Cancer is a leading cause of death in the United States, with incidence on the rise. The American Cancer Society predicts there will be over two million new cases and 611,720 deaths from cancer in 2024, largely driven by breast, colorectal, lung and prostate cancer, the four most common types of cancer.
As statistics show, there’s a pressing need for early cancer detection and more effective cancer treatments. We know that cancer screening plays an important role in early detection and that diagnosing cancer at an early stage increases the chance of successful treatment.
Artificial intelligence (AI) is currently used in cancer diagnostics, but what is the role of AI in cancer treatment? We’ll answer that question and explore how AI is shaping personalized therapies and improving patient outcomes.
To understand the role of AI in oncology, it’s essential first to understand the nature of the disease and what we mean by personalized medicine.
Cancer is a disease in which some cells in the body rapidly grow and divide uncontrollably. It’s considered a genetic disease caused by mutations in genes that control how cells grow and multiply. It is a hugely complex and heterogeneous disease, meaning there are distinct differences in tumor characteristics even within the same tumor type.
As such, tumors can be difficult to predict regarding how they will respond to treatment and grow. Cancer patients with the same treatment and tumor may have different responses and curative effects. This highlights the need for more precise treatment methods and personalized treatment plans to improve cancer care.
Helpful tip: Learn more about the history, causes, and advancements in cancer.
In personalized or precision medicine, clinicians use specific information on genomic abnormalities, proteins, and other substances found in a person's body to develop a tailored treatment plan or approach.
In oncology, personalized medicine involves looking at genes or proteins in an individual’s cancer cells; with this information, the oncologist can create a personalized treatment plan. Certain genetic mutations may help predict how well that cancer will respond to a specific treatment or give information regarding prognosis. The clinician may refer to genetic testing or biomarkers. An example of a bio-marker is HER-2 in breast cancer. If this marker is present, a treatment can be given to target this.
Personalized medicine is also useful in cancer diagnosis and prevention. It can help detect who might be at risk of cancer and help diagnose cancers early.
A patient-centered approach is important in oncology, with cancer being such a complex disease. With AI and personalized medicine, clinicians can give cancer patients more accurate data and information regarding treatment options, response and outcome prediction. Shared decision-making is key. It would enable clinicians to weigh the risks and benefits more accurately and avoid treatments that could cause considerable harm through toxic side effects.
Artificial intelligence is accelerating personalized medicine, cancer research, and cancer care. It has numerous applications throughout the field of oncology, from cancer detection and diagnosis to the characterization of tumors and their environment, drug discovery, and prediction of treatment response and outcomes.
AI is computer science that makes intelligent machines behave like intelligent humans. It uses a set of theories, algorithms, and computing powers to carry out tasks such as decision-making, reasoning, and language understanding. In healthcare, AI uses software, algorithms, and machine learning to understand complex health and medical data.
This is a subset of artificial intelligence. Machine learning (ML) uses data and algorithms to imitate how humans learn, allowing machines to perform complex tasks without explicit human instruction. The machine gradually improves performance over time and aims to create “intelligent” systems to make accurate predictions from various data. These models and algorithms can potentially be useful in cancer diagnostics and predicting outcomes.
This subset of machine learning uses multi-layered neural networks to replicate the complex decision-making of the human brain. Deep learning plays a role in healthcare through image recognition applications to support radiologists in analyzing and assessing medical images. For instance, Ezra uses AI to assist in identifying possible lesions on prostate MRIs and to enhance MRI images acquired at higher speeds.
Helpful tip: Learn more about Ezra’s applications of AI.
The use of AI algorithms in cancer treatment is varied with the ultimate aim of improving precision or personalized medicine. The prospect of using AI to predict and treat cancer is relatively new. In this digital era, clinicians are becoming aware of the need to use AI technologies, such as machine learning models and deep learning, to accelerate cancer care and improve outcomes.
In oncology, accurately estimating which treatment regimens will be most effective and least toxic for each patient is challenging. Estimates are usually based on molecular, genetic, and other tumor features, and AI could play a pivotal role in overcoming this challenge.
It is particularly relevant for certain immunotherapy drugs, which have variable response rates. The use of AI in this domain is growing, and AI models have been used to predict responses to immunotherapy in advanced melanoma patients. Furthermore, AI models have been used to predict response to neoadjuvant chemotherapy (chemotherapy given before surgery) from PET/MRI scans in advanced breast cancer patients. The predictive accuracy was reported to be better than some conventional methods.
Assessing response to chemotherapy accurately is vital as it allows clinicians to stop and alter the treatment regimen in a timely manner if the patient has not had a satisfactory response.
There’s also evidence that AI tools are useful in identifying prognostic biomarkers that can help guide treatment decisions. A study published by The Lancet has shown that deep learning models have successfully detected a clinically useful prognostic biomarker in colorectal cancer. This could help decipher which patients would be suitable for certain treatments.
AI can use information to provide more precise predictions on disease forecasts. It can use data based on clinical assessments, investigations, scans performed, patient medical histories, and other information to predict disease prognosis, progression, and patient survival more accurately.
Machine learning algorithms have been shown to give accurate prognostic information on skin cancer, equivalent to that of a skilled dermatologist. Furthermore, deep learning techniques through image analysis at multiple time points have been shown to successfully predict prognosis, survival, and pathological response in patients with locally advanced lung cancer treated with radiation therapy.
Helpful tip: Find out more about the current use of AI in diagnosing lung cancer.
The method of discovering and developing new cancer drugs is costly and time-consuming. Artificial intelligence has the potential to accelerate this process due to the availability of large datasets and access to genetic analysis and imaging technologies. AI models have been used to predict potential cancer drug targets in liver cancer and breast cancer. AI systems are also helpful in analyzing large datasets to identify subsets that may be relevant for a particular predictive tool.
The potential role of AI in drug discovery, development, screening, and manufacturing is extensive. It’s hoped that with the latest AI technologies, new drugs can be developed more quickly and inexpensively while being safer and more effective for patients.
Effective and robust AI models require data that accurately represents the entire human population. Otherwise, it could lead to bias. We know that socioeconomic factors, as well as race and gender, impact disease risk and course. Existing datasets are biased towards certain races and ethnicities, so moving forward, it will be crucial for datasets to represent the full diverse population accurately.
Another issue is data privacy, confidentiality, and accessibility. High-quality and large quantities of data are the foundation for an effective AI model. Access to private and confidential patient data is an ongoing issue. Restricting confidential data may in turn, lead to less robust and accurate AI models.
Healthcare data is collected on various platforms, often with controlled access. An appropriate balance needs to be struck between privacy protection, technological development, and the interests of individuals, the public, and corporations.
The use of artificial Intelligence is growing fast within the field of oncology. However, its current use is mostly applied to cancer diagnosis and screening. The role of AI in cancer treatment is relatively new and not yet established in practice, but the potential applications are vast. From aiding the development of new drugs to predicting treatment response, disease course, and prognosis, it could accelerate personalized medicine growth.
AI can reshape the future of oncology and care for cancer patients. Early detection, however, is key. That includes taking charge of your health before you even have symptoms.
Screening is the best defense against cancer, and the Ezra Full Body Scan uses cutting-edge advancements in AI to enhance medical images and generate reader-friendly reports following image analysis. The scan takes just one hour and gives information on up to 13 organs.
Be proactive and consider booking an Ezra Full Body MRI today.
