Developing machine learning systems worthy of trust for infection science: a requirement for future implementation into clinical practice.

Infection science is a discipline of healthcare which includes clinical microbiology, public health microbiology, mechanisms of microbial disease, and antimicrobial countermeasures. The importance of infection science has become more apparent in recent years during the SARS-CoV-2 (COVID-19) pandemic and subsequent highlighting of critical operational domains within infection science including the hospital, clinical laboratory, and public health environments to prevent, manage, and treat infectious diseases. However, as the global community transitions beyond the pandemic, the importance of infection science remains, with emerging infectious diseases, bloodstream infections, sepsis, and antimicrobial resistance becoming increasingly significant contributions to the burden of global disease. Machine learning (ML) is frequently applied in healthcare and medical domains, with growing interest in the application of ML techniques to problems in infection science. This has the potential to address several key aspects including improving patient outcomes, optimising workflows in the clinical laboratory, and supporting the management of public health. However, despite promising results, the implementation of ML into clinical practice and workflows is limited. Enabling the migration of ML models from the research to real world environment requires the development of trustworthy ML systems that support the requirements of users, stakeholders, and regulatory agencies. This paper will provide readers with a brief introduction to infection science, outline the principles of trustworthy ML systems, provide examples of the application of these principles in infection science, and propose future directions for moving towards the development of trustworthy ML systems in infection science.

Machine learning pipeline for blood culture outcome prediction using Sysmex XN-2000 blood sample results in Western Australia.

BACKGROUND: Bloodstream infections (BSIs) are a significant burden on the global population and represent a key area of focus in the hospital environment. Blood culture (BC) testing is the standard diagnostic test utilised to confirm the presence of a BSI. However, current BC testing practices result in low positive yields and overuse of the diagnostic test. Diagnostic stewardship research regarding BC testing is increasing, and becoming more important to reduce unnecessary resource expenditure and antimicrobial use, especially as antimicrobial resistance continues to rise. This study aims to establish a machine learning (ML) pipeline for BC outcome prediction using data obtained from routinely analysed blood samples, including complete blood count (CBC), white blood cell differential (DIFF), and cell population data (CPD) produced by Sysmex XN-2000 analysers. METHODS: ML models were trained using retrospective data produced between 2018 and 2019, from patients at Sir Charles Gairdner hospital, Nedlands, Western Australia, and processed at Pathwest Laboratory Medicine, Nedlands. Trained ML models were evaluated using stratified 10-fold cross validation. RESULTS: Two ML models, an XGBoost model using CBC/DIFF/CPD features with boruta feature selection (BFS) , and a random forest model trained using CBC/DIFF features with BFS were selected for further validation after obtaining AUC scores of [Formula: see text] and [Formula: see text] respectively using stratified 10-fold cross validation. The XGBoost model obtained an AUC score of 0.76 on a internal validation set. The random forest model obtained AUC scores of 0.82 and 0.76 on internal and external validation datasets respectively. CONCLUSIONS: We have demonstrated the utility of using an ML pipeline combined with CBC/DIFF, and CBC/DIFF/CPD feature spaces for BC outcome prediction. This builds on the growing body of research in the area of BC outcome prediction, and provides opportunity for further research.

The foundations of aetiology: a common language for infection science.

With the adoption of infection science as an umbrella term for the disciplines that inform our ideas of infection, there is a need for a common language that links infection's constituent parts. This paper develops a conceptual framework for infection science from the major themes used to understand causal relationships in infectious diseases. The paper proposes using the four main themes from the Principia Aetiologica to classify infection knowledge into four corresponding domains: Clinical microbiology, Public health microbiology, Mechanisms of microbial disease and Antimicrobial countermeasures. This epistemology of infection gives form and process to a revised infection ontology and an infectious disease heuristic. Application of the proposed epistemology has immediate practical implications for organization of journal content, promotion of inter-disciplinary collaboration, identification of emerging priority themes, and integration of cross-disciplinary areas such as One Health topics and antimicrobial resistance. Starting with these foundations, we can build a coherent narrative around the idea of infection that shapes the practice of infection science.

Plasma cfDNA predictors of established bacteraemic infection.

