated supply chain service platform to automatically initiate the drug delivery process; user authorization methods include, but are not limited to: clicking the "Confirm Replenishment" button on the smart terminal touchscreen, explicitly responding "yes" or "agree" via voice command, completing electronic signature confirmation on a remote client, or pre-setting an automatic authorization strategy (such as enabling one-click ordering by default after three consecutive reminders for the same drug); medication plan information comes from structured prescription data synchronized from the cloud data platform to the main control unit, including the generic name of the drug, specifications, single dose, daily frequency, total treatment duration, and currently executed medication. The process of generating an order with one click includes: the main control unit calls the local order template, fills in the drug ID, the quantity to be replenished (rounded up to the smallest sales unit based on the remaining amount of the treatment course, such as 1 box or 2 bottles), the user's delivery address (extracted from the health record), the medical insurance payment identifier, and the timestamp of the associated medication record; the generated order information is encrypted and encapsulated, and then sent to the preset supply chain service platform (such as pharmaceutical e-commerce API, regional center pharmacy system, or DTP pharmacy interface) via HTTPS or a dedicated API interface; this process does not rely on manual input, all fields are automatically filled by the system, and it supports failure retry and abnormal alarms (such as displaying error codes and retry buttons on the terminal when the network is interrupted or the interface authentication fails).

[0079] Through the above technical solutions, this application achieves real-time perception and dynamic response of the inventory in the temporary drug storage area: Due to the establishment of an inventory detection unit, the main control unit can continuously acquire physical inventory status signals; by binding the inventory judgment logic with preset thresholds and linking it with a multi-channel reminder mechanism, interruptions in subsequent drug replenishment operations due to inventory depletion are avoided; with explicit user authorization, replenishment reminders are deeply integrated with structured medication plan information, automatically generating standardized orders that conform to clinical norms and commercial rules, and directly connecting to the supply chain service platform, thus opening up a fully automated pathway from home terminal to drug delivery; ultimately forming a closed-loop drug supply mechanism of "inventory monitoring—threshold triggering—user confirmation—intelligent ordering—platform integration—logistics initiation," significantly improving the timeliness, accuracy, and age-appropriateness of drug replenishment in chronic disease management scenarios.

[0080] In one optional embodiment, this application also provides an AI-powered medical care, rehabilitation, and medication management service system, wherein the smart terminal is equipped with a voice interaction unit; the voice interaction unit includes a microphone, a speaker, and a local voice processing chip, and supports both online and offline working modes; The voice interaction unit is configured to respond to voice wake-up, accept user commands through natural voice dialogue, and execute corresponding operations.

[0081] The voice interaction unit is located on the front or top panel of the smart terminal cabinet. The microphone is used to collect the voice signals emitted by the user, and the speaker is used to play system feedback voice, reminder information, and dialogue response content. The local voice processing chip integrates a voice wake-up engine and a lightweight speech recognition (ASR) model. It is configured to complete keyword wake-up (such as preset wake words like "nurse" or "medical cabinet") and basic command semantic parsing locally on the device. It can respond to common commands such as "open medicine box A", "how many times have I taken my medicine today", and "measure blood pressure" without relying on a network connection. The chip supports firmware upgrades, and its model parameters can be adaptively fine-tuned according to different dialects, speech speeds, and accents. It can also continuously monitor wake words in a low-power state.

[0082] The voice interaction unit supports both online and offline working modes. In offline mode, all voice wake-up, edge ASR, and simple intent understanding are completed in the local voice processing chip, ensuring privacy and real-time response. This is suitable for home scenarios with no network coverage, high network latency, or data sensitivity. In online mode, the voice interaction unit uploads the pre-processed voice features locally to the cloud data platform via the remote communication unit. The high-precision ASR and Natural Language Understanding (NLU) models deployed in the cloud complete complex semantic parsing and multi-turn dialogue management, and return the structured instructions to the main control unit for execution. This supports complex and context-sensitive instructions such as "retrieving the blood pressure data after taking medication three times last week" and "reminding me to reduce medication if my blood pressure is higher than 140 tomorrow."

[0083] The voice interaction unit and the main control unit communicate via a standard serial bus (such as UART or I2C). 2 C) Connection: After receiving structured instructions from the voice interaction unit, the main control unit calls the corresponding functional module to perform the operation. For example, if the instruction "Open the No. 3 medicine storage unit" is parsed, an unlock signal is sent to the locking mechanism of that unit; if the instruction "Query today's medication records" is parsed, the corresponding behavioral data is retrieved from the local cache or cloud data platform and a summary is broadcast by the speaker; if the instruction "Start blood pressure measurement" is parsed, the data interface of the health monitoring module is activated and the user is prompted to connect an external blood pressure monitor.

