Image Photoplethysmography (IPPG) signals are easily disturbed by noise during acquisition. To address the issue, this study proposes a denoising diffusion probability model for IPPG (DDPM-IPPG). This model eliminates baseline drift and noise through diffusion and reverse diffusion stages, and improves the signal-to-noise ratio and heart rate accuracy. First, Gaussian noise is gradually added to the photoplethysmography (PPG) signal during the diffusion phase to create a noise sequence. A noise predictor based on a nonlinear fusion module and a bridging module is trained. Subsequently, in the reverse diffusion phase, the well-trained noise predictor is employed to perform step-by-step denoising on the initially extracted IPPG signal. Through this denoising, a signal with high signal-to-noise ratio is recovered. The model proposed in this paper is validated and compared with current mainstream algorithms on the PURE, UBFC-IPPG, UBFC-Phys,and MMPD datasets. The experimental results show that DDPM-IPPG improves the signal-to-noise ratio by 1.06 dB on the PURE dataset comparing with the existing highest-precision extraction method. The mean absolute error of heart rate decreases by 0.24 bpm. The root mean square error of heart rate decreases by 0.41 bpm. On the UBFC-IPPG dataset, the signal-to-noise ratio is improved by 1.50 dB. The proposed DDPM-IPPG model has achieved the current advanced level in eliminating baseline drift and noise from IP-PG signals, enabling a more precise approximation of the true signals and providing a more reliable data foundation for physiological health assessment and telemedicine monitoring.

Blood oxygen saturation $(\text{SpO}_{2})$ is a critical vital sign, defined as the percentage of oxygenated hemoglobin relative to total hemoglobin in arterial blood. $\text{SpO} \mathrm{O}_{2}$ monitoring facilitates the early detection of physiological abnormalities and plays a significant role in disease prevention. This paper presents a non-contact measurement method for $S p O_{2}$ estimation under ambient illumination. Facial videos are captured using a smartphone camera, and remote photoplethysmography (rPPG) is employed to extract pulse wave signals. Red and blue channel signals are subsequently separated from the rPPG data. Wavelet decomposition and independent component analysis (ICA) are applied for signal denoising. The final $\text{SpO}_{2}$ values are computed based on the processed signals. Experimental results demonstrate that the proposed method achieves a mean absolute error (MAE) of 0.94% in $\text{SpO}_{2}$ estimation, satisfying the clinical requirement of $\pm 2 \%$ error tolerance. The proposed approach offers high measurement accuracy and robust stability, providing an effective solution for non-contact blood oxygen saturation monitoring.

Non-contact anxiety recognition methods based on facial videos offer a convenient and efficient approach for large-scale mental health screening. Most existing studies on anxiety recognition based on facial videos have mainly used facial expressions or remote photoplethysmography (rPPG). However, anxiety states manifest through multiple behavioral and physiological dimensions, such as fine-grained eye movement trajectories, pupil diameter, eye-blink, and rPPG signals. Therefore, single-view approaches do not fully capture information related to anxiety. To address this limitation, we propose a novel multi-view anxiety recognition framework based on facial videos. First, a multi-fusion attention net (MFA-Net) model is developed to estimate pupil diameter under complex facial video conditions, enabling precise reconstruction of pupil information. Then, we designed a multi-view fusion network (STFMF-Net) that integrates a Hybrid Fusion Module to effectively fuse the spatiotemporal and time-frequency features from eye movement signals, pupil diameter, eye-blink, head movement, and rPPG. The proposed method was evaluated on the UBFC-Phys public dataset. Experimental results indicate that our approach achieves an accuracy of 97.39% and an F1 score of 97.19%, outperforming the state-of-the-art (SOTA) by 3.06 and 2.80 percentage points, respectively. Therefore, the results highlight the effectiveness of multi-view fusion for non-contact, high-precision anxiety recognition and provide a promising technical solution for early anxiety assessment.

