Bedridden patient sign measurement and medication auxiliary system based on multi-sensor fusion

Abstract

The invention discloses a bedridden patient sign measurement and medication assisting system based on multi-sensor fusion, which belongs to the technical field of medical instruments and comprises a multi-sensor fusion measurement module, a self-adaptive calibration compensation module, an intelligent evaluation decision module, a medication assisting calculation module and a closed-loop feedback optimization module. A pressure sensor array, an ultrasonic sensor and an acceleration sensor are used for cooperatively measuring the height and weight of a patient, a deformation compensation model and a body position correction model are adopted for accurate calibration, a BMI index is automatically calculated, a VTE risk score, a falling risk score and a self-care ability score are generated, and the medication dosage is intelligently recommended according to medicine characteristics and patient physique. And continuously optimizing system parameters through an incremental learning algorithm to form a measurement-evaluation-medication-optimization deep coupling closed-loop system, so that the physical sign measurement precision, clinical evaluation efficiency and medication safety of the bedridden patient are remarkably improved, and the nursing labor intensity and the medical error risk are reduced.

Description

Technical Field

[0001] This invention belongs to the field of medical device technology, specifically relating to a system for measuring vital signs and assisting medication administration in bedridden patients based on multi-sensor fusion. Background Technology

[0002] With the aging population and the increasing number of patients with chronic diseases, the medical care needs of bedridden patients are growing. Because bedridden patients are unable to stand, traditional methods of measuring height and weight are difficult to implement, yet height and weight data are crucial for BMI assessment, medication dosage calculation, nutritional management, and clinical risk assessment.

[0003] Currently, the measurement of height and weight for bedridden patients in clinical practice mainly relies on manual methods. Nurses use a tape measure to measure the patient's height while lying down and a bedside scale or hanging scale to measure their weight. This method suffers from problems such as low measurement accuracy, cumbersome operation, the need for multiple people to work together, and the risk of causing secondary injury to the patient. Existing portable measuring devices are mostly single-function devices, requiring nurses to manually calculate BMI values, check medication dosages, and perform risk assessments after measuring height and weight, lacking intelligent and integrated solutions.

[0004] In recent years, wearable devices and bedside monitoring technologies have developed rapidly, and wearable devices are increasingly used in intensive care. However, challenges remain in measurement accuracy, interpretability of data processing algorithms, and integration with clinical decision-making processes.

[0005] In drug dosage calculation, weight-based dosing has become standard clinical practice, particularly in pediatrics, oncology, and intensive care. Weight-based drug dosage calculation significantly reduces medication errors and adverse drug reactions; however, problems such as inaccurate weight measurement, dosage calculation errors, and insufficient consideration of patients' special circumstances still exist in clinical practice. For obese patients, it is necessary to differentiate between total weight, ideal weight, and lean body mass to guide medication, and current systems lack this refined dosage calculation capability.

[0006] In clinical risk assessment, VTE scores, fall scores, and self-care ability scores are important tools for inpatient management. Research published in the Journal of Thrombosis and Haemostasis shows that the Padua and Caprini scores are mainstream tools for VTE risk assessment; however, these scoring tools typically require nurses to manually input multiple parameters and perform calculations using tables, which is inefficient and prone to errors. A 2020 study published in the journal Sensors proposed a radar-based non-contact vital sign monitoring method, but it was not integrated with clinical scoring systems.

[0007] The main problems with existing technologies include: insufficient accuracy in measuring the height and weight of bedridden patients; lack of automated BMI and medication dosage calculation functions; inefficiency due to the need for manual calculation of various clinical scores; disconnect between measurement, assessment, and medication administration with a lack of data linkage; and a lack of intelligent optimization mechanisms based on historical data. These problems lead to a heavy workload for nursing staff, a high risk of medical errors, and insufficient patient safety.

[0008] Therefore, there is an urgent need to develop an integrated and intelligent system for measuring vital signs and assisting medication administration in bedridden patients. This system should be able to accurately measure height and weight, automatically calculate BMI and medication dosage, complete multi-dimensional clinical scoring, and continuously optimize system performance through data feedback. This would improve nursing efficiency, reduce medical errors, and ensure patient safety. Summary of the Invention

[0009] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a multi-sensor fusion-based system for measuring vital signs and assisting medication in bedridden patients. Through five core technologies—multi-sensor collaborative measurement, adaptive calibration compensation, intelligent assessment and decision-making, precise medication calculation, and closed-loop feedback optimization—it achieves accurate measurement of height and weight, automatic BMI calculation, multi-dimensional clinical risk assessment, and intelligent medication assistance for bedridden patients. This forms a deeply coupled closed-loop collaborative system of measurement, assessment, medication, and optimization, solving the technical problems of low measurement accuracy, low assessment efficiency, and poor medication safety in the care of bedridden patients.

[0010] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0011] A multi-sensor fusion-based system for measuring vital signs and assisting medication administration in bedridden patients includes:

[0012] A multi-sensor fusion measurement module, housed in a portable measuring device, includes a pressure sensor array, an ultrasonic sensor, and an accelerometer. The pressure sensor array collects pressure distribution data of the contact area between the bedridden patient's body and the mattress. The ultrasonic sensor measures the straight-line distance from the patient's head to the soles of their feet. The accelerometer detects the posture angle data of the measuring device. The multi-sensor fusion measurement module uses a weighted fusion algorithm to comprehensively process the pressure distribution data, straight-line distance data, and posture angle data to generate initial values ​​for height and weight.

[0013] The adaptive calibration and compensation module, connected to the multi-sensor fusion measurement module, establishes a deformation compensation model and a body position correction model based on mattress material parameters, patient body position parameters, and environmental temperature and humidity parameters. It performs nonlinear compensation on the initial values ​​of height and weight. The deformation compensation model dynamically calculates the mattress deformation based on the pressure distribution characteristics, and the body position correction model performs geometric correction on the measured values ​​based on the patient's torso bending angle, generating calibrated accurate height and weight values.

[0014] The intelligent assessment and decision-making module, connected to the adaptive calibration and compensation module, calculates the BMI index based on accurate height and weight values. It then automatically generates VTE risk scores, fall risk scores, and self-care ability scores using a multi-level assessment algorithm, based on multi-dimensional patient information such as BMI, age, medical history, activity level, and medication history. Simultaneously, it generates assessment confidence parameters to indicate the reliability of the assessment results.

[0015] The medication-aid calculation module connects to the intelligent assessment and decision-making module. It receives accurate weight values, BMI index, and assessment confidence parameters. Based on drug type attributes and patient physiological characteristics, it calculates the recommended drug dose using a multi-mode dosing algorithm. The multi-mode dosing algorithm includes the overall weight algorithm, the ideal weight algorithm, and the lean body mass algorithm. The system automatically selects the appropriate algorithm based on the drug's fat solubility and the patient's BMI index, and generates dosage ranges and medication prompts.

