AI for cancer diagnosis and treatment

You’ve been anxiously waiting for the test results to come back. Life seems to be on pause while you are desperately hoping and praying it is benign. All other daily activities are placed in sharp perspective by the health scare.

If treatment is needed, your prayers shift to hoping it was caught early, and that treatment is both effective and free of side-effects. Post treatment payers are then for recovery and continued remission.

Theresa May will today announce that “The diagnosis of cancer and other diseases in the UK can be transformed by using artificial intelligence”. In fact artificial intelligence (AI) has the potential to answer our prayers during almost every part of the process: to make detection earlier, more decisive, treatments more effective, with fewer side-effects and better rates of remission.

Big data – the key underpinning technology for AI – means using thousands or millions of data points to enable artificial learning systems to explore and deduce relationships between cause and effect. By studying large amounts of data from the population as a whole, or from target groups, or even large amounts of data from a single person over time, these systems can build an understanding of individuals and groups.

In human terms, the AI systems learn how to;

  • Understand more about people like you – your health-related behaviours and their risks.
  • Know you better as an individual – what is normal for you, and how have you changed.
  • Model and track the progression of diseases and conditions in the population.
  • Treat diseases like yours earlier with higher success rates, fewer side-effects and better outcomes.
  • And yes; to do all of the above at lower cost.

An AI system can get to know you from your data, and most importantly will get to know how you are changing. Disease markers sometimes take months or years to become visible to the naked eye, but AI-based monitoring can identify fine-grained and correlated changed in an individuals’ daily life patterns. These patterns again reveal effectiveness of treatments, updated if necessary to the millisecond (rather than after every GP visit).

These tools are no substitute for the skill, experience and dedication of a good GP – at least not within the near future – but could become crucial in advising your local family doctor and consultant alike. After all, AI can know you better, observe you more closely, track you more frequently, direct your treatment more effectively, and monitor your outcome more objectively than any human.

Yet when you are faced with the long wait, when you receive that dreaded news, and then when living after cancer, you’ll need the human face of your family doctor more than ever.

Professor Ian McLoughlin
Medway School of Computing
Data Science Research Group

See press release here