Marco A. F. Pimentel

Gaussian Processes for Personalized e-Health Monitoring With Wearable Sensors

Advances in wearable sensing and communications infrastructure have allowed the widespread development of prototype medical devices for patient monitoring. However, such devices have not penetrated into clinical practice, primarily due to a lack of research into “intelligent” analysis methods that are sufficiently robust to support large-scale deployment. Existing systems are typically plagued by large false-alarm rates, and an inability to cope with sensor artifact in a principled manner. This paper has two aims: 1) proposal of a novel, patient-personalized system for analysis and inference in the presence of data uncertainty, typically caused by sensor artifact and data incompleteness; 2) demonstration of the method using a large-scale clinical study in which 200 patients have been monitored using the proposed system. This latter provides much-needed evidence that personalized e-health monitoring is feasible within an actual clinical environment, at scale, and that the method is capable of improving patient outcomes via personalized healthcare.

Predictive Monitoring of Mobile Patients by Combining Clinical Observations With Data From Wearable Sensors

The majority of patients in the hospital are ambulatory and would benefit significantly from predictive and personalized monitoring systems. Such patients are well suited to having their physiological condition monitored using low-power, minimally intrusive wearable sensors. Despite data-collection systems now being manufactured commercially, allowing physiological data to be acquired from mobile patients, little work has been undertaken on the use of the resultant data in a principled manner for robust patient care, including predictive monitoring. Most current devices generate so many false-positive alerts that devices cannot be used for routine clinical practice. This paper explores principled machine learning approaches to interpreting large quantities of continuously acquired, multivariate physiological data, using wearable patient monitors, where the goal is to provide early warning of serious physiological determination, such that a degree of predictive care may be provided. We adopt a one-class support vector machine formulation, proposing a formulation for determining the free parameters of the model using partial area under the ROC curve, a method arising from the unique requirements of performing online analysis with data from patient-worn sensors. There are few clinical evaluations of machine learning techniques in the literature, so we present results from a study at the Oxford University Hospitals NHS Trust devised to investigate the large-scale clinical use of patient-worn sensors for predictive monitoring in a ward with a high incidence of patient mortality. We show that our system can combine routine manual observations made by clinical staff with the continuous data acquired from wearable sensors. Practical considerations and recommendations based on our experiences of this clinical study are discussed, in the context of a framework for personalized monitoring.

Multitask Gaussian Processes for Multivariate Physiological Time-Series Analysis

Gaussian process (GP) models are a flexible means of performing nonparametric Bayesian regression. However, GP models in healthcare are often only used to model a single univariate output time series, denoted as single-task GPs (STGP). Due to an increasing prevalence of sensors in healthcare settings, there is an urgent need for robust multivariate time-series tools. Here, we propose a method using multitask GPs (MTGPs) which can model multiple correlated multivariate physiological time series simultaneously. The flexible MTGP framework can learn the correlation between multiple signals even though they might be sampled at different frequencies and have training sets available for different intervals. Furthermore, prior knowledge of any relationship between the time series such as delays and temporal behavior can be easily integrated. A novel normalization is proposed to allow interpretation of the various hyperparameters used in the MTGP. We investigate MTGPs for physiological monitoring with synthetic data sets and two real-world problems from the field of patient monitoring and radiotherapy. The results are compared with standard Gaussian processes and other existing methods in the respective biomedical application areas. In both cases, we show that our framework learned the correlation between physiological time series efficiently, outperforming the existing state of the art.

The association between the neutrophil-to-lymphocyte ratio and mortality in critical illness: an observational cohort study.

IntroductionThe neutrophil-to-lymphocyte ratio (NLR) is a biological marker that has been shown to be associated with outcomes in patients with a number of different malignancies. The objective of this study was to assess the relationship between NLR and mortality in a population of adult critically ill patients.MethodsWe performed an observational cohort study of unselected intensive care unit (ICU) patients based on records in a large clinical database. We computed individual patient NLR and categorized patients by quartile of this ratio. The association of NLR quartiles and 28-day mortality was assessed using multivariable logistic regression. Secondary outcomes included mortality in the ICU, in-hospital mortality and 1-year mortality. An a priori subgroup analysis of patients with versus without sepsis was performed to assess any differences in the relationship between the NLR and outcomes in these cohorts.ResultsA total of 5,056 patients were included. Their 28-day mortality rate was 19%. The median age of the cohort was 65 years, and 47% were female. The median NLR for the entire cohort was 8.9 (interquartile range, 4.99 to 16.21). Following multivariable adjustments, there was a stepwise increase in mortality with increasing quartiles of NLR (first quartile: reference category; second quartile odds ratio (OR) = 1.32; 95% confidence interval (CI), 1.03 to 1.71; third quartile OR = 1.43; 95% CI, 1.12 to 1.83; 4th quartile OR = 1.71; 95% CI, 1.35 to 2.16). A similar stepwise relationship was identified in the subgroup of patients who presented without sepsis. The NLR was not associated with 28-day mortality in patients with sepsis. Increasing quartile of NLR was statistically significantly associated with secondary outcome.ConclusionThe NLR is associated with outcomes in unselected critically ill patients. In patients with sepsis, there was no statistically significant relationship between NLR and mortality. Further investigation is required to increase understanding of the pathophysiology of this relationship and to validate these findings with data collected prospectively.