[0084] The corresponding operations performed by the voice interaction unit include, but are not limited to: triggering the unlocking and sound and light reminders of a specified smart medicine storage unit; calling the health monitoring module to start physiological parameter measurement; querying and voice-reading historical medication records and recent blood pressure / blood sugar trends; initiating video consultation requests with remote clients; broadcasting restocking reminders or early warning information; controlling auxiliary functions such as cabinet lighting and screen brightness; and responding to open health consultations, such as "dietary precautions for patients with hypertension" and "common side effects of aspirin," with answers sourced from a local pre-built knowledge base or a cloud-based medical Q&A API.

[0085] Through the above technical solutions, this application achieves deep coupling between the voice interaction unit and the hardware architecture and business logic of the smart terminal: by setting up independent microphones, speakers and local voice processing chips, the integrity and controllability of the voice input and output link are guaranteed; by supporting online / offline dual-mode operation, real-time response, data privacy and semantic understanding depth are taken into account; by establishing a deterministic communication interface with the main control unit and mapping it to specific functional modules, it is ensured that natural voice commands can be accurately converted into actual control actions on physical components such as the lock control mechanism, health monitoring module, and remote communication unit; ultimately, without increasing the user's operational burden, the system accessibility and user-friendliness of elderly and mobility-impaired users are significantly improved, providing reliable human-machine interface support for building an accessible and highly compliant family medical, elderly care and rehabilitation closed loop.

[0086] Meanwhile, the embodiments of this application integrate local AI voice interaction, one-click video consultation, and can automatically trigger drug delivery based on inventory, forming a complete, full-process, closed-loop, one-stop service system including "consultation-monitoring-diagnosis-nursing-rehabilitation-chronic disease management-supply chain services-settlement and payment".

[0087] Given the systemic deficiencies in existing chronic disease home management technologies, such as unknowable medication behavior, disconnect between health data and treatment behavior, lack of integration between medical services and the home environment, and passive interruption of material supply, this application provides an AI-powered medical, elderly care, rehabilitation, and medication management service method. This method uses the AI-powered medical, elderly care, rehabilitation, and medication management service system described in the above embodiments and includes the following steps: Step 1: At the preset medication time, the main control unit of the smart terminal sends an unlock command to the target smart medication storage unit and activates an audio-visual reminder; The "preset medication time" refers to a time point pre-configured according to the user's individual medication plan. This can be a fixed time every day (e.g., 08:00, 12:00, 20:00) or a non-fixed time that is dynamically adjusted periodically (e.g., every other day, Wednesday / Saturday mornings). This time information is stored in the local database of the main control unit or synchronously distributed by the cloud data platform, and can be modified and confirmed through a remote client or voice interaction unit. The "target intelligent medication storage unit" is one of the physically isolated, electrically addressable, and independently controllable intelligent medication storage units. Its unit identifier corresponds to a type of medicine or a type of medication task. The "unlock command" is a digital control signal sent by the main control unit to the unit's locking mechanism to release its mechanical locking state. The "audio-visual reminder" includes a buzzer sound prompt and a status indicator light prompt, both triggered synchronously to enhance the perception reliability of elderly users in a multimodal manner.

[0088] In one alternative implementation, the time triggering method is as follows: the main control unit has a built-in real-time clock module that continuously compares the current system time with the locally stored medication plan table. When a match is found, an instruction is immediately generated and issued. In another alternative implementation, the time triggering method is as follows: the master control unit receives a dynamic scheduling instruction from the cloud data platform, which carries the updated medication time and the corresponding unit identifier. The master control unit then overwrites the local plan and executes subsequent actions accordingly. Furthermore, the time triggering method is as follows: an emergency medication reminder is temporarily inserted based on the user's daily health monitoring data (such as abnormally high blood pressure). The main control unit, in conjunction with the rule engine, determines whether the triggering conditions are met, and automatically adds an unlocking and reminder process when the conditions are met.

[0089] Step 2: Detect the user's medication retrieval action using the medication retrieval sensing sensor configured in the target intelligent medication storage unit; The "medication retrieval sensor" is a physical sensing component installed inside the target intelligent medicine storage unit or on the lid structure. It is used to respond to physical events such as opening of the medicine compartment, displacement of the lid, changes in the weight of the medicine, or changes in contact pressure. This sensor does not rely on high-computing methods such as image recognition or voice recognition. It only outputs discrete switch signals or analog change signals to indicate whether the user has completed a valid medication retrieval action. The "user medication retrieval action" refers to the physical operation behavior that occurs within a specified time window after the sound and light reminder is activated and meets the preset threshold conditions. Its validity is confirmed by the main control unit based on the original sensor signal after simple filtering and logical judgment.

