Venipuncture training and assessment model

Abstract

The application provides a venipuncture training and examination model simulation arm module, which simulates the forearm structure of an adult, and is sequentially provided inside a simulation arm body from outside to inside with a simulation skin layer made of flexible high polymer elastic material and detachably installed on the outer surface of the simulation arm, a simulation fat layer provided inside the simulation skin layer and used for simulating the buffer resistance of subcutaneous fat tissue, a simulation muscle layer provided inside the simulation fat layer and used for supporting the arm shape and providing puncture resistance, and a simulation venous vessel network embedded between the simulation fat layer and the simulation muscle layer according to the real anatomical running, wherein the simulation skin layer, the simulation fat layer, the simulation muscle layer and the simulation venous vessel network are all configured as detachable modules, and can be replaced with combinations of different thickness, hardness and venous elasticity parameters according to the training requirements.

Description

Technical Field

[0001] This invention belongs to the field of intravenous puncture training, specifically relating to an intravenous puncture training and assessment model. Background Technology

[0002] Venipuncture is one of the most common and fundamental invasive procedures in nursing and clinical work. Currently, venous puncture models used for teaching and training mainly fall into two categories:

[0003] Single-layer material + single vein model; most commercial training arms use a single layer of silicone skin and several ordinary rubber tubes to simulate veins. The skin thickness, elasticity, and color are basically fixed, and the vein position is also fixed. Trainees mainly train in the single scenario of "finding a vein and getting blood return after insertion." These models have a simple structure and lack replaceable multi-layered tissue structures. They also cannot accurately simulate complex situations such as loose skin, thinning fat, and muscle atrophy in the elderly, or venous sclerosis in chemotherapy patients, or vein collapse in critically ill patients. The training content remains at a basic introductory level.

[0004] Simulated wrist and arm models with simple sensors; some newer models attempt to incorporate pressure sensors or simple switches to trigger indicator lights or sounds when a blood vessel is punctured, but these typically only detect whether the tube has been punctured. The number of sensors is limited, their distribution is coarse, and they lack precise data acquisition of key process parameters such as needle insertion angle, depth, force, speed, and hand stability. These models still primarily focus on "result prompts" rather than "process monitoring." Teaching evaluation relies mainly on teacher observation and subjective scoring, lacking traceable data support for student progress and types of errors.

[0005] The problems with simulated skin include rapid wear and tear and high maintenance costs. Existing models typically use ordinary silicone as the skin and blood vessel material. Under high-frequency training conditions, pinholes accumulate continuously, easily leading to obvious leakage and densely visible pinholes, requiring frequent replacement of skin sheaths and intravenous tubing. For universities and hospitals with heavy training workloads, the high maintenance costs and frequent downtime for replacements severely impact continuous teaching.

[0006] Training scenarios are limited, and bedside / interactive training capabilities are lacking. Most existing devices can only be placed on a table or simulated bedside as a "dummy arm," and trainees' interaction with the "patient" remains largely at the level of imagination. A very few products support attaching a local puncture pad to the human body, but these are usually just flat pads that cannot reproduce the complete three-dimensional structure of the forearm, and lack blood circulation and multi-point sensing support, making them difficult to use for systematic bedside puncture training.

[0007] There is a lack of systematic control and intelligent scoring. While some publicly available technologies mention using sensor data for evaluation, these are mostly localized attempts, such as only calculating success rates or roughly recording puncture time. No comprehensive solution has yet been found that integrates multi-layered tissue structures, controllable blood circulation, multiple sensors and self-healing materials, desktop / wearable dual-mode operation, and complex state machine control with multi-index scoring algorithms into a closed-loop training-assessment system.

[0008] Therefore, the existing technology has at least the following key problems that have not yet been solved: it is impossible to systematically simulate multiple typical patient scenarios (elderly, post-chemotherapy, critically ill with low blood pressure, etc.) and their corresponding venous physiological characteristics on a single model;

[0009] The current system suffers from several drawbacks. First, it cannot provide objective, multimodal, high-resolution monitoring of the entire puncture process, hindering scientific quantitative scoring. Second, the model is prone to wear and tear and requires frequent maintenance, failing to meet the demands of high-frequency training. Third, it cannot quickly switch between standardized desktop assessment scenarios and interactive real-person wearing scenarios, resulting in a fragmented training ecosystem. Fourth, it lacks a comprehensive solution that organically unifies structural design, sensor layout, blood circulation control, self-healing mechanisms, and intelligent scoring algorithms. This invention addresses these issues by providing a long-term, high-frequency, and intelligent venous puncture training and assessment model through the collaborative design of a multi-layered modular simulated tissue structure, a multi-scenario vein module, a sensor-control-scoring closed loop, and a material self-healing and dual-mode switching mechanism. Summary of the Invention

[0010] The purpose of this invention is to provide a long-term, high-frequency, and intelligent venous puncture training and assessment model through the collaborative design of a multi-layered modular simulated tissue structure, a multi-scenario vein module, a sensing-control-scoring closed loop, and a material self-healing and dual-mode switching mechanism; including:

[0011] The simulated arm module, which mimics the structure of an adult forearm, consists of the following components arranged sequentially from the outside to the inside of the simulated arm body:

[0012] The simulated skin layer is made of flexible polymer elastic material and can be detachably installed on the outer surface of the simulated arm.

[0013] The simulated fat layer is placed inside the simulated skin layer to simulate the buffering resistance of subcutaneous fat tissue;

[0014] The simulated muscle layer is located inside the simulated fat layer to support the shape of the arm and provide resistance to puncture.

[0015] The simulated venous network is embedded between the simulated fat layer and the simulated muscle layer according to the actual anatomical course. The simulated skin layer, simulated fat layer, simulated muscle layer and simulated venous network are all constructed as detachable modules, which can be replaced with different combinations of thickness, hardness and venous elasticity parameters according to training needs.

[0016] The simulated blood circulation system includes a reservoir, a micro-liquid pump, a one-way valve, and an elastic buffer element installed in a desktop base. The reservoir is connected to the simulated venous network through a flexible tube. Under the control of the control module, the micro-liquid pump provides simulated blood with adjustable pressure and flow to the simulated venous network. The one-way valve and elastic buffer element are used to stabilize the circulation pressure and prevent backflow.

[0017] The desktop base and quick-connect mechanism are described. The desktop base is a box structure that houses the simulated blood circulation system and control module. The upper surface of the base is provided with a positioning bracket and a quick-connect interface that integrates electrical connections and liquid channels. The simulated arm module can be mechanically fixed to the base, connected to the liquid channel, and connected to the electrical signal through the quick-connect interface. The quick-connect interface has a foolproof and self-sealing structure to enable quick assembly and disassembly of the simulated arm module and leak-free liquid switching.

[0018] A multi-sensor module is used for multi-modal monitoring of the puncture procedure, including:

[0019] Pressure and flow sensors are placed on a simulated venous network or its connecting pipes to detect changes in venous pressure and flow over time.

[0020] An array of tactile / position sensors, placed at different depths in the simulated skin and simulated fat layers, is used to detect the needle tip's contact with the skin, the depth of insertion, and its relationship with the blood vessel position.

[0021] An attitude sensor mounted on the puncture needle or needle handle is used to detect the spatial tilt angle, direction, and motion stability of the needle.

[0022] Temperature sensors are placed near the simulated skin layer and simulated veins to monitor the local temperature of the model;

[0023] The control and evaluation module, located in a desktop base or portable control box, is electrically connected to the multi-sensor module and the simulated blood circulation system, and is used for:

[0024] The state machine process executes the following steps: startup self-test, standby, needle tip contact detection, needle insertion monitoring, blood vessel positioning and blood return detection, puncture success / failure determination, exit, and material self-repair control.

[0025] The micro liquid pump automatically adjusts its output pressure, valve opening and closing status, the tension of simulated skin and blood vessels, and the working status of heating elements based on sensor signals such as pressure, flow rate, position, posture, and temperature, in order to simulate physiological phenomena such as venous filling, blood vessel rolling, and collapse in different patients.

[0026] After the puncture, multiple characteristic indicators such as needle insertion angle, puncture force, needle insertion time / speed, blood return waiting time, number of attempts and hand stability are normalized and a comprehensive score is calculated according to preset weights.

[0027] The venipuncture training and assessment model also includes a self-healing control unit, which controls an embedded heating element to locally heat the simulated skin layer and / or simulated venipuncture material to the target temperature and maintain it for the required time when the puncture damage meets a preset threshold. This allows the self-healing polymer material to complete the needle hole healing during the heating and cooling process, thereby restoring the model's airtightness.

[0028] Furthermore, the simulated skin layer has at least three replaceable skin modules, which are used to simulate the skin mode of the elderly, the skin mode of long-term chemotherapy patients, and the skin mode of critically ill patients. The different skin modules have differences in thickness, elasticity, surface texture and color, as well as thermal conductivity. When the corresponding training mode is loaded, the control module adjusts the target temperature of the skin surface and the baseline pressure in the vein at the same time.

[0029] Furthermore, the simulated venous network includes multiple sets of pluggable venous modules. Each set of venous modules corresponds to different tube diameters, wall thicknesses, and material hardness, and is used to simulate normal veins, veins prone to collapse in the elderly, and hardened veins, respectively. The control module uses pressure sensors to detect the vascular collapse process and provide corresponding prompts in the corresponding training mode by setting different initial filling pressures and collapse trigger times.

[0030] Furthermore, the quick-connect mechanism of the desktop base includes a mechanical locking structure and a self-sealing liquid quick connector. The mechanical locking structure is used to limit the installation of the simulated arm body in a posture with the elbow slightly bent and the palm facing upward. The self-sealing liquid quick connector automatically closes when the simulated arm is not inserted and automatically opens to connect the reservoir and the venous network after the simulated arm is inserted and positioned.

[0031] Furthermore, the model further includes a wearable mode switching structure, which includes:

[0032] The arc-shaped groove and soft padding on the back of the simulated arm body are designed to fit the surface of a real forearm.

[0033] At least two adjustable straps are used to bind the simulated arm body to the real forearm;

[0034] The portable control box is connected to the main body of the simulated arm via cables and catheters. The control module automatically switches to wearable mode by recognizing the interface status, reducing the basal pressure in the vein to a preset value, turning off the body's audio-visual prompts, and enabling wireless communication, sending sensor data and scoring results to an external terminal.

[0035] Furthermore, the tactile / position sensor array in the multi-sensor module is arranged in multiple layers along the thickness direction of the simulated skin layer and simulated fat layer, and is arranged in a ring around the simulated vein. It is used to generate different combinations of trigger signals in sequence when the needle tip contacts the skin, penetrates to different depths, and presses against the outer wall of the blood vessel. The control module determines the depth of the needle tip and its relative position to the blood vessel based on the array signals.

[0036] Furthermore, the control module uses data from the attitude sensor to calculate the angle between the puncture needle and the simulated skin, and sets an angle deviation threshold. When the needle insertion angle is detected to exceed the preset range, a prompt message is issued in training mode, and points are deducted from the angle index during scoring.

[0037] Furthermore, in the state of blood vessel positioning and blood return detection, the control module comprehensively determines whether the needle tip has pierced the anterior and posterior walls of the vein based on the instantaneous pressure change detected by the pressure sensor, the flow change waveform detected by the flow sensor, and the triggering sequence of the tactile / position sensor array. When puncture is determined, an alarm is triggered and recorded as a serious error event.