[0016] The closed-loop feedback optimization module, connected to the medication assistance calculation module, collects patient medication records, adverse reaction data, and clinical outcome data, establishes a parameter optimization library, and dynamically adjusts the sensor weight coefficients of the weighted fusion algorithm, the compensation parameters of the deformation compensation model, and the scoring threshold of the multi-level evaluation algorithm through incremental learning algorithms. The optimized parameters are then fed back to the multi-sensor fusion measurement module and the adaptive calibration compensation module, forming a deeply coupled closed-loop feedback path of measurement-evaluation-medication-optimization.

[0017] Compared with the prior art, the present invention has the following beneficial effects:

[0018] This invention forms a complete closed-loop system of measurement-assessment-medication-optimization by deeply coupling a multi-sensor fusion measurement module, an adaptive calibration and compensation module, an intelligent assessment and decision-making module, a medication assistance calculation module, and a closed-loop feedback optimization module, thereby realizing integrated intelligent management of vital sign measurement and medication assistance for bedridden patients.

[0019] The multi-sensor fusion measurement module of this invention overcomes the problem of insufficient measurement accuracy of a single sensor by working together with a pressure sensor array, an ultrasonic sensor, and an accelerometer. The weighted fusion algorithm dynamically adjusts the weights according to the data quality of each sensor, improving the accuracy and stability of the measurement. Compared with traditional manual measurement, the height measurement error is reduced to within ±1.0cm, and the weight measurement error is reduced to within ±0.5kg.

[0020] The adaptive calibration compensation module of this invention establishes a deformation compensation model and a body position correction model, and performs dynamic compensation for differences in mattress material and changes in patient body position. It solves the systematic errors caused by mattress deformation and body curvature in bed measurements, and the measurement accuracy is improved by 15% to 20% after compensation. The improvement in measurement accuracy is particularly significant for obese patients and patients with scoliosis.

[0021] The intelligent assessment and decision-making module of this invention integrates four major clinical assessment tools: BMI calculation, VTE score, fall assessment, and self-care ability assessment. It achieves automated scoring through multi-level assessment algorithms, reducing the original manual assessment time of 5 to 10 minutes to less than 30 seconds, improving assessment efficiency by more than 90%. At the same time, it adds assessment confidence parameters to provide a reliable reference for clinical decision-making and reduce assessment bias and human error.

[0022] The medication-aid calculation module of this invention adopts a multi-mode dosage algorithm, which automatically selects the overall weight algorithm, ideal weight algorithm, or lean body mass algorithm according to the drug characteristics and the patient's physical condition. This avoids the possibility of drug overdose or underdose that may be caused by traditional single algorithms. In particular, for obese patients with a BMI greater than 28, the ideal weight algorithm can reduce the risk of drug overdose by 40% to 50%, thereby improving the safety and effectiveness of medication.

[0023] The closed-loop feedback optimization module of this invention continuously optimizes system parameters by collecting medication records and clinical outcome data and using incremental learning algorithms, thereby achieving self-evolution of system performance. As the usage time increases and data accumulates, the measurement accuracy, assessment accuracy, and rationality of dosage recommendations continue to improve. After accumulating data from 300 patients, the measurement accuracy can be further improved by 5% to 8%, and the assessment accuracy can be improved by 10% to 15%.

[0024] The various modules of this invention form a deep coupling relationship: the output of the multi-sensor fusion measurement module is the input of the adaptive calibration compensation module, the accurate data after calibration is the basis of the intelligent assessment and decision-making module, the assessment results and weight data jointly drive the medication assistance calculation module, the medication record is fed back to the closed-loop feedback optimization module, and the optimized parameters are fed back to the measurement and calibration module, forming a complete closed loop of positive transmission, performance evaluation, reverse feedback and parameter adjustment. The synergistic effect of each module produces a nonlinear synergy of 1+1>2, and the overall performance of the system far exceeds the simple superposition of the modules running independently.

[0025] The portable measuring device of the present invention adopts a split structure. The main unit and the sensor pad are connected by a connecting cable. The sensor pad can be placed directly under the bed sheet without moving the patient. Nursing staff can carry the main unit to quickly switch between different beds, which improves work efficiency. The whole machine weighs less than 2.0kg, supports single-person operation, and reduces the intensity of nursing labor.

[0026] This invention encrypts and stores patient privacy data through a data security module, using the AES256 encryption algorithm and TLS1.3 protocol, which complies with medical data security standards. The data access log recording function supports data traceability and auditing, meeting the requirements of medical quality management and legal compliance.

[0027] The abnormal detection and early warning function of this invention can automatically identify measurement abnormalities, nutritional risks, and fall risks, and promptly send early warning information to the terminal of the responsible nurse. This realizes the transformation from passive monitoring to proactive early warning, which helps to intervene early, reduce the incidence of adverse events, and improve the level of patient safety.

[0028] In summary, this invention, through the deep coupling of five core technologies—multi-sensor fusion, adaptive compensation, intelligent assessment, precision medication, and closed-loop optimization—constructs a complete closed-loop system encompassing measurement, assessment, medication, and optimization. This system enables intelligent, integrated, and precise management of vital sign measurement and medication assistance for bedridden patients, significantly improving measurement accuracy, assessment efficiency, and medication safety. It also reduces nursing workload and the risk of medical errors, demonstrating significant clinical application value and promising prospects for wider adoption. Attached Figure Description

[0029] Figure 1 This is a schematic diagram of the overall structure of the system of the present invention;

[0030] Figure 2 This is a schematic diagram of the structure of the multi-sensor fusion measurement module of the present invention;

[0031] Figure 3 This is a flowchart of the adaptive calibration compensation module of the present invention;

[0032] Figure 4 This is a flowchart of the evaluation process for the intelligent evaluation and decision-making module of this invention;

[0033] Figure 5 This is a flowchart of the algorithm selection for the medication assistance calculation module of this invention. Detailed Implementation

[0034] Please refer to the attached document. Figures 1-5 The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0035] Reference Figure 1This invention provides a multi-sensor fusion-based system for measuring vital signs and assisting medication administration in bedridden patients. It constructs a deeply coupled closed-loop collaborative system encompassing measurement, assessment, medication administration, and optimization. The entire system includes a multi-sensor fusion measurement module 1, an adaptive calibration and compensation module 2, an intelligent assessment and decision-making module 3, a medication assistance calculation module 4, a closed-loop feedback optimization module 5, and a display and interaction module 6. These modules are connected via a data bus, forming a tight parameter-level and state-level coupling relationship.

[0036] Reference Figure 2 The multi-sensor fusion measurement module 1, housed in the portable measurement device, includes a pressure sensor array 11, an ultrasonic sensor 12, an accelerometer 13, and a data fusion unit 14. The pressure sensor array 11 contains 20×30 pressure sensing points with a spacing of 2.0 cm. Each sensing point has a measurement range of 0 to 500 g / cm² and a sampling frequency of 50 Hz, used to collect pressure distribution data in the contact area between the bedridden patient's body and the mattress. The ultrasonic sensor 12 uses a 40 kHz ultrasonic transmitter and receiver, with a measurement range of 50 cm to 220 cm and a measurement accuracy of ±0.5 cm, used to measure the straight-line distance from the patient's head to their feet. The accelerometer 13 is a triaxial accelerometer with a measurement range of ±2 g, used to detect the attitude angle data of the measurement device, ensuring that the ultrasonic sensor is perpendicular to the patient's body.

