Filtering and smoothing methods are used to produce an accurate estimate of the state of a time-varying system based on multiple observational inputs (data). Interest in these methods has exploded in recent years, with numerous applications emerging in fields such as navigation, aerospace engineering, telecommunications, and medicine. This compact, informal introduction for graduate students and advanced undergraduates presents the current state-of-the-art filtering and smoothing methods in a unified Bayesian framework. Readers learn what non-linear Kalman filters and particle filters are, how they are related, and their relative advantages and disadvantages. They also discover how state-of-the-art Bayesian parameter estimation methods can be combined with state-of-the-art filtering and smoothing algorithms. The book’s practical and algorithmic approach assumes only modest mathematical prerequisites. Examples include MATLAB computations, and the numerous end-of-chapter exercises include computational assignments. MATLAB/GNU Octave source code is available for download at www.cambridge.org/sarkka, promoting hands-on work with the methods.

贝叶斯滤波与平滑 (Bayesian filtering and smoothing)

Pathophysiology and treatment of cerebral edema in traumatic brain injury.

Measurement of intracranial pressure (ICP) can be invaluable in the management of critically ill patients. Cerebrospinal fluid is produced by the choroid plexus in the brain ventricles (a set of communicating chambers), after which it circulates through the different ventricles and exits into the subarachnoid space around the brain, where it is reabsorbed into the venous system. If the fluid does not drain out of the brain or get reabsorbed, the ICP increases, which may lead to brain damage or death. ICP elevation accompanied by dilatation of the cerebral ventricles is termed hydrocephalus, whereas ICP elevation accompanied by normal or small ventricles is termed idiopathic intracranial hypertension. Objective We performed a comprehensive literature review on how to measure ICP invasively and noninvasively. Approach This review discusses the advantages and disadvantages of current invasive and noninvasive approaches. Main results Invasive methods remain the most accurate at measuring ICP, but they are prone to a variety of complications including infection, hemorrhage and neurological deficits. Ventricular catheters remain the gold standard but also carry the highest risk of complications, including difficult or incorrect placement. Direct telemetric intraparenchymal ICP monitoring devices are a good alternative. Noninvasive methods for measuring and evaluating ICP have been developed and classified in five broad categories, but have not been reliable enough to use on a routine basis. These methods include the fluid dynamic, ophthalmic, otic, and electrophysiologic methods, as well as magnetic resonance imaging, transcranial Doppler ultrasonography (TCD), cerebral blood flow velocity, near-infrared spectroscopy, transcranial time-of-flight, spontaneous venous pulsations, venous ophthalmodynamometry, optical coherence tomography of retina, optic nerve sheath diameter (ONSD) assessment, pupillometry constriction, sensing tympanic membrane displacement, analyzing otoacoustic emissions/acoustic measure, transcranial acoustic signals, visual-evoked potentials, electroencephalography, skull vibrations, brain tissue resonance and the jugular vein. Significance This review provides a current perspective of invasive and noninvasive ICP measurements, along with a sense of their relative strengths, drawbacks and areas for further improvement. At present, none of the noninvasive methods demonstrates sufficient accuracy and ease of use while allowing continuous monitoring in routine clinical use. However, they provide a realizable ICP measurement in specific patients especially when invasive monitoring is contraindicated or unavailable. Among all noninvasive ICP measurement methods, ONSD and TCD are attractive and may be useful in selected settings though they cannot be used as invasive ICP measurement substitutes. For a sufficiently accurate and universal continuous ICP monitoring method/device, future research and developments are needed to integrate further refinements of the existing methods, combine telemetric sensors and/or technologies, and validate large numbers of clinical studies on relevant patient populations.

https://iopscience.iop.org/article/10.1088/1361-6579/aa7256/pdf

Invasive and noninvasive means of measuring intracranial pressure: a review.

Physiology of the cerebrospinal fluid

OBJECTIVE\n: Intracranial hypertension after severe head injury is associated with case fatality, but there is no sound evidence that monitoring of intracranial pressure (ICP) and targeted management of cerebral perfusion pressure (CPP) improve outcome, despite widespread recommendation by experts in the field. The purpose was to determine the effect of ICP/CPP-targeted intensive care on functional outcome and therapy intensity levels after severe head injury.\n\n\nDESIGN\n: Retrospective cohort study with prospective assessment of outcome.\n\n\nSETTING\n: Two level I trauma centers in The Netherlands from 1996 to 2001.\n\n\nPATIENTS\n: Three hundred thirty-three patients who had survived and remained comatose for >24 hrs, from a total of 685 consecutive severely head-injured adults.\n\n\nINTERVENTIONS\n: In center A (supportive intensive care), mean arterial pressure was maintained at approximately 90 mm Hg, and therapeutic interventions were based on clinical observations and computed tomography findings. In center B (ICP/CPP-targeted intensive care), management was aimed at maintaining ICP <20 mm Hg and CPP >70 mm Hg. Allocation to either trauma center was solely based on the site of the accident.\n\n\nMEASUREMENTS AND MAIN RESULTS\n: We measured extended Glasgow Outcome Scale after >/=12 months. Patient characteristics were well balanced between the centers. ICP monitoring was used in zero of 122 (0%) and 142 of 211 (67%) patients in centers A and B, respectively. In-hospital mortality rate was 41 (34%) vs. 69 (33%; p = .87). The odds ratio for a more favorable functional outcome following ICP/CPP-targeted therapy was 0.95 (95% confidence interval, 0.62-1.44). This result remained after adjustment for potential confounders. Sedatives, vasopressors, mannitol, and barbiturates were much more frequently used in center B (all p < .01). The median number of days on ventilator support in survivors was 5 (25th-75th percentile, 2-9) in center A vs. 12 (7-19) in center B (p < .001).\n\n\nCONCLUSIONS\n: ICP/CPP-targeted intensive care results in prolonged mechanical ventilation and increased levels of therapy intensity, without evidence for improved outcome in patients who survive beyond 24 hrs following severe head injury.