[0090] In one alternative implementation, the detection method is as follows: the drug dispensing sensor is a piezoelectric thin film sensor, which is attached to the bottom of the load-bearing plate of the medicine compartment. When the user opens the box and takes out the medicine, causing a slight deformation of the load-bearing plate, the sensor outputs a voltage pulse signal. The main control unit performs dual threshold judgment on the amplitude and duration of the pulse to confirm that the medicine is dispensing effectively. In another optional implementation, the detection method is as follows: the medicine dispensing sensor is an infrared through-beam photoelectric switch, which is installed on both sides of the medicine compartment opening. When the user's hand enters the detection area after the box is opened and blocks the light path, an on / off signal is generated. The main control unit combines the opening duration and signal stability to determine the validity of the action. Furthermore, the detection method is as follows: the medicine dispensing sensor is a combination of a miniature Hall switch and a magnet. The magnet is embedded in the lid, and the Hall switch is fixed to the cabinet frame. When the opening angle of the lid exceeds a preset threshold (e.g., 30°), the switch action is triggered, and the main control unit determines that the intention to dispense medicine has been converted into actual operation.

[0091] Step 3: After a valid medication retrieval signal is detected, the main control unit generates behavioral data containing a timestamp, the corresponding unit identifier, and medication retrieval confirmation information; Among them, the "valid drug retrieval signal" is a Boolean confirmation result output by the main control unit based on the sensor signals obtained above and after local logic judgment. This result meets the time constraint (occurring within the predetermined window period after the reminder is activated), signal characteristic constraint (amplitude, duration, trend of change, etc.), and anti-accidental touch constraint (excluding instantaneous jitter or environmental interference); the "time stamp" is generated by the real-time clock module built into the main control unit at the moment of confirmation, with an accuracy of not less than 1 second; the "corresponding unit identifier" is the unique code of the target smart drug storage unit in the system, in the format of a string or integer number, consistent with the identifier used in the lock control command; the "drug retrieval action confirmation information" is a structured field, the content of which includes at least "confirmation status = success" and optional "drug retrieval quantity estimation level (such as 'full retrieval', 'partial retrieval', 'not retrieval')". This information does not rely on weighing or image recognition, but is inferred based on sensor response mode and historical behavior statistical model.

[0092] Step 4: Upload the behavioral data to the cloud data platform to build a medication behavior record; "Upload" refers to the main control unit packaging the generated behavioral data into a standard JSON or XML format message via wired or wireless communication links (such as Wi-Fi, 4G / 5G, Ethernet), and sending it to the designated interface of the cloud data platform after authentication and encrypted transmission. "Medication behavior record" is a structured event record added to the user's health record by the cloud data platform after receiving the data. Its fields include at least: user ID, device ID, unit identifier, timestamp, confirmation status, and source terminal type. This record serves as the basic data source for subsequent correlation analysis, compliance statistics, and early warning generation, and has a complete audit trail.

[0093] Step 5: Obtain physiological parameter measurement data collected by external medical testing equipment through the health monitoring module configured on the smart terminal; The "Health Monitoring Module" is a functional unit in the smart terminal dedicated to connecting, identifying, and receiving data from external devices. Its hardware interfaces include USB Type-A / B, Bluetooth 5.0 and above protocol stacks, and dedicated medical device proprietary wireless protocol adapter circuits. "External Medical Testing Devices" refers to portable home testing devices that comply with national medical device registration standards, such as electronic blood pressure monitors, blood glucose meters, thermometers, and pulse oximeters. Their output data formats follow IEEE 11073, HL7 FHIR, or manufacturer-specific specifications. "Physiological Parameter Measurement Data" refers to the numerical measurement results and metadata natively output by the device, such as "systolic blood pressure = 138 mmHg, diastolic blood pressure = 86 mmHg, heart rate = 72 bpm, measurement time = 2024-06-01T09:15:22". This data is not subject to algorithm correction or unit conversion during transmission; only a local timestamp generated by the main control unit and a device binding identifier are added.

[0094] In one alternative implementation, the acquisition method is as follows: the health monitoring module automatically scans and establishes a BLE connection with the paired blood pressure monitor via Bluetooth, and actively retrieves the latest set of measurement data after the detection device completes the measurement and enters broadcast mode; In another alternative implementation, the acquisition method is as follows: the health monitoring module identifies the connected blood glucose meter as a CDC type device through the USB interface, calls the standard CDC driver to read the ASCII format measurement message in its serial port buffer, and parses out the key parameter fields; Furthermore, the acquisition method is as follows: the health monitoring module is pre-loaded with multiple mainstream device communication protocol templates, and automatically matches the protocol stack according to the VID / PID of the access device to complete the entire process of handshake, authentication and data synchronization.