[0037] The data fusion unit 14 runs a weighted fusion algorithm to comprehensively process pressure distribution data, straight-line distance data, and attitude angle data. The weighted fusion algorithm employs an adaptive weight allocation strategy, dynamically adjusting the weight coefficients based on the signal-to-noise ratio (SNR) and measurement stability of each sensor's data. Specifically, the system first calculates the SNR of each sensor's data, defined as the ratio of the effective signal amplitude to the noise standard deviation. For the pressure sensor array, the effective signal is the main peak value of the pressure distribution, and the noise is the pressure fluctuation at the edge sensing points. For the ultrasonic sensor, the effective signal is the echo peak value, and the noise is baseline drift. For the accelerometer, the effective signal is the gravitational acceleration component, and the noise is vibration interference.

[0038] The system calculates normalized weighting coefficients based on the signal-to-noise ratio. In a preferred embodiment, the weighting coefficient for the pressure sensor is set to 0.5, the weighting coefficient for the ultrasound sensor is set to 0.4, and the weighting coefficient for the accelerometer is set to 0.1, with the sum of the three weighting coefficients always equal to 1.0. When an abnormality is detected in the data of a certain sensor, such as the ultrasound sensor losing its echo due to patient movement, the system automatically reduces the weight of that sensor to 0.1, while increasing the weight of the pressure sensor to 0.7 and the weight of the accelerometer to 0.2 to ensure measurement continuity.

[0039] The initial weight value is obtained through a pressure sensor array. The system integrates and sums the pressure values ​​at all sensing points, estimating the patient's weight based on the contact area and pressure distribution characteristics. The initial height value is obtained through an ultrasonic sensor. The system measures the straight-line distance from the top of the head to the sole of the foot and performs geometric corrections based on the posture angle provided by the accelerometer. In a preferred embodiment, if the tilt angle of the measuring device is θ, the actual height is equal to the measured distance multiplied by cosθ. The data fusion unit 14 weights and fuses the pressure integral value, ultrasonic distance, and posture angle according to weighting coefficients to generate the initial height and initial weight values.

[0040] Reference Figure 3 The adaptive calibration compensation module 2 is connected to the multi-sensor fusion measurement module 1. It receives the initial values ​​of height and weight, and establishes a deformation compensation model and a position correction model based on the mattress material parameters, patient position parameters and environmental temperature and humidity parameters to perform nonlinear compensation on the initial measurement values.

[0041] The deformation compensation model compensates for the deformation of the mattress caused by the patient's weight. Mattress deformation leads to an underestimation of the weight measured by the pressure sensor array, and also causes the patient's body to sink, resulting in an underestimation of the height measured by ultrasound. The deformation compensation model establishes a non-linear mapping relationship between mattress deformation, weight, and pressure distribution area. In this embodiment, the system first calculates the contact area based on the pressure distribution data. The contact area is defined as the number of sensing points with pressure values ​​greater than the threshold of 50 g / cm² multiplied by a unit area of ​​2.0 cm × 2.0 cm. The system estimates the mattress deformation based on the contact area and an initial weight value.

[0042] The deformation compensation model employs a segmented compensation strategy. When the initial weight is less than 50kg, the mattress deformation is small, and the compensation coefficient is set between 1.02 and 1.05. The specific value is determined based on the mattress material's firmness: 1.02 for firm mattresses, 1.03 for medium-firm mattresses, and 1.05 for soft mattresses. When the initial weight is between 50kg and 80kg, the mattress deformation is moderate, and the compensation coefficient is set between 1.05 and 1.08. When the initial weight is greater than 80kg, the mattress deformation is large, and the compensation coefficient is set between 1.08 and 1.12. The calibrated weight value equals the initial weight multiplied by the compensation coefficient.

[0043] The postural correction model corrects for height measurement errors caused by patient body curvature. Bedridden patients may have postural problems such as scoliosis and lower limb flexion, resulting in ultrasound measurements of straight-line distances that are less than their actual height. The postural correction model detects the spinal curvature angle by analyzing the pressure distribution curve. The system integrates and projects the pressure distribution along the patient's longitudinal axis to form a one-dimensional pressure curve, and estimates the spinal curvature angle based on the degree of curvature of the curve.

[0044] In this embodiment, when the spinal curvature angle is less than 5°, the patient's position is considered standard and no correction is required. When the curvature angle is between 5° and 15°, the correction amount is 0.5% to 2.0% of the measured value, and the specific correction amount is calculated based on the linear interpolation of the curvature angle. For example, when the curvature angle is 10°, the correction amount is 1.25%, that is, the calibrated height value is equal to the initial height value multiplied by 1.0125. When the curvature angle is greater than 15°, the system determines that the patient's position is not standard and prompts the nursing staff on the display screen to adjust the patient's position and remeasure.

[0045] Furthermore, the adaptive calibration compensation module 2 also considers the impact of ambient temperature and humidity on sensor performance. The measurement accuracy of both the pressure sensor and the ultrasonic sensor is affected by temperature; therefore, the system integrates temperature and humidity sensors to monitor ambient temperature and humidity in real time. When the temperature is below 15°C or above 35°C, the system performs temperature compensation on the measured values. The compensation formula is determined based on the temperature coefficient provided by the sensor manufacturer. In a preferred embodiment, for every 1°C change in temperature, the pressure sensor measurement value changes by 0.1%, and the ultrasonic sensor measurement value changes by 0.05%. The system calculates the temperature compensation amount and corrects the measured values ​​based on the difference between the real-time temperature and the standard temperature of 25°C.

[0046] The adaptive calibration compensation module 2 outputs calibrated, precise height and weight values, which serve as the basis for subsequent evaluations and calculations. Internally, the module maintains a calibration parameter library, storing compensation coefficients for different mattress types and patient weight ranges. The system automatically matches the appropriate compensation coefficient based on actual measurements, achieving personalized and precise calibration.

[0047] Reference Figure 4 The intelligent assessment and decision-making module 3 is connected to the adaptive calibration and compensation module 2. It receives accurate height and weight values ​​and combines them with multi-dimensional information such as the patient's age, gender, medical history, activity level, and medication history. Through a multi-level assessment algorithm, it automatically generates BMI index, VTE risk score, fall risk score, and self-care ability score.

[0048] BMI is calculated using a standard formula: BMI equals weight divided by the square of height, where weight is measured in kg and height in m. The system determines a patient's nutritional status based on BMI: a BMI less than 18.5 indicates malnutrition, 18.5 to 23.9 is normal, 24.0 to 27.9 is overweight, and 28.0 or higher is obese. The system displays BMI in different colors on the screen: green for normal, yellow for overweight, and red for obesity or malnutrition, facilitating quick identification by nursing staff.

