Search

Background: The COVID-19 pandemic has accelerated adoption of remote consulting in healthcare. Despite opportunities posed by telemedicine, most hypertension services in Europe have suspended ambulatory blood pressure monitoring (ABPM). Methods: We examined the process and performance of remotely delivered ABPM using two methodologies: firstly, a Failure Modes and Effects Analysis (FMEA) and secondly, a quantitative analysis comparing ABPM data from a subgroup of 65 participants of the Screening for Hypertension in the INpatient Environment (SHINE) diagnostic accuracy study. The FMEA was performed over seven sessions from February to March 2021, with a multidisciplinary team comprising a patient representative, a research coordinator with technical expertise and four research clinicians. Results: The FMEA identified a single high-risk step in the remote ABPM process. This was cleaning of monitoring equipment in the context of the COVID-19 pandemic, unrelated to the remote setting. A total of 14 participants were scheduled for face-to-face ABPM appointments, before the UK March 2020 COVID-19 lockdown; 62 were scheduled for remote ABPM appointments since emergence of the COVID-19 pandemic between November 2020 and August 2021. A total of 65 (88%) participants completed ABPMs; all obtained sufficient successful measurements for interpretation. For the 10 participants who completed face-to-face ABPM, there were 402 attempted ABPM measurements and 361 (89%) were successful. For the 55 participants who completed remote ABPM, there were 2516 attempted measurements and 2214 (88%) were successful. There was no significant difference in the mean per-participant error rate between face-to-face (0.100, SD 0.009) and remote (0.143, SD 0.132) cohorts (95% CI for the difference -0.125 to 0.045 and two-tailed P-value 0.353). Conclusions: We have demonstrated that ABPM can be safely and appropriately provided in the community remotely and without face-to-face contact, using video technology for remote fitting appointments, alongside courier services for delivery of equipment to participants.

There is a growing emphasis being placed on the potential for cuffless blood pressure (BP) estimation through modelling of morphological features from the photoplethysmogram (PPG) and electrocardiogram (ECG). However, the appropriate features and models to use remain unclear. We investigated the best features available from the PPG and ECG for BP estimation using both linear and non-linear machine learning models. We conducted a clinical study in which changes in BP ($$\Delta$$ Δ BP) were induced by an infusion of phenylephrine in 30 healthy volunteers (53.8% female, 28.0 (9.0) years old). We extracted a large and diverse set of features from both the PPG and the ECG and assessed their individual importance for estimating $$\Delta$$ Δ BP through Shapley additive explanation values and a ranking coefficient. We trained, tuned, and evaluated linear (ordinary least squares, OLS) and non-linear (random forest, RF) machine learning models to estimate $$\Delta$$ Δ BP in a nested leave-one-subject-out cross-validation framework. We reported the results as correlation coefficient ($$\rho _p$$ ρ p ), root mean squared error (RMSE), and mean absolute error (MAE). The non-linear RF model significantly ($$p p < 0.05 ) outperformed the linear OLS model using both the PPG and the ECG signals across all performance metrics. Estimating $$\Delta$$ Δ SBP using the PPG alone ($$\rho _p$$ ρ p = 0.86 (0.23), RMSE = 5.66 (4.76) mmHg, MAE = 4.86 (4.29) mmHg) performed significantly better than using the ECG alone ($$\rho _p$$ ρ p = 0.69 (0.45), RMSE = 6.79 (4.76) mmHg, MAE = 5.28 (4.57) mmHg), all $$p < 0.001$$ p < 0.001 . The highest ranking features from the PPG largely modelled increasing reflected wave interference driven by changes in arterial stiffness. This finding was supported by changes observed in the PPG waveform in response to the phenylephrine infusion. However, a large number of features were required for accurate BP estimation, highlighting the high complexity of the problem. We conclude that the PPG alone may be further explored as a potential single source, cuffless, blood pressure estimator. The use of the ECG alone is not justified. Non-linear models may perform better as they are able to incorporate interactions between feature values and demographics. However, demographics may not adequately account for the unique and individualised relationship between the extracted features and BP.

Circadian rhythms in blood pressure (BP) may in some cases be indicative of an increased risk of adverse cardiovascular events. However, current methods for assessing these rhythms can be disruptive to sleep, work, and daily activities. Features of the photoplethysmogram (PPG), which can be non-invasively and unobtrusively recorded, have been suggested as surrogate measures of BP. This work investigates the presence of a circadian rhythm in these features and evaluates their potential to classify nocturnal BP patterns. 742 patients who were discharged home from the ICU were selected from the MIMIC-III database. Our results show that a number of PPG features exhibit a clear and observable circadian rhythm. Of the 19 features evaluated, the circadian rhythms of 5 features outperformed heart rate (HR) in terms of correlation with the circadian rhythm of SBP ( ). We also present evidence that a metric combining the PPG features significantly improves BP phenotype classification accuracy. Clinical Relevance-This work suggests that a combined metric of PPG features may be able to accurately assess an individual's circadian rhythm of BP.