With the intensification of global aging and the increasing number of individuals in “sub-health” conditions, the importance of vital sign monitoring in disease prevention, diagnosis, and treatment is becoming increasingly prominent. This article designs and implements a Small-scale health monitoring system aimed at proactive health management, intended to support health management decisions through continuous monitoring of physiological indicators. The core hardware of the system uses an STM32 microcontroller integrated with small biomedical sensors to collect physiological signals such as electrocardiogram, respiration, pulse, and blood oxygen levels. The system is designed to prioritize small-scale through low-power circuit design and the integration of multiple parameters to achieve long-duration continuous monitoring. The collected data is wirelessly transmitted to a host computer for visualization, intelligent analysis, and storage. Experimental results indicate that the system can monitor heart rate, respiratory rate, blood oxygen saturation, blood perfusion index, and heart rate variability in real-time, with accuracy levels of $\pm 2$ bpm, $\pm 1$ bpm, $\pm 2\%$, and $\pm 5$ ms, respectively. The system's power consumption is below 10m W, with a battery life exceeding 24 hours, meeting the low power consumption requirements of wearable devices. The wearable health monitoring system designed in this study features characteristics such as multi-parameter integration, miniaturization, low power consumption, and intelligent analysis, making it suitable for fields such as smart elderly care and exercise monitoring.

Falling has become one of the greatest threats to the physical health of the elderly. Accurately predicting the risk of falling and timely and effective detection of falls are very important. The application of wearable devices in fall detection is receiving increasing attention. This paper uses the small biomedical sensor mpu9150 to collect three-axis acceleration data of the human body, and then uses neural networks to train a fall detection model based on the three-axis acceleration. Additionally, simulated fall experiments were conducted on 30 volunteers in a laboratory environment to build a fall database. The experimental results show that the accuracy of the fall detection system is 95.8%. The system designed in this paper has the features of wearability, miniaturization, low power consumption, and high accuracy, and is suitable for multiple fields such as sports monitoring and high-risk operations.

Blood pressure is an important physiological signal of the human body and serves as a crucial reference standard for the prevention of cardiovascular diseases, making blood pressure monitoring particularly significant. In response to the current challenges of blood pressure measurement, such as difficulty and insufficient accuracy, this paper utilizes Photoplethymography (PPG) to acquire physiological parameters and combines deep learning algorithms to predict blood pressure values. First, the PPG signals in the database are preprocessed by denoising and cutting single-cycle signals, followed by feature extraction. Finally, a convolutional neural network (CNN) algorithm is used to build a PPG-based blood pressure prediction model. Experimental results show that the CNN model proposed in this paper has a mean error (ME) of $\mathbf{9 4 9 5} \mathbf{~ m m H g}$ for systolic blood pressure (SBP) and 7347 mmHg for diastolic blood pressure (DBP), both within the ±5mmHg range. Additionally, in terms of standard deviation (SD) for SBP and DBP, the CNN model has values of 4.0165 mmHg and 2.7941 mmHg, respectively, both less than 8 mmHg. The results meet the international standards of AAMI, indicating high accuracy and reliability in practical applications.

Emotion recognition finds broad applications across psychology, computer science, and artificial intelligence. However, the intricacies of emotional states pose challenges, rendering single modality-based emotion recognition approaches less robust. In this study, we introduce a multi-view features fusion algorithm leveraging convolutional neural networks (CNNs) for enhanced emotion detection. Initially, imaging photoplethysmography (IPPG) signals are extracted from face videos. Subsequently, we employ heart rate variability (HRV) for feature extraction and deploy branch convolutional neural networks to achieve multi-view representation of emotional attributes within IPPG and facial video signals. Our methodology was validated using the DEAP public dataset, demonstrating that our approach attained accuracies of 72.37% and 70.82% for arousal and valence dimensions, respectively. Notably, our multi-view strategy enhances emotion recognition accuracy by 7.23% and 5.31% for arousal and valence, respectively, when contrasted with methodologies relying solely on facial expressions. This advancement underscores our method's capability to capture multimodal emotional expressions through facial videos without the necessity for additional sensors, thus significantly elevating the precision and robustness of emotion recognition endeavors.

Traditional hand rehabilitation devices present a challenge in providing personalized training that can lead to finger movements exceeding the safe range, resulting in secondary injuries. To address this issue, we introduce a soft rehabilitation training glove with the function of safety and personalization, which can allow patients to select training modes based on rehabilitation and provide real-time monitoring, as well as feedback on finger movement data. The inner glove is equipped with bending sensors to access the maximum/minimum angle of finger movement and to provide data for the safety of rehabilitation training. The outer glove contains flexible drivers, which can drive fingers for different modes of rehabilitation training. As a result, the rehabilitation glove can drive five fingers to achieve maximum extension/flexion angles of 15.65°/85.97°, 15.34°/89.53°, 16.78°/94.27°, 15.59°/88.82°, and 16.73°/88.65°, from thumb to little finger, respectively, and the rehabilitation training frequency can reach six times per minute. The safety evaluation result indicated an error within ±6.5° of the target-motion threshold. The reliability assessment yielded a high-intra-class correlation coefficient value (0.7763–0.9996). Hence, the rehabilitation glove can achieve targeted improvement in hand function while ensuring safety.

