Skip to main content
Home » Oncology » Why the way forward in oncology is data-driven
Innovations in Oncology 2019

Why the way forward in oncology is data-driven

iStock / Getty Images Plus / Jeng_Niamwhan

Dr Tom Mikkelsen

Medical Director of the Precision Medicine Program at the Henry Ford Health System, Detroit, USA

A data-driven approach to healthcare has the potential to speed up diagnosis and guide more effective, personalised treatments. It could have huge benefits in the field of oncology.

Some cancers can be difficult to detect, largely because symptoms are not always obvious in their early stages.

“Many cancers that arise in deep-seated organs — such as the pancreas — may not announce themselves clinically until they are well-advanced,” says Dr Tom Mikkelsen, Medical Director of the Precision Medicine Program at the Henry Ford Health System in Detroit, USA. “Until now, physicians have had to witness a change in symptoms over time, or notice a pattern that announces itself as cancerous.”

Personalised and more efficient diagnosis

However, data-driven medicine — powered by artificial intelligence (AI) — means that healthcare is entering a dynamic new era: one that is personalised, more efficient and with the potential to make a positive impact on diagnosis and treatment. As such, it could have huge benefits in the field of oncology.

Liquid biopsy is in its early stages, but is an incredibly exciting development.

There’s certainly no shortage of medical data to analyse. “On the DNA side, there’s an avalanche of information,” says Dr Mikkelsen. “This includes data about the mutations a tumour carries, in what fraction of the cell the mutation persists when the tumour reoccurs, and how the tumour has escaped therapy.” On top of this there’s more standard medical information about a patient’s contact with their physician over the years, and lab results such as blood tests, etc.

Possibility of better outcomes for patients

Close analysis of personal medical data may help speed diagnosis and trigger earlier interventions which are more likely to result in a positive outcome for individual patients. For example, someone who may be genetically predisposed to certain types of cancer might be surveyed more closely, either with intensive screening or — hopefully soon — with a new diagnostic methodology called liquid biopsy, which can detect mutations from cancer cells in the blood. “Liquid biopsy is in its early stages, but is an incredibly exciting development,” says Dr Mikkelsen.

A data-driven approach can also result in more effective treatments. “If specific mutations are matched with specific drugs, it means that patients are more likely to receive therapy that they will respond to positively, and without adverse side-effects,” says Dr Mikkelsen. “Essentially it allows for a more tailored and focussed treatment, based on information about an individual’s specific cell mutations.”

This breakthrough has been made possible thanks to artificial intelligence algorithms which can discover patterns in vast amounts of existing medical data, allowing medical professionals to make better decisions and more informed predictions. Dr Mikkelsen urges a note of caution, however, because this technology is only as good as how we ‘human’ train it and thanks to the quality of data we put into it.

Need for collaboration on a global scale

“AI is not magic,” he says. “Machine learning can only be done with valid, quality, reproducible, uniform data gleaned from large populations. So it might sound mundane, but it’s critical for healthcare professionals to ensure that all data — from clinical examinations right through to how medical records are documented — is standardised.”

He also stresses the need for a global approach to data-driven strategies. “The teams involved in this research are international but they all need to be talking the same language,” he says. “Their collaboration is crucial because no single institution can gather enough information on, say, a rare cancer cell mutation that only exists in one or two per cent of a population. They have to work together to share their knowledge in order to properly identify those patients and so improve the value of therapy.”

Next article