[0038] Furthermore, the scoring module employs a linear weighted scoring model, normalizing features such as needle insertion angle error, puncture force, needle insertion time or speed, blood return waiting time, number of attempts, and hand stability into a score fi ranging from 0 to 1, and then calculating the score according to S=Σw. i ·f i The comprehensive score S is calculated in the following way, where wi is a preset or trained and optimized weight coefficient.

[0039] Furthermore, the scoring module uses machine learning methods to optimize the weighting coefficients w. i The scoring threshold is adaptively adjusted, specifically by using historical training data and expert human ratings as samples, and optimizing w using a supervised learning algorithm. i This minimizes the error between the model's output score and the expert score, thereby achieving iterative optimization of the scoring model.

[0040] Compared with existing venipuncture training models, this invention achieves substantial technical progress in at least the following aspects:

[0041] This invention utilizes a multi-layered, replaceable tissue structure with various vein modules to construct realistic simulations across multiple scenarios. Instead of simply embedding a tube within a silicone block, it constructs a multi-layered, replaceable structure encompassing skin, fat, muscle, and veins. Through combinations of these four modules, it achieves engineered representations of venous condition differences in various typical populations, including the elderly, chemotherapy patients, and critically ill patients with low blood pressure. Compared to existing single-layer models, this invention allows for efficient switching between multiple training scenarios on a single platform, significantly improving the coverage and realism of the training content.

[0042] This invention integrates a standardized desktop setup with interactive live-action wearable technology, bridging different training levels. While existing technologies often separate desktop models and wearable pads into two distinct systems, this invention utilizes a quick-connect interface and a portable control box to design a complete simulated forearm module that can be fixed to a desktop base for standardized assessments or detached and strapped to a live forearm for bedside interactive training. The control module automatically identifies the mode and adjusts pressure supply, communication, and prompting strategies, allowing trainees to seamlessly transition from standard operation to more clinically relevant scenarios using the same equipment, creating a continuous training gradient – ​​a synergistic effect difficult to achieve with existing devices.

[0043] Deep fusion of multiple sensors enables a leap from "only looking at the result" to "quantifying the entire process." This invention deploys various sensors, including pressure, flow, tactile / position, posture, and temperature sensors, in a simulated tissue and circulatory system. Through the state machine of the control module, these sensor information are mapped one-to-one with the puncture steps, enabling the system to identify key events such as contact, needle insertion, approaching the blood vessel, puncturing the blood vessel, blood return, penetrating the vessel, and blood vessel collapse. Compared to traditional models that can only determine "whether there is blood return," this invention obtains high temporal resolution, multi-dimensional, full-process data, providing a data foundation for objective teaching assessment, skills analysis, and teaching improvement.

[0044] The built-in scoring model and AI-based scalable mechanism transform assessment from subjective experience to data-driven evaluation. This invention not only collects process data but also implements a configurable multi-indicator weighted scoring model in the control module. This model converts indicators such as needle insertion angle, force, time, blood return speed, number of attempts, and stability into quantitative scores, outputting a total score and sub-item analysis. This transforms the previous "expert-based scoring" model into a "data-driven objective evaluation." Furthermore, by optimizing weights and thresholds through machine learning, the scoring system can continuously self-adjust with data accumulation, adapting to the needs of different departments and teaching objectives. This collaborative evolution mechanism of "equipment + algorithm" significantly expands the intelligence level of the training system.

[0045] Self-healing materials and a self-controlled repair process address the durability bottleneck under high-frequency training. At the material level, this invention uses polymer materials with reversible cross-linked structures or self-healing microcapsules as the simulated skin and vein materials. Under the self-repair control of the control module, the material's self-healing is actively triggered through a heating-insulating-cooling process, allowing for the periodic repair of pinhole micro-damage and maintaining the skin and blood vessels' seal and appearance integrity. Compared to traditional "use and replace" skin sleeves, this invention significantly extends the lifespan of a single assembly, reduces consumable costs, and ensures the simulation quality of each training session.

[0046] This invention employs a multi-dimensional, synergistic approach integrating structure, sensing, control, materials, and algorithms, achieving a holistic effect far beyond simple additive integration. It doesn't merely add sensors, materials, or scoring functions to existing models; instead, it employs a comprehensive, collaborative design across multiple dimensions, including simulated tissue structure design, multi-scenario vein modules, blood circulation control, multi-sensor fusion, self-healing materials, and scoring algorithms. Multi-layered replaceable tissue provides the foundation for multiple scenarios, sensing and blood control ensure dynamic controllability, self-healing materials guarantee long-term usability, and scoring and AI optimization create a closed loop for training and assessment. This cross-structure-control-material-algorithm combination is unprecedented in existing technologies, and its overall technical effect far surpasses the effect of simply adding individual components. In summary, this invention significantly outperforms existing intravenous puncture training devices in five dimensions: simulation depth, multi-scenario reproduction, process monitoring, intelligent evaluation, and long-term usability. It possesses outstanding substantive features and significant advancements, providing medical students and nurses with a highly realistic, data-rich, and sustainably usable intravenous puncture training and assessment platform. Attached Figure Description

[0047] Figure 1 This is a block diagram of the overall system structure of the present invention.

[0048] Figure 2 This is a schematic diagram showing the connection relationship between the desktop base, the simulated arm assembly, and the simulated blood circulation fluid circuit of the present invention. Figure 3 This is a schematic diagram of the layout of infusion, drainage and pressure detection points in the simulated arm assembly of the present invention. Figure 4 This is a schematic diagram showing the connection relationship between the control unit of the present invention and the infusion pump, the drainage pump, the mixing / heating device, the pressure / flow sensor, and the air valve / pump. Figure 5 This is a schematic diagram of the infusion, maintenance, drainage, and parameter iteration control process of the present invention. Detailed Implementation

[0049] The intravenous puncture training and assessment model provided by this invention simulates the structure of an adult forearm, comprising two main parts: a simulated arm body and a desktop base. The simulated arm body employs a modular, multi-layered structure, including (from the outside in) a simulated skin layer, a simulated fat layer, a simulated muscle layer, and a simulated venous network arranged anatomically. Each layer is connected by adhesive or mechanical fixation and is designed as a replaceable module to adapt to different training scenarios. For example, by replacing skin or vascular modules with different hardness, different venous conditions can be simulated in elderly patients, patients undergoing long-term chemotherapy, or patients in shock.

[0050] Simulated skin layer: Made of soft medical-grade silicone or thermoplastic elastomer material, with a surface texture and skin tone similar to real skin. The thickness and firmness of the skin layer can be adjusted according to the training mode: for example, a thinner, wrinkled, and loose skin is used in the elderly mode; a slightly thicker and less elastic skin is used in the chemotherapy patient mode, with a firmer feel to reflect the sensation of hardening; the skin in the critical care mode is made paler and has a lower initial temperature to simulate the cold and pale skin of a shock patient. The simulated skin layer is tightly fixed to the outside of the arm body by means of buckles, zippers, etc., and its tension and looseness can be adjusted and it is easy to remove and replace.

[0051] Simulated fat layer: Located inside the skin layer, it is made of medium- to low-hardness silicone gel or soft foam material, simulating the soft touch and initial puncture resistance of subcutaneous adipose tissue. The thickness of the fat layer varies, generally from a few millimeters to about 1 centimeter: the fat layer is thinner in elderly or thin patients; the fat layer is thicker in young or obese patients, providing significant cushioning resistance. This layer is fixed between the skin and muscle layers by adhesive or interlocking methods.

[0052] Simulated muscle layer: Located just beneath the fat layer, this layer is made of high-rigidity silicone or dense foam material, simulating the elasticity and tension of forearm muscle tissue and providing shape support for the arm. The muscle layer determines the puncture feel: a firmer muscle layer increases needle resistance and provides a "pushing" sensation, while a softer layer reduces resistance, making it easier for the needle to penetrate. Different muscle modules with varying hardness / thickness can be used in different modes: for example, the muscle layer is softer and thinner in the elderly mode, while it is firmer in the young and healthy mode to increase resistance. The muscle layer usually contains a supporting skeleton or rods to maintain the shape of the forearm and the joint bending angle (e.g., the arm is slightly bent, and the palm is facing upwards), preventing excessive deformation of the simulated arm during forceful puncture.

[0053] Simulated venous network: Simulated venous tubing made of highly elastic silicone or latex material is pre-embedded and distributed within the muscle / fat layer according to real anatomical locations. The venous tubing is approximately 3-4 mm in diameter, with thin, flexible walls, and can be filled with simulated blood. The course of major veins matches that of real people (such as the basilic vein and cephalic vein in the elbow crease, and the veins in the mid-forearm). The simulated vessels are installed in a modular manner, facilitating the replacement of vessels with different characteristics: the elderly mode uses thinner-walled, more easily flattened venous tubing, which is more loosely fixed in the tissue and prone to rolling and slipping; the sclerotic vessel mode uses tubing with slightly thicker walls and higher rigidity to simulate vascular fibrosis and reduced elasticity; the critical low-pressure mode uses a smaller diameter tubing and lower initial internal pressure, making vessel collapse more likely. Both ends of the simulated vessels are connected to a reservoir and a fluid pump via flexible catheters, forming a closed simulated blood circulation loop. A one-way valve is installed in the catheter to control the unidirectional circulation of fluid. After the needle successfully pierces the vein, the simulated "blood" in the reservoir will flow back along the needle tube under pressure. The trainee can clearly see the backflow of blood at the end of the syringe or catheter, verifying a successful puncture. After the needle is withdrawn, the simulated blood vessel automatically closes the puncture hole due to its own elasticity, preventing simulated blood leakage. The same site can be punctured repeatedly without significant leakage or leaving a visible needle mark.

[0054] Simulated Blood Circulation System: The base houses a reservoir to hold simulated blood (such as an aqueous solution with added red dye and preservatives to prevent spoilage). A miniature liquid pump provides continuous circulation power and intravenous static pressure. One-way valves and elastic buffers in the tubing work together to produce relatively smooth fluid flow. In normal mode, the controller sets the baseline intravenous pressure at approximately 12–15 cmH2O to ensure venous filling; in elderly or hypotensive modes, the internal pressure can be reduced to 8–10 cmH2O to simulate low blood volume; for simulated arterial puncture training, the pressure can be significantly increased and pulse fluctuations can be superimposed. The entire fluid system is sealed and safe, and combined with the self-sealing properties of the simulated blood vessels, it ensures no leakage after needle removal. The application of self-healing materials allows the model to withstand hundreds of punctures without leakage and with minimal needle marks on the surface, not affecting subsequent training results.

[0055] Desktop Base: The base is a robust box-like support structure that integrates the control unit, power supply, motor / hydraulic pump, valves, and other components. The bottom of the base has anti-slip pads or optional vacuum suction cups and clamps to secure the model to the desktop, preventing it from sliding when the trainee applies force. The top surface of the base features a quick-connect interface and positioning bracket for the simulated arm: the interface includes a multi-core electrical connector and a liquid line connector, featuring a foolproof design (specific shapes and notches ensure correct alignment). When the simulated arm assembly is inserted into the base interface, the electrical connector simultaneously connects all sensor signal lines and actuator cables, and the liquid connector automatically connects the circulation loop (an internal valve automatically closes to prevent liquid leakage when removed). Mechanical locking securely fixes the simulated arm in the positioning slot of the base, ensuring the arm is installed in a standard position (e.g., elbow slightly bent, palm facing upwards at an angle to the base surface), consistent with the patient's posture during clinical puncture. Installation and removal require no tools; a simple plug and press is all that's needed. During training, the base powers the various modules within the simulated arm and transmits sensor data to the main controller via control cables, achieving overall coordination.