[0049] The multi-level assessment algorithm includes a base scoring layer, a weighted scoring layer, and a risk grading layer. The base scoring layer assigns initial scores to each risk factor, the weighted scoring layer dynamically adjusts the weights of each risk factor based on the individual characteristics of the patient, and the risk grading layer maps the total score into three levels: low risk, medium risk, and high risk.

[0050] The VTE risk score integrates the Caprini scoring system, which includes factors such as age, surgical history, malignancy, varicose veins, obesity, and bedridden time. Each factor is assigned a score ranging from 1 to 5 points based on its risk level. In this embodiment, the system adds mattress pressure distribution characteristics as a supplementary factor. By analyzing the symmetry of the pressure distribution and local pressure peaks, the system assesses the patient's lower limb blood circulation status. If the pressure distribution is highly asymmetrical or there are abnormally high pressure areas, it suggests potential circulatory disorders and an increased risk of VTE. The system then adds an additional 1 to 2 points to the baseline Caprini score.

[0051] The system employs an innovative VTE risk adaptive assessment algorithm, the specific formula of which is as follows:

[0052] ,

[0053] in, The overall risk score for VTE. The number of Caprini rating factors. For the first Adaptive weights of each factor For the first The basic score of each factor This is the pressure distribution influence coefficient, with a value ranging from 0.1 to 0.3. Pressure distribution asymmetry is defined as the ratio of the average pressure difference between the left and right sides to the total average pressure. In a preferred embodiment, The value is 0.2, when If the score is greater than 0.2, an additional 1 point will be awarded. If the score is greater than 0.4, an additional 2 points will be awarded.

[0054] Adaptive weights The calculation takes into account individual patient characteristics. For elderly patients, the weight of the age factor is increased by 10% to 20%. For obese patients, the weight of the obesity factor is increased by 15% to 25%. For long-term bedridden patients, the weight of the bed rest time factor is increased by 20% to 30%. The weight adjustment adopts normalization to ensure that the sum of all weights is always equal to 1.0. In a preferred embodiment, for a 75-year-old patient with a BMI of 30 who has been bedridden for more than 7 days, the weight of age is 0.3, the weight of obesity is 0.25, the weight of bed rest time is 0.25, and the sum of the weights of other factors is 0.2.

[0055] The system classifies risk based on the total VTE score, with scores of 0 to 2 indicating low risk, 3 to 4 indicating intermediate risk, and 5 or higher indicating high risk. For high-risk patients, the system automatically generates preventative recommendations, including the use of intermittent pneumatic compression devices, early mobilization, and anticoagulant prophylaxis, and sends warning information to the responsible nurse's terminal via a wireless communication interface.

[0056] The fall risk assessment integrates the Morse scoring system, which includes six factors: fall history, secondary diagnoses, assistive devices, intravenous infusion, gait, and mental status. In this embodiment, balance test data is added as a supplementary factor. The balance test assesses a patient's limb coordination and balance control by analyzing changes in pressure distribution as the patient moves in bed. Irregular movement of the center of pressure or excessively rapid movement indicates poor balance and an increased risk of falls.

[0057] The system employs an innovative multi-factor weighted assessment algorithm for fall risk, the specific formula of which is as follows:

[0058] ,

[0059] in, For the overall risk score of falling, The number of Morse rating factors. For the first The weights of each factor, For the first The basic score of each factor The balancing capacity influence coefficient ranges from 0.15 to 0.25. The balance capability index is calculated using the standard deviation of the pressure center's movement trajectory; a larger standard deviation indicates a poorer balance capability. The higher the value, the better. In a preferred embodiment, The value is 0.2, when When the value is greater than 15, an additional 10 points are awarded. If the value is greater than 30, an additional 15 points will be awarded.

[0060] The system classifies risks based on the total fall score: 0-24 points indicate low risk, 25-50 points indicate medium risk, and 51 points and above indicate high risk. For high-risk patients, the system automatically generates prevention recommendations, including bedside safety rails, non-slip mats, regular rounds, and assistance with toileting, and sends alerts to the responsible nurse's terminal.

[0061] The Barthel Index is used to assess a patient's self-care ability. It evaluates 10 activities of daily living, including eating, bathing, dressing, toileting, and walking. Each activity is assigned a score from 0 to 10 based on its level of independence, for a total score of 100. A higher score indicates stronger self-care ability. In this embodiment, the system incorporates patient activity monitoring data for dynamic assessment. Activity monitoring uses a pressure sensor array to track the frequency and amplitude of the patient's movements in bed over a long period, assessing the patient's voluntary activity ability. If the activity level is significantly lower than the average for the same age group, it suggests a possible decline in self-care ability, and the system accordingly lowers the Barthel score.

[0062] The system employs an innovative time-series evaluation algorithm for self-care ability, the specific formula of which is as follows:

[0063] ,

[0064] in, For a moment Self-care ability score The base score for the Barthel Index. This is the activity level attenuation coefficient, with a value ranging from 0.1 to 0.3. For a moment The patient's activity level was calculated by counting the number of times pressure distribution changed per unit time. The activity level is set based on the patient's age and health status, serving as a reference. It is the base of the natural logarithm. In a preferred embodiment, The value is 0.2. For a 70-year-old patient, Set to 10 activities per hour, based on actual activity levels. much smaller When the index term approaches 1, the self-care ability score is... Significantly lower than the base score .

[0065] The system categorizes patients based on their self-care ability score: 60 points or above indicates mild dependence, 40-59 points indicates moderate dependence, 20-39 points indicates severe dependence, and below 20 points indicates complete dependence. For patients with moderate or higher dependence, the system automatically generates nursing care plan suggestions, including measures such as assisting with eating, assisting with turning over, skin care, and psychological support.

[0066] The intelligent assessment and decision-making module 3 calculates the assessment confidence parameter while generating various scores. Assessment confidence reflects the reliability of the assessment results and is influenced by data completeness, sensor accuracy, and algorithm applicability. The system assigns an information completeness score based on the completeness of patient information, a measurement reliability score based on sensor measurement stability, and an algorithm applicability score based on the matching degree between patient characteristics and the algorithm training set. The weighted average of these three scores is the assessment confidence, ranging from 0 to 1.0. A higher confidence score indicates a more reliable assessment result. In a preferred embodiment, the information completeness weight is 0.4, the measurement reliability weight is 0.3, and the algorithm applicability weight is 0.3. When the assessment confidence is below 0.6, the system prompts nursing staff on the display screen to pay attention to the uncertainty of the assessment results and suggests combining clinical observation for comprehensive judgment.

[0067] Reference Figure 5 The medication assistance calculation module 4 is connected to the intelligent assessment and decision-making module 3. It receives accurate weight value, BMI index and assessment confidence parameters, and calculates the recommended drug dose based on drug type attributes and patient physiological characteristics using a multi-mode dosing algorithm.

[0068] The multi-modal dosing algorithm includes the overall weight algorithm, the ideal body weight algorithm, and the lean body mass algorithm. The system automatically selects the appropriate algorithm based on the drug's lipid solubility and the patient's BMI. Drug lipid solubility information is obtained from a drug database, which stores information such as lipid solubility parameters, therapeutic window parameters, and renal function impact parameters for commonly used drugs.