[0095] Step Six: Automatically associate and match the physiological parameter measurement data with the medication records within the corresponding time period to form correlation data between medication and physiological indicators; Users can customize the "corresponding time period," for example, by using the physiological parameter measurement time as a baseline, looking back 30 minutes and extending forward 90 minutes to form a time window. This window covers the typical drug onset time range, ensuring that the associated medication records have clinical interpretability. "Medication records" refer to all behavioral data entries uploaded to the cloud data platform whose timestamps fall within this window. "Automatic association and matching" is a data fusion operation performed by the main control unit or the cloud data platform, without relying on manual intervention. Its matching criteria are only the overlap of timestamps and the consistency of user IDs. "Associated data" are newly generated data objects, structurally based on physiological parameter measurement data, nesting and referencing one or more successfully matched medication behavior records to form a one-to-many mapping relationship of "one measurement - multiple medications," supporting subsequent aggregation and analysis of clinical indicators such as blood pressure trends and blood glucose fluctuations according to the medication regimen dimension.

[0096] Step 7: Receive user health records and alert information from the cloud data platform via a remote client for medical staff or caregivers to view; The "remote client" is an application running on mobile terminals or PCs, possessing user authentication, hierarchical data access control, and visualization rendering capabilities; the "user health record" is a unified data view formed by the cloud data platform integrating medication behavior records, physiological parameter measurement data, related data, and basic health information (age, medical history, allergy history), supporting display in various formats such as calendar, charts, and lists; the "early warning information" is a structured prompt generated by the cloud data platform based on a preset rule engine or a lightweight AI model, including but not limited to: continuous missed dose warnings, physiological indicator exceeding threshold warnings, medication and indicator abnormal correlation warnings (such as blood pressure rising instead of falling after a certain medication), and insufficient inventory warnings; all warnings are marked with the trigger time, related data ID, and confidence level.

[0097] Step 8: Based on health records or early warning information, perform remote intervention operations via a remote client; "Remote intervention operations" refer to controllable actions initiated by medical staff or caregivers through remote clients, which affect the user terminal or system backend. These actions include: sending voice / text reminders to the user terminal, adjusting medication plans and synchronizing them to the main control unit, initiating video consultation requests, retrieving historical data to generate assessment reports, authorizing the supply chain service platform to execute replenishment orders, and pushing consultation summaries to community doctor workstations. All intervention operations are logged in a complete log, including the operator's identity, operation time, operation content, and feedback on the execution results.

[0098] This application organically connects the previously isolated four aspects of medication management, health monitoring, risk warning, and remote intervention by triggering precise reminders and unlocking at preset medication times, achieving closed-loop confirmation of behavior using medication detection sensors, automatically linking multi-source data based on timestamps, and establishing a professional care intervention path through a remote client. Furthermore, by leveraging a cloud data platform as a unified data hub, a two-way controllable data and command flow is formed between the main control unit, health monitoring module, and remote client, ultimately constructing a four-in-one AI-powered medical, elderly care, and rehabilitation service methodology covering "family self-management—real-time terminal response—cloud-based intelligent analysis—remote medical and nursing collaboration."

[0099] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of an AI-based medical care, rehabilitation, and drug management service method.

[0100] In another aspect, this application also provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiments, or may exist independently and not assembled into the electronic device. The aforementioned computer-readable storage medium stores one or more programs that, when used by one or more processors, execute the AI-based medical and elderly care and drug management service method described in this application.

[0101] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order shown in the operation diagrams. For example, depending on the functions / operations involved, two consecutively shown blocks may actually be executed substantially simultaneously, or the blocks may sometimes be executed in reverse order. Furthermore, the embodiments presented and described in the flowcharts of this application are provided by way of example to provide a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented herein. Alternative embodiments are contemplated in which the order of various operations is changed and sub-operations described as part of a larger operation are executed independently.

[0102] The embodiments of this application have been described in detail above with reference to the accompanying drawings. However, this application is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of this application. Furthermore, unless otherwise specified, the embodiments and features described in the embodiments of this application can be combined with each other.

