In recent years there has been rapid growth in the use of machine learning (ML) and artificial intelligence (AI) to analyse the vast amounts of data being generated in all areas of society. ML and AI approaches are increasingly being employed in health research, and sometimes are even explicitly called for by funders. While ML and AI may offer advantages over traditional statistical methods for certain tasks, they tend to use complex algorithms and are often regarded as analytical “black boxes”. Hence these novel methods pose new challenges of statistical literacy for producers and consumers of health research. Such challenges may be compounded by an underlying lack of communication and understanding between the statistics and ML/AI communities. Are statisticians slow to embrace the potential of ML and AI methods, and guilty of trying to protect their role in collaborative research in the face of competition from other fields? Or are they rightly concerned about the fundamentals of robust, reproducible and interpretable research being forgotten in the clamour for applying fashionable methods understood by few? This meeting will explore issues in statistical literacy raised by the growth of ML and AI in health research.
- Gary Collins, Professor of Medical Statistics at the University of Oxford and co-lead of the TRIPOD-AI initiative: “Methodological and reporting issues in machine learning based clinical prediction models”
- Aldo Faisal, Professor of AI & Neuroscience at Imperial College London & Director, UKRI Centre for Doctoral Training in AI for Healthcare: “AI for Healthcare = Data + Clinical Agency”
- Xiao Liu, Ophthalmologist at the University of Birmingham and co-lead of the SPIRIT-AI and CONSORT-AI initiative: “SPIRIT-AI & CONSORT-AI: Reporting Guidelines for Clinical Trials involving AI interventions”
- Mario Moroso, Assistant Director, NIHR Central Commissioning Facility
RSS Medical Section & NIHR Statistics Group