[0069] The total weight recalculation method is applicable to dosage calculations for water-soluble drugs, which are mainly distributed in the blood and extracellular fluid, and their dosage is directly proportional to total weight. The dosage calculation formula is: dosage equals standard dose multiplied by patient weight divided by standard weight 70 kg. In a preferred embodiment, for the aminoglycoside antibiotic gentamicin, the standard dose is 3 mg / kg daily. If the patient's weight is 80 kg, the recommended dose is 240 mg daily, divided into three intravenous injections of 80 mg each.

[0070] The ideal weight algorithm is applicable to dosage calculations for fat-soluble drugs. These drugs tend to accumulate in adipose tissue, and calculating the dosage based on total weight in obese patients could lead to overdose. The ideal weight calculation formula is determined based on height and gender. For men, the ideal weight equals 50 kg plus 0.9 multiplied by the portion of their height exceeding 150 cm; for women, it equals 45 kg plus 0.9 multiplied by the portion of their height exceeding 150 cm. For example, a male patient who is 170 cm tall has an ideal weight of 50 kg plus 0.9 multiplied by 20 cm, which equals 68 kg. The dosage calculation formula is: dosage equals standard dose multiplied by ideal weight divided by standard weight 70 kg. In a preferred embodiment, for the benzodiazepine sedative diazepam, the standard dose is 0.1 mg / kg daily. If a patient's actual weight is 95 kg and their BMI is 32, classifying them as obese, the system calculates their ideal weight as 68 kg. Therefore, the recommended dose is 6.8 mg daily, instead of the 9.5 mg calculated based on total weight, thus avoiding the risk of respiratory depression due to overdose.

[0071] The lean body mass algorithm is suitable for dosage calculation of drugs with a narrow therapeutic window. The effective dose of these drugs is close to the toxic dose, requiring more precise dosage calculations. Lean body mass is defined as total body weight minus fat weight, which is estimated using BMI. The formula for calculating lean body mass is as follows:

[0072] ,

[0073] in, Lean body mass For the overall weight, This refers to the Body Mass Index (BMI). This formula is applicable to patients with a BMI between 20 and 35. When the BMI is less than 20, lean body mass is approximately equal to total weight. When the BMI is greater than 35, a more complex body composition analysis method is required.

[0074] In a preferred embodiment, for the antitumor drug doxorubicin, the standard dose is 60 mg / m² per cycle. Based on body surface area, for a patient weighing 85 kg, 165 cm tall, with a BMI of 31.2, and a systematically calculated lean mass of 61.4 kg, the body surface area calculated using the Mosteller formula is 1.86 m², and the recommended dose is 112 mg per cycle. If the body surface area is calculated based on total weight as 2.05 m², the dose would reach 123 mg, which may increase the risk of cardiotoxicity due to excessively high doses.

[0075] The system automatically matches the algorithm mode based on the lipid solubility parameters and treatment window parameters in the drug database. When the drug's lipid solubility parameter LogP is greater than 2.0, the ideal weight algorithm is used first. When the drug's therapeutic index is less than 3.0, the lean body mass algorithm is used first. When the patient's BMI is greater than 28, the system automatically switches to either the ideal weight algorithm or the lean body mass algorithm to avoid drug overdose. The system displays the selected algorithm type and calculation basis on the screen, facilitating physician review and decision-making.

[0076] The dosage calculation results output by the medication assistance calculation module 4 include recommended dosage values, safe dosage ranges, and suggested dosing intervals. The safe dosage range is determined based on the drug instructions and clinical guidelines, generally ranging from 80% to 120% of the recommended dosage. Suggested dosing intervals are determined based on the drug's half-life and renal function. The system integrates a renal function assessment algorithm, calculating the glomerular filtration rate (GFR) based on the patient's age, gender, and serum creatinine level. When renal function is abnormal, the dosing interval is automatically adjusted; for example, if the GFR is less than 30 mL / min, the dosing interval is adjusted from three times daily to once daily to avoid drug accumulation and toxicity.

[0077] The system also provides medication tips, including precautions before medication, common adverse reactions, and drug interactions. These are extracted from the drug database and displayed on the screen to help doctors and nurses fully understand medication information and reduce the risk of medication errors.

[0078] The closed-loop feedback optimization module 5 is connected to the medication assistance calculation module 4. It collects patient medication records, adverse reaction data and clinical outcome data, establishes a parameter optimization library, dynamically adjusts system parameters through incremental learning algorithms, and sends the optimized parameters back to the multi-sensor fusion measurement module 1 and the adaptive calibration compensation module 2, forming a deeply coupled closed-loop feedback path of measurement-evaluation-medication-optimization.

[0079] The closed-loop feedback optimization module 5 collects data including: medication records, which record the types, dosages, routes of administration, and times of administration of medications actually used by the patient; adverse reaction data, which record adverse reactions such as nausea and vomiting, skin rashes, allergies, and abnormal liver and kidney function that occurred after medication, as well as their severity; and clinical outcome data, which record the patient's treatment effects, length of hospital stay, and occurrence of complications. This data is automatically acquired from the hospital information system via a wireless communication interface, or it can be manually entered by nursing staff through the interactive display module.

[0080] The parameter optimization library stores historical patient measurement data, assessment results, medication regimens, and clinical outcomes, forming a structured dataset. The dataset is categorized and indexed according to patient characteristics such as age, gender, diagnosis, and BMI for easy retrieval and analysis. In this embodiment, the parameter optimization library is stored using the relational database MySQL. The data tables include a patient basic information table, a measurement record table, an assessment record table, a medication record table, and a clinical outcome table, all linked by a unique patient identifier.

[0081] The incremental learning algorithm employs a sliding window mechanism, maintaining medication records and clinical outcome data for the most recent 300 patients. This mechanism ensures the system always optimizes parameters based on the latest data, preventing older data from interfering with current decisions. When the number of patients in the parameter optimization pool exceeds 300, the system automatically deletes the oldest data record, maintaining a constant window size.

[0082] The system calculates the deviation between the actual clinical outcome and the predicted result. Deviation is defined as the degree of match between the predicted risk level and the actual event. For example, if the system predicts a patient has a low risk of VTE, but the patient actually experiences a VTE event, then a prediction deviation exists. The system calculates the prediction accuracy rate, and triggers a parameter update when the accuracy rate falls below 90%. For medication dosage calculation, deviation is defined as the difference between the recommended dose and the final doctor's prescription dose. Parameter updates are triggered when the proportion of cases with a difference exceeding 20% ​​is greater than 10%.

[0083] The parameter updates employ gradient descent to progressively adjust the sensor weight coefficients and scoring thresholds. Gradient descent is an iterative optimization algorithm that calculates the gradient of the loss function with respect to each parameter and updates the parameters along the negative gradient direction, thereby gradually reducing the loss function. In this embodiment, the loss function is defined as the sum of squares of the prediction bias.