Effect of Intracranial Pressure Monitoring and Targeted Intensive Care on Functional Outcome after Severe Head Injury

Since Monro published his observations on the nature of the contents of the intracranial space in 1783 there has been investigation of the unique relationship between the contents of the skull and the intracranial pressure (ICP). This is particularly true following traumatic brain injury (TBI), where it is clear that elevated ICP due to the underlying pathological processes is associated with a poorer clinical outcome. Consequently, there is considerable interest in monitoring and manipulating ICP In patients with TBI. The two techniques most commonly used in clinical practice to monitor ICP are via an intraventricular or intraparenchymal catheter with a microtransducer system. Both of these techniques are invasive and are thus associated with complications such as haemorrhage and infection. For this reason, significant research effort has been directed towards development of a non-invasive method to measure ICP. These include imaging based studies using computed tomography (CT) and magnetic resonance imaging (MRI), transcranial Doppler sonography (TCD), near-infrared spectroscopy (NIRS), tympanic membrane displacement (TMD), visual-evoked potentials (VEPs), measurements of optic nerve sheath diameter (ONSD) and other measurements of the optic nerve, retina, pupil and ophthalmic artery. The principle aims of ICP monitoring in TBI are to allow early detection of secondary haemorrhage or ischaemic processes and to guide therapies that limit intracranial hypertension and optimise cerebral perfusion. However, information from the ICP value and the ICP waveform can also be used to estimate intracranial compliance, assess cerebrovascular pressure reactivity and attempt to forecast future episodes of intracranial hypertension.

Monitoring of intracranial pressure in patients with traumatic brain injury.

Objectives: We have previously demonstrated a relationship between transcranial bioimpedance (TCB) measurements and intracranial pressure (ICP) in an animal model of raised ICP. The primary objective of this study was to explore the relationship between non-invasive bioelectrical impedance measurements of the brain and skull and ICP in traumatic brain injury (TBI) patients.

Transcranial Bioimpedance Measurement as a Non-invasive Estimate of Intracranial Pressure

Abstract Neurocritical care patients are a complex patient population, and to aid clinical decision-making, many models and scoring systems have previously been developed. More recently, techniques from the field of machine learning have been applied to neurocritical care patient data to develop models with high levels of predictive accuracy. However, although these recent models appear clinically promising, their interpretability has often not been considered and they tend to be black box models, making it extremely difficult to understand how the model came to its conclusion. Interpretable machine learning methods have the potential to provide the means to overcome some of these issues but are largely unexplored within the neurocritical care domain. This article examines existing models used in neurocritical care from the perspective of interpretability. Further, the use of interpretable machine learning will be explored, in particular the potential benefits and drawbacks that the techniques may have when applied to neurocritical care data. Finding a solution to the lack of model explanation, transparency, and accountability is important because these issues have the potential to contribute to model trust and clinical acceptance, and, increasingly, regulation is stipulating a right to explanation for decisions made by models and algorithms. To ensure that the prospective gains from sophisticated predictive models to neurocritical care provision can be realized, it is imperative that interpretability of these models is fully considered. 

https://link.springer.com/content/pdf/10.1007/s12028-022-01504-4.pdf

Demystifying the Black Box: The Importance of Interpretability of Predictive Models in Neurocritical Care

Advances in technology has transformed clinical medicine; electronic patient records routinely store clinical
notes, internet-enabled mobile apps support self-management of chronic diseases, point-of-care testing enables
laboratory tests to be performed outside of hospital environments, patient treatment can be delivered over wide
geographic areas and wireless sensor networks are able to collect and send physiological data. Increasingly,
this technology leads to the development of large databases of sensitive electronic patient information. There
is public interest into the secondary use of this data; many concerns are voiced about the involvement of private
companies and the security and privacy of this data, but at the same time, these databases present a valuable
source of clinical information which can drive health informatics and clinical research, leading to improved
patient treatment. In this position paper, we argue that for health informatics projects to be successful, public
concerns over the secondary use of patient data need to be addressed in the design and implementation of the
technology and conduct of the research project.

Sharing of Big Data in Healthcare: Public Opinion, Trust, and Privacy Considerations for Health Informatics Researchers

In this study we collect and annotate data from 27 intensive care unit (ICU) patients from the Southern General Hospital in Glasgow. Two models are compared for the detection, removal and cleaning of artifact in the vital signs data, namely the Factorial Switching Linear Dynamical System (FSLDS) and the Discriminative Switching Linear Dynamical System (DSLDS). We also consider a combination of the two, called the α-mixture (as described in sec. 7.3). Three types of artifactual events are considered: blood sample, damped trace (in the arterial line), and suction events. The area under ROC curve (AUC) scores for the detection of these events are: blood sample 0.95, damped trace: 0.79, suction 0.64 (α-mixture), with similar results for the FSLDS and DSLDS. The system is able run in realtime, and we discuss issues that had to be addressed to achieve this.

https://homepages.inf.ed.ac.uk/ckiw/projects/adult_icu/CSOreport230915.pdf

Christopher Hawthorne

Papers

Monitoring of intracranial pressure in patients with traumatic brain injury.

Transcranial Bioimpedance Measurement as a Non-invasive Estimate of Intracranial Pressure

Demystifying the Black Box: The Importance of Interpretability of Predictive Models in Neurocritical Care

Sharing of Big Data in Healthcare: Public Opinion, Trust, and Privacy Considerations for Health Informatics Researchers

Detecting artifactual events in vital signs monitoring data