Using machine learning for predicting intensive care unit resource use during the COVID-19 pandemic in Denmark

The COVID-19 pandemic has put massive strains on hospitals, and tools to guide hospital planners in resource allocation during the ebbs and flows of the pandemic are urgently needed. We investigate whether machine learning (ML) can be used for predictions of intensive care requirements a fixed number of days into the future. Retrospective design where health Records from 42,526 SARS-CoV-2 positive patients in Denmark was extracted. Random Forest (RF) models were trained to predict risk of ICU admission and use of mechanical ventilation after n days (n = 1, 2, …, 15). An extended analysis was provided for n = 5 and n = 10. Models predicted n-day risk of ICU admission with an area under the receiver operator characteristic curve (ROC-AUC) between 0.981 and 0.995, and n-day risk of use of ventilation with an ROC-AUC between 0.982 and 0.997. The corresponding n-day forecasting models predicted the needed ICU capacity with a coefficient of determination (R2) between 0.334 and 0.989 and use of ventilation with an R2 between 0.446 and 0.973. The forecasting models performed worst, when forecasting many days into the future (for large n). For n = 5, ICU capacity was predicted with ROC-AUC 0.990 and R2 0.928, and use of ventilator was predicted with ROC-AUC 0.994 and R2 0.854. Random Forest-based modelling can be used for accurate n-day forecasting predictions of ICU resource requirements, when n is not too large.

The critical importance of mask seals on respirator performance: An analytical and simulation approach

Filtering facepiece respirators (FFRs) and medical masks are widely used to reduce the inhalation exposure of airborne particulates and biohazardous aerosols. Their protective capacity largely depends on the fraction of these that are filtered from the incoming air volume. While the performance and physics of different filter materials have been the topic of intensive study, less well understood are the effects of mask sealing. To address this, we introduce an approach to calculate the influence of face-seal leakage on filtration ratio and fit factor based on an analytical model and a finite element method (FEM) model, both of which take into account time-dependent human respiration velocities. Using these, we calculate the filtration ratio and fit factor for a range of ventilation resistance values relevant to filter materials, 500-2500 Pa∙s∙m-1, where the filtration ratio and fit factor are calculated as a function of the mask gap dimensions, with good agreement between analytical and numerical models. The results show that the filtration ratio and fit factor are decrease markedly with even small increases in gap area. We also calculate particle filtration rates for N95 FFRs with various ventilation resistances and two commercial FFRs exemplars. Taken together, this work underscores the critical importance of forming a tight seal around the face as a factor in mask performance, where our straightforward analytical model can be readily applied to obtain estimates of mask performance.

The addition of a humidifier device to a circuit and its impact on home ventilator performance: a bench study

INTRODUCTION AND OBJECTIVES

Humidification and non-invasive ventilation are frequently used together, despite the lack of precise recommendations regarding this practice. We aimed to analyse the impact of active external and built-in humidifiers on the performance of home ventilators, focusing on their pressurization efficacy and their behaviour under different inspiratory efforts.

METHODS

We designed a bench study of a lung simulator programmed to emulate mechanical conditions similar to those experienced by real respiratory patients and to simulate three different levels of inspiratory effort: five different commonly used home NIV devices and active humidifiers attached to the latter (internal or "built-in") or to the circuit (external). To test ventilator pressurization under different humidification and effort settings, pressure-time products in the first 300ms and 500ms of the respiratory cycle were calculated in the 45 situations simulated. Inferential statistical analysis was performed.

RESULTS

A significant reduction of PTP 300 and PTP 500 was observed with the external humidifier in three of the devices. The same pattern was noted for another device with an internal humidifier, and only one device showed no significant changes. This impact on pressurization was commonly higher under high inspiratory effort.

CONCLUSIONS

These results indicate the need to monitor pressure changes in the use of external humidification devices in some home NIV ventilators.

Improved lung recruitment and oxygenation during mandatory ventilation with a new expiratory ventilation assistance device: A controlled interventional trial in healthy pigs

BACKGROUND

In contrast to conventional mandatory ventilation, a new ventilation mode, expiratory ventilation assistance (EVA), linearises the expiratory tracheal pressure decline.

OBJECTIVE

We hypothesised that due to a recruiting effect, linearised expiration oxygenates better than volume controlled ventilation (VCV). We compared the EVA with VCV mode with regard to gas exchange, ventilation volumes and pressures and lung aeration in a model of peri-operative mandatory ventilation in healthy pigs.

