How Can AI Help Develop New Cancer Diagnosis & Research?

The rise of AI has been exponential over the last few years and it is only gathering pace.

Tools such as algorithm recommendations, task streamlining, search engine queries, customer service and generative chatbots are all commonplace – and for the field of oncology, AI can prove particularly beneficial.

AI can save researchers time when analysing data and modelling information, improving accuracy, speed and personalisation of service across the care continuum. 

How is AI used in medicine?

Currently, AI use in medicine and healthcare is still in its disruptive stage. It is transforming research approaches and applied patient care, beyond being simply a helpful admin tool for management.

Large language models (LLMs) such as ChatGPT have been tested for use in diagnosing internal medicine cases. Meanwhile, a DeepMind and Google model has been shown to deliver “safe and helpful answers” to patients’ and medical professionals’ questions. 

These examples work via a question-and-answer method, using written prompts to produce results. 

The LLMs in these cases have been trained on previous medical data so they can provide the most accurate responses. The expectation is that using AI in this way will become more common. 

How can AI be used in diagnosis and detection of cancer?

Another type of AI could be useful in diagnosis that does not involve patient interaction. Deep learning algorithms have the ability to analyse scans. 

These AI tools can look at MRIs, CTs and X-rays to pinpoint anything out of the ordinary. They can quickly and accurately detect tumours and cell masses. By comparing recent scans with previous versions, they can detect growth too.

AI can do this quickly, with potential to detect even small abnormalities in the scans. Early detection can be supported and care delivered responsively at a cancer treatment centre. 

How is AI being used to develop new cancer treatments?

Researchers developing new drugs and formulations for treatment need to deal with large quantities of data. The same goes for scientists investigating genetic markers and predicting prevalence and incidence of certain cancers. 

Using AI to analyse the data can shorten the time needed for research and development of new drugs. That means patients could access new treatments faster.

Recently, the AI tool AlphaFold2 has been used to successfully model protein structures to enable the discovery of new cancer drugs. 

Predicting how particular proteins will react to drugs requires large amounts of data processing. AI use in research means the process can not only be quicker but cheaper too, allowing the treatments to reach clinical trials and therefore patients sooner. 

How is AI being used in radiotherapy?

Broadly speaking, AI is being used across the entire radiotherapy workflow, improving precision, speeding up cancer care and reducing manual workload, as well as allowing tailored treatment for patients.

Key applications include:

• Autocontouring and segmentation, automatically outlining tumours on CT, MRI and PET scans
• Treatment planning and dose optimisation, forecasting distributions and optimising radiation delivery
• Adaptive radiotherapy, where treatment plans are adjusted based on real-time information
• Prognosis, helping to predict patient treatment responses

Could AI support cancer surgeries?

In the same way that AI can accurately detect masses of abnormal cells, it can also map them for use in surgery. Planning the approach to removing cancerous cells is often a complicated task that AI can potentially streamline. 

Advanced surgical techniques using robotics can be guided by real-time mapping and differentiation of the tissue so that margins are reduced, leaving more healthy tissue intact.

Surgeries can be more precise and less invasive, potentially leading to better outcomes for the patient and a quicker recovery.  

What could the future hold for AI in cancer treatments?

While there is no doubt that AI can be a useful tool, medical professionals will still be needed in the future. Patient care is far more than diagnosis and treatment. 

What AI tools lack is the human ability to empathise and show compassion for patients. The complexities of patient care, particularly when it comes to cancer treatments, require critical thinking, which goes beyond the scope of current AI models. 

There are also concerns about potential bias, safety and ethics when it comes to using LLMs for medical diagnosis. These are flaws in the system that need to be fixed before technologies are rolled out. 

The future is more likely to be a collaboration between AI and humans. The technology could be a way to provide more efficient diagnosis and treatment plans, but it should be balanced with a human approach to patient care. 

At Stingray, we look at AI as a co-pilot, handling data-heavy administrative tasks to allow medical professionals to focus on complex decision-making and clinical nuance, a seamless integration of this technology into existing workflows to deliver precision medicine at scale.