[0056] Wearable Mode Switching: This model supports detaching the simulated arm component from the base and strapping it to the forearm of a real volunteer for scenario training. The switching steps are as follows: First, turn off the device power and liquid pump, release the latch on the base, and pull the simulated arm out of the base interface (the electrical and liquid interfaces are automatically disconnected and sealed upon removal). Then, attach the simulated arm component to the predetermined location on the real forearm (usually the volunteer's non-dominant arm, aligning the simulated vein with its anatomical position in the elbow crease area). The back of the simulated arm has soft padding and curved grooves to conform to the curvature of the real arm for increased comfort and stability. Use the included adjustable straps (Nylon Velcro or elastic straps with anti-slip silicone strips on the inside) to secure the proximal and distal ends of the simulated component to the volunteer's arm, ensuring a moderate tightness that prevents slippage without affecting blood circulation. Once secured, the simulated vein module conforms to the real limb, providing the trainee with a discreetly installed simulated blood vessel on the arm of the "patient" they are facing for practice.

[0057] Simultaneously, the control and fluid supply functions within the base are switched to the portable control box. The portable control box is a small, portable device containing the main controller, battery pack, micro-pump, and valves, connected to the simulator arm assembly via extension cables and conduits (using the same plugs and connectors as the base). The trainee can wear the control box on their person or have an assistant hold it beside them. Once connected, turning on the control box power will power the system back to standby mode. The controller can automatically identify the current mode (e.g., by using detection pins on the connection interface to determine whether it's in base mode or portable mode) and make corresponding adjustments.

[0058] Reduce intravenous pressure: In a wearable setting, appropriately reduce intravenous pressure and flow rate to prevent accidental leakage during volunteer movements and to simulate low blood pressure caused by patient stress or hypotension.

[0059] Switch communication method: Enable wireless communication module (such as Bluetooth or WiFi) to transmit data to the teacher's computer / tablet to replace the wired connection of the base station and avoid the connection affecting the activities of the students.

[0060] Turn off the on-board prompting device: To ensure the realism of the scenario, the indicator lights or buzzers on the model can be automatically silenced and turned off in wearable mode, and instead send prompts only to the instructor's terminal to prevent volunteers or trainees from being disturbed by unrealistic prompts during training.

[0061] Adjusting sensor parameters: Considering the increased environmental interference when worn by real people, the controller can slightly adjust the sensor sensitivity and filtering algorithm to avoid false alarms caused by subtle movements of volunteers. For example, the threshold for pinpoint contact detection can be relaxed, or motion noise filtering can be added to the posture sensor data.

[0062] After the above switching, the model can be used in dynamic scenarios where a live person plays the role of a "patient." Trainees need to interact with live volunteers according to clinical guidelines (such as verifying identity and providing reassurance), and then perform a puncture operation on the arm equipped with the simulation module. Throughout the process, the sensors of the simulation module record the operation data in real time and transmit it wirelessly to the instructor's device so that the trainee's operation can still be monitored and scored. The mode switching mechanism has been optimized to allow for rapid (approximately 1-2 minutes) and reliable connection between independent use on a desktop and use by a live person, without affecting the model's functionality or the continuity of the assessment.

[0063] Sensor layout and detection principle

[0064] The model integrates various types of sensors to monitor key parameters of the puncture procedure in real time. These sensors are distributed across different layers of the simulated arm and the circulatory system, and their installation locations and working principles are as follows:

[0065] Pressure sensors are installed at multiple locations within the simulated vascular circuit to monitor pressure changes within the fluid path. A typical location is at the outlet of the reservoir or at the connection between the pump and the simulated blood vessel, measuring the overall supply pressure. Another location can be placed on the simulated blood vessel wall or within a small cavity connected to it, sensing local vascular pressure. When the needle tip presses against the skin and contacts the blood vessel, the increased local tissue pressure is transmitted to the blood vessel, producing subtle pressure changes. At the moment the needle punctures the blood vessel, the intravascular pressure experiences instantaneous fluctuations (possibly rising first and then falling). If the needle fails to enter the blood vessel, leading to tissue leakage, the intravascular pressure may continuously decrease, approaching zero. All of these signals are captured by the pressure sensors and converted into electrical signals, which are then sent to the controller. The pressure sensors are small, highly sensitive MEMS piezoresistive sensors, connected to the blood vessel via a thin catheter or directly attached to the blood vessel surface, enabling precise detection of pressure surges. The controller analyzes the pressure-time curve to determine key events during the puncture process: such as contact with the blood vessel (slight pressure increase), puncture of the blood vessel (sudden rise followed by a sharp drop and stabilization), and vascular collapse (pressure gradually approaches zero).

[0066] Flow sensor: Installed in a simulated blood circulation tubing (such as the tubing from a miniature pump to the inlet of a simulated blood vessel), it is used to detect fluid velocity and cumulative flow. A miniature turbine flow meter or ultrasonic flow meter can be used. During normal circulation, the flow sensor provides a baseline reading; when backflow or leakage occurs, the flow reading will exhibit characteristic changes: When the needle successfully penetrates the vein and backflows, some fluid flows out of the circulation via the needle tube, causing a temporary decrease or fluctuation in the main circuit flow; if the needle does not enter the blood vessel and fluid seeps into the tissue, the pump output may be obstructed, causing a decrease in flow or abnormal fluctuations. The controller combines data from both the pressure and flow sensors to accurately determine the puncture result: for example, a significant increase in pressure without a corresponding change in flow indicates that the needle tip is pressing on the blood vessel and has not penetrated, preventing fluid from flowing out (the blood vessel may be blocked or the needle position may be incorrect); if the pressure drops suddenly and the flow decreases, it indicates that fluid has momentarily flowed out through the needle (backflow has occurred).

[0067] Position / Tactile Sensor Array: Multiple miniature pressure sensing units are arranged at different depths under the skin of the simulated arm, forming a layered matrix to sense the position, depth, and contact with blood vessels of the needle tip. For example, a first layer of flexible pressure sensing film is embedded at a depth of about 5 mm under the skin, and a second layer of sensing units is arranged at a deeper depth (e.g., 10 mm); in addition, tactile sensors are also attached in a ring around the outer wall of the simulated vein. As the needle tip penetrates layer by layer, the sensing units at different depths are triggered sequentially: the surface unit detects pressure on the skin surface, indicating that the needle tip has contacted the skin; after the pressure is applied and the skin is pierced, the surface pressure is quickly released and the subsurface unit begins to sense the presence of the needle tip, indicating that the needle has entered the subcutaneous fat layer; further penetration compresses the deep units, indicating that the needle tip has reached the vicinity of the muscle layer. If the needle tip is just touching the outer wall of the simulated blood vessel without piercing it, the sensors surrounding the blood vessel will detect a local pressure signal peak. This tactile array is essentially a multi-point pressure sensing network. By analyzing which units are triggered and the triggering sequence, the controller can calculate the current depth of the needle tip and its position relative to the blood vessel. For example, if the sensing unit above the blood vessel is not triggered but one side of the unit has a signal, it indicates that the needle tip is deviating from the central axis of the blood vessel. The combination of multi-layer sensor data can help determine whether the needle tip is aligned with the target vein and whether the insertion depth is appropriate.

[0068] Attitude sensor: A miniature inertial measurement unit (IMU) containing a three-axis gyroscope and a three-axis accelerometer is installed on the puncture needle or its handle to measure the needle's tilt angle, orientation, and acceleration in space in real time. The IMU data is fused and calculated to obtain the needle's tilt angle and orientation angle relative to the horizontal plane, thereby estimating the needle tip's entry angle relative to the simulated skin surface. During initialization, the controller calibrates the IMU's zero-position orientation with a reference plane on the simulated arm surface, thus obtaining the actual needle entry angle (e.g., the angle between the needle axis and the skin plane) during operation. In embodiments where necessary, an auxiliary camera can also be installed above the model to identify the spatial attitude of the needle (marked points on the needle) using computer vision, cross-calibrating it with the IMU data to improve angle measurement accuracy. The attitude sensor also records the speed and stability of the needle's movement: the needle tip advance speed is estimated by integrating the acceleration signal, and the amplitude of hand tremors is assessed by high-frequency changes in the gyroscope. Therefore, the system can acquire quantitative information such as "needle tip tilt angle 20°, 5° to the left of the direction of blood vessel" for scoring, and can also be used for real-time prompts during training (such as issuing warnings when the angle deviation is too large). In addition, the subtle hand tremor data captured by the IMU can also reflect the stability and tension level of the trainee during operation.

[0069] Temperature sensors, embedded within the simulated skin layer and near simulated blood vessels, monitor local temperature. On one hand, the model is equipped with surface heating elements to simulate a patient's body temperature. The temperature sensors provide feedback signals, which the controller uses to adjust the heating power to maintain the skin surface temperature at approximately 37°C (slight adjustments are made in different modes, such as approximately 35°C in the elderly mode and approximately 34°C in the critical care mode to simulate hypothermia). On the other hand, in the material self-healing mode, these sensors monitor the temperature distribution during the heating process, ensuring that the skin or blood vessels are heated to the threshold that triggers self-healing (typically around 50°C) without excessively high temperatures damaging the material. The temperature sensors utilize thermistors or digital temperature chips, are small in size for easy embedding within the silicone layer, and do not affect tactile sensation. Multi-point temperature monitoring enables the controller to implement closed-loop control of the heating process, precisely and safely executing the healing operation of the simulated material.

[0070] All the aforementioned sensors are connected to the interface circuitry inside the simulated arm via flexible flat cables, and ultimately connected to the main controller via quick-connect plugs. For ease of maintenance, some sensors are designed as pluggable sub-modules (e.g., the simulated blood vessel module has its own pressure sensor and connector). Replacing this module updates the sensor, reducing calibration work. Upon power-up, the controller runs a self-test and calibration procedure: sequentially reading and zeroing the static zero point of each pressure sensor, calibrating the IMU reference attitude, and resetting the tactile array baseline value. If a sensor is found to be unresponsive or its reading exceeds a reasonable range, the system will mark the fault state, issue an alarm via indicator light / interface, and prohibit entry into training mode to prevent erroneous data from interfering with operational safety. Only when all sensors are functioning normally is the trainee allowed to begin puncture practice. Through the combination of these sensors, the system obtains comprehensive information during the puncture process, providing a reliable basis for subsequent control feedback, success determination, and scoring evaluation.

[0071] Control logic flow (state machine)

[0072] The model's main control unit runs dedicated control software to monitor and intelligently determine the entire puncture training process. A phased finite state machine control logic is employed, dividing the puncture operation into several states and setting clear trigger conditions and system responses. The main states and processes are as follows:

[0073] 1. Mode settings and startup status

[0074] After the device is powered on or reset, it enters the startup state. At this time, the system performs a self-test initialization: checking the working status of each sensor and actuator, initializing parameters (such as motor zero position, sensor calibration), and establishing communication with the host computer. If the self-test passes, the indicator light will show normal, and the system will load the default training mode or wait for the user to select a mode; if the self-test fails, a red light will illuminate or an error message will be displayed on the interface, and the system will remain locked until the fault is resolved.