Background: Ambulatory blood pressure monitoring (ABPM) has become less frequent in primary care since the COVID-19 pandemic, with home blood pressure monitoring (HBPM) often the preferred alternative; however, HBPM cannot measure night-time blood pressure (BP) and patients whose night-time BP does not dip or rises (reverse-dipping) have poorer cardiovascular outcomes. Aim: To investigate the importance of measuring night-time BP when assessing individuals for hypertension. Design and Setting: Retrospective cohort study of two patient populations; the first being hospital patients admitted to four UK acute hospitals and the second being participants of the BP-Eth study recruited from 28 UK GP practices. Method: Using blood pressure data collected for the two cohorts, we studied three systolic blood pressure (SBP) phenotypes (dipper, non-dipper and reverse-dipper). Results: Among the hospital cohort (n=21,739) 48.9% patients were ‘reverse-dippers’, with an average day-night SBP difference of -8.0 mmHg. Among the community cohort (n=585) 10.8% patients were ‘reverse-dippers’ with an average day-night SBP difference of -8.5 mmHg. Non-dipper and reverse-dipper phenotypes both have lower day-time SBP and higher night-time SBP than the dipper phenotype. Average day-time SBP was lowest in the reverse-dipping phenotype (6.5 mmHg and 6.8 mmHg lower than for the dipper phenotype in the hospital and community cohorts, respectively), thus placing them at risk of undiagnosed, or masked hypertension. Conclusion: Not measuring night-time BP puts all groups other than dippers at risk of failure to identify hypertension. We recommend that GPs should offer ABPM to all patients aged ≥60 as a minimum, when assessing for hypertension.

Importance Inadequate management of elevated blood pressure (BP) is a significant contributing factor to maternal deaths. Self-monitoring of BP in the general population has been shown to improve the diagnosis and management of hypertension; however, little is known about its use in pregnancy. Objective To determine whether self-monitoring of BP in higher-risk pregnancies leads to earlier detection of pregnancy hypertension. Design, Setting, and Participants Unblinded, randomized clinical trial that included 2441 pregnant individuals at higher risk of preeclampsia and recruited at a mean of 20 weeks’ gestation from 15 hospital maternity units in England between November 2018 and October 2019. Final follow-up was completed in April 2020. Interventions Participating individuals were randomized to either BP self-monitoring with telemonitoring (n = 1223) plus usual care or usual antenatal care alone (n = 1218) without access to telemonitored BP. Main Outcomes and Measures The primary outcome was time to first recorded hypertension measured by a health care professional. Results Among 2441 participants who were randomized (mean [SD] age, 33 [5.6] years; mean gestation, 20 [1.6] weeks), 2346 (96%) completed the trial. The time from randomization to clinic recording of hypertension was not significantly different between individuals in the self-monitoring group (mean [SD], 104.3 [32.6] days) vs in the usual care group (mean [SD], 106.2 [32.0] days) (mean difference, −1.6 days [95% CI, −8.1 to 4.9]; P = .64). Eighteen serious adverse events were reported during the trial with none judged as related to the intervention (12 [1%] in the self-monitoring group vs 6 [0.5%] in the usual care group). Conclusions and Relevance Among pregnant individuals at higher risk of preeclampsia, blood pressure self-monitoring with telemonitoring, compared with usual care, did not lead to significantly earlier clinic-based detection of hypertension. Trial Registration ClinicalTrials.gov Identifier: NCT03334149

Rapid antigen tests in the form of lateral flow devices (LFDs) allow testing of a large population for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). To reduce the variability in device interpretation, we show the design and testing of an artifical intelligence (AI) algorithm based on machine learning. The machine learning (ML) algorithm is trained on a combination of artificially hybridized LFDs and LFD data linked to quantitative real-time PCR results. Participants are recruited from assisted test sites (ATSs) and health care workers undertaking self-testing, and images are analyzed using the ML algorithm. A panel of trained clinicians is used to resolve discrepancies. In total, 115,316 images are returned. In the ATS substudy, sensitivity increased from 92.08% to 97.6% and specificity from 99.85% to 99.99%. In the self-read substudy, sensitivity increased from 16.00% to 100% and specificity from 99.15% to 99.40%. An ML-based classifier of LFD results outperforms human reads in assisted testing sites and self-reading.

