Mukunthan Tharmakulasingam

Predicting Antimicrobial Resistance (AMR) from genomic data has important implications for human and animal healthcare, and especially given its potential for more rapid diagnostics and informed treatment choices. With the recent advances in sequencing technologies, applying machine learning techniques for AMR prediction have indicated promising results. Despite this, there are shortcomings in the literature concerning methodologies suitable for multi-drug AMR prediction and especially where samples with missing labels exist. To address this shortcoming, we introduce a Rectified Classifier Chain (RCC) method for predicting multi-drug resistance. This RCC method was tested using annotated features of genomics sequences and compared with similar multi-label classification methodologies. We found that applying the eXtreme Gradient Boosting (XGBoost) base model to our RCC model outperformed the second-best model, XGBoost based binary relevance model, by 3.3% in Hamming accuracy and 7.8% in F1-score. Additionally, we note that in the literature machine learning models applied to AMR prediction typically are unsuitable for identifying biomarkers informative of their decisions; in this study, we show that biomarkers contributing to AMR prediction can also be identified using the proposed RCC method. We expect this can facilitate genome annotation and pave the path towards identifying new biomarkers indicative of AMR.

Predicting Antimicrobial Resistance (AMR) from genomic sequence data has become a significant component of overcoming the AMR challenge, especially given its potential for facilitating more rapid diagnostics and personalised antibiotic treatments. With the recent advances in sequencing technologies and computing power, deep learning models for genomic sequence data have been widely adopted to predict AMR more reliably and error-free. There are more than 30 different types of AMR; therefore, any practical AMR prediction system must be able to identify multiple AMRs present in a genomic sequence. Unfortunately, most genomic sequence datasets do not have all the labels marked, thereby making a deep learning modelling approach challenging owing to its reliance on labels for reliability and accuracy. This paper addresses this issue by presenting an effective deep learning solution, Mask-Loss 1D convolution neural network (ML-ConvNet), for AMR prediction on datasets with many missing labels. The core component of ML-ConvNet utilises a masked loss function that overcomes the effect of missing labels in predicting AMR. The proposed ML-ConvNet is demonstrated to outperform state-of-the-art methods in the literature by 10.5%, according to the F1 score. The proposed model's performance is evaluated using different degrees of the missing label and is found to outperform the conventional approach by 76% in the F1 score when 86.68% of labels are missing. Furthermore, the proposed ML-ConvNet is established with an explainable artificial intelligence (XAI) pipeline, thereby making it ideally suited for hospitals and healthcare settings where model interpretability is an essential requirement.

Background Human, animal, and environmental health are increasingly threatened by the emergence and spread of antibiotic resistance. Inappropriate use of antibiotic treatments commonly contributes to this threat, but it is also becoming apparent that multiple, interconnected environmental factors can play a significant role. Thus, a One Health approach is required for a comprehensive understanding of the environmental dimensions of antibiotic resistance and inform science-based decisions and actions. The broad and multidisciplinary nature of the problem poses several open questions drawing upon a wide heterogeneous range of studies. Objective This study seeks to collect and catalogue the evidence of the potential effects of environmental factors on the abundance or detection of antibiotic resistance determinants in the outdoor environment, i.e., antibiotic resistant bacteria and mobile genetic elements carrying antibiotic resistance genes, and the effect on those caused by local environmental conditions of either natural or anthropogenic origin. Methods Here, we describe the protocol for a systematic evidence map to address this, which will be performed in adherence to best practice guidelines. We will search the literature from 1990 to present, using the following electronic databases: MEDLINE, Embase, and the Web of Science Core Collection as well as the grey literature. We shall include full-text, scientific articles published in English. Reviewers will work in pairs to screen title, abstract and keywords first and then full-text documents. Data extraction will adhere to a code book purposely designed. Risk of bias assessment will not be conducted as part of this SEM. We will combine tables, graphs, and other suitable visualisation techniques to compile a database i) of studies investigating the factors associated with the prevalence of antibiotic resistance in the environment and ii) map the distribution, network, cross-disciplinarity, impact and trends in the literature.

An artificial intelligence-assisted low-cost portable device for the rapid detection of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is presented here. This standalone temperature-controlled device houses tubes designed for conducting reverse transcription loop-mediated isothermal amplification (RT-LAMP) assays. Moreover, the device utilises tubes illuminated by LEDs, an in-built camera, and a small onboard computer with automated image acquisition and processing algorithms. This intelligent device significantly reduces the normal assay run time and removes the subjectivity associated with operator interpretation of colourimetric RT-LAMP results. To further improve this device's usability, a mobile app has been integrated into the system to control the LAMP assay environment and to visually display the assay results by connecting the device to a smartphone via Bluetooth. This study was undertaken using similar to 5000 images produced from the similar to 200 LAMP amplification assays using the prototype device. Synthetic RNA and a small panel of positive and negative SARS-CoV-2 patient samples were assayed for this study. State-of-the-art image processing and artificial intelligence algorithms were applied to these images to analyse them and to select the most efficient algorithm. The template matching algorithm for image extraction and MobileNet CNN architecture for classification results provided 98.0% accuracy with an average run time of 20 min to confirm the endpoint result. Two working points were chosen based on the best compromise between sensitivity and specificity. The high sensitivity point has a sensitivity value of 99.12% and specificity value of 70.8%, while at the high specificity point, the sensitivity is 96.05% and specificity 93.59%. Furthermore, this device provides an efficient and cost-effective platform for non-health professionals to detect not only SARS-CoV-2 but also other pathogens in resource-limited laboratories, factories, airports, schools, universities, and homes.