Image: Dr Stephen-John Sammut
Unless you’ve been living under a rock for the past few years, you’ll no doubt have heard the bold assertions that artificial intelligence (AI) will eventually render most of us jobless. As computer-coded programs and algorithms become increasingly sophisticated, some people argue that they are threatening to – or, in some cases, already able to – outperform humans. After all, subsets of AI, known as machine learning and deep learning, make it possible for computers to learn from data to refine their performance and analyse huge quantities of complicated data.
However, cynics argue that humans possess skills and attributes that are unattainable to machines, including emotional intelligence, anticipation, judgement, creativity and the ability to build relationships. As a result, they say, there are limits to what AI can achieve, and most jobs will remain in human hands.
It remains to be seen exactly how far AI can go. What can’t be refuted, though, is its potential impact on healthcare. Beyond the more obvious uses for this technology, such as completing administrative tasks, it could also lead to improvements in patient care. For instance, in cancer, AI has been shown to be useful in diagnosing the disease, determining the treatments that are most likely to be effective for different patients and indicating the likely outlook.
The Institute of Cancer Research, London, has played a part in demonstrating the use of AI in these areas. Much of this research has suggested how the technology could be harnessed to improve the diagnosis and treatment of breast cancer.
Demonstrating how AI can improve the accuracy of diagnosis
In 2019, a team of ICR researchers successfully used AI to uncover five new types of breast cancer. They applied machine learning to molecular data and gene sequences from breast tumours, then used the trained AI model to identify types of disease with particular patterns of response to treatment. In doing so, they found that there were crucial differences among cancers that scientists had previously been including under one umbrella.
In the future, clinicians could use this or a similar model to help them make diagnoses with a higher degree of accuracy and completeness. In the meantime, researchers must work to build a better knowledge of the complexities of each type and subtype of breast cancer, and of how each one responds to different treatments. Armed with this information and a comprehensive diagnosis, clinicians will be able to identify the optimal treatment approach for each patient, leading to better outcomes.
Our researchers have already made great progress in understanding and treating breast cancer, but we are working tirelessly to continue to improve outcomes and quality of life for patients.
Learn more about our ground-breaking breast cancer research here.
Using AI to maximise treatment effectiveness
One researcher who believes strongly in the power of AI to improve patient outcomes is Dr Stephen-John Sammut, who joined the ICR in November 2022 as a Clinician Scientist and leader of the Cancer Dynamics Group in the Breast Cancer Now Toby Robins Research Centre. His research uses AI to predict how cancer will respond to treatment and to spot early signs that the disease is becoming resistant to drugs.
Dr Sammut was previously at the University of Cambridge, where he completed his medical oncology training. He was then awarded a postdoctoral lectureship in breast cancer and a PhD in breast cancer genomics at the Cancer Research UK Cambridge Institute.
Last year, Dr Sammut published his advances in personalised breast cancer medicine in a landmark paper in Nature, which was cited as one of the top 10 cancer research publications by the European Association for Cancer Researchers in 2022.
In the study, Dr Sammut and his colleagues took samples from 180 people with breast cancer and used AI to demonstrate that response to treatment was dependent on the characteristics of the tumour ecosystem – meaning both the malignant cells themselves and the environment around them.
Dr Sammut says: “Unfortunately, in the clinic, we are unable to predict response to cancer treatments. While some patients see exceptional results and are cured, many are less fortunate. But what if we could predict which patients will respond well to treatment before we even start? That's where I come in.”
At the ICR, Dr Sammut has continued to focus on developing machine learning methods that teach computers to predict cancer’s response to treatment by feeding them data about the wider tumour ecosystem. He also uses data acquired from cancers during chemotherapy to stay ahead of drug resistance and to detect it if it emerges during treatment.
Predicting outcomes using AI
Other ICR-led work has resulted in a tool called PhenMap – short for phenotype mapping – which integrates biological data with data on patient outlooks and treatment responses. The team responsible for developing this tool, led by Dr Anguraj Sadanandam, published a framework outlining how to apply it, using breast cancer datasets as an example.
Once PhenMap or a similar tool gains regulatory approval, clinicians will be able to use it to aid diagnosis. PhenMap groups patients either into discrete subgroups or as individuals along a spectrum, allowing clinicians to pinpoint the exact stage of a patient’s cancer. Alongside other patient information, this should help them provide an estimate of how and when the tumour is likely to progress.
Although there will inevitably be a delay in getting effective AI tools licensed for patient use, the ICR is well-placed to validate them in the lab and to develop clinical trials – working closely with our partner, The Royal Marsden NHS Foundation Trust, to ensure that they will have benefits in the clinical setting.
Dr Sammut says: “The ICR is truly passionate and committed to making a difference to the lives of the cancer patients I see in the clinic, and this is one of the main reasons why I work here. The ICR’s association with The Royal Marsden allows for unparalleled translational science, meaning the latest research advances can be quickly translated into tangible benefits for our patients.”
comments powered by