[0084] The update formula for the sensor weighting coefficients is as follows:

[0085] ,

[0086] in, For the updated number Each sensor weighting coefficient This is the current weighting coefficient. This represents the learning rate, with a value ranging from 0.01 to 0.05. For loss function, The partial derivative of the loss function with respect to the weight coefficients is calculated using the chain rule. In a preferred embodiment, The value is set to 0.02, which is a moderate update step size, ensuring both convergence speed and avoiding excessive oscillation.

[0087] The scoring threshold update formula is similar; the system adjusts the risk stratification threshold based on the actual VTE incidence and fall incidence. For example, if the actual VTE incidence in the intermediate-risk group is significantly higher than expected, the system lowers the cutoff threshold between intermediate and high risk, classifying more patients as high-risk and thus improving the coverage of preventative measures.

[0088] The parameter update cycle is 7 to 14 days. The system analyzes the data in the parameter optimization library weekly, calculates the deviation, and determines whether a parameter update is necessary. If the deviation exceeds a threshold for two consecutive cycles, the system performs a parameter update. If the deviation is within the threshold range, the system keeps the current parameters unchanged. After a parameter update, the system records the parameter evolution trajectory, including the update time, parameter value before the update, parameter value after the update, and changes in deviation, storing this information in the parameter optimization library and supporting parameter version rollback. If a parameter update is found to have caused a performance degradation, the system can revert to the previous parameter version.

[0089] The closed-loop feedback optimization module 5 transmits the optimized parameters back to the multi-sensor fusion measurement module 1 and the adaptive calibration compensation module 2. For the multi-sensor fusion measurement module 1, the system updates the sensor weight coefficients of the weighted fusion algorithm, making the measurement results more accurate. For the adaptive calibration compensation module 2, the system updates the compensation coefficients of the deformation compensation model and the correction parameters of the body position correction model, making the calibration effect more precise. Parameter transmission is achieved through a data bus, and standard communication protocols are used between modules to ensure the real-time performance and reliability of parameter transmission.

[0090] Through the continuous action of the closed-loop feedback optimization module 5, the system achieves a complete closed loop of measurement-assessment-medication-optimization. The output of the multi-sensor fusion measurement module 1 serves as the input to the adaptive calibration compensation module 2. The accurate data after calibration forms the basis of the intelligent assessment decision module 3. The assessment results and weight data jointly drive the medication-assisted calculation module 4. Medication records are fed back to the closed-loop feedback optimization module 5, and the optimized parameters are transmitted back to the measurement and calibration modules, forming a complete closed loop of forward transmission, performance evaluation, reverse feedback, and parameter adjustment. Not only is there data flow between the modules, but there is also deep coupling at the parameter and state levels. The output of one module directly affects the performance of the next, and the results of the subsequent module in turn affect the parameters of the preceding module, achieving a non-linear growth effect of mutual promotion, synergistic enhancement, and collaborative optimization.

[0091] The portable measuring device of this invention adopts a split structure, including a main unit 61, a sensor pad 62, and a connecting cable 63. The main unit 61 integrates a display interaction module 6 and a data processing unit. The display interaction module 6 includes a 7-inch touchscreen display and a wireless communication interface. The touchscreen display has a resolution of 1280×800 pixels, supports multi-touch, and uses a partitioned design: the left area displays real-time measurement values, the middle area displays scoring results and trend curves, and the right area displays medication suggestions and precautions. The wireless communication interface supports dual-mode communication via WiFi and Bluetooth. WiFi is used for data exchange with the hospital information system, and Bluetooth is used for quick pairing with doctor workstations and nurse terminals.

[0092] The data processing unit employs an ARM Cortex-A53 quad-core processor with a clock speed of 1.5GHz, 2GB of RAM, and 32GB of storage. Running a Linux operating system, it handles all computational tasks, including data fusion, calibration compensation, evaluation calculations, and closed-loop optimization. The unit has a built-in 10000mAh lithium battery, providing over 8 hours of continuous operation on a full charge, sufficient for a single work shift. The main unit's casing is made of medical-grade ABS plastic, making it waterproof and shockproof, meeting IP54 protection standards. Its surface is coated with an antibacterial layer and can be disinfected with 75% alcohol. The main unit is equipped with a handle for easy one-handed holding and carrying by nursing staff. Weighing only 1.8kg, it is lightweight and easy to use.

[0093] The sensor pad 62 comprises a pressure sensor array 11, an ultrasonic sensor 12, and an accelerometer 13. It measures 80cm x 60cm and is 8mm thick, made of medical-grade silicone material that is soft, comfortable, and breathable. The sensor pad can be placed directly under the bed sheet without moving the patient, ensuring their comfort. The edges of the sensor pad have anti-slip textures to prevent displacement during measurement. Positioning markings are printed on the surface of the sensor pad to indicate the correct placement of the head and feet, ensuring proper measurement posture.

[0094] The connecting cable 63 is 2.5m long, with one end connected to the data interface of the host unit 61 and the other end connected to the interface of the sensor pad 62. Internally, the connecting cable contains a power cable and a data cable. The power cable supplies power to the sensor pad, and the data cable transmits sensor data to the host unit. The connecting cable features a shielded design, providing strong anti-interference performance and ensuring stable and reliable data transmission. The connecting cable connector uses an aviation plug, which is easy to plug and unplug, has reliable contact, and a service life exceeding 10,000 plug-and-unplug cycles.

[0095] When nurses use the system of this invention for measurement, they first place the sensor pad 62 under the patient's bed sheet, ensuring that the patient's head and feet are aligned with the markings on the sensor pad. Then, they turn on the main unit 61, and the system automatically enters the measurement interface. Nurses input the patient's basic information on the touch screen, including name, age, gender, diagnosis, etc., or this information can be automatically obtained from the hospital information system via a wireless communication interface. After the information is entered, they click the "Start Measurement" button. The system then activates the multi-sensor fusion measurement module 1. The pressure sensor array 11 collects pressure distribution data, the ultrasonic sensor 12 measures height data, the accelerometer 13 detects posture angles, and the data fusion unit 14 performs weighted fusion of the data to generate initial height and weight values.

[0096] The adaptive calibration compensation module 2 receives initial height and weight values, and performs compensation corrections based on mattress type and patient position to generate accurate height and weight values. The entire measurement process takes approximately 30 seconds. After measurement, the touch screen displays the results in real time, with height accuracy ±1.0cm and weight accuracy ±0.5kg.

[0097] The intelligent assessment and decision-making module 3 automatically calculates the BMI index and, combined with the patient's age, medical history, and other information, generates a VTE risk score, a fall risk score, and a self-care ability score. The assessment process takes approximately 10 seconds, and the results are displayed in chart form in the central area of ​​the touch screen, with different risk levels indicated by different colors for easy understanding.

[0098] The medication dosage calculation module 4 provides doctors with medication dosage calculation assistance based on accurate weight and BMI. Nurses select the medication name on the touchscreen display, and the system automatically calculates the recommended dosage, displays the safe dosage range and suggested dosing intervals, and shows the calculation results on the right side of the touchscreen display. After reviewing the recommended dosage, doctors can confirm or modify it. The final dosage is then sent to the hospital information system via a wireless communication interface to generate an electronic medical order.