[0075] During the mode selection phase, instructors or trainees can choose preset training modes (such as "intravenous infusion for elderly patients," "routine intravenous blood collection for adults," and "puncture for critically ill patients in the ICU") through the software interface. Upon receiving the mode command, the controller enters the parameter setting sub-state: based on the selected mode, it retrieves the corresponding settings from the built-in parameter table, including simulated vascular target pressure, vascular wall elasticity coefficient, collapse judgment threshold time, skin target temperature, and force feedback threshold. Then, the controller drives the actuator to complete the preparatory actions.

[0076] Adjust the output of the liquid pump to gradually bring the pressure inside the simulated blood vessel to the set value and stabilize it at that level.

[0077] If the model has an adjustable tension mechanism, the fixation of the simulated skin and veins can be relaxed or tightened as needed (for example, in the elderly mode, the fixation of the skin and blood vessels can be slightly relaxed to make them easier to slide; in the normal mode, the veins can be moderately tightened to keep them fixed; in the special low-pressure mode, the blood vessels may be slightly pulled to reduce the risk of collapse).

[0078] The surface heater is activated to slowly heat the skin surface to a simulated body temperature range (e.g., around 36°C). The temperature sensor provides continuous feedback, and the controller switches to constant temperature control once the target temperature is reached.

[0079] Switch the main control software state machine to standby mode and provide a prompt (such as the interface displaying "Ready" or the indicator light turning to standby signal) to wait for the student to start operating.

[0080] 2. Standby mode

[0081] In standby mode, the system is in idle monitoring mode. All sensors poll in real time, but the actuators (such as pump flow rate and robotic arms) remain in a steady state and do not perform any active actions. The display screen or host computer interface will display "System ready, waiting for puncture." At this time, the trainee can perform routine pre-puncture preparations (such as checking the needle, disinfecting the dummy skin, and putting on gloves). The standby state will continue until signs of the trainee starting puncture are detected or a clear start command is received.

[0082] The conditions for triggering the standby state transition can be varied. In manual mode, the student typically presses the start button or foot switch to notify the system to enter the next state. In intelligent detection mode, the system can automatically detect the start of the puncture action, such as the event of the needle tip contacting the skin. When the tactile sensor array detects continuous pressure on the skin surface accompanied by local temperature fluctuations (contact body temperature), it can be determined that the needle tip has touched the model skin, meeting the conditions for starting the operation. At this time, the controller switches the state from standby to contact detection state.

[0083] 3. Needle tip contact detection status

[0084] When the trainee's needle tip first touches the simulated skin surface, the model enters contact detection mode. The trigger is primarily provided by sensors: when the pressure sensor reading on the subcutaneous surface exceeds a preset micro-threshold, it is considered that the needle tip has made contact with the skin. The controller records the contact timestamp and immediately provides feedback signals to remind the trainee: for example, the interface displays "Needle tip has touched skin," or a prompt sound is emitted. This confirms to the trainee that the correct location has been touched and marks the official start of the puncture. The system then takes several measures to facilitate subsequent operations:

[0085] If the model is equipped with an electric propulsion / force feedback device, the controller will now drive the needle pusher into a compliant standby mode, which reduces resistance to the needle, allowing the learner to maintain autonomous control over the needle tip without applying excessive additional force. This is to prevent excessive damping of the device at the moment of contact from affecting the learner's feel.

[0086] The system begins to closely monitor the continuity of pressure and tactile signals. If the contact is unstable (the sensor signal appears / disappears repeatedly, indicating that the needle tip may be about to leave), the controller can display a message on the interface such as "Needle tip not in stable contact, please keep your hand steady," instructing the trainee to keep the needle tip in contact with the skin before proceeding further.

[0087] Once in contact mode, if the needle tip remains in place for too long without further insertion (e.g., more than 5 seconds without triggering a deeper sensor), the controller will issue a gentle reminder in training mode: "Please insert the needle slowly." This avoids excessive hesitation among trainees and helps them develop decisive and smooth operating habits in clinical practice (prolonged lingering increases patient discomfort).

[0088] When the system detects that the student continues to apply pressure, causing the needle tip to break through the skin (the pressure of the surface sensor suddenly drops and the secondary sensor begins to give a signal), it is determined that the needle tip has penetrated the subcutaneous tissue, and the system status changes to the needle insertion state.

[0089] 4. Needle insertion state (needle insertion is performed)

[0090] After the needle tip pierces the skin and enters the subcutaneous tissue, the model enters the needle insertion state. During this stage, the needle is allowed to advance into the target tissue. The system primarily monitors and records various parameters during the insertion process, while also providing auxiliary feedback as needed. Entering the needle insertion state is indicated by a sudden drop in the superficial skin sensor signal and the triggering of the subsurface sensor array, or by user confirmation of "insertion achieved" on the interface.

[0091] During needle insertion, the controller performs the following control and recording actions:

[0092] Depth monitoring: Based on the layers triggered by the tactile array sensors, the depth range of needle tip penetration is estimated in real time. For example, if the 5mm and 10mm layers are triggered sequentially, it indicates that the needle tip is advancing from the fat layer to the muscle layer. The controller provides this depth information to the interface for display (approximate insertion depth can be displayed in training mode) and also stores it in the log for scoring analysis.

[0093] Angle Monitoring: The posture sensor continuously uploads needle axis tilt angle data. The controller compares the real-time angle with an ideal reference angle (e.g., approximately 20° to the skin plane). If the deviation is significant, the training mode interface will highlight the current angle value in red or indicate "Needle insertion angle too steep / too flat"; the examination mode will not provide this indication, but will record the magnitude of the deviation for scoring. This encourages trainees to maintain a standardized insertion angle, avoiding excessively flat angles that could lead to subcutaneous penetration or excessively steep angles that could puncture blood vessels.

[0094] Speed ​​and Force Monitoring: The controller estimates the needle tip advance speed based on IMU acceleration data and infers changes in insertion resistance based on pressure sensor readings. If an abnormally high needle tip speed is detected (indicating a possible forceful insertion), the system will immediately issue a voice prompt, "Insertion too fast, please slow down"; if the speed is too slow and there is a prolonged pause, it will prompt, "Insertion too slow, be decisive but steady." Simultaneously, if the monitored force / pressure signal shows a sudden spike, indicating excessive force, the controller will also alert the trainee to adjust the force in training mode. All this data is meticulously recorded for later quality assessment.

[0095] Force feedback control (if applicable): For models equipped with mechanical resistance control, the controller dynamically adjusts the resistance based on sensor data during needle insertion. Typically, resistance is reduced at the moment of skin puncture to allow for smooth needle entry, and thereafter, during subcutaneous advancement, the system maintains appropriate damping to simulate tissue resistance. When the force sensor detects a significant change in resistance (e.g., a brief period of emptiness upon puncturing the fascia / blood vessel wall), the controller adjusts the feedback force accordingly, allowing the trainee to feel the change in resistance to closely resemble the feel of a real puncture.

[0096] During this stage, the system does not interfere with the trainee's actual operating path, but focuses on recording objective data and providing appropriate assistance. As the needle tip gradually approaches the depth of the simulated blood vessel, the system prepares to enter the next critical phase—blood vessel localization and blood return detection.

[0097] 5. Vascular location and blood return detection status

[0098] When the needle tip reaches the vicinity of the target vein, the system enters the vessel localization / blood return detection state. This is the most critical stage in the puncture process, where the system needs to determine whether the needle tip has successfully entered the blood vessel and detected blood return. Typical conditions that trigger this state include: the needle tip depth reaching the preset vessel layer range, or the sensor detecting a suspected vessel contact signal.

[0099] Once in this state, the controller combines data from multiple sensors to determine the puncture result:

[0100] Position and orientation determination: The controller first checks the spatial relationship between the needle tip and the blood vessel. If the tactile array indicates that the needle tip is directly above the blood vessel and has reached the appropriate depth, the blood return monitoring process begins. If the needle tip is deviated (e.g., there is a signal from the sensors around the blood vessel but no signal directly above it, or the IMU shows that the needle direction is inconsistent with the blood vessel axis), it is suspected that the needle is not aligned with the blood vessel's central axis. In this case, the system can prompt "The needle may not be aligned with the blood vessel; try fine-tuning the direction" in training mode; in assessment mode, it silently records possible deviations.

[0101] Initial contact with a blood vessel: If the needle tip is against the outer wall of the blood vessel but has not yet penetrated it, the pressure sensor may detect a transient, slight increase in pressure (due to the deformation of the blood vessel under pressure), and the units of the tactile array surrounding the blood vessel will also give a pressure signal peak. These signs indicate that the needle tip has found the location of the blood vessel but has not entered the lumen. The system will remain in this judgment state, allowing the trainee to make a decision: to push forward slightly more to puncture the blood vessel, or, if the position is not ideal, to partially withdraw / adjust the angle and try to penetrate again.

[0102] Successful puncture of the vein: As the trainee advances the needle further to pierce the vein wall, the system will capture a series of success signals:

[0103] Sudden Resistance Change: The moment the needle pierces the blood vessel, the force feedback resistance decreases abruptly, and the trainee clearly feels a sense of emptiness as if "breaking through" the cavity. The controller detects the instantaneous and violent fluctuations in the force / pressure signal (e.g., the local pressure first rises and then falls when a blood vessel ruptures).

[0104] Pressure Changes: After the simulated blood vessel lumen opens, the internal pressure will momentarily drop or fluctuate. The pressure sensor reading will correspondingly drop sharply to near the actual pressure inside the vein, and the fluctuation may be detected as it tends towards a new equilibrium (if simulated blood returns to the lumen later, the pressure will stabilize at a slightly lower new value).

[0105] Blood return signal: The controller immediately triggers the simulated blood return process—opening the one-way valve or increasing the pump pressure to quickly push a portion of the simulated blood in the reservoir along the simulated blood vessel into the syringe. The trainee can visually see the red liquid flowing back in the syringe or catheter (i.e., the "blood return" sign appears). At the same time, the flow sensor will detect a brief decrease in the main circuit flow, matching the amount of blood diverted.

[0106] Once the above signal combination is detected, the controller determines that the needle tip has successfully entered the blood vessel lumen, and the puncture has achieved its target. The system immediately records the successful puncture time (the time interval from the needle tip contacting the skin to the first detection of blood return) and switches to the successful puncture status.

[0107] Continuous blood return monitoring: After successful detection, the controller does not immediately terminate monitoring but continues to observe for several seconds to ensure continuous and stable blood return, rather than occasional air bubbles. If the simulated blood return continues for several seconds and the pressure / flow reading is stable, it proves that the needle is indeed in the blood vessel lumen and has not slipped out. During this time, the system will keep the needle position undisturbed (or lock the current position if there is a robotic arm) to prevent the needle tip from slipping out of the blood vessel due to excited movement by the trainee.

[0108] Once the puncture is confirmed to be successful, the system provides multiple feedback mechanisms: the interface highlights "Puncture Successful!", the indicator light changes from yellow to green, and a notification sound is played to provide positive reinforcement. At this point, the model enters a success-holding state, and the trainee can proceed with subsequent procedures according to clinical steps (such as releasing the tourniquet, connecting the infusion device, and securing the needle). The model will maintain a small, constant pressure within the blood vessel to continuously and slowly push simulated blood into the needle, allowing the trainee to observe blood return and practice the next step. The scoring timer stops recording the puncture time the moment of success.