Introduction. Increased plasma cell-free DNA (cfDNA) has been reported for various diseases in which cell death and tissue/organ damage contribute to pathogenesis, including sepsis. Gap Statement. While several studies report a rise in plasma cfDNA in bacteraemia and sepsis, the main source of cfDNA has not been identified. Aim. In this study, we wanted to determine which of nuclear, mitochondrial or bacterial cfDNA is the major contributor to raised plasma cfDNA in hospital subjects with bloodstream infections and could therefore serve as a predictor of bacteraemic disease severity. Methodology. The total plasma concentration of double-stranded cfDNA was determined using a fluorometric assay. The presence of bacterial DNA was identified by PCR and DNA sequencing. The copy numbers of human genes, nuclear ß globin and mitochondrial MTATP8, were determined by droplet digital PCR. The presence, size and concentration of apoptotic DNA from human cells were established using lab-on-a-chip technology. Results. We observed a significant difference in total plasma cfDNA from a median of 75 ng ml-1 in hospitalised subjects without bacteraemia to a median of 370 ng ml-1 (P=0.0003) in bacteraemic subjects. The copy numbers of nuclear DNA in bacteraemic also differed between a median of 1.6 copies µl-1 and 7.3 copies µl-1 (P=0.0004), respectively. In contrast, increased mitochondrial cfDNA was not specific for bacteraemic subjects, as shown by median values of 58 copies µl-1 in bacteraemic subjects, 55 copies µl-1 in other hospitalised subjects and 5.4 copies µl-1 in healthy controls. Apoptotic nucleosomal cfDNA was detected only in a subpopulation of bacteraemic subjects with documented comorbidities, consistent with elevated plasma C-reactive protein (CRP) levels in these subjects. No bacterial cfDNA was reliably detected by PCR in plasma of bacteraemic subjects over the course of infection with several bacterial pathogens. Conclusions. Our data revealed distinctive plasma cfDNA signatures in different groups of hospital subjects. The total cfDNA was significantly increased in hospital subjects with laboratory-confirmed bloodstream infections comprising nuclear and apoptotic, but not mitochondrial or bacterial cfDNAs. The apoptotic cfDNA, potentially derived from blood cells, predicted established bacteraemia. These findings deserve further investigation in different hospital settings, where cfDNA measurement could provide simple and quantifiable parameters for monitoring a disease progression.

Broad-spectrum in vitro activity of macrophage infectivity potentiator inhibitors against Gram-negative bacteria and Leishmania major.

BACKGROUND: The macrophage infectivity potentiator (Mip) protein, which belongs to the immunophilin superfamily, is a peptidyl-prolyl cis/trans isomerase (PPIase) enzyme. Mip has been shown to be important for virulence in a wide range of pathogenic microorganisms. It has previously been demonstrated that small-molecule compounds designed to target Mip from the Gram-negative bacterium Burkholderia pseudomallei bind at the site of enzymatic activity of the protein, inhibiting the in vitro activity of Mip. OBJECTIVES: In this study, co-crystallography experiments with recombinant B. pseudomallei Mip (BpMip) protein and Mip inhibitors, biochemical analysis and computational modelling were used to predict the efficacy of lead compounds for broad-spectrum activity against other pathogens. METHODS: Binding activity of three lead compounds targeting BpMip was verified using surface plasmon resonance spectroscopy. The determination of crystal structures of BpMip in complex with these compounds, together with molecular modelling and in vitro assays, was used to determine whether the compounds have broad-spectrum antimicrobial activity against pathogens. RESULTS: Of the three lead small-molecule compounds, two were effective in inhibiting the PPIase activity of Mip proteins from Neisseria meningitidis, Klebsiella pneumoniae and Leishmania major. The compounds also reduced the intracellular burden of these pathogens using in vitro cell infection assays. CONCLUSIONS: These results indicate that Mip is a novel antivirulence target that can be inhibited using small-molecule compounds that prove to be promising broad-spectrum drug candidates in vitro. Further optimization of compounds is required for in vivo evaluation and future clinical applications.

Development, deployment and in-field demonstration of mobile coronavirus SARS-CoV-2 Nucleic acid amplification test.