Making Big Data Useful for Health Care: A Summary of the Inaugural MIT Critical Data Conference

With growing concerns that big data will only augment the problem of unreliable research, the Laboratory of Computational Physiology at the Massachusetts Institute of Technology organized the Critical Data Conference in January 2014. Thought leaders from academia, government, and industry across disciplines—including clinical medicine, computer science, public health, informatics, biomedical research, health technology, statistics, and epidemiology—gathered and discussed the pitfalls and challenges of big data in health care. The key message from the conference is that the value of large amounts of data hinges on the ability of researchers to share data, methodologies, and findings in an open setting. If empirical value is to be from the analysis of retrospective data, groups must continuously work together on similar problems to create more effective peer review. This will lead to improvement in methodology and quality, with each iteration of analysis resulting in more reliability.

Modelling physiological deterioration in post-operative patient vital-sign data

Patients who undergo upper-gastrointestinal surgery have a high incidence of post-operative complications, often requiring admission to the intensive care unit several days after surgery. A dataset comprising observational vital-sign data from 171 post-operative patients taking part in a two-phase clinical trial at the Oxford Cancer Centre, was used to explore the trajectory of patients’ vital-sign changes during their stay in the post-operative ward using both univariate and multivariate analyses. A model of normality based vital-sign data from patients who had a “normal” recovery was constructed using a kernel density estimate, and tested with “abnormal” data from patients who deteriorated sufficiently to be re-admitted to the intensive care unit. The vital-sign distributions from “normal” patients were found to vary over time from admission to the post-operative ward to their discharge home, but no significant changes in their distributions were observed from halfway through their stay on the ward to the time of discharge. The model of normality identified patient deterioration when tested with unseen “abnormal” data, suggesting that such techniques may be used to provide early warning of adverse physiological events.

Gaussian process regression in vital-sign early warning systems

The current standard of clinical practice for patient monitoring in most developed nations is connection of patients to vital-sign monitors, combined with frequent manual observation. In some nations, such as the UK, manual early warning score (EWS) systems have been mandated for use, in which scores are assigned to the manual observations, and care escalated if the scores exceed some pre-defined threshold. We argue that this manual system is far from ideal, and can be improved using machine learning techniques. We propose a system based on Gaussian process regression for improving the efficacy of existing EWS systems, and then demonstrate the method using manual observation of vital signs from a large-scale clinical study.

Heart beat detection in multimodal physiological data using a hidden semi-Markov model and signal quality indices.

Accurate heart beat detection in signals acquired from intensive care unit (ICU) patients is necessary for establishing both normality and detecting abnormal events. Detection is normally performed by analysing the electrocardiogram (ECG) signal, and alarms are triggered when parameters derived from this signal exceed preset or variable thresholds. However, due to noisy and missing data, these alarms are frequently deemed to be false positives, and therefore ignored by clinical staff. The fusion of features derived from other signals, such as the arterial blood pressure (ABP) or the photoplethysmogram (PPG), has the potential to reduce such false alarms. In order to leverage the highly correlated temporal nature of the physiological signals, a hidden semi-Markov model (HSMM) approach, which uses the intra- and inter-beat depolarization interval, was designed to detect heart beats in such data. Features based on the wavelet transform, signal gradient and signal quality indices were extracted from the ECG and ABP waveforms for use in the HSMM framework. The presented method achieved an overall score of 89.13% on the hidden/test data set provided by the Physionet/Computing in Cardiology Challenge 2014: Robust Detection of Heart Beats in Multimodal Data.

Gaussian process clustering for the functional characterisation of vital-sign trajectories

Recognition of complex trajectories in multivariate time-series data requires effective models and representations for the analysis and matching of functional data. In this work, we introduce a new representation that allows for matching of noisy, and unevenly-sampled trajectories, and we explore whether this representation may be used to characterise the state of health of a patient based on vital-sign data. We model the evolution of each vital-sign trajectory using multivariate Gaussian process regression, and we introduce a similarity measurement for the comparison of latent functions based on the local likelihood of the points in each trajectory. The similarity measurement is then used for recognising known trajectories and identifying unknown trajectories as would be required for identifying “abnormal” vital-sign time-series. We test our approach using a dataset that contains vital-sign observational data collected from a cohort of 154 patients who are recovering from gastrointestinal surgery. We show that our approach is able to discriminate between abnormal patient trajectories corresponding to those who deteriorated physiologically and were admitted to a higher level of care, from those belonging to patients who had no clinically relevant events.

Probabilistic estimation of respiratory rate using Gaussian processes

The presence of respiratory information within the electrocardiogram (ECG) signal is a well-documented phenomenon. We present a Gaussian process framework for the estimation of respiratory rate from the different sources of modulation in a single-lead ECG. We propose a periodic covariance function to model the frequency- and amplitude-modulation time series derived from the ECG, where the hyperparameters of the process are used to derive the respiratory rate. The approach is evaluated using data taken from 40 healthy subjects each with 2 hours of monitoring, containing ECG and respiration waveforms. Results indicate that the accuracy of our proposed method is comparable with that of existing methods, but with the advantages of a principled probabilistic approach, including the direct quantification of the uncertainty in the estimation.