We investigate the problem of automatic cardiomegaly diagnosis. We approach this by developing classifiers using multimodal data enhanced by two image-derived digital biomarkers, the cardiothoracic ratio (CTR) and the cardiopulmonary area ratio (CPAR). The CTR and CPAR values are estimated using segmentation and detection models. These are then integrated into a multimodal network trained simultaneously on chest radiographs and ICU data (vital sign values, laboratory values and metadata). We compare the predictive power of different data configurations with and without the digital biomarkers. There was a negligible performance difference between the XGBoost model containing only CTR and CPAR (accuracy 81.4%, F1 0.859, AUC 0.810) and black-box models which included full images (ResNet-50: accuracy 81.9%, F1 0.874, AUC 0.767; Multimodal: 81.9%, F1 0.873, AUC 0.768). We concluded that models incorporating domain knowledge-based digital biomarkers CTR and CPAR provide comparable performance to black-box multimodal approaches with the former providing better clinical explainability.

Introduction Physical activity (PA) interventions have a promising role in the management of gestational diabetes mellitus (GDM). Digital technologies can support PA at scale and remotely. The protocol describes a study designed to determine the feasibility and acceptability of a complex intervention; known as +Stay Active. +Stay Active combines motivational interviewing with a bespoke behaviour change informed smartphone application (Stay-Active) to augment PA levels in women with GDM. Methods and analysis This is a non-randomised feasibility study using a mixed methods approach. Participants will be recruited from the GDM antenatal clinic at the Women Centre, John Radcliffe Hospital, Oxford. Following baseline assessments (visit 1) including self-reported and device determined PA assessment (wearing a wrist accelerometer), women will be invited to participate in an online motivational interview, then download and use the Stay-Active app (Android or iOS) (visit 2). Women will have access to Stay-Active until 36 weeks gestation, when engagement and PA levels will be reassessed (visit 3). The target sample size is 60 women. Primary outcomes are recruitment and retention rates, compliance and assessment of participant engagement and acceptability with the intervention. Secondary outcomes are assessment of blood glucose control, self-reported and device determined assessment of PA, usage and structured feedback of participant’s attitudes to +Stay Active, assessment of health costs and description of maternal and neonatal outcomes. This study will provide key insights into this complex intervention regarding engagement in smartphone technology and the wearing of accelerometers. These data will inform the development of a randomised controlled trial with refinements to intervention implementation. Ethics and dissemination The study has received a favourable opinion from South Central—Hampshire B Research Ethics Committee; REC reference: 20/SC/0342. Written informed consent will be obtained from all participants. Findings will be disseminated through peer-reviewed journals, conferences and seminar presentations. Trial registration number NCT11366562.

There is an increasing need for fast and accurate transfer of readings from blood glucose metres and blood pressure monitors to a smartphone mHealth application, without a dependency on Bluetooth technology. Most of the medical devices recommended for home monitoring use a seven-segment display to show the recorded measurement to the patient. We aimed to achieve accurate detection and reading of the seven-segment digits displayed on these medical devices using an image taken in a realistic scenario by a smartphone camera. A synthetic dataset of seven-segment digits was developed in order to train and test a digit classifier. A dataset containing realistic images of blood glucose metres and blood pressure monitors using a variety of smartphone cameras was also created. The digit classifier was evaluated on a dataset of seven-segment digits manually extracted from the medical device images. These datasets along with the code for its development have been made public. The developed algorithm first preprocessed the input image using retinex with two bilateral filters and adaptive histogram equalisation. Subsequently, the digit segments were automatically located within the image by two techniques operating in parallel: Maximally Stable Extremal Regions (MSER) and connected components of a binarised image. A filtering and clustering algorithm was then designed to combine digit segments to form seven-segment digits. The resulting digits were classified using a Histogram of Orientated Gradients (HOG) feature set and a neural network trained on the synthetic digits. The model achieved 93% accuracy on digits found on the medical devices. The digit location algorithm achieved a F1 score of 0.87 and 0.80 on images of blood glucose metres and blood pressure monitors respectively. Very few assumptions were made of the locations of the digits on the devices so that the proposed algorithm can be easily implemented on new devices.

Additional file 1. Collaborating Centres and supplementary data table.

Additional file 3. Final process evaluation interview guide.

Additional file 2. Baseline process evaluation interview guide.

