Sara Lombardi

Objective . Vascular ageing (VA) is an age-related process characterised by vessel stiffening and altered endothelial function. Recognised as a major predictor of cardiovascular events, VA remains under-assessed in clinical practice due to limitations of current measurement systems. Photoplethysmography (PPG) emerges as a promising non-invasive, cost-effective tool, but its application to VA is hindered by signal complexity and lack of standardisation. This study investigated the relationship between VA and PPG waveform morphology through combined in vitro and in vivo analyses. Approach . In vitro , PPG signals were acquired from vascular phantoms with 3 stiffness levels. in vivo , PPG signals from 31 healthy subjects in two age groups ( < 30 years and > 50 years) were analysed, assuming increased stiffness with age. A novel skew-Gaussian decomposition model was developed to reconstruct pulses and extract morphological features. In vitro feature differences were assessed using the Kruskal–Wallis test; in vivo data were analysed with an XGBoost classifier to discriminate age groups. Main results . The skew-Gaussian model exhibited excellent waveform reconstruction ( R mean 2 > 0.98 ). In vitro , 23 of 60 features differed significantly across stiffness levels, particularly between the stiffest phantom and the others. Amplitude and energy parameters correlated negatively with stiffness, indicating reduced vascular compliance. Temporal features related to wave reflection correlated positively, suggesting altered signal dynamics. in vivo , the XGBoost classifier achieved 87% mean accuracy, with 83% sensitivity and 92% specificity in differentiating age groups. Significance . Feature relevance analysis confirmed the physiological significance of the extracted parameters, supporting the potential of PPG as a non-invasive and scalable technology for monitoring VA.

New technologies, including artificial intelligence (AI), offer significant opportunities to improve blood glucose level (BGL) estimation systems, potentially enhancing care and quality of life for diabetic patients. This study aimed to assess the accuracy of BGL estimation using photoplethysmographic signal (PPG) analysis and AI methods by comparing various studies in terms of population, PPG signal acquisition and analysis, AI approaches, and BGL estimation performance. A systematic search was conducted in Scopus, Web of Science, Embase, PubMed and CINAHL databases. Conference proceedings and book chapters were included, excluding other gray literature, focusing on English-language studies published from 2010 to February 2024. Only publications concerning PPG signal analysis using AI algorithms for noninvasive estimation of BGL in patients with diabetes were considered. Of 48 identified articles, 24 were reviewed in full text, and 5 were deemed eligible. These studies varied in methodology (populations, devices, AI solutions) and evaluation metrics. However, all studies used Clarke error grid or Parkes error grid, with over 98% of estimates falling into clinically acceptable zones A or B. Current research confirm that PPG-based BGL estimation is feasible and accurate. Further studies are needed to overcome existing limitations and make this procedure available, accurate, and easy to perform.

Augmented and mixed reality in the medical field is becoming increasingly important. The creation and visualization of digital models similar to reality could be a great help to increase the user experience during augmented or mixed reality activities like surgical planning and educational, training and testing phases of medical students. This study introduces a technique for enhancing a 3D digital model reconstructed from cone-beam computed tomography images with its real coloured texture using an Intel D435 RGBD camera. This method is based on iteratively projecting the two models onto a 2D plane, identifying their contours and then minimizing the distance between them. Finally, the coloured digital models were displayed in mixed reality through a Microsoft HoloLens 2 and an application to interact with them using hand gestures was developed. The registration error between the two 3D models evaluated using 30,000 random points indicates values of: 1.1 ± 1.3 mm on the x-axis, 0.7 ± 0.8 mm on the y-axis, and 0.9 ± 1.2 mm on the z-axis. This result was achieved in three iterations, starting from an average registration error on the three axes of 1.4 mm to reach 0.9 mm. The heatmap created to visualize the spatial distribution of the error shows how it is uniformly distributed over the surface of the pointcloud obtained with the RGBD camera, except for some areas of the nose and ears where the registration error tends to increase. The obtained results indicate that the proposed methodology seems effective. In addition, since the used RGBD camera is inexpensive, future approaches based on the simultaneous use of multiple cameras could further improve the results. Finally, the augmented reality visualization of the obtained result is innovative and could provide support in all those cases where the visualization of three-dimensional medical models is necessary.

