Job description

Job summary

Applications are invited for a Research Associate position within a project that strives to develop new methodologies to address the current gaps in our knowledge and lead to establishing a tool that will assess diet-induced non-communicable disease (NCD) risk. The project is funded by EU HORIZON and UKRI and titled "CoDiet: Combating diet related non-communicable disease through enhanced surveillance" and involves 17 partners from 10 countries. Work package (WP) 1 concerns the development of AI driven literature searching tools to bring clear understanding of the large global literature of the physiological and metabolic links between diet and NCD. This WP is lead by academic investigators from Imperial College London (Dr Joram Posma, Dr Marek Rei, Prof Gary Frost) and the University of Leicester (Dr Tim Beck), and involves multiple EU partners.

This position involves developing standardised data formats (e.g. processing articles from various sources), biomedical named entity recognition (bioNER) algorithms (e.g. what are the reported phenotypes, genotypes, methodologies), text summarization (e.g. to focus on summarising key sentences and results identified from the main text rather than the abstract) and interactive visualisation of results in online team-based annotation software (a consortium effort to curate a high-quality training and test set).

Duties and responsibilities

The topic area of diet and its relationship with development of NCD risk factors is very broad. It would be very ambitious to systematically review the literature across these broad topic areas and have the results available to be used within the project using traditional dietary assessment. Therefore, we will adopt a new AI based literature searching methodology. This will allow high quality literature review but in a fraction of the time of traditional reviews using the ongoing developments in the field of (biomedical) natural language processing (NLP).

You will be responsible to create automated systems for crawling and processing large numbers of medical articles from various online sources, apply existing bioNER algorithms for annotation of key terms and sentences, and fine-tune the algorithms using feedback from a team of expert annotators within the CoDiet consortium. These fine-tuned algorithms will then be deployed on the remaining literature (we estimate this to be >20,000 relevant articles) in fully automated fashion. For compute, you will have access to dedicated GPU resources in the Department of Computing. The combination of the separate algorithm outputs will enable articles to be summarised and their relevance judged based on the search query, thereby enabling CoDiet to review and summarise a much larger portion of the literature than would be possible using current manual methods. The goal of CoDiet is to review the full available literature using the initial search terms to provide the most comprehensive overview of relationships between diet, NCDs and physiological processes to help other WPs prioritise their analytical and data science tools. You will create a live, interactive graph that summarises the literature and connections between entities that will facilitate the knowledge transfer, and write papers to disseminate the work to a wider audience.

The role will require regular contributions to a variety of normal academic tasks, including management of research projects, positively interacting and communicating with internal and external partners, developing research proposals, supervision of Master and PhD students, and demonstrating research excellence and leadership to more junior members of the Research Group. You will lead on the workflow and dataset creation, and collaborate with two other postdocs to be recruited for WP1 and with CoDiet staff from partners in Spain, Ireland and Greece. You will receive strong mentoring in a stimulating multidisciplinary academic research environment, with access to internationally leading academics, industrial partners, research computing facilities, and networks.

Essential requirements

• Research Associate: Hold a PhD in computer science, biostatistics or a closely related field, or equivalent research, industrial or commercial experience


• Research Assistant: Near completion of a PhD in computer science, biostatistics or a closely related field, or equivalent research, industrial or commercial experience

*Candidates who have not yet been officially awarded their PhD will be appointed as a Research Assistant.

• Experience with implementing NLP applications in Python


• Basic understanding of biological and medical terminology


• Productive track record of publishing high-quality papers in peer-reviewed journals, or in conference proceedings


• Well organised, attention to detail, critical thinking, and ability to work as part of a team

Further information

This is a full time and fixed term post for 24 months, with an anticipated start date in March 2023. Later start dates in 2023 are also possible, depending on the circumstances of the candidate. You will be based at the Hammersmith Hospital Campus (East Acton).

Candidates who have not yet been officially awarded their PhD will be appointed as a Research Assistant within the salary range £38,194 - £41,388 per annum.

Another 18-month position in this project will become available at Imperial with a start date of 1 August 2023 and will be centered around relation extraction (e.g. relation between phenotypes and genotypes within sentences and table data), and question-answering (e.g. extracting the relevant sentences based on a research question). A separate post in WP1 will be based at Leicester. These three posts have a close working relationship.

For informal enquiries on the vacancy please contact Dr Joram Posma (Section of Bioinformatics, Department of Metabolism, Digestion and Reproduction) at jmp111@imperial.ac.uk.

The College is a proud signatory to the San-Francisco Declaration on Research Assessment (DORA), which means that in hiring and promotion decisions, we evaluate applicants on the quality of their work, not the journal impact factor where it is published. For more information, see https://www.imperial.ac.uk/research-and-innovation/about-imperial-research/research-evaluation/

The College believes that the use of animals in research is vital to improve human and animal health and welfare. Animals may only be used in research programmes which are ultimately aimed towards finding new treatments and making scientific and medical advances, and where there are no satisfactory or reasonably practical alternatives to their use. Imperial is committed to ensuring that, in cases where this research is deemed essential, all animals in the College’s care are treated with full respect, and that all staff involved with this work show due consideration at every level. Find out more about animal research at Imperial.

Documents

• Final JD - MED03732 - GS.pdf