[0109] 6. Handling of puncture failure and abnormalities

[0110] If no blood return signal is detected after the needle reaches the depth of the blood vessel, and further advancement or adjustment by the trainee still shows no success, the system will determine that the attempt has failed after a certain judgment time threshold, i.e., the puncture has failed. The controller will then enter the failure handling branch, analyze the cause of failure based on sensor data, and provide prompts to help the trainee improve. Failure scenarios mainly include:

[0111] Missed Vessel: The needle tip has penetrated beyond the vein but there is no blood return. The pressure sensor does not detect any significant insertion fluctuation, and the tactile sensor signal around the vessel is also weak. This usually indicates that the needle tip has deviated from the vessel location (i.e., "missed"). The needle tip may have passed beside the vessel or stopped in the fascia layer below the vessel. The controller determines that the needle tip has not hit the vessel lumen based on sensor feedback. In teaching mode, the interface displays "Vessel lumen not found, may have deviated from the target, please adjust the angle or reselect the position and try again"; at the same time, the system records a failed attempt. If the needle is still in the body and the student wishes to try fine-tuning (e.g., gently changing direction to find the vessel), the system allows continuous monitoring of the blood return signal without removing the needle and does not immediately determine the end of the process. However, if the needle has been partially or completely withdrawn, or the student voluntarily gives up, the system enters the exit process to prepare for restarting (see the exit status section below).

[0112] Excessive puncture: The needle tip penetrates the blood vessel too forcefully, exiting through the vessel from the other side. This may manifest as a brief backflow of blood immediately after puncturing the vessel, which then stops as the needle has pierced the anterior and posterior walls of the vein. Pressure sensors will record abnormal patterns: pressure may briefly rise and then rapidly drop to near zero (the vessel loses normal pressure after puncture and rupture), and tactile sensor arrays may indicate the needle has penetrated far beyond the vessel layer. The controller determines this as excessive insertion (penetration). The system immediately illuminates a red warning light and sounds an alarm: "Warning: Suspected vessel puncture!" The trainee should immediately stop further insertion. In training mode, the system may suggest the trainee slowly pull back the needle tip a little to observe if backflow occurs (in reality, slightly pulling back the needle tip can sometimes return it to the vessel lumen, salvaging the puncture). If backflow occurs after pulling back, the system will switch to a successful state but record this excessive insertion as a deduction of points. In assessment mode, excessive puncture is considered a serious error; even if the attempt is later successful, it will result in a significant deduction of points or a failing grade.

[0113] Vascular Collapse: In specific low-pressure or fragile vascular simulation modes, if the needle tip fails to enter the blood vessel for an extended period and continues to compress it, simulated vascular collapse may occur. The system sets a timeout threshold (e.g., the needle tip repeatedly probes the vascular layer for more than 5-10 seconds without blood return). At this time, the controller automatically determines that the vascular has closed and collapsed due to prolonged pressure. The system then issues a prompt sound and warns on the interface: "No blood return detected, the vascular may have collapsed." Simultaneously, it automatically executes a series of actions to simulate the collapse effect: immediately shutting off the fluid pump and related valves, reducing the pressure in the simulated vein segment to zero, and temporarily flattening the lumen of the vessel by gently pulling on the internal pull wire. This way, even if the trainee is in the correct position for the vascular location, there will be no more blood return. The system marks this attempt as a failure and records the "vascular collapse" event. In training scenarios, instructors can choose to have the trainee change the puncture site or rest for a while before trying again (corresponding to real clinical procedures), and can use controls to restore the pressure of the model vascular to normal to reset the collapse state; in formal assessments, vascular collapse means that the trainee has failed to complete the puncture within the limited time, and the system will end the operation and determine it as a failure. Such timeout collapses during scoring will result in severe point deductions.

[0114] In the event of the aforementioned failures, the system possesses a certain degree of fault tolerance and retry mechanism. For non-irreversible errors (such as failing to hit a blood vessel), the trainee can choose not to remove the needle but make minor adjustments to the angle / depth and attempt blood return again. The system will continue to monitor without immediately counting the failure. Only when the trainee completely removes the needle and prepares to insert it again will the system consider the attempt complete and count the number of failures. For recoverable errors (such as timely retraction after slight puncture), the system records the error process while determining the final success, deducting points accordingly in the scoring rather than directly classifying it as a failure. Safety is paramount throughout the entire anomaly handling process: if a dangerous trend occurs (such as inserting too deeply towards a critical area deep within the model), the system would rather terminate the operation early and issue a warning than risk continuing.

[0115] Regardless of success or failure, the system will enter an exit / completion state after each puncture attempt. The exit and scoring process is described below.

[0116] 7. Exit status (needle removal and operation completed)

[0117] When the trainee decides to end the puncture (either by successfully obtaining blood return or by abandoning the procedure due to failure), the needle will be removed from the simulated arm. The system can detect needle removal in several ways: one is by the sudden complete disappearance of the tactile sensor signal, indicating the needle tip has left the tissue; the other is by the trainee notifying them via an interface button that "needle removal complete." Upon entering the exit state, the controller guides the trainee through safe procedures:

[0118] If simulating a clinical scenario, the tourniquet should be loosened before removing the needle, and the system interface will provide a timely reminder (this reminder can be enabled by the instructor as needed).

[0119] The moment the needle is completely withdrawn from the skin, the simulated blood vessel closes the puncture site using an integrated one-way valve and self-sealing material, theoretically preventing any simulated blood leakage. Even so, training requires trainees to apply pressure to the puncture site with a cotton ball / gauze for a few moments to prevent bleeding from a real patient. On the model, when a pressure sensor / tactile array detects continuous pressure on the skin surface after needle withdrawal, the controller recognizes the pressure application and prompts the trainee to press for at least a certain number of seconds. These detailed operations do not affect the scoring in the current embodiment, but the model allows for expansion to record such data for teaching feedback.

[0120] If the model is equipped with a robotic arm automatic puncture device, in the exit state, the robotic arm will be driven to slowly and safely withdraw the needle completely from the simulated arm and move it back to the initial safe position.

[0121] Once the needle is withdrawn and moves a safe distance from the skin surface, the controller releases the locks on all moving parts and marks the system as "operation completed." In cases of failure where retry is allowed, the system will automatically reset the necessary parameters to prepare for a return to standby mode for the trainee to try again. If no further retrying is desired or the maximum number of attempts has been reached, the system will enter the scoring state.

[0122] The exit process also includes: stopping the micropumps and heaters, restoring the model to a resting state; and recording all events and data logs of this operation to memory. For simulation training environments, after exiting, the model can perform fluid circuit resets (such as opening the loop bypass valve to release residual pressure) and blood simulation fluid recovery for future use. Simultaneously, any worn parts (such as disposable simulated skin patches) will be prompted for replacement at this stage. After completing the exit process, the system will enter the scoring and evaluation phase.

[0123] 8. Scoring Status

[0124] Once the puncture procedure is completed and the scoring process begins, the system starts to assess the quality of the procedure based on a pre-defined algorithm. The controller extracts feature parameters from sensor data and event logs recorded in previous stages, inputs them into the scoring model to calculate a comprehensive score, and generates a detailed feedback report.

[0125] The scoring is calculated automatically in the background, requiring no student intervention. The scoring algorithm uses a weighted summation model, and the total score can be expressed as:

[0126] Among them, f i w is the normalized score of the i-th feature (value between 0 and 1). i These are the weight coefficients corresponding to this feature. Each w... i The sum of the scores corresponds to the total score (e.g., 100 points). The scores for each feature f i Calculated based on the raw data collected by the sensors, reflecting the relative excellence of this indicator; each weight w i The percentage of this indicator in the overall score can be adjusted according to the training focus.

[0127] Indicator 1: Needle entry angle error f1

[0128] Needle insertion angle error measures the difference between the angle at which the needle tip penetrates the skin and the standard requirement. Let the actual insertion angle detected by the attitude sensor be θ0, and the ideal insertion angle be θ0 (e.g., approximately 15°), then the angle deviation is defined as: Δθ = |θ−θ0|

[0129] Let the maximum allowable deviation be Δθ max (Above this value, 0 points are awarded), then the score for the angle item is calculated using linear normalization: When the actual angle equals the ideal angle, Δθ = 0, and a full score of f1 = 1f is obtained; when the deviation reaches or exceeds the upper limit Δθmax, f1 = 0. For example, with an ideal angle of 15° and an allowable deviation of ±5°: when the actual angle is 15°, f1 = 1; when the actual angle is 20° or 10°, the deviation is 5°, which can be recorded as 0 points or a very low score depending on the boundary situation. This indicator encourages trainees to maintain the correct needle insertion angle, avoiding angles that are too flat (prone to subcutaneous slippage) or too steep (prone to puncturing blood vessels).

[0130] Indicator 2: Puncture force control f2

[0131] Puncture force control reflects whether the maximum force applied during needle insertion is appropriate. Let F be the maximum pressure recorded by the pressure sensor during the puncture process. max Set a safe and effective force threshold F. thr (For example, 2.0N, usually a force lower than this is sufficient to puncture a vein without causing significant damage), and set a completely undesirable upper limit F. max (bad) (e.g., 4.0N; anything exceeding this is considered a range and scores 0).

[0132] The following scoring rules can be used as an example:

[0133] F max ≤F thr At that time, if the force was deemed appropriate, full marks were awarded: f2=1

[0134] F max >F thr And F max <F max (bad) At that time, points are deducted linearly based on the degree of excess, for example:

[0135]

[0136] When F max ≥F max (bad) At that time, score 0: f2=0

[0137] For example: If F thr =2.0 N, F max (bad) =4.0 N, when F max When = 2.5 N,

[0138] f2≈=0.75

[0139] This indicator is mainly used to prevent trainees from using too much force. It also serves as a reminder that if the force is too weak and the puncture fails, it will be reflected in other indicators (such as the number of attempts).

[0140] Indicator 3: Needle insertion speed / time f3

[0141] The needle insertion speed / time evaluation assesses whether the speed at which the needle tip travels from penetrating the skin to reaching the target depth is appropriate. Let the ideal needle insertion time range be [T]. min ,T max (For example, puncture completed in 1-3 seconds is considered appropriate), the actual needle insertion time is T. in .

[0142] A typical segmented scoring method is as follows:

[0143] When T min ≤T in ≤T max Full marks will be awarded for the time being:

[0144] f3=1

[0145] When T in <T min or T in >T max When this happens, the threshold can be decreased linearly based on the degree of deviation from the nearest threshold, for example:

[0146] Among them, T ref For the most recent threshold (T) min or T max ), T lim This is the maximum tolerable deviation time (which can be determined by the specific design). Too fast or too slow needle insertion will cause f3 to decrease, thus prompting the trainee to maintain a reasonable operating rhythm while ensuring safety.

[0147] Indicator 4: Blood Return Waiting Time (f4) The blood return waiting time is the time from when the needle tip enters the blood vessel to when blood return is detected, reflecting the accuracy and efficiency of vessel localization. Let t be the time from the start of puncture to the detection of blood return. blood Let the maximum tolerable waiting time be T. max (blood) (e.g., 5 seconds; if it exceeds this time, it is considered unsuccessful).

[0148] A linear normalized form can be used:

[0149] For example: if blood is drawn within 1 second, f4≈1; if blood is drawn within 4 seconds, f4=1−4 / 5=0.2; if there is no blood return after 5 seconds, f4=0. This indicator reflects the efficiency of "one-shot success" and is usually given a high weight.

[0150] Indicator 5: Number of Attempts (f5) The number of attempts refers to the number of independent punctures performed to achieve success. A successful puncture is counted as 1 attempt, a failed attempt followed by a second puncture is counted as 2 attempts, and so on. Let N be the maximum allowed number of attempts. max The actual number of attempts is n. The score can decrease linearly:

[0151] Thus, when n=1 (success on the first try), f5=1; when n=N max When f5=0 exceeds N max If the attempt fails again, it is also recorded as 0. This item usually has a higher weight to highlight the importance of "success on the first try".