[0099] The closed-loop feedback optimization module 5 runs continuously in the background, periodically collecting medication records and clinical outcome data, updating the parameter optimization library, and optimizing system parameters. Nurses and doctors do not need to monitor the optimization process; it is completed automatically, ensuring continuous improvement in measurement and evaluation performance.

[0100] The interactive display module 6 supports one-click generation of assessment reports. These reports include basic patient information, measurement results, scoring results, medication recommendations, and nursing suggestions. The reports are in PDF format and can be sent to the doctor's workstation or printed via a wireless communication interface. As part of the patient's medical record, the assessment report is used for clinical decision-making and quality management.

[0101] The interactive display module 6 also includes multi-level access control. Regular nurses can view measurement results and assessment reports; responsible nurses can modify patient information and confirm medication recommendations; doctors can adjust assessment parameters and review system-recommended dosages; and system administrators can configure system parameters and export historical data. Access control is implemented through username and password login, with each user account corresponding to a specific access level. The system records user operation logs, including login time, operation content, and data modification records, supporting auditing and traceability.

[0102] The system of this invention also includes a data security module for encrypted storage and transmission of patient privacy data. Data encryption employs the AES256 encryption algorithm, with the encryption key set and changed periodically by the system administrator. The key length is 256 bits, providing high encryption strength and complying with medical data security standards. Wireless communication uses the TLS1.3 protocol to establish an encrypted channel, preventing data from being stolen or tampered with during transmission. The system includes a data access log recording function, recording the user, time, and content of each data query and modification operation. Log data is stored separately, cannot be deleted or tampered with, and supports data traceability and auditing. Local data is retained for 90 days; expired data is automatically archived to a cloud server. The cloud server uses redundant backups to ensure data security and reliability.

[0103] The system also includes an anomaly detection and early warning function. When the measured value fluctuates beyond a reasonable range, such as a change in height or weight exceeding 5%, the system automatically triggers a retest process, prompting nursing staff to check whether the sensor is placed correctly and whether the patient's posture is standard, and to remeasure and confirm the results. When the BMI index indicates that the patient is severely malnourished or obese, the system generates nutritional management recommendations, including nutritional risk screening, nutritional assessment, and nutritional support plans, and sends them to the nutrition department for consultation. When the VTE risk score or fall risk score reaches a high-risk level, the system automatically sends an early warning message to the responsible nurse's terminal. The early warning message includes the risk type, score, recommended intervention measures, and emergency contact information. After receiving the early warning, the responsible nurse must confirm and take appropriate measures within 15 minutes. The system records the early warning sending time and confirmation time for quality management and performance evaluation.

[0104] The system supports personalized configuration of warning thresholds, allowing different departments and patient groups to set different warning standards based on their specific circumstances. For example, the fall risk threshold is set low in the orthopedic ward, the VTE risk threshold is set low in the cardiology ward, and all risk thresholds in the intensive care unit are set low to ensure timely intervention for high-risk patients. Warning threshold configurations are set by the head nurse or attending physician and take effect after review by the system administrator.

[0105] Based on Example 1, this example provides an optimized solution specifically optimized for the measurement and assessment of obese patients.

[0106] For obese patients with a BMI greater than 30, the mattress deformation is significant, resulting in a larger deviation in the weight measured by the pressure sensors. This embodiment adds a secondary compensation mechanism to the deformation compensation model. First, a primary compensation is calculated based on the initial weight and contact area. Then, a secondary compensation is calculated based on the unevenness of the pressure distribution. The unevenness of the pressure distribution is defined as the ratio of the standard deviation to the average value of the pressure values ​​at each sensing point. A greater unevenness indicates higher local pressure peaks, more irregular mattress deformation, and a higher compensation coefficient required.

[0107] The formula for calculating the secondary compensation coefficient is as follows:

[0108] ,

[0109] in, This is the secondary compensation coefficient. This is a primary compensation coefficient, determined based on the initial body weight and contact area. The unevenness influence factor ranges from 0.05 to 0.15. The standard deviation of the pressure distribution. This represents the average pressure distribution. This refers to non-uniformity. In a preferred embodiment, The value is 0.1. When the non-uniformity is 0.3, the secondary compensation coefficient increases by 3% relative to the primary compensation coefficient.

[0110] For medication dosage calculation in obese patients, this embodiment adds a fat correction factor to the ideal weight algorithm. Studies have shown that the volume of distribution of certain fat-soluble drugs increases in obese patients, and simply using ideal weight to calculate the dosage may lead to underdosing. The fat correction factor is calculated based on the drug's fat solubility and the patient's BMI, and the corrected dose lies between the ideal weight dose and the total weight dose.

[0111] The formula for calculating fat correction dose is as follows:

[0112] ,

[0113] in, This is the recommended dose after fat correction. This is the dosage calculated based on ideal body weight. The dose is calculated based on total weight. The fat-correcting factor, with a value ranging from 0.2 to 0.5, is determined based on the drug's LogP value and the patient's BMI. A higher LogP value indicates greater fat solubility of the drug. The higher the value, the higher the BMI and the greater the body fat percentage. The larger the value, the better. In a preferred embodiment, for a moderately lipid-soluble drug with a LogP of 3.0, and an obese patient with a BMI of 32, The value is 0.4. If the ideal body weight dose is 100mg and the total body weight dose is 140mg, then the fat correction dose is 116mg.

[0114] This embodiment provides an application scenario in which the system is applied to a postoperative rehabilitation ward to continuously monitor and evaluate bedridden patients after surgery.

[0115] Postoperative patients have a higher risk of VTE and falls, requiring close monitoring. In this embodiment, the system measures and assesses patients daily, recording changes in weight, BMI, and risk score, and generating trend curves. Weight changes reflect the patient's nutritional status and edema; continuous weight loss suggests malnutrition or muscle atrophy, while rapid weight gain suggests edema or ascites. The system sets a weight change threshold; when a daily weight change exceeds 2 kg or a 7-day weight change exceeds 5 kg, a nutritional consultation request is automatically sent.

[0116] Changes in risk scores reflect patient recovery progress and complication risk. A decrease in fall risk score indicates improved mobility, while an increase in VTE risk score suggests the need for enhanced preventative measures. The system dynamically adjusts the nursing plan based on changes in risk scores. For patients with persistently rising risk scores, the frequency of rounds is increased, health education is strengthened, and physician consultations are sought to adjust the treatment plan when necessary.

[0117] In this embodiment, the system interfaces with the hospital's rehabilitation management platform, uploading measurement data and assessment results to the platform. The platform integrates rehabilitation assessment information from a multidisciplinary team, including the physical therapist's motor function assessment, the nutritionist's nutritional assessment, and the psychotherapist's psychological assessment, to form a comprehensive rehabilitation assessment report and develop a personalized rehabilitation plan for the patient.