Introduction. The evolving SARS-CoV-2 coronavirus pandemic presents a series of challenges to clinical diagnostic services. Many proprietary PCR platforms deployed outside centralised laboratories have limited capacity to upscale when public health demands increase. We set out to develop and validate an open-platform mobile laboratory for remote area COVID-19 diagnosis, with a subsequent field trial.Gap Statement. In regional Western Australia, molecular diagnostic support is limited to near point-of-care devices. We therefore aimed to demonstrate open-platform capability in a rapidly deployable format within the context of the COVID-19 pandemic.Methodology. We compared, selected and validated components of a SARS-CoV-2 RT-PCR assay in order to establish a portable molecular diagnostics laboratory. The optimal combination of PCR assay equipment, reagents and consumables required for operation to national standards in regional laboratories was identified. This comprised RNA extraction and purification (QuickGene-480, Kurabo, Japan), a duplex RT-PCR assay (Logix Smart COVID-19, Co-Diagnostics, USA), a Myra liquid handling robot (Biomolecular Systems, Australia) and a magnetic induction thermal cycler (MIC, Biomolecular Systems).Results The 95 and 99% limits of detection were 1.01 copies µl-1 (5.05 copies per reaction) and 2.80 copies µl-1 (14.00 copies per reaction) respectively. The Co-Diagnostics assay amplified both SARS-CoV-1 and -2 RNA but showed no other cross reactivity. Qualitative results aligned with the reference laboratory SARS-CoV-2 assay (sensitivity 100% [95 % CI 96.48-100%], specificity 100% [95% CI 96.52-100%]). In field trials, the laboratory was operational within an hour of arrival on-site, can process up to 36 samples simultaneously, produces results in two and a half hours from specimen reception, and performed well during six consecutive runs during a 1 week deployment.Conclusion. Our mobile laboratory enables an adaptive response to increased test demand, and unlike many proprietary point-of-care PCR systems, rapid substitution with an alternative assay if gene targets change or reagent supply chains fail. We envisage operation of this RT-PCR assay as a standby capability to meet varying regional test demands under public health emergency operations guidance.

JMM Profile: Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2).

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) virus is the cause of an infection known as coronavirus infectious disease 2019 (COVID-19). COVID-19 has become a global source of morbidity, mortality and social disruption since its emergence in East Asia in late 2019 and subsequent pandemic spread. Typical symptoms include cough, sore throat, fever, and sudden loss of taste and smell. Persistent, post-infection sequelae have been noted in a minority of cases. Severe complications and deaths occur mostly in older adults. Laboratory confirmation can be performed by viral RNA and antigen detection in nasal swabs or by detecting specific neutralizing antibodies. There is no effective and approved antiviral treatment, but several vaccines with favourable safety and efficacy profiles are being used in mass vaccination programmes. Vaccine-based COVID control should be seen as an addition to existing hygiene measures such as physical distancing, increased hand hygiene, cough etiquette, and barrier protection with personal protective equipment for frontline healthcare workers and other high-risk professions.

2020: the year of living cautiously.

2020 was the year when microbiology burst onto the world stage, not just as the science of small living things, but as the prism through which we understood global events. Clinical logic suffered under pressure arising from an urgent need to confirm or exclude severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection. This is a generation's Hobbesian moment in which the public concern for safety and security from infection outweighs the pursuit of personal freedom. The strangeness of a world in which a minute particle wields superhuman power has generated its list of unlikely heroes and mendacious villains. As the year comes to an end, there are glimmers of light amid the gloom: the prospect of an effective vaccine, and life after the pandemic.

Estimating COVID Risk During a Period of Pandemic Decline.

Background: Many parts of the world that succeeded in suppressing epidemic coronavirus spread in 2020 have been caught out by recent changes in the transmission dynamics of SARS-CoV-2. Australia's early success in suppressing COVID-19 resulted in lengthy periods without community transmission. However, a slow vaccine rollout leaves this geographically isolated population vulnerable to leakage of new variants from quarantine, which requires internal travel restrictions, disruptive lockdowns, contact tracing and testing surges. Methods: To assist long term sustainment of limited public health resources, we sought a method of continuous, real-time COVID-19 risk monitoring that could be used to alert non-specialists to the level of epidemic risk on a sub-national scale. After an exploratory data assessment, we selected four COVID-19 metrics used by public health in their periodic threat assessments, applied a business continuity matrix and derived a numeric indicator; the COVID-19 Risk Estimate (CRE), to generate a daily spot CRE, a 3 day net rise and a seven day rolling average. We used open source data updated daily from all Australian states and territories to monitor the CRE for over a year. Results: Upper and lower CRE thresholds were established for the CRE seven day rolling average, corresponding to risk of sustained and potential outbreak propagation, respectively. These CRE thresholds were used in a real-time map of Australian COVID-19 risk estimate distribution by state and territory. Conclusions: The CRE toolkit we developed complements other COVID-19 risk management techniques and provides an early indication of emerging threats to business continuity.