Facial emotion recognition is a crucial process for many applications and is still unresolved. Historically, emotional recognition has usually been achieved through artificial intelligence techniques such as Convolutional Neural Networks. However, this approach is quite expensive in terms of computational power and complexity. To alleviate this problem, we propose a lightweight CNN for facial emotion recognition, called Custom Lightweight CNN-based Model (CLCM), based on the well-known MobileNetV2 architecture. Performance was evaluated on four public datasets, FER-2013, RAF-DB, AffectNet, and CK+, where seven facial emotions were detected. The CLCM model was compared to the well-known MobileNetV2 and ShuffleNetV2 architectures. The CLCM performance results were close to or better than the more complex models. Specifically, in the FER-2013 dataset, CLCM achieved an accuracy of 63%, while MobileNetV2 was 58%, and ShuffleNetV2 65%. In the RAF-DB dataset, CLCM showed 84%, MobileNetV2 73%, and ShuffleNetV2 80%. Lastly, in the AffectNet dataset, MobileNetV2 and ShuffleNetV2 showed an accuracy of 57%, while CLCM showed 54%. The results obtained from this study establish CLCM as one of the efficient models in FER. Although CLCM is a smaller model in terms of parameters (2.3 Million) compared to MobileNetV2 (3.5 Million) and ShuffleNetV2 (3.9 Million), it showed good results in almost all analyses. The reduced CLCM’s computational power allows its application in enhanced human-computer interaction, affective computing, and personalized user experience, especially for real-world scenarios such as psychological and medical assessment, automotive (real-time driver emotional state), and vulnerable individual care where limited resource systems are commonly employed and real-time and reliable response are strongly recommended.

The early identification of microvascular changes in patients with Coronavirus Disease 2019 (COVID-19) may offer an important clinical opportunity. This study aimed to define a method, based on deep learning approaches, for the identification of COVID-19 patients from the analysis of the raw PPG signal, acquired with a pulse oximeter. To develop the method, we acquired the PPG signal of 93 COVID-19 patients and 90 healthy control subjects using a finger pulse oximeter. To select the good quality portions of the signal, we developed a template-matching method that excludes samples corrupted by noise or motion artefacts. These samples were subsequently used to develop a custom convolutional neural network model. The model accepts PPG signal segments as input and performs a binary classification between COVID-19 and control samples. The proposed model showed good performance in identifying COVID-19 patients, achieving 83.86% accuracy and 84.30% sensitivity (hold-out validation) on test data. The obtained results indicate that photoplethysmography may be a useful tool for microcirculation assessment and early recognition of SARS-CoV-2-induced microvascular changes. In addition, such a noninvasive and low-cost method is well suited for the development of a user-friendly system, potentially applicable even in resource-limited healthcare settings.

Purpose: Sepsis is a life-threatening organ dysfunction. It is caused by a dysregulated immune response to an infection and is one of the leading causes of death in the intensive care unit (ICU). Early detection and treatment of sepsis can increase the survival rate of patients. The use of devices such as the photoplethysmograph could allow the early evaluation in addition to continuous monitoring of septic patients. The aim of this study was to verify the possibility of detecting sepsis in patients from whom the photoplethysmographic signal was acquired via a pulse oximeter. In this work, we developed a deep learning-based model for sepsis identification. The model takes a single input, the photoplethysmographic signal acquired by pulse oximeter, and performs a binary classification between septic and nonseptic samples. To develop the method, we used MIMIC-III database, which contains data from ICU patients. Specifically, the selected dataset includes 85 septic subjects and 101 control subjects. The PPG signals acquired from these patients were segmented, processed and used as input for the developed model with the aim of identifying sepsis. The proposed method achieved an accuracy of 76.37% with a sensitivity of 70.95% and a specificity of 81.04% on the test set. As regards the ROC curve, the Area Under Curve reached a value of 0.842. The results of this study indicate how the plethysmographic signal can be used as a warning sign for the early detection of sepsis with the aim of reducing the time for diagnosis and therapeutic intervention. Furthermore, the proposed method is suitable for integration in continuous patient monitoring.