Physics-informed machine learning (PIML) has emerged as a promising new approach for simulating complex physical and biological systems that are governed by complex multiscale processes for which some data are also available. In some instances, the objective is to discover part of the hidden physics from the available data, and PIML has been shown to be particularly effective for such problems for which conventional methods may fail. Unlike commercial machine learning where training of deep neural networks requires big data, in PIML big data are not available. Instead, we can train such networks from additional information obtained by employing the physical laws and evaluating them at random points in the space–time domain. Such PIML integrates multimodality and multifidelity data with mathematical models, and implements them using neural networks or graph networks. Here, we review some of the prevailing trends in embedding physics into machine learning, using physics-informed neural networks (PINNs) based primarily on feed-forward neural networks and automatic differentiation. For more complex systems or systems of systems and unstructured data, graph neural networks (GNNs) present some distinct advantages, and here we review how physics-informed learning can be accomplished with GNNs based on graph exterior calculus to construct differential operators; we refer to these architectures as physics-informed graph networks (PIGNs). We present representative examples for both forward and inverse problems and discuss what advances are needed to scale up PINNs, PIGNs and more broadly GNNs for large-scale engineering problems.

Resistance to carbapenems in human pathogens is a growing clinical and public health concern. The carbapenems are in an antimicrobial class considered last-resort, they are used to treat human infections caused by multidrug-resistant Enterobacterales, and they are classified by the World Health Organization as ‘High Priority Critically Important Antimicrobials’. The presence of carbapenem-resistant Enterobacterales (CREs) of animal-origin is of concern because targeted studies of Canadian retail seafood revealed the presence of carbapenem resistance in a small number of Enterobacterales isolates. To further investigate this issue, a risk profile was developed examining shrimp and salmon, the two most important seafood commodities consumed by Canadians and Escherichia coli, a member of the Enterobacterales order. Carbapenem-resistant E. coli (CREc) isolates have been identified in shrimp and other seafood products. Although carbapenem use in aquaculture has not been reported, several classes of antimicrobials are utilised globally and co-selection of antimicrobial-resistant microorganisms in an aquaculture setting is also of concern. CREs have been identified in retail seafood purchased in Canada and are currently thought to be uncommon. However, data concerning CRE or CREc occurrence and distribution in seafood are limited, and argue for implementation of ongoing or periodic surveillance.

Scrub typhus is a common bacterial infection in Asia caused by Orientia tsutsugamushi. This serological cohort study estimated the incidence of infection in a rural population in South India. Participants were enrolled through systematic sampling in 46 villages at baseline, and revisited the following year. Blood samples were tested for IgG antibodies using ELISA, followed by indirect immunofluorescence assays (IFA) in those positive for ELISA at both rounds. A case was defined as sero-conversion (ELISA), or at least a 4-fold titre increase (IFA), between the two time points. In addition to crude incidence rate estimates, we used piecewise linear rates across calendar months, with rates proportional to the monthly incidence of local hospital cases to address seasonality and unequal follow-up times. Of 402 participants, 61.7% were female. The mean age was 46.7 years, (range 13–88). 21 participants showed evidence for serological infection. The estimated incidence was 4.4 per 100 person-years (95% CI 2.8–6.7). The piecewise linear rates approach resulted in a similar estimate of 4.6 per 100 person years (95% CI 2.9–6.9). Considering previous estimates of symptomatic scrub typhus incidence in the same study population, only about 2–5% of infections may result in clinically relevant disease.