DESIGN

Controlled interventional trial.

SETTING

Animal operating facility at a university medical centre.

ANIMALS

A total of 16 German Landrace hybrid pigs.

INTERVENTION

The lungs of anaesthetised pigs were ventilated with the EVA mode (n=9) or VCV (control, n=7) for 5 h with positive end-expiratory pressure of 5 cmH2O and tidal volume of 8 ml kg. The respiratory rate was adjusted for a target end-tidal CO2 of 4.7 to 6 kPa.

MAIN OUTCOME MEASURES

Tracheal pressure, minute volume and arterial blood gases were recorded repeatedly. Computed thoracic tomography was performed to quantify the percentages of normally and poorly aerated lung tissue.

RESULTS

Two animals in the EVA group were excluded due to unstable ventilation (n=1) or unstable FiO2 delivery (n=1). Mean tracheal pressure and PaO2 were higher in the EVA group compared with control (mean tracheal pressure: 11.6 ± 0.4 versus 9.0 ± 0.3 cmH2O, P < 0.001 and PaO2: 19.2 ± 0.7 versus 17.5 ± 0.4 kPa, P = 0.002) with comparable peak inspiratory tracheal pressure (18.3 ± 0.9 versus 18.0 ± 1.2 cmH2O, P > 0.99). Minute volume was lower in the EVA group compared with control (5.5 ± 0.2 versus 7.0 ± 1.0 l min, P = 0.02) with normoventilation in both groups (PaCO2 5.4 ± 0.3 versus 5.5 ± 0.3 kPa, P > 0.99). In the EVA group, the percentage of normally aerated lung tissue was higher (81.0 ± 3.6 versus 75.8 ± 3.0%, P = 0.017) and of poorly aerated lung tissue lower (9.5 ± 3.3 versus 15.7 ± 3.5%, P = 0.002) compared with control.

CONCLUSION

EVA ventilation improves lung aeration via elevated mean tracheal pressure and consequently improves arterial oxygenation at unaltered positive end-expiratory pressure (PEEP) and peak inspiratory pressure (PIP). These findings suggest the EVA mode is a new approach for protective lung ventilation.

Neuromuscular disorders and chronic ventilation

Morbidity and mortality have decreased in patients with neuromuscular disease due to implementation of therapies to augment cough and improve ventilation. Infants with progressive neuromuscular disease will eventually develop respiratory complications as a result of muscle weakness and their inability to compensate during periods of increased respiratory loads. The finding of nocturnal hypercapnia is often the trigger for initiating non-invasive ventilation and studies have shown that its use not only may improve sleep-disordered breathing, but also that it may have an effect on daytime function, symptoms related to hypercapnia, and partial pressure of CO₂. It is important to understand the respiratory physiology of this population and to understand the benefits and limitations of assisted ventilation.

Anesthesia and critical care ventilator modes: past, present, and future

Mechanical ventilators have evolved from basic machines to complicated, electronic, microprocessing engines. Over the last 2 decades, ventilator capabilities and options for critical care and anesthesia ventilators have rapidly advanced. These advances in ventilator modalities--in conjunction with a better understanding of patient physiology and the effects of positive pressure ventilation on the body--have revolutionized the mechanical ventilation process. Clinicians today have a vast array of mechanical ventilator mode options designed to match the pulmonary needs of the critically ill and anesthetized patient. Modes of mechanical ventilation continue to be based on 1 of 2 variances: volume-based or pressure-based. The wording describing the standard ventilatory modes on select present-day ventilators has changed, yet the basic principles of operation have not changed compared with older ventilators. Anesthesia providers need to understand these ventilator modes to best care for patients. This literature review encompasses a brief history of mechanical ventilation and current modes available for anesthesia and critical care ventilators, including definitions of each mode, definitions of the various descriptive labels given each mode, and techniques for optimizing and meeting the ventilator needs of the patient while avoiding complications in the surgical and critical care patient.

[Neurally adjusted ventilatory assist: a revolution of mechanical ventilation?]

Neurally adjusted ventilatory assist or NAVA is a new assisted ventilatory mode which, in comparison with pressure support, leads to improved patient-ventilator synchrony and a more variable ventilatory pattern. It also improves arterial oxygenation. With NAVA, the electrical activity of the diaphragm is recorded through a nasogastric tube equipped with electrodes. This electrical activity is then used to pilot the ventilator. With NAVA, the patient's respiratory pattern controls the ventilator's timing of triggering and cycling as well as the magnitude of pressurization, which is proportional to inspiratory demand. The effect of NAVA on patient outcome remains to be determined through well-designed prospective studies.