[0118] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A multi-sensor fusion-based system for measuring vital signs and assisting medication administration in bedridden patients, characterized in that, include: The multi-sensor fusion measurement module includes a pressure sensor array, an ultrasonic sensor, and an accelerometer. The pressure sensor array is used to collect pressure distribution data of the contact area between the bedridden patient's body and the mattress. The ultrasonic sensor is used to measure the straight-line distance data from the top of the patient's head to the soles of their feet. The accelerometer is used to detect the posture angle data of the measuring device. The multi-sensor fusion measurement module uses a weighted fusion algorithm to comprehensively process the pressure distribution data, straight-line distance data, and posture angle data to generate initial values ​​for height and weight. An adaptive calibration and compensation module, connected to the multi-sensor fusion measurement module, establishes a deformation compensation model and a position correction model based on mattress material parameters, patient position parameters, and environmental temperature and humidity parameters. This module performs nonlinear compensation on the initial height and weight values. The deformation compensation model dynamically calculates the mattress deformation based on pressure distribution characteristics, and the position correction model geometrically corrects the measured values ​​based on the patient's body curvature angle, generating calibrated, accurate height and weight values. The intelligent assessment and decision-making module, connected to the adaptive calibration and compensation module, calculates the BMI index based on the precise height and weight values. Based on multi-dimensional patient information including BMI, age, medical history, activity level, and medication history, it automatically generates VTE risk scores, fall risk scores, and self-care ability scores through a multi-level assessment algorithm. Simultaneously, it generates assessment confidence parameters to indicate the reliability of the assessment results. The medication-aid calculation module is connected to the intelligent assessment and decision-making module. It receives the precise weight value, BMI index, and assessment confidence parameters. Based on the drug type attributes and patient physiological characteristics, it calculates the recommended drug dose using a multi-mode dosing algorithm. The multi-mode dosing algorithm includes an overall weight algorithm, an ideal weight algorithm, and a lean body mass algorithm. It automatically selects the appropriate algorithm based on the drug's fat solubility and the patient's BMI index, and generates a dosage range and medication prompts. The closed-loop feedback optimization module, connected to the medication assistance calculation module, collects patient medication records, adverse reaction data, and clinical outcome data, establishes a parameter optimization library, and dynamically adjusts the sensor weight coefficients of the weighted fusion algorithm, the compensation parameters of the deformation compensation model, and the scoring threshold of the multi-level evaluation algorithm through an incremental learning algorithm. The optimized parameters are then fed back to the multi-sensor fusion measurement module and the adaptive calibration compensation module, forming a deeply coupled closed-loop feedback path of measurement-evaluation-medication-optimization.

2. The system according to claim 1, characterized in that, It also includes a display interaction module, which is connected to the intelligent assessment and decision-making module and the medication assistance calculation module. The module includes a touch screen and a wireless communication interface. The touch screen displays the accurate height value, accurate weight value, BMI index, various risk scores and recommended drug dosage in real time. The wireless communication interface is used to exchange data with the hospital information system. The weighted fusion algorithm adopts an adaptive weight allocation strategy, which dynamically adjusts the weight coefficients according to the signal-to-noise ratio and measurement stability of each sensor data. The weight coefficients for pressure sensors range from 0.4 to 0.6, for ultrasonic sensors from 0.3 to 0.5, and for accelerometers from 0.1 to 0.

2. The sum of the three weight coefficients is always equal to 1.

0.

3. The system according to claim 1, characterized in that, The deformation compensation model establishes a nonlinear mapping relationship between mattress deformation and body weight and pressure distribution area. It adopts a piecewise compensation strategy. When the body weight is less than 50 kg, the compensation coefficient is 1.02 to 1.05; when the body weight is between 50 kg and 80 kg, the compensation coefficient is 1.05 to 1.08; and when the body weight is greater than 80 kg, the compensation coefficient is 1.08 to 1.

12. The body position correction model corrects the height measurement value by detecting the spinal curvature angle. When the spinal curvature angle is less than 5°, no correction is required; when the curvature angle is between 5° and 15°, the correction amount is 0.5% to 2.0% of the measured value.

4. The system according to claim 1, characterized in that, The multi-level assessment algorithm includes a basic scoring layer, a weighted scoring layer, and a risk grading layer. The basic scoring layer assigns initial scores to each risk factor. The weighted scoring layer dynamically adjusts the weights of each risk factor based on the individual characteristics of the patient. The risk grading layer maps the total score to three levels: low risk, medium risk, and high risk. The VTE risk score integrates the Caprini scoring standard and adds mattress pressure distribution characteristics as a supplementary factor. The fall risk score integrates the Morse scoring standard and adds balance ability test data as a supplementary factor.

5. The system according to claim 1, characterized in that, In the multi-mode dosing algorithm, the overall weight algorithm is suitable for dosing calculation of water-soluble drugs, the ideal weight algorithm is suitable for dosing calculation of fat-soluble drugs, and the lean body weight algorithm is suitable for dosing calculation of drugs with narrow therapeutic windows. The algorithm mode is automatically matched according to the fat solubility parameters and therapeutic window parameters in the drug database. When the patient's BMI index is greater than 28, the ideal weight algorithm or the lean body weight algorithm is preferred.

6. The system according to claim 1, characterized in that, The incremental learning algorithm uses a sliding window mechanism to maintain the medication records and clinical outcome data of the most recent 300 patients, calculates the deviation between the actual clinical outcome and the predicted result, and triggers parameter updates when the deviation exceeds a preset threshold of 10%. The parameter updates use the gradient descent method to gradually adjust the sensor weight coefficients and scoring thresholds, with an update step size of 0.01 to 0.05 and an update cycle of 7 to 14 days.

7. The system according to claim 1, characterized in that, The portable measuring device adopts a split structure, including a main unit, a sensor pad, and a connecting cable. The main unit integrates the display interaction module and the data processing unit. The sensor pad contains the pressure sensor array. The sensor pad is 5mm to 10mm thick and is placed under the bed sheet. The main unit is equipped with a handle and a drop-proof shell. The entire device weighs less than 2.0kg.

8. The system according to claim 1, characterized in that, The display and interaction module includes multi-level permission management functions. Ordinary nurses can view measurement results and assessment reports, responsible nurses can modify patient basic information and confirm medication recommendations, and doctors can adjust assessment parameters and review system-recommended doses. The display interface adopts a partitioned design: the left area displays real-time measurement values, the middle area displays scoring results and trend curves, and the right area displays medication recommendations and precautions.

9. The system according to claim 1, characterized in that, The system also includes a data security module for encrypting and storing and transmitting patient privacy data. It uses the AES256 encryption algorithm to encrypt sensitive data such as height and weight, and the TLS1.3 protocol to ensure wireless communication security. The system is equipped with a data access log recording function to record the user, time and content of each data query and modification operation.

10. The system according to claim 1, characterized in that, The system also includes an anomaly detection and early warning function. When the measured value fluctuates beyond a reasonable range, a retest process is automatically triggered. When the BMI index shows that the patient is in a state of severe malnutrition or obesity, nutritional management recommendations are generated. When the VTE risk score or fall risk score reaches a high risk level, an early warning message is automatically sent to the responsible nurse's terminal.

Images