Claims

1. An AI-powered medical, elderly care, rehabilitation, and medication management service system, characterized in that: include: Smart terminals and cloud data platforms; The intelligent terminal includes a cabinet and a main control unit. The cabinet contains multiple intelligent medicine storage units, which are physically isolated from each other and electrically addressable and independently controllable. Each intelligent medicine storage unit is equipped with a locking mechanism, a status indicator component, and a medicine retrieval sensing sensor. The main control unit is configured to: send an unlock command to the corresponding target intelligent drug storage unit at a preset medication time and trigger an audio-visual reminder; after obtaining a valid medication retrieval signal through the medication retrieval sensing sensor, generate behavioral data containing a timestamp, target unit identifier, and medication retrieval confirmation information, and upload the behavioral data to the cloud data platform to form a closed-loop drug management mechanism from medication reminder, unlock control, medication retrieval confirmation to behavioral recording.

2. The AI-powered medical, elderly care, rehabilitation, and medication management service system according to claim 1, characterized in that, The smart terminal also includes a health monitoring module; The health monitoring module is equipped with a data interface for connecting to external medical testing equipment to obtain physiological parameter measurement data generated by the equipment, and sending the physiological parameter measurement data to the main control unit. The main control unit is configured to automatically associate and match the received physiological parameter measurement data with the medication records within the corresponding time range, and generate and output an associated data packet containing medication information and physiological indicator information.

3. The AI-powered medical, elderly care, rehabilitation, and medication management service system according to claim 2, characterized in that, It also includes a remote client that communicates with the cloud data platform; The remote client is configured to be accessible to healthcare professionals or caregivers to perform at least one of the following operations: View the corresponding user's health record; Receive early warning information from the system; Remote intervention operations are performed based on the health records or early warning information.

4. The AI-powered medical, elderly care, rehabilitation, and medication management service system according to claim 3, characterized in that, The smart terminal also includes a remote communication unit; The remote communication unit is used to establish a communication connection between the smart terminal and the remote client to support remote consultation interaction based on the user's health record.

5. The AI-powered medical, elderly care, rehabilitation, and medication management service system according to claim 1, characterized in that, The smart terminal also includes a temporary medicine storage area; the temporary medicine storage area is located inside the cabinet and is used to temporarily store medicines to be replenished to each smart medicine storage unit.

6. The AI-powered medical, elderly care, rehabilitation, and medication management service system according to claim 5, characterized in that, The system also includes a predictive replenishment unit; The predictive replenishment unit is configured to build a drug consumption prediction model based on historical drug usage data and use the model to predict future stockout times. The predictive replenishment unit is also configured to automatically trigger a linkage with the cloud service platform when the drug inventory reaches a preset predictive alarm threshold, generating prescription renewal or replenishment suggestion information; the prescription renewal or replenishment suggestion information includes at least the medication plan currently being implemented by the user.

7. The AI-powered medical, elderly care, rehabilitation, and medication management service system according to claim 5, characterized in that, The temporary storage area for tonics is equipped with a quantity detection unit. The main control unit is configured to receive the inventory signal sent by the inventory detection unit, and automatically generate and send a replenishment reminder to the user when it is determined that the current inventory is lower than a preset threshold. The system is configured to: after obtaining user authorization, generate an order with one click based on the replenishment reminder and medication plan information, and send the order information to the associated supply chain service platform to automatically start the drug delivery process.

8. The AI-powered medical, elderly care, rehabilitation, and medication management service system according to claim 1, characterized in that, The AI-powered medical, elderly care, and drug management service system according to claim 1 is characterized in that the intelligent terminal is equipped with a voice interaction unit; the voice interaction unit includes a microphone, a speaker, and a local voice processing chip, and supports both online and offline working modes; The voice interaction unit is configured to: respond to voice wake-up, accept user commands through natural voice dialogue, and execute corresponding operations.

9. A method for AI-powered medical care, rehabilitation, and medication management services, characterized in that, Using the AI-powered medical, elderly care, rehabilitation, and medication management service system according to any one of claims 1 to 8 includes the following steps: At the preset medication time, the main control unit of the smart terminal sends an unlock command to the target smart medication storage unit and activates an audio-visual reminder; The user's medication retrieval action is detected by the medication retrieval sensing sensor configured in the target intelligent medication storage unit; After a valid medication retrieval signal is detected, the main control unit generates behavioral data containing a timestamp, a corresponding unit identifier, and medication retrieval confirmation information. The behavioral data is uploaded to a cloud data platform to construct a medication behavior record; The health monitoring module configured on the smart terminal acquires physiological parameter measurement data collected by external medical testing equipment. The physiological parameter measurement data is automatically associated and matched with the medication records within the corresponding time period to form the correlation data between medication and physiological indicators; The user's health records and early warning information are received from the cloud data platform via a remote client, which can be viewed by medical staff or caregivers. Based on the health records or early warning information, remote intervention operations are performed through the remote client.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method of claim 9.

Images