Chronic ventilator need in the community: a 2005 pediatric census of Massachusetts

OBJECTIVES

The purpose of this study was to describe the population of children with chronic mechanical ventilation in Massachusetts and their patterns of medical care.

PATIENTS AND METHODS

Investigators surveyed all of the Massachusetts home ventilator clinics, pediatric pulmonary services, hospital-based pediatric services for special health care needs, insurers, home care vendors, nursing agencies, the Massachusetts Department of Public Health, selected individual providers, and rehabilitation and long-term care facilities providing services to children with chronic respiratory support needs. Support was defined as daily use of noninvasive, negative-pressure, or invasive/transtracheal ventilators. Subsequent matching of demographic data, including date of birth, zip code, and gender supported maximal census yield without duplications. Geographic information systems were used to create distribution maps and estimate distances between children with chronic mechanical ventilator needs and key resources.

RESULTS

A total of 197 children were identified as requiring chronic mechanical respiratory support in Massachusetts in 2005, which was a nearly threefold increase in this population in the 15-year interval since the last census. Congenital or perinatal-acquired neurologic or neuromuscular disorders constituted the majority of primary diagnoses (n = 107 [54%]). Chronic lung disease attributed to prematurity represented only 7% of the sample.

CONCLUSIONS

Children receiving chronic mechanical respiratory support are a growing population. The shift in underlying diagnoses from pulmonary disease to neurogenic respiratory insufficiency has implications for hospital and community-based providers from all disciplines in extending services to the home setting. Barriers encountered when performing this study, however, reflect an overall lack of coordination among the many individuals and agencies involved in their care. Coordinated and centralized care efforts require a clear and managed flow of information; census reports such as this one are only the beginning. Direct needs assessments and quality-of-life surveys from families are needed to design and implement programmatic changes and advocacy efforts.

[Intensive care medicine -- update 2005]

This manuscript gives a review about important studies addressing problems in intensive care medicine that have been published in journals focussing on critical care medicine and surgery in 2005. Only clinical studies are included in this review, mostly meta-analyses, randomized controlled trials and a few important or interesting observational studies. In addition to describing major results a critical appraisal of each study is undertaken, which, however, is neither comprehensive nor complete. It is merely intended to address some important aspects for the reader who should be stimulated to go deeper into one or the other topic or study. The publication of the new CPR-guidelines of the American Heart Association and the European Resuscitation Council as well as the newly developed SAPS III score to predict intensive care unit outcome are among the outstanding topics. Several randomized trials and meta-analyses deal with aspects of drug therapy of septic patients. Some important and relevant findings have been reported with respect to the efficiency of the open-lung concept, non-invasive ventilation, the use of heat and moisture exchanger filters compared to active humidifiers and of closed systems for endotracheal suctioning. The role of immuno-nutrition in adults and children as well as of early enteral nutrition can be defined more clearly. Whether corticosteroids should be used in the treatment of severe traumatic brain injury can be definitely answered now. There are some new insights reported into the management of patients infected or contaminated with MRSA in the intensive care unit. Last but not least an impressive study shows that not only the newest therapeutic developments but the stringent use of the already known treatment options may result in dramatic improvements of patient outcome.

Anesthesia Ventilators: Better Options for Children

Mechanical ventilation of pediatric patients in the operating room is challenging. Infants require significantly smaller tidal volumes than adults and changes in delivered volume that would be clinically insignificant for an adult patient, can produce unintended hyper- or hypoventilation in children. The consequences of these unintended ventilation changes can produce hypoxemia, hypercarbia, or barotrauma. This article discusses unique aspects of pediatric ventilation in the operating room, limitations of traditional anesthesia machine technology, the features of modern anesthesia ventilators that circumvent these limitations, and presents several comparison studies.

[The research actualities and developing trend of nitric oxide (NO) inhalation systems]

This paper discusses the limitations of current NO inhalation systems, based on the research in collocation of NO, inspection of NO/NO2 and synchronous working of NO inhalation systems with ventilators. And then, the developing trend of NO inhalation systems is put forward here too.

Understanding and implementing advances in ventilator capabilities

PURPOSE OF REVIEW

To review the changes in mechanical ventilation technology over the past year and identify areas that provide a benefit.