[0152] Indicator 6: Hand stability f6

[0153] Hand stability is used to evaluate the smoothness of needle control during operation. Let δ be the needle tip jitter metric calculated from the attitude sensor, and let δ be the maximum permissible jitter. max It can be in the following form:

[0154] When the operation is smooth and there is almost no excessive shaking, δ≈0, then f6→1; when the shaking is close to or exceeds δ... max At that time, f6 is close to 0. This indicator reflects the trainee's hand-eye coordination and tension control ability.

[0155] The weights w of the above features i It can be flexibly set according to the training or assessment mode, as long as the sum of the weights corresponds to the preset total score (e.g., 100 points). For example, in the basic training stage, the weights of needle insertion time and hand stability can be appropriately reduced, while in the assessment or advanced training, the weights of "success on the first attempt" and hand stability can be increased.

[0156] For example, during basic training, the focus can be on success rate, with high weights assigned to the number of attempts and recovery time, while the weights for angle or speed requirements are slightly lower; formal assessments, on the other hand, comprehensively evaluate all aspects, assigning appropriate weights to each item. Through a weighted scoring model, the system transforms quantitative data from the operation process into an objective comprehensive score, judging both success and failure as well as reflecting technical details.

[0157] After the scoring is calculated, the system will display the total score and scores for each sub-indicator on the screen or host computer interface, along with text feedback. For example: "Puncture successful, total score 78 points. Among them: excellent needle insertion angle (full marks), slightly excessive puncture force (deduct 1 point), moderate needle insertion speed (deduct a small number of points), slow blood return (deduct a large number of points), successful on the second attempt (deduct half a point), good operator hand stability (deduct a small number of points)." Through this feedback, trainees can clearly understand which aspects they performed well in and which aspects need improvement. All scoring results and related data are also automatically stored in the system database for instructors to review historical records or use for teaching research.

[0158] 9. Repair Status

[0159] After scoring is complete, the system enters a repair state, used to reset and maintain the model and prepare it for the next operation. Under normal circumstances, the repair state automatically performs the following tasks:

[0160] Mechanical reset: Returns all movable parts to their initial positions. For example, if the robotic arm has moved, it is moved back to its safe original position; the simulated arm posture is adjusted to a standard human position; any remaining locks are released.

[0161] Sensor reset: Clears or resets the data in the sensor cache. For example, it slowly depressurizes each pressure sensor to zero, resets the IMU's reference orientation, and resets the baseline of the tactile array. This establishes a clean measurement reference for the next puncture.

[0162] Fluid circuit maintenance: If simulated blood leakage or air bubbles entered during the previous puncture, the controller can drive the pump to circulate at low speed for a period of time during the repair phase to expel air and replenish fluid, ensuring the blood vessel is filled and free of air bubbles. If the vascular module has been punctured multiple times during this operation and there is a slight risk of leakage, the system will prompt the instructor to check or replace the vascular module.

[0163] Exit or switch modes: If you need to change the training mode or end the training after this training session, the recovery mode can disable special parameters enabled in the current mode (such as turning off low temperature mode heating, restoring normal blood pressure, etc.), or allow users to select the next mode while keeping the system in standby mode.

[0164] Data logging: The final result of this operation and any fault codes are written to storage. For example, the number of attempts, success or failure, scores for each indicator, and whether any abnormalities such as sensor failure or blood vessel collapse occurred are recorded for future analysis.

[0165] Material self-healing check: The system determines whether material self-repair needs to be performed based on preset strategies (see the self-repair mechanism section below for details). For example, it counts whether the number of punctures in a certain area has reached a threshold, or whether the pressure sensor has detected a slow leakage trend. If necessary, the repair process is automatically triggered at this stage.

[0166] The repair phase typically lasts only a short time. Once completed, the system returns to standby mode, indicating that a new puncture cycle can begin. If a serious anomaly occurs during the procedure (such as a persistent and unrecoverable sensor malfunction), the repair phase may keep the system in fail-safe mode, awaiting manual maintenance. Under normal circumstances, with a well-executed repair process, the model can quickly recover to its initial good condition, providing a stable and consistent starting point for the next trainee's practice.

[0167] In summary, the finite state machine process described above ensures the orderly connection, rigorous monitoring, and safety control of each stage from start-up preparation to puncture completion. Each state has clearly defined entry conditions, system responses, and exit conditions, making the puncture training process stable and reliable. Furthermore, all critical events are recorded for scoring and evaluation, achieving safe, efficient, and measurable teaching and training.

[0168] Exceptions and Fault Tolerance Design

[0169] During complex puncture procedures, various unexpected and abnormal situations may occur. To improve system robustness and safety, this invention designs robust fault-tolerant logic to detect and handle common anomalies:

[0170] Puncture Interruption: If the trainee suddenly stops the procedure during puncture—for example, withdrawing the needle before it reaches the blood vessel, or pausing halfway through insertion for an extended period—the system will determine that a puncture interruption has occurred. Specifically, the tactile / posture sensors detect that needle advancement has stopped for more than a preset time and no blood return is detected, possibly accompanied by a signal of the needle gradually retracting. Once the controller recognizes this situation, it freezes the current scoring timer and records the "puncture interruption" event. The interface displays "Operation Interrupted" and suggests that the trainee safely remove the needle and prepare to retry. If the needle is still in the tissue, the system enters an exit procedure, guiding the trainee to remove the needle according to protocol (e.g., issuing a voice prompt "Please slowly remove the needle and apply pressure to stop the bleeding"). After needle removal, the system can automatically return to standby mode, allowing a new attempt to begin. This interruption will be counted in the number of attempts and used as a deduction factor in the final score. Through this fault-tolerant handling, even if the trainee abandons the procedure midway, the system can smoothly and safely end the current process, preparing for another attempt.

[0171] Sensor Failure: If any critical sensor malfunctions during training (e.g., pressure sensor readings are significantly unreasonable or stop updating, attitude sensor data is lost), the system will immediately enter a safety fault handling sub-process. The controller first pauses the current operation: stopping the liquid pump and any force feedback actions, and locking the robotic arm (if any) to prevent dangerous actions caused by erroneous sensing information. The interface then displays a warning: "Sensor malfunction, please check the equipment," specifying the type of faulty sensor. The system logs the fault and attempts basic automatic recovery measures: such as reinitializing the sensor module, switching to a redundant sensor (if available), or adjusting and recalibrating the sensor drive parameters. If the fault recovers automatically within a short time and the data returns to normal, the system will prompt the trainee to continue (resuming the timer from the pause point if necessary); if the sensor continues to malfunction and cannot be recovered, the system terminates the training: instructing the trainee to remove the needle to end the operation, and the model remains in fail-safe mode, no further action is permitted. It can only be restarted after manual inspection and component replacement. This strategy ensures that the system does not continue without reliable sensor data, thus eliminating potential risks.

[0172] Vessel collapse timeout: This anomaly has been described in the previous section on handling puncture failures, specifically addressing the potential collapse of a vessel due to prolonged needle failure. In terms of error tolerance, once the collapse criteria are met (e.g., entering the blood return detection state and exceeding the threshold without blood return), the system automatically treats it as a failure and simulates a collapse to alert the trainee. It's worth noting that this timeout event is specifically marked as "Vessel collapse leading to failure" during recording and scoring. The scoring model considers the blood return time of this attempt as a maximum value (exceeding the allowable range), therefore the f4 index receives 0 points, and the f5 attempt count is increased by one failure. The system will also indicate the reason in the interface feedback (e.g., "Vessel collapse, it is recommended to try again at a different site"). By incorporating real-world clinical scenarios of vessel collapse into the model and implementing error tolerance, trainees can learn decision-making strategies for handling similar situations (e.g., timely stopping, changing vessels, etc.).

[0173] Anomalies in Scoring Data: During the scoring phase, if incomplete raw sensor data or outliers exceeding the physical limits are detected, the system will activate scoring fault-tolerance logic to minimize the impact of data defects on trainee evaluations. First, the controller performs a rationality check on each feature parameter: for example, the needle angle should be within 0–90°, the applied force should be positive and not exceed the equipment's upper limit, and the needle insertion time should not be negative. If any data is missing or invalid, that feature is skipped or a default value is used. If data anomalies would severely affect the safety assessment (e.g., a pressure sensor reading 0 throughout but actual resistance is expected), the system will add a note to the scoring result, such as "Sensor data anomaly, scoring is for reference only," or, if necessary, not calculate the total score but only provide the operation record. Simultaneously, the anomaly is recorded in detail in the log for future analysis and improvement by engineers. After fault-tolerance processing, the controller uses the remaining reliable data to complete the scoring calculation, providing an evaluation result as close to the actual level as possible. Thus, even if some sensors malfunction, the system can still provide a general evaluation of the trainee's operation without interrupting the training process.

[0174] For all the aforementioned abnormal events, whether it's puncture interruption, equipment failure, or collapse timeout, the system has clear judgment conditions and handling strategies to ensure that the model can be safely shut down or smoothly withdrawn under abnormal circumstances, and to promptly notify trainees and instructors to take appropriate measures. This fault-tolerant design significantly enhances the system's robustness, enabling it to ensure trainee operational safety, protect model hardware, and provide effective teaching feedback as much as possible even in various unexpected situations.

[0175] Puncture procedure scoring system

[0176] As mentioned above, the model of this invention is equipped with a comprehensive scoring system that objectively and quantitatively evaluates each puncture procedure using data from multiple sensors. The scoring mathematical model, indicator weight settings, and example cases are further explained below.

[0177] Scoring Indicators and Weighting

[0178] The scoring system consists of several key indicators, each corresponding to a specific quality element of the puncture procedure. Commonly used scoring indicators and their key focus areas are as follows:

[0179] Needle insertion angle: Whether the angle between the needle tip and the skin surface is within a reasonable range (e.g., 15° ± 5°). Too flat an angle makes it easy for the needle to slide under the skin, while too steep an angle makes it easy to puncture blood vessels or penetrate tissue. This indicator encourages trainees to use a standard needle insertion angle to improve success rate and safety.

[0180] Puncture force: Is the needle force appropriate? Ideally, the force used to puncture the skin and blood vessels should be neither too strong (to prevent damage to deep structures) nor too weak (so that the skin or blood vessel wall cannot be penetrated). This indicator encourages trainees to practice controlling the force in their hands to form a smooth and even puncture force curve, rather than quickly and roughly inserting the needle or hesitating and not using enough force.

[0181] Needle insertion speed: Is the insertion speed or time appropriate? In principle, the insertion should be decisive and steady from the moment the needle tip touches the blood vessel, avoiding dragging or being too hasty. The speed of needle insertion is usually complementary to the force applied, and also reflects the trainee's skill level and psychological state (too slow may indicate tension and hesitation, too fast may indicate impatience and recklessness).

[0182] Hand stability: The degree of hand steadiness during the puncture process. This includes whether the needle tip's movement trajectory is straight, and whether there is any lateral swaying or vertical shaking. Good stability indicates that the trainee has good hand-eye coordination and can accurately control the direction of the needle; conversely, if the needle tip shakes significantly, not only will the success rate decrease, but it is also easy to cause additional tissue trauma.