About 70% of Parkinson's disease patients have the initial symptoms of tremors at the end of upper limbs in the clinic, which seriously affects the normal work and life of patients. The severity of Parkinson's disease patients is evaluated clinically by doctors based on their experience, lacking objective evaluation criteria. It is particularly important to study an objective and fast tremor assessment method to assist doctors in the diagnosis and treatment of Parkinson's disease. In this paper, a recognition system of Parkinson's patients' hand function tremor based on machine learning is designed. Firstly, the acceleration sensor is used to collect the hand tremor signal, and then the median and band-pass filters are used to remove the noise. Next, the time-domain and frequency-domain characteristics of the tremors signal are extracted. Finally, BP neural network algorithm is used to classify the tremor degree into three categories. 12 volunteers were selected to carry out the system function experiment, and the results show that the system can achieve the classification of hand tremors, with an accuracy rate of 84.5%. The Parkinson's patient's hand tremor evaluation system designed in this paper has the advantages of low cost, small size, comfortable wearing, and high accuracy. It can assist clinical rehabilitation training and help doctors formulate scientific and reasonable rehabilitation training programs.

基于成像式光电容积描记技术(Imaging Photoplethysmography,IPPG)的情绪识别具有非接触式和客观性等优势.但基于IPPG信号的情绪识别还处于探索阶段,因IPPG信号质量容易受到影响,识别情绪的准确率还有待提高.为解决上述问题,选择额头作为感兴趣区域,通过小波变换和带通滤波提高信号质量,提取心率变异性特征,并使用支持向量机、K近邻、决策树和随机森林对特征进行分类.在DEAP数据库上进行实验,结果表明:支持向量机的情绪识别效果最好,在唤醒维度上的准确率为 61.09%,在效价维度上的准确率为 53.31%.该研究在远程医疗、人机交互等领域有广阔的应用前景.

The 3D quantitative analysis of facial morphology is of importance in plastic surgery (PS), which could help surgeons design appropriate procedures before conducting the surgery. We propose a system to simulate and guide the shaping effect analysis, which could produce a similar but more harmonious face simulation. To this end, first, the depth camera based on structured light coding is employed for facial 3D data acquisition, from which the point cloud data of multiple facial perspectives could be obtained. Next, the cascade regression tree algorithm is used to extract the esthetic key points of the face model and to calculate the facial features composed of the key points, such as the nose, chin, and eyes. Quantitative facial esthetic indexes are offered to doctors to simulate PS. Afterward, we exploit a face mesh metamorphosis based on finite elements. We design several morphing operators, including augmentation, cutting, and lacerating. Finally, the regional deformation is detected, and the operation effect is quantitatively evaluated by registering the 3D scanning model before and after the operation. The test of our proposed system and the simulation of PS operations find that the measurement error of facial geometric features is 0.458 mm, and the area is 0.65 mm(2). The ratings of the simulation outcomes provided by panels of PS prove that the system is effective. The manipulated 3D faces are deemed more beautiful compared to the original faces respecting the beauty canons such as facial symmetry and the golden ratio. The proposed algorithm could generate realistic visual effects of PS simulation. It could thus assist the preoperative planning of facial PS.

Hand dysfunctions in Parkinson's disease include rigidity, muscle weakness, and tremor, which can severely affect the patient's daily life. Herein, a multimodal sensor glove is developed for quantifying the severity of Parkinson's disease symptoms in patients' hands while assessing the hands' multifunctionality. Toward signal processing, various algorithms are used to quantify and analyze each signal: Exponentially Weighted Average algorithm and Kalman filter are used to filter out noise, normalization to process bending signals, K-Means Cluster Analysis to classify muscle strength grades, and Back Propagation Neural Network to identify and classify tremor signals with an accuracy of 95.83%. Given the compelling features, the flexibility, muscle strength, and stability assessed by the glove and the clinical observations are proved to be highly consistent with Kappa values of 0.833, 0.867, and 0.937, respectively. The intraclass correlation coefficients obtained by reliability evaluation experiments for the three assessments are greater than 0.9, indicating that the system is reliable. The glove can be applied to assist in formulating targeted rehabilitation treatments and improve hand recovery efficiency.