RECENT FINDINGS

The literature demonstrates a continued effort to improve patient ventilator synchrony though the development of new triggering and cycling methods. These techniques include using new signals and using closed loop techniques to respond to changes in patient breathing pattern. New modes of ventilation continue to be introduced, often without proof of efficacy. Fortunately, clinicians have developed alterations to new modes that improve utility and they continue to study these techniques clinically to determine appropriate use. Monitoring the patient remains an important area of investigation, with a flurry of activity surrounding pressure volume curves of the respiratory system. Finally, new ventilators have been introduced that combine high-end performance with small size and weight, while providing an on-board source of air.

SUMMARY

Mechanical ventilation is ubiquitous to intensive care. Advances in ventilator technology are rapid, and clinicians must keep abreast of changes in ventilator performance and application.

Patient ventilation and oxygenation. A new day is dawning

Maintaining an airway and providing adequate ventilation and oxygenation to the patient can be challenging in the prehospital environment. Ventilation and oxygenation are a complex series of interactions between the patient, EMS providers and emergency airway equipment. Routine ventilation techniques carry significant risk of long-term complications. New ventilatory equipment is available to perform this function and provide verification of its effectiveness. This opportunity for improved patient care is available to all EMS providers and sets new standards for delivery of ventilation and oxygenation. The technical methods available to EMS personnel vary considerably and are reviewed in this article.

Newer techniques of mechanical ventilation: an overview

The introduction of newer, state-of-the-art, microprocessor controlled ventilator systems provides clinicians with opportunities to apply a number of advanced ventilatory modalities which were not previously available for treating newborns. Some of these techniques will need further scientific evaluation in controlled trials, but this should not preclude their use in clinical settings, as their safety has already been proved by "standard setters" for use in neonates. There is a firm physiological rationale for their use, and individual centres have already acquired substantial experience in the application of these modalities. The trend towards increasing sophistication and greater versatility is likely to continue, and clinicians involved in the care of sick newborn infants must keep abreast of these developments.

Newer modes of mechanical ventilation for the neonate

For decades, the overwhelming majority of infants requiring mechanical ventilation for respiratory failure were treated with standard time-cycled, pressure-limited intermittent mandatory ventilation. Technologic advances in the 1990s brought forth sophisticated transducers and microprocessor-based mechanical ventilators that enabled implementation of many newer modes of mechanical ventilation. Some of these are volume-targeted rather than pressure-targeted, and many allow an element of patient control of the ventilator, including initiation and termination of inspiration and control of flow. Some modes are even hybrids, combining the best features of both pressure-targeted and volume-targeted modes. This article reviews the principles and salient clinical features of the newer ventilatory modes for newborns with respiratory failure.

[Modalities of mechanical ventilation]

Mechanical ventilation improves the symptoms and reduces complications of acute respiratory failure. Recent advances in microprocessor technology have increased the sophistication of mechanical ventilators, thus leading to new ventilation modalities. This article describes the ventilation modalities available, grouping them as conventional, alternative and new modalities. Conventional ventilation includes the most widely used modalities, alternative ventilation includes less frequently used modalities, and new ventilation modalities include recently introduced options that are available on the latest-generation mechanical ventilators.

Non-invasive ventilation--mechanisms of benefit

Over the last few years there has been growing interest in the use of non-invasive ventilation (NIV) in the management of ventilatory failure in acute on chronic ventilatory failure and in stable hypercapnic patients. To understand why assisted ventilation should be effective in improving gas exchange even during spontaneous breathing a basic understanding of the pathophysiology of ventilatory failure is required and this is discussed. The etiology of ventilatory failure is likely to be multifactorial with different factors assuming greater or lesser degrees of importance even in patients with the same condition. Indeed in an individual patient there may be differences at various stages of the illness. The same is true for the mechanism of benefit from NIV. With the current state of knowledge during NIV the aim should be to rest the respiratory muscles, control nocturnal hypoventilation and improve sleep quality. In some individuals severe symptoms, as a consequence of disturbed sleep, may occur and be symptomatically improved by non-invasive ventilation during sleep.

Lungs at home

More than 100 years of progress in providing home care for sufferers of a variety of chronic respiratory disorders is reviewed. Patients with tuberculosis, polio, chronic respiratory insufficiency from chronic obstructive pulmonary disease (COPD), restrictive ventilatory disorders, neuromuscular disorders and sleep apnoea can receive most if not all of their care at home with the application of modern technology. Major advances in portable and stationary oxygen systems and mechanical ventilators are reviewed. The advantages of home care are economic, social and spiritual.