[0183] Blood return waiting time: How quickly blood return is seen after successful insertion into the blood vessel. This indicator directly reflects the accuracy of puncture positioning. If the needle tip just enters the lumen, blood return is basically immediate; if the position is at the edge, or only part of the needle tip is in the blood vessel, adjustment may be needed or there may be a delay in seeing blood; if the position is incorrect, there will be no blood return for a long time. The shorter the time, the better. Generally, blood return should be seen within a few seconds; otherwise, it indicates that the operation positioning needs improvement.

[0184] Success on the first attempt: Whether the puncture was completed on the first try. This refers to the number of attempts, as explained in the aforementioned F5 indicator. Successful venipuncture on the first attempt is the gold standard, and therefore carries significant weight in the scoring. The system rigorously records the number of times the needle is withdrawn and reinserted; each additional attempt results in a substantial deduction from the total score, encouraging trainees to strive for success on the first try.

[0185] In addition to the quantitative indicators mentioned above, some qualitative or manual assessment elements (not automatically detected by this model) can also be considered in actual teaching: such as aseptic technique (whether disinfection, wearing gloves, and needle disposal are correct), patient communication (pre-procedure explanation and reassurance), and post-puncture care (whether pressure hemostasis is timely and correct, and whether dressing is standardized). Instructors can score these aspects through observation, and together with the model's automatic scoring results, they constitute a comprehensive evaluation of the trainees. The model currently focuses on objective technical operations, but an interface can be expanded to allow instructors to input the aforementioned manual scoring elements.

[0186] The weight of each automatic scoring indicator in the total score w iThe score needs to be reasonably set so that it comprehensively reflects the skill level. For example, the weighting can be determined by referring to the opinions of clinical experts or by statistically analyzing a large amount of training data: assuming a full score of 100 points, the success rate is assigned 30 points as the most important indicator, operational stability 20 points, needle insertion angle 15 points, force control 15 points, needle insertion speed 10 points, and blood return time 10 points. This is just an example configuration, and it can be adjusted according to the teaching objectives. For basic training for beginners, the focus can be on success or failure, with less weight given to details such as angle and speed; for assessments of advanced learners, a balanced development in all aspects is desired, with relatively even weighting for each aspect.

[0187] Scoring Calculation Method

[0188] After the operation is completed, the system calculates the normalized score f for each indicator. i (The score range is 0 to 1, or converted to a 0 to 100-point scale), then weighted summation is performed to obtain the total score S. As stated in the formula above: Each f i The calculation method converts the raw sensor data into a score between 0 and 1 based on a pre-defined evaluation function. For example, the angle deviation Δθ is converted into angle f using a linear function. Peak force F max The force f is obtained through a piecewise linear function. Insertion time T in The velocity f is obtained relative to the ideal range using a linear or trapezoidal function. The recovery time t... blood The recovery rate (f) is obtained through linear decrease. The number of attempts (n) is obtained using the aforementioned formula for the number of attempts (f). The stability fluctuation (δ) is obtained through linear decrease for the stability of f. Each term of f... i A value of 0 indicates that the result is far below the requirement, while 1 indicates that the result fully meets or exceeds the requirement. Weight w i The value is determined by the proportion of the corresponding item in the total score (it can be a decimal between 0 and 1 whose sum is 1, then S is multiplied by 100 to get the percentage score; or the weights are directly allocated to the full score of 100, whose sum is 100, and S is the final score).

[0189] Example rating case

[0190] The scoring process is illustrated below with a specific example. Suppose a student succeeds after two attempts at puncture; the system records the following data:

[0191] First attempt: needle insertion angle of about 15° (ideal angle), maximum puncture force of 1.5N (less than the 2.0N threshold), but no blood return was observed at the needle tip (failure and withdrawal).

[0192] Second attempt: needle insertion angle 15°, maximum force 1.8N, moderate insertion speed (approximately 2 seconds to complete from skin penetration to target depth), blood return was observed after about 4.5 seconds of insertion, total number of attempts 2. During the puncture, there was slight hand tremor but not obvious (the attitude sensor measured needle tip angle tremor within ±2°).

[0193] Based on the above data, the system calculates the score:

[0194] Angle (15°): Ideal 15°, actual 15° deviation of 0° is within the allowable range. f-angle = 1.0 (full marks). If the angle weight is 15 points, then this item will receive 15 points.

[0195] Strength (1.8N): Threshold 2.0N, 1.8N is within the threshold and moderate, estimated strength f≈0.9 (close to full marks, slightly deducted). If the strength weight is 15 points, the score is approximately 13.5 points.

[0196] Speed ​​(2 seconds): Ideally within the range of 1-3 seconds, 2 seconds is considered slightly below average. The speed f can be set between 0.8 and 0.9 (slightly slower than optimal but within an acceptable range). Let f ≈ 0.8. With a weight of 10 points, this yields 8 points.

[0197] Healing time (4.5 seconds): close to the 5-second limit, which is considered slow healing. f-healing is approximately 0.1 (very low, as it only succeeds when almost at the edge of failure). It scores 2 points out of a weight of 20, a significant deduction.

[0198] Number of attempts (2): It took two attempts to succeed. According to the aforementioned algorithm, the number of attempts is approximately 0.5 (equivalent to deducting half the score). With a weight of 20 points, you get 10 points.

[0199] Stability (±2° jitter): Relatively stable, f_stability≈0.85. With a weight of 20 points, it scores 17 points.

[0200] The sum of the scores is: 15 + 13.5 + 8 + 2 + 10 + 17 ≈ 65.5 points (rounded to approximately 66 points). The system will convert and display this total score, for example, "66 / 100 points". It will also generate detailed scoring rules for each indicator: perfect angle, near-perfect strength, acceptable speed, significant deduction for long recovery time, half-point deduction for two attempts, and slight deduction for good stability. From this, the student can see that the main reasons for the deductions are "failure on the first attempt" and "slow recovery," which are precisely the areas that need improvement.

[0201] If the weights are adjusted under different training requirements, the score will change accordingly. For example, for basic exercises, the weights of stability and speed can be reduced while the weights of success rate can be increased, which may improve the total score in the above case (because success, although slower, is eventually completed). The system also supports switching between different scoring criteria, but the principle is that regardless of how the weights are allocated, the scoring model transparently lets trainees know how they performed in each individual section.

[0202] Rating upgrade

[0203] The scoring algorithm of this invention employs clear rules and weight allocation, and is already capable of objectively evaluating the quality of puncture procedures. However, with the accumulation of large amounts of training data and the development of artificial intelligence technology, the scoring system can be further upgraded to an AI-driven model to improve the accuracy and adaptability of the scoring. Future improvement directions include:

[0204] Introducing machine learning models: Supervised learning models are trained using extensive historical puncture data to evaluate the quality of procedures. For example, curves showing the changes in force and angle over time during needle insertion can be collected and input into a neural network or random forest model, allowing it to learn to map these temporal patterns to expert scores or success probabilities. Machine learning models can capture many complex nonlinear features (such as the frequency of subtle hand tremors, the shape of the force curve, etc.) that are difficult for simple rules to fully cover. A well-trained AI model can supplement or upgrade existing rule-based scoring, providing more refined and fair scoring. As the sample data increases, the model's evaluation of various operating styles becomes increasingly accurate and robust.

[0205] Dynamic weight optimization: A data-driven approach is used to automatically adjust the weights (wi) of each indicator. For example, by analyzing massive training records, if certain indicators are found to be more correlated with final success (such as hand stability contributing significantly to success rate), the system can suggest increasing their weights; conversely, the weights of less relevant indicators can be decreased. This can be achieved through optimization techniques such as genetic algorithms, allowing the system to experiment with different weight combinations in a simulated environment to maximize the consistency between the scoring results and expert human scoring. Ultimately, an adaptive weight configuration scheme is formed, optimizing the scoring criteria for different levels or training objectives, making it more scientific and fair. In other words, the scoring system itself can "learn" how to more reasonably evaluate trainees.

[0206] Integrating More Intelligent Evaluation Indicators: AI technology enables models to combine data from multiple dimensions for comprehensive evaluation. For example, computer vision analysis of trainee operation videos can be introduced to extract additional features: such as whether the trainee's needle-holding posture is correct, whether the arm is tense and trembling during needle insertion, and changes in the volunteer's facial expressions (reflecting the patient's pain level); voice analysis can also be used to determine whether the trainee's tone of voice in communicating with the patient is appropriate. This new information can be input into the AI ​​model along with existing sensor data to provide a more comprehensive evaluation. Furthermore, by visually tracking the three-dimensional relationship between the needle tip and the blood vessel, it is possible to determine whether the entire needle insertion trajectory is smooth and close to the blood vessel, and whether there are any unnecessary adjustments. These can all serve as a basis for improving the scoring. With the integration of these intelligent perceptions, the scoring system will not only examine whether the mechanical movements are standardized, but also, to a certain extent, assess the trainee's soft skills such as clinical thinking and humanistic care.

[0207] Through the above upgrade approach, the future scoring system will gradually evolve from the current rule-based static model to a dynamic, self-evolving intelligent assessment assistant. On the one hand, it will retain the current objective quantitative indicator system as a foundation; on the other hand, it will utilize AI to continuously refine the model and expand the evaluation dimensions, making the scoring results more consistent with the actual clinical skills performance. Ultimately, the system is expected to reach an evaluation level close to that of human experts, accurately distinguishing between high-level operations and problematic operations in various complex situations, and accordingly tailoring personalized training improvement suggestions for each trainee. This evolution from rules to intelligence will significantly improve the scientific rigor and effectiveness of clinical skills assessments and provide technological reserves for future training and scoring of more complex medical procedures.

[0208] Simulation material self-healing mechanism

[0209] Frequent puncture practice can cause extensive needle prick damage to the simulated skin and blood vessels of the model. To improve model durability and reduce the frequency of consumable replacement, this invention employs self-healing technology in the simulation materials, enabling the micro-needle puncture damage to the skin and blood vessels to heal automatically under specific conditions.

[0210] Self-healing material design

[0211] The simulated skin layer and vein wall materials incorporate polymeric elastomers with dynamically reversible bonds. When the material is damaged by a needle puncture, the molecular chain movement can be reactivated through heating or other means, causing the ruptured surfaces to re-bond. For example, Diels-Alder reversible covalent bonds or multiple hydrogen bonds are introduced into silicone rubber-based materials. These bonds maintain the material's strength at room temperature, break at a certain temperature (such as 50°C), causing the material to soften and flow, and then recombine and solidify upon cooling. [Some commercial simulated arms have already used similar self-healing silicone, showing no obvious needle holes or leakage after hundreds of punctures]. Simulated blood vessels also use highly elastic, self-closing tubing (such as latex tubing that instantly contracts and closes the needle hole after puncture) to further reduce the risk of leakage.

[0212] Tests have shown that after repeated punctures at the same site dozens of times, the skin and blood vessel materials used in this model can be completely repaired with a single self-healing treatment, making the needle holes virtually invisible to the naked eye, preventing internal leakage, and restoring the model's performance almost to its original state. This self-healing property greatly extends the model's lifespan: traditional models often require replacement after a period of use when the skin / blood vessels become covered with needle holes and begin to leak, while this model can "heal" itself and can be used continuously for hundreds of times while maintaining good condition.

