COVID-19 Detection Using Photoplethysmography and Neural Networks

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.

Automatic COVID-19 severity assessment from HRV

COVID-19 is known to be a cause of microvascular disease imputable to, for instance, the cytokine storm inflammatory response and the consequent blood coagulation. In this study, we propose a methodological approach for assessing the COVID-19 presence and severity based on Random Forest (RF) and Support Vector Machine (SVM) classifiers. Classifiers were applied to Heart Rate Variability (HRV) parameters extracted from photoplethysmographic (PPG) signals collected from healthy and COVID-19 affected subjects. The supervised classifiers were trained and tested on HRV parameters obtained from the PPG signals in a cohort of 50 healthy subjects and 93 COVID-19 affected subjects, divided into two groups, mild and moderate, based on the support of oxygen therapy and/or ventilation. The most informative feature set for every group's comparison was determined with the Least Absolute Shrinkage and Selection Operator (LASSO) technique. Both RF and SVM classifiers showed a high accuracy percentage during groups' comparisons. In particular, the RF classifier reached 94% of accuracy during the comparison between the healthy and minor severity COVID-19 group. Obtained results showed a strong capability of RF and SVM to discriminate between healthy subjects and COVID-19 patients and to differentiate the two different COVID-19 severity. The proposed method might be helpful for detecting, in a low-cost and fast fashion, the presence and severity of COVID-19 disease; moreover, these reasons make this method interesting as a starting point for future studies that aim to investigate its effectiveness as a possible screening method.

COVID-19 detection using a model of photoplethysmography (PPG) signals

Objective:Coronavirus disease 2019 (COVID-19) targets several tissues of the human body; among these, a serious impact has been observed in the microvascular system. The aim of this study was to verify the presence of photoplethysmographic (PPG) signal modifications in patients affected by COVID-19 at different levels of severity.

Approach: The photoplethysmographic signal was evaluated in 93 patients with COVID-19 of different severity (46: grade 1; 47: grade 2) and in 50 healthy control subjects. A pre-processing step removes the long-term trend and segments of each pulsation in the input signal. Each pulse is approximated with a model generated from a multi-exponential curve, and a Least Squares fitting algorithm determines the optimal model parameters. Using the parameters of the mathematical model, three different classifiers (Bayesian, SVM and KNN) were trained and tested to discriminate among healthy controls and patients with COVID, stratified according to the severity of the disease. Results are validated with the leave-one-subject-out validation method.

Main results: Results indicate that the fitting procedure obtains a very high determination coefficient (above 99% in both controls and pathological subjects). The proposed Bayesian classifier obtains promising results, given the size of the dataset, and variable depending on the classification strategy. The optimal classification strategy corresponds to 79% of accuracy, with 90% of specificity and 67% of sensibility.

Significance:The proposed approach opens the possibility of introducing a low cost and non-invasive screening procedure for the fast detection of COVID-19 disease, as well as a promising monitoring tool for hospitalized patients, with the purpose of stratifying the severity of the disease.

Classifying sepsis from photoplethysmography

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.

Cardiovascular Dynamics in COVID-19: A Heart Rate Variability Investigation

COVID-19 is known to be a cause of microvascular disease due, for example, to the cytokine storm inflammatory response and the result of blood coagulation. This study reports an investigation on Heart Rate Variability (HRV) extracted from photoplethysmography (PPG) signals measured from healthy subjects and COVID-19 affected patients. We aimed to determine a statistical difference between HRV parameters among subjects' groups. Specifically, statistical analysis through Mann-Whitney U Test (MWUT) was applied to compare 42 dif-ferent parameters extracted from PPG signals of 143 subjects: 50 healthy subjects (i.e. group 0) and 93 affected from COVID-19 patients stratified through increasing COVID severity index (i.e. groups 1 and 2). Results showed significant statistical differences between groups in several HRV parameters. In particular, Multiscale Entropy (MSE) analysis provided the master key in patient stratification assessment. In fact, MSE11, MSE12, MSE15, MSE16, MSE17, MSE18, MSE19 and MSE20 keep statistical significant difference during all the comparisons between healthy subjects and patients from all the pathological groups. Our preliminary results suggest that it could be possible to distinguish between healthy and COVID-19 affected subjects based on cardiovascular dynamics. This study opens to future evaluations in using machine learning models for automatic decision-makers to distinguish between healthy and COVID-19 subjects, as well as within COVID-19 severity levels. Clinical Relevance - This establishes the possibility to distin-guish healthy subjects from COVID-19 affected patients basing on HRV parameters monitored non invasively by PPG.

The CloudUPDRS smartphone software in Parkinson's study: cross-validation against blinded human raters

Digital assessments of motor severity could improve the sensitivity of clinical trials and personalise treatment in Parkinson’s disease (PD) but have yet to be widely adopted. Their ability to capture individual change across the heterogeneous motor presentations typical of PD remains inadequately tested against current clinical reference standards. We conducted a prospective, dual-site, crossover-randomised study to determine the ability of a 16-item smartphone-based assessment (the index test) to predict subitems from the Movement Disorder Society-Unified Parkinson’s Disease Rating Scale part III (MDS-UPDRS III) as assessed by three blinded clinical raters (the reference-standard). We analysed data from 60 subjects (990 smartphone tests, 2628 blinded video MDS-UPDRS III subitem ratings). Subject-level predictive performance was quantified as the leave-one-subject-out cross-validation (LOSO-CV) accuracy. A pre-specified analysis classified 70.3% (SEM 5.9%) of subjects into a similar category to any of three blinded clinical raters and was better than random (36.7%; SEM 4.3%) classification. Post hoc optimisation of classifier and feature selection improved performance further (78.7%, SEM 5.1%), although individual subtests were variable (range 53.2–97.0%). Smartphone-based measures of motor severity have predictive value at the subject level. Future studies should similarly mitigate against subjective and feature selection biases and assess performance across a range of motor features as part of a broader strategy to avoid overly optimistic performance estimates.