When is liquid oxygen really needed?

Liquid oxygen is a synonym for portable oxygen, as it combines a big cylinder with an easy-to-fill portable unit, suitable for exercise and use out-doors. The main drawback is its high cost, inherent in a home delivery system, which discouraged many nations from its introduction. The best candidates are patients able to move, who are still active and do not have psychological reticence to its use in public. Transtracheal systems and the advantage of a round the clock treatment and a reduction of flow rate, crucial both to lengthen the autonomy of portable units and to avoid flows higher than 4 L.min-1, which cannot be maintained. Finally, patients on liquid oxygen usually have a better adherence to treatment, mainly compared to those using a concentrator, possibly improving its effectiveness, which is notoriously dependent on total usage per day.

Introducing new mechanical ventilation technology: the hospital perspective

The introduction of innovative mechanical ventilation technology requires a careful assessment of the cost-benefit relationship of the new technology. Justification for purchase can be sometimes be made on the basis of improved patient safety and physiologic response. Improved patient outcome is much more difficult to demonstrate. It is clear in today's health-care environment that outcome-based research is needed before new technology is introduced to the marketplace, if that new technology is to become part of the average community-hospital complement of mechanical ventilators.

Mechanical ventilation innovations: the manufacturer's perspective

Manufacturers develop products that fit the corporate vision and maximize their return on investment. The expense and time required by the FDA regulatory approval process have a negative impact on product innovation. I propose the following approach: Reduce the amount of documentation required for a PMA or 510(k). Reduce approval times through cooperative interaction among manufacturers, clinicians, and the FDA throughout the product-development process. Allow independent agencies to provide product approvals. Efficacy assessment guidelines should be a function of the level of risk and the claims made for the marked device.

New generation ventilators

Desirable features of new generation intensive care ventilators include the ability to ventilate a wide range of patient sizes, an uncomplicated control panel, an appropriate but not excessive variety of ventilatory patterns, adequate patient monitoring and alarm functions, and simplicity of cleaning and routine maintenance. Examples of currently available ventilators include the Servo 900-C, CPU-1, Engstrom Erica, Bear 5, Drager EV-A and Hamilton Veolar. The incorporation of microcomputer control into some of these ventilators has resulted in improved flexibility and a limited number of automatic responses to detected patient changes. However, the function of components provided to allow spontaneous ventilation, such as demand valves, requires considerable improvement. Current trends in ventilator design include further refinement of computer control and the provision of graphic displays showing the results of continuous sophisticated analysis of respiratory function. The extent to which these developments will prove clinically useful will require careful evaluation.

A new anaesthetic breathing system combining Mapleson A, D and E principles. A simple apparatus for low flow universal use without carbon dioxide absorption

A new simple anaesthetic breathing system is described which has been designed to incorporate into a single system advantages of Mapleson A, D and E type systems. Coaxial and non-coaxial versions are available. The system can be used for adults, children or neonates and allows both spontaneous or controlled ventilation with low fresh gas flows at all times. As a Mapleson A system for spontaneous respiration a considerable saving of anaesthetic gases and vapours is achieved since, for adults, the new system requires a lower fresh gas flow even than that for the Magill. For children breathing spontaneously the system requires only one third of the fresh gas flow necessary for the Jackson Rees modification of Ayre's T-piece. For controlled ventilation the system behaves as a modified Mapleson D/E (Bain type) system with the advantage of predictable CO2 tensions and good humidification. The system is safe, simple in design and operation, and is easily sterilized. Further it offers low resistance to expiration and facilitates scavenging at all times which, with low anaesthetic gas flows, permits complete theatre pollution control. Its potential application in academic and rural environments and major advantages over the circle absorber system are discussed.

Ventilation and ventilators

The history of ventilation is reviewed briefly and recent developments in techniques of ventilation are discussed. Operating features of ventilators have changed in the past few years, partly as the result of clinical progress; yet, technology appears to have outstripped the clinician's ability to harness it most effectively. Clinical discipline and training of medical staff in the use of ventilators could be improved. The future is promising if clinician and designer can work together closely. Ergonomics of ventilators and their controls and the provision of alarms need special attention. Microprocessors are likely to feature prominently in the next generation of designs.