Skin temperature has long been used as a natural indicator of vascular diseases in the extremities. Considerable correlation between oscillations in skin surface temperature and oscillations of skin blood flow has previously been demonstrated. We hypothesised that the impairment of blood flow in stenotic (subcutaneous) peripheral arteries would influence cutaneous temperature such that, by measuring gradients in the temperature distribution over skin surfaces, one may be able to diagnose or quantify the progression of vascular conditions in whose pathogenesis a reduction in subcutaneous blood perfusion plays a critical role (e.g. peripheral artery disease). As proof of principle, this study investigates the local changes in the skin temperature of healthy humans (15 male, 30.0 ± 5.2 years old, BMI 25.1 ± 2.2 kg/m 2 ) undergoing two physical challenges designed to vary their haemodynamic status. Skin temperature was measured in four central regions (forehead, neck, chest, and left shoulder) and four peripheral regions (left upper arm, forearm, wrist, and hand) using an infrared thermal camera. We compare inter-region patterns. Median temperature over the peripheral regions decreased from baseline after both challenges (maximum decrease: -2.09 ± 0.41 °C at 60 s after exercise; p = 0.0001 and -0.58 ± 0.14 °C at 180 s of cold-water immersion; p = 0.0013). Median temperature over the central regions showed no significant changes. Our results show that the non-contact measurement of perfusion-related changes in peripheral temperature from infrared video data is feasible. Further research will be directed towards the thermographic study of patients with symptomatic peripheral vascular disease.

Aim The National Early Warning System (NEWS) is based on vital signs; the Laboratory Decision Tree Early Warning Score (LDT-EWS) on laboratory test results. We aimed to develop and validate a new EWS (the LDTEWS:NEWS risk index) by combining the two and evaluating the discrimination of the primary outcome of unanticipated intensive care unit (ICU) admission or in-hospital mortality, within 24 hours. Methods We studied emergency medical admissions, aged 16 years or over, admitted to Oxford University Hospitals (OUH) and Portsmouth Hospitals (PH). Each admission had vital signs and laboratory tests measured within their hospital stay. We combined LDT-EWS and NEWS values using a linear time-decay weighting function imposed on the most recent blood tests. The LDTEWS:NEWS risk index was developed using data from 5 years of admissions to PH, and validated on a year of data from both PH and OUH. We tested the risk index’s ability to discriminate the primary outcome using the c-statistic. Results The development cohort contained 97,933 admissions (median age = 73 years) of which 4,723 (4.8%) resulted in in-hospital death and 1,078 (1.1%) in unanticipated ICU admission. We validated the risk index using data from PH (n = 21,028) and OUH (n = 16,383). The risk index showed a higher discrimination in the validation sets (c-statistic value (95% CI)) (PH, 0.901 (0.898-0.905); OUH, 0.916 (0.911–0.921)), than NEWS alone (PH, 0.877 (0.873–0.882); OUH, 0.898 (0.893–0.904)). Conclusions The LDTEWS:NEWS risk index increases the ability to identify patients at risk of deterioration, compared to NEWS alone.

Vital signs recorded at the hospital bedside manually by clinical staff are key indicators of patient physiology and may be used to track patient deterioration. The low frequency of vital-sign observations by clinical staff (every 4, 8 or 12 hours) makes it difficult to determine the underlying distribution for each vital sign. In this paper we demonstrate how a Bayesian approach may be used to estimate the unknown parameters of vital sign data.

Objectives To calculate fractional inspired oxygen concentration (FiO2) thresholds in ward patients and add these to the National Early Warning Score (NEWS). To evaluate the performance of NEWS-FiO2 against NEWS when predicting in-hospital death and unplanned intensive care unit (ICU) admission. Methods A multi-centre, retrospective, observational cohort study was carried out in five hospitals from two UK NHS Trusts. Adult admissions with at least one complete set of vital sign observations recorded electronically were eligible. The primary outcome measure was an ‘adverse event’ which comprised either in-hospital death or unplanned ICU admission. Discrimination was assessed using the Area Under the Receiver Operating Characteristic curve (AUROC). Results A cohort of 83,304 patients from a total of 271,363 adult admissions were prescribed oxygen. In this cohort, NEWS-FiO2 (AUROC 0.823, 95% CI 0.819–0.824) outperformed NEWS (AUORC 0.811, 95% CI 0.809–0.814) when predicting in-hospital death or unplanned ICU admission within 24 h of a complete set of vital sign observations. Conclusions NEWS-FiO2 generates a performance gain over NEWS when studied in ward patients requiring oxygen. This warrants further study, particularly in patients with respiratory disorders.

Knowledge of the pathological instabilities in the breathing pattern can provide valuable insights into the cardiorespiratory status of the critically-ill infant as well as their maturation level. This paper is concerned with the measurement of respiratory rate (RR) in premature infants. We compare the rates estimated from (a) the chest impedance pneumogram, (b) the ECG-derived respiratory rhythms, and (c) the PPG-derived respiratory rhythms against those measured in the reference standard of breath detection provided by attending clinical staff during 165 manual breath counts. We demonstrate that accurate RR estimates can be produced from all sources for RR in the 40 - 80 bpm (breaths per min) range. We also conclude that the use of indirect methods based on the ECG or the PPG poses a fundamental challenge in this population due to their poor behaviour at fast breathing rates (upwards of 80 bpm).