Recently there has been huge interest in the application of artificial intelligence (AI) to medicine, and there is accumulating evidence that AI can assist clinicians to make better clinical decisions or even replace human judgement in certain areas of healthcare. This is due to the increasing availability of healthcare data and rapid development of big data analytic methods. These AI technologies fall into several groups, including classical machine learning techniques, neural networks, and deep learning techniques: however, the rate of change of this research area means that new approaches are continuously being developed.
- Project Lead: Dr Owain Jones
- Senior CanTest Lead: Dr Fiona Walter
- Others involved: Professor Willie Hamilton, Professor Niek de Wit, Professor Hardeep Singh, Professor Stephen Duffy (QMUL)
Additional funding from the Cancer Policy Research Unit – led by Prof Stephen Duffy (QMUL)
Aims & objectives
In this scoping review we will focus on which modalities of AI for early detection and diagnosis of cancer are ready for possible implementation in the NHS.
We will include:
- AI-driven cancer risk prediction tools for the symptomatic population consulting in primary care;
- AI-driven pattern recognition systems for use across the NHS (e.g. applied to dermoscopy images to identify possible melanomas)
We will explore the AI techniques that are being utilised and the clinical areas in which they are being employed. We will also explore the stage of development that each technology has reached, the diagnostic accuracy and/or health-economic assessments that have been undertaken, and any implementation barriers that have been encountered.
Methodology & activity
We will follow scoping review methodology, and search Medline, Embase, Scopus and Web of Science from 1st January 2000 to 31st December 2018 for relevant published studies. The search will be manually extended by: hand-searching reference lists of included articles, hand-searching of key journals, hand-searching of existing networks (e.g. the International Skin Imaging Collaboration), organisations (e.g. Google, Microsoft, IBM etc) and conferences.
Title and abstract screening will be completed by at least two reviewers independently. Data extraction will then also be undertaken independently by at least two reviewers. Quality assessment of the included studies will utilise the Joanna Briggs Institute Critical Appraisal Tools. We will work with Professor van der Schaar’s research group to provide expertise on machine learning and contribute to interpretation of findings.
Outputs & impact
As this review is a scoping study, we will aim to present a broad overview of all material reviewed, without necessarily synthesizing the results. This stage will be refined depending on the data identified during the data extraction stage but will likely take two approaches:
- organising the literature thematically based on intervention types, to give an overview of the different interventions that are in development;
- summarizing any numerical data using tables and charts.
The wider impact of the study will be to provide an overview of the current state of artificial intelligence research in primary care, and to identify any technologies that are potentially ready for implementation in clinical practice. For technologies that are not yet ready for implementation but demonstrate potential, the review will identify the future steps needed to develop these technologies to the point where they will be ready for implementation.
The findings will underpin further workstreams evaluating the diagnostic accuracy, acceptability and cost-effectiveness of AI approaches for early detection and diagnosis of cancer. The review may also identify next steps in the development of other AI technologies, and the evidence needed for implementation in primary care.