An Imperial-led project will use AI to provide more accurate and up-to-date information on breast cancer to aid clinical decision-making.

Imperial experts, including Senior Lecturer in Colorectal Surgery Dr James Kinross, Professor in Computational Logic Professor Francesca Toni, Research Fellow in Intelligent Data Processing and Curation and Lead of the Data Observatory at the Data Science Institute (DSI) Dr Ovidiu Serban, and NIHR Clinical Lecturer and Vascular Surgeon in the Department of Surgery and Cancer Dr Guy Martin have secured £567,000 in funding to be used to tackle the cancer infodemic.

“This project represents a promising tool for use in the future of information retrieval research and data quality in medicine. We are currently validating our AI pipeline for cancer research, while also looking at AI-driven summarisation and fact-checking.” Dr Ovidiu Serban Research Fellow in Intelligent Data Processing and Curation

The newly funded project, INDICATE, will use AI to meaningfully improve clinical decision-making in relation to breast cancer by developing a system to provide a real-time analysis of health data from both published and non-peer-reviewed social data sets.

The aim of the project is to use their new platform, CURADR, to develop the most accurate and up-to-date guidance for breast cancer treatment and to filter out any misinformation.

According to the DSI’s Dr Ovidiu Serban: “This project represents a promising tool for use in the future of information retrieval research and data quality in medicine. We are currently validating our AI pipeline for cancer research, while also looking at AI-driven summarisation and fact-checking.”

Funded by UKRI, INDICATE has been created through a collaborative network of Imperial College London, AWS, NICE and the BMJ. 

The INDICATE project was part of £13 million funding that Imperial secured from UKRI. Read the full Imperial News Story here. 

Combatting a cancer infodemic

Every year, over 250,000 people in England are diagnosed with cancer, and 130,000 die due to the disease. However, the overabundance of information relating to cancer, often called the ‘cancer infodemic’, remains a significant barrier to the identification of novel cancer therapies, biomarkers and early detection strategies.

This is because the speed and volume of clinical evidence is overwhelming and almost impossible to accurately condense into the best available guidance.

Additionally, the growth of medical mis- and disinformation and the ability of modern AI-supported natural language models such as ChatGPT to create readable but factually flawed text-based outputs, stresses the need for autonomous medical literature summarisation that is robust and fit for purpose.

These challenges present significant hurdles for medical publishers, which must continuously understand and summarise emerging data on novel therapies for clinicians and healthcare providers.

INDICATE is a solution to these challenges; it is a medical infodemic engine that performs AI-supported peer review of medical literature for generating continuous, systematic reviews of health-related data.

Learning from COVID-19: REDASA

INDICATE uses technology created initially to tackle the overabundance of information as a result of the COVID-19 pandemic.

The computational tool, known as REDASA (REaltime Data Synthesis and Analysis), uses deep learning combined with the knowledge of medical experts to filter large amounts of complex information to extract only the most relevant and important parts to be used for clinical guidance.

The study, published in the Journal of Medical Interest Research, used REDASA to create one of the world’s largest and most up-to-date sources of COVID-19-related evidence, consisting of over 104,000 documents.

Keeping humans in the loop

The REDASA’s design adopts a user-friendly, human-in-the-loop methodology by embedding an efficient, user-friendly curation platform into a natural language processing search engine.

This means that human experts are involved in the decision-making process along with AI components: the automated system filters down the information to a manageable amount, humans then check the results and assess the quality of the selected work and another AI component then verifies that the experts are consistent.

The INDICATE project will use the same human-in-the-loop technology, implemented through a new platform called CURADR. CURADR allows a community of clinical experts to set and rank clinical research questions, and to peer review multiple research questions across various clinical domains.

With uses from COVID-19 to cancer, the funding will help develop this novel data synthesis method to enable clinicians to give the best available treatments for a range of healthcare applications moving forward.

-

Find out more about the UKRI funding bid by visiting their website. 

Read more about REDASA in this Imperial News Story.