为了满足人工智能医疗对生物医学工程人才实践的需求,开发了一套集教学、科研、创新实践为一体的生物医学信号智能处理一体化实验平台.该平台包含生物医学信号采集硬件系统、层次递进式实验教学内容和立体化教学资源,集生物医学信号采集、处理和特征识别于一体,串联起生物医学工程核心知识点.该平台还提供基于心电信号的睡眠呼吸暂停综合征检测的教学案例,并用于教学实践.教学实践表明,该平台提高了学生实践和创新能力,加深了专业认知度,提升了科研素养,从而能更好地适应目前生物医疗产业升级的需求.

为了适应人工智能医疗领域对生物医学人才培养的需求,以人才培养模式的组成要素为切入点,剖析了目前地方高校"人工智能+生物医学"人才培养的现状和面临的困境.从培养目标、课程体系、师资队伍等教学要素与环节进行分析和探讨,提出了改革策略,包括重新定位深度融合人工智能的生物医学人才培养目标,优化深度融合人工智能的生物医学复合型培养课程体系,构建深度融合人工智能的"校企医"协同育人体系,提出"请进来+走出去+讲互通"的高水准的师资队伍建设思路.以长春理工大学为试点,充分发掘专业潜力,形成鲜明的"人工智能+生物医学"特色,以期为生物医学人才培养模式的更新和完善提供参考.

In order to objectively and quantitatively evaluate the facial morphology of patients in plastic surgery, the key points of facial aesthetics are extracted by interaction and specified face database, and the geometric features are calculated to obtain accurate quantitative face data, and the aesthetic evaluation model is established. Face assessment, eyebrows and eyes overall assessment and nose assessment were carried out for the subjects. The results showed that the model could objectively give the facial score of the subjects, find out the deviation from the standard, and provide objective and effective guidance for further plastic surgery.

在新文科建设背景下,"人工智能+法学"成为理工科高校复合型法学人才培养的创新选择.理工科高校"人工智能+法学"人才培养正处于初步探索阶段,在培养方案制定、教学模式改革与教学评价优化三方面仍有待提升,应当通过制定行之有效且符合"人工智能+法学"特点的培养方案、创新科技驱动法学教育模式以及构建智能差异化的法学教学测评体系,不断完善"人工智能+法学"人才培养体系.

以工程教育认证理念为指导原则,结合"以学生为中心""成果导向"和"持续改进"的认证理念,提出系统督导、闭环督导、纵向督导等督导内容和方法,拓宽本科教学督导工作内容的深度和广度,为本科工程教育专业认证提供有价值的信息,推进本科教学的高质量发展.

随着我国教育改革的深化,信息化管理是高校实验室发展的必然趋势.针对传统生物实验室暴露出的仪器药品管理不规范、信息孤岛等诸多问题,设计并开发了高校生物实验室仪器药品信息管理平台.该平台以C/S模式构建,采用VS2017集成开发环境,结合MySQL关系型数据库,实现了实验室的安全和规范化管理,提高了实验室的工作效率和管理水平,促进了实验室对外交流与信息共享,在学生创新和实践能力的培养发面发挥了重要作用.

生物医用金属材料在植入人体后,要求必须对人体无毒、无致敏、无不良反应,因此需要在临床前进行生物相容性试验.生物相容性是医用金属材料能否进行临床使用的重要指标.本文通过细胞毒性实验和动物体内毒性实验这两个方面对细晶粒铝合金金属的生物相容性进行测试,实验结果表明,细晶粒铝合金不具有细胞毒性,并且在动物体内无毒副作用.

Non-contact detection of various physiological parameters has attract great attention. In this paper, a method of estimating physiological parameters based on imaging photoplethysmography from videos of people's faces recorded by mobile phone is proposed. First, a "wavelet transform-principal component analysis-blind source separation" algorithm is proposed to extract the video's RGB three-channel pulse wave signal with a high signal-to-noise ratio. Then, the green channel signal is processed separately in the frequency and the time domains to estimate heart and respiratory rates. The pulse wave signals of the red and blue channels are processed, and combined with the oxygen saturation detected by an oximeter to perform data fitting, the best linear equation for estimating the oxygen saturation value from the facial video is found. Finally, the error of the estimation results of various physiological parameters under natural light is compared, and the estimation results of each parameter under three lighting environments are analyzed. The results show that under the three lighting environments, the average error of heart rate detection is 0.551 2 time/min, the average error of respiration rate is -0.632 1 time/min , and the average error of oxygen saturation is -0.2743%. In summary, the non-contact physiological parameter estimation method proposed in this paper is highly accurate, universally applicable and stable. Its estimation results are highly consistent with the measurement result of standard instruments, which meets the needs of daily physiological parameter measurement.