[0213] Self-repair process

[0214] The model's built-in controller and heating device can perform a material self-healing process during training breaks. A typical process is as follows:

[0215] Triggering conditions: The system determines when to initiate self-repair based on preset strategies. For example, it may be triggered after a certain number of punctures at the same simulated vein site (e.g., 20 times); after each training session; or when the pressure sensor detects a slow decrease in pressure within the blood vessel (suspected micro-leakage). Instructors can also manually select "Execute Self-Repair" on the interface. The system must be in idle standby mode and no students are currently interacting with the system during triggering.

[0216] Preparation: Upon entering self-healing mode, the controller first takes safety measures: stopping the liquid pump and reducing the pressure inside the simulated blood vessel to zero (to prevent leakage due to liquid expansion or pressure during heating); releasing the tension and fixation of the simulated skin and blood vessels, allowing them to relax naturally (this facilitates the alignment and closure of the needle holes). The system then displays a message on the interface or indicator lights: "Material repair in progress, please do not operate."

[0217] Heating via Heat: The controller activates a flexible heater embedded beneath the skin and around blood vessels to heat the concentrated puncture area to a predetermined temperature (e.g., 50°C). The heating process is controlled in a closed loop by a temperature sensor, with a slow temperature increase to avoid overshoot. Once the target temperature is reached, heating is maintained for a period of time (e.g., 30 seconds to 1 minute) to allow the material to fully soften and flow, filling and healing all the tiny needle puncture channels. During this process, the system continuously monitors temperature changes to ensure they remain within a safe range (heating is immediately stopped and an alarm is triggered if the temperature rises abnormally).

[0218] Cooling and Shaping: After heating and heat preservation are completed, the controller turns off the heater, allowing the material to cool naturally or turning on a micro-fan to accelerate heat dissipation. During the temperature drop, the broken bonds in the polymer material recombine, and the material gradually regains its elastic strength. The filled pinholes fuse with the surrounding material after solidification. The controller waits for the temperature to drop to near body temperature or room temperature to ensure that the material is fully shaped.

[0219] Resumption of Operation: After cooling is complete, the controller readjusts the tension of the simulated skin and blood vessels to the training state; if there is a cavitation in the fluid path, simulated blood is replenished and air bubbles are removed; normal venous pressure is re-established. Then, the self-repair mode is exited, and the indicator lights return to the ready display.

[0220] The entire self-healing process typically takes a few minutes and can be scheduled during training breaks or after a course, with minimal impact on trainees. During self-healing, the model is paused and unavailable, but this is a necessary maintenance step. Through regular automatic healing, the model can be restored to an ideal "zero-damage" state before each batch of trainees begins practice, ensuring a consistent experience for all trainees and the reliability of the model's functionality.

[0221] It's worth noting that self-healing is not an unlimited number of complete repairs. Once a module material experiences a significant performance decline after hundreds of punctures and multiple self-healing attempts (e.g., significantly reduced skin elasticity, repeated local collapse and leakage of blood vessels), the system will record this information and prompt for manual replacement of the module. However, compared to the frequent replacements of traditional methods, this invention significantly extends the component replacement cycle and reduces teaching costs.

[0222] In the future, more advanced self-healing technologies can be introduced to further improve the durability of models. For example, microcapsule repair agents can be dispersed in silicone (the capsules rupture upon needle puncture, releasing repair fluid to fill the needle hole), or ultraviolet light-triggered self-healing materials can be used (healing is promoted by irradiation with specific light waves after puncture). These improvements are all within the scope of the present invention, requiring only changes to materials and local hardware without affecting the overall design concept.

[0223] In summary, the intravenous puncture training model of this invention, through the aforementioned structural design, control logic, and scoring system, achieves a highly realistic operational experience and scientifically quantifiable skills assessment. Trainees can repeatedly practice puncture techniques in a risk-free environment, receiving timely and intuitive feedback and improvement suggestions from basic steps to handling abnormal situations. The model's multi-mode use (desktop independent or worn by a person), rich sensor monitoring, and self-healing materials ensure the flexibility, realism, and economy of teaching. Its scoring and assessment mechanism provides a basis for objective assessment and can be continuously upgraded with the integration of artificial intelligence technology. The various components and method steps of this invention can be adjusted or equivalently replaced according to specific needs, such as adding new sensor types or changing material formulations. Any modifications that do not deviate from the core ideas of this invention fall within the scope of protection of this invention.

Claims

1. A training and assessment model for intravenous puncture, characterized in that, include: The simulated arm module, which mimics the structure of an adult forearm, consists of the following components arranged sequentially from the outside to the inside of the simulated arm body: The simulated skin layer is made of flexible polymer elastic material and can be detachably installed on the outer surface of the simulated arm. The simulated fat layer is placed inside the simulated skin layer to simulate the buffering resistance of subcutaneous fat tissue; The simulated muscle layer is located inside the simulated fat layer to support the shape of the arm and provide resistance to puncture. The simulated venous network is embedded between the simulated fat layer and the simulated muscle layer according to the actual anatomical course. The simulated skin layer, simulated fat layer, simulated muscle layer and simulated venous network are all constructed as detachable modules, which can be replaced with different combinations of thickness, hardness and venous elasticity parameters according to training needs. The simulated blood circulation system includes a reservoir, a micro-liquid pump, a one-way valve, and an elastic buffer element installed in a desktop base. The reservoir is connected to the simulated venous network through a flexible tube. Under the control of the control module, the micro-liquid pump provides simulated blood with adjustable pressure and flow to the simulated venous network. The one-way valve and elastic buffer element are used to stabilize the circulation pressure and prevent backflow. The desktop base and quick-connect mechanism are described. The desktop base is a box structure that houses the simulated blood circulation system and control module. The upper surface of the base is provided with a positioning bracket and a quick-connect interface that integrates electrical connections and liquid channels. The simulated arm module can be mechanically fixed to the base, connected to the liquid channel, and connected to the electrical signal through the quick-connect interface. The quick-connect interface has a foolproof and self-sealing structure to enable quick assembly and disassembly of the simulated arm module and leak-free liquid switching. A multi-sensor module is used for multi-modal monitoring of the puncture procedure, including: Pressure and flow sensors are placed on a simulated venous network or its connecting pipes to detect changes in venous pressure and flow over time. An array of tactile / position sensors, placed at different depths in the simulated skin and simulated fat layers, is used to detect the needle tip's contact with the skin, the depth of insertion, and its relationship with the blood vessel position. An attitude sensor mounted on the puncture needle or needle handle is used to detect the spatial tilt angle, direction, and motion stability of the needle. Temperature sensors are placed near the simulated skin layer and simulated veins to monitor the local temperature of the model; The control and evaluation module, located in a desktop base or portable control box, is electrically connected to the multi-sensor module and the simulated blood circulation system, and is used for: The state machine process executes the following steps: startup self-test, standby, needle tip contact detection, needle insertion monitoring, blood vessel positioning and blood return detection, puncture success / failure determination, exit, and material self-repair control. The micro liquid pump automatically adjusts its output pressure, valve opening and closing status, the tension of simulated skin and blood vessels, and the working status of heating elements based on sensor signals such as pressure, flow rate, position, posture, and temperature, in order to simulate physiological phenomena such as venous filling, blood vessel rolling, and collapse in different patients. After the puncture, multiple characteristic indicators such as needle insertion angle, puncture force, needle insertion time / speed, blood return waiting time, number of attempts and hand stability are normalized and a comprehensive score is calculated according to preset weights. The venipuncture training and assessment model also includes a self-healing control unit, which controls an embedded heating element to locally heat the simulated skin layer and / or simulated venipuncture material to the target temperature and maintain it for the required time when the puncture damage meets a preset threshold. This allows the self-healing polymer material to complete the needle hole healing during the heating and cooling process, thereby restoring the model's airtightness.

2. The intravenous puncture training and assessment model according to claim 1, characterized in that, The simulated skin layer has at least three replaceable skin modules, which are used to simulate the skin mode of the elderly, the skin mode of long-term chemotherapy patients, and the skin mode of critically ill patients. The different skin modules differ in thickness, elasticity, surface texture and color, as well as thermal conductivity. When the corresponding training mode is loaded, the control module adjusts the target temperature of the skin surface and the baseline pressure in the vein at the same time.

3. The intravenous puncture training and assessment model according to claim 1, characterized in that, The simulated venous network includes multiple sets of pluggable venous modules. Each set of venous modules corresponds to different tube diameters, wall thicknesses, and material hardness, and is used to simulate normal veins, veins prone to collapse in the elderly, and sclerotic veins, respectively. The control module uses pressure sensors to detect the vascular collapse process and provide corresponding prompts by setting different initial filling pressures and collapse trigger times in the corresponding training mode.

4. The intravenous puncture training and assessment model according to claim 1, characterized in that, The quick-connect mechanism of the desktop base includes a mechanical locking structure and a self-sealing liquid quick connector. The mechanical locking structure is used to limit the installation of the simulated arm body in a posture with the elbow slightly bent and the palm facing upward. The self-sealing liquid quick connector automatically closes when the simulated arm is not inserted and automatically opens to connect the reservoir and the venous network after the simulated arm is inserted and positioned.

5. The intravenous puncture training and assessment model according to claim 1, characterized in that, The model further includes a wearable mode switching structure, which includes: The arc-shaped groove and soft padding on the back of the simulated arm body are designed to fit the surface of a real forearm. At least two adjustable straps are used to bind the simulated arm body to the real forearm; The portable control box is connected to the main body of the simulated arm via cables and catheters. The control module automatically switches to wearable mode by recognizing the interface status, reducing the basal pressure in the vein to a preset value, turning off the body's audio-visual prompts, and enabling wireless communication, sending sensor data and scoring results to an external terminal.

6. The intravenous puncture training and assessment model according to claim 1, characterized in that, The tactile / position sensor array in the multi-sensor module is arranged in multiple layers along the thickness direction of the simulated skin layer and simulated fat layer, and is arranged in a ring around the simulated vein. It is used to generate different combinations of trigger signals when the needle tip contacts the skin, penetrates to different depths, and presses against the outer wall of the blood vessel. The control module determines the depth of the needle tip and its relative position to the blood vessel based on the array signals.

7. The intravenous puncture training and assessment model according to claim 5, characterized in that, The control module uses data from the attitude sensor to calculate the angle between the puncture needle and the simulated skin, and sets an angle deviation threshold. When the needle insertion angle is detected to exceed the preset range, a prompt message is issued in training mode and points are deducted from the angle index during scoring.

8. The intravenous puncture training and assessment model according to claim 7, characterized in that, In the state of blood vessel positioning and blood return detection, the control module comprehensively determines whether the needle tip has pierced the anterior and posterior walls of the vein based on the instantaneous pressure change detected by the pressure sensor, the flow change waveform detected by the flow sensor, and the triggering sequence of the tactile / position sensor array. When puncture is determined, an alarm is triggered and recorded as a serious error event.

9. The intravenous puncture training and assessment model according to claim 1, characterized in that, The scoring module employs a linear weighted scoring model, normalizing features such as needle insertion angle error, puncture force, needle insertion time or speed, blood return waiting time, number of attempts, and hand stability into a score fi ranging from 0 to 1, and then applying the formula S=Σw. i ·f i The comprehensive score S is calculated in the following way, where wi is a preset or trained and optimized weight coefficient.

10. A training and assessment model for intravenous puncture according to claim 9, characterized in that, The scoring module uses machine learning methods to evaluate the weight coefficients w. i The scoring threshold is adaptively adjusted, specifically by using historical training data and expert human ratings as samples, and optimizing w using a supervised learning algorithm. i This minimizes the error between the model's output score and the expert score, thereby achieving iterative optimization of the scoring model.

Images