Can artificial intelligence help detect early-stage cancer?

Artificial intelligence (AI) might be under heavy scrutiny these days, but you’ve probably been using AI well before you started using ChatGPT. When you purchase something online, more than likely you’ll see ads for similar products elsewhere on the web. Your Google searches are typically based on some of your past searches. You also use AI anytime you need a quick translation of a foreign language (via European Parliament).

AI is used in the healthcare industry to assist in reviewing and analyzing health information, according to a 2023 article in the Journal of Multidisciplinary Healthcare. One subset of AI is machine learning, which uses specific algorithms that learn from patient data and then make continual improvements without explicit programming. Researchers are now creating AI algorithms based on past images of people with various types of cancer. These AI models also rely on patients’ medical history and other patient data (such as age, family history, or lifestyle) to predict and identify early stages of cancer.

How AI works to detect cancer

According to the National Cancer Institute, researchers build AI algorithms by telling the model what cancer looks like on an image such as an MRI. Then the model learns what a healthy organ or tissue looks like so it can distinguish what is cancerous and what is not. The algorithm eventually teaches itself how to interpret new images based on past images and determine someone’s risk of cancer. 

For example, a 2023 study in Radiology tested several breast cancer risk models by comparing the data from women who had negative and positive mammograms in 2016. Typically, doctors will rely on the Breast Cancer Surveillance Consortium (BCSC) to determine a person’s risk for breast cancer, and it’s based on family history, age, whether a person has had children, breast density, and other measures. The study found that the AI algorithms did a better job at predicting the five-year risk of breast cancer than the BCSC risk model. The researchers said that AI could possibly detect any missed cancers and identify precancerous breast tissue.

AI can identify cancer more accurately and efficiently

AI is used to help detect cancer in real-time during surgeries and screenings, according to the 2023 article in the Journal of Multidisciplinary Healthcare. If you’re having brain surgery, AI can identify any malignant tissue during the surgery. During a colonoscopy, AI can help diagnose tissue and recognize any blind areas during the screening. 

Radiologists sometimes disagree over their interpretation of MRI scans, but AI algorithms could interpret images more reliably and quickly. Although multiparametric MRI gives a more accurate picture of the prostate gland, it can take years for radiologists to learn how to read them. Researchers have developed an AI model to detect prostate cancer in a multiparametric MRI that can also help train radiologists how to read the images.

Radiomics is a deep learning technique that looks at medical images and detects cancers that the human eyes might not be able to see. The AI learns patterns and traits of certain diseases and matches them with a patient’s genetic data. This technique was used in a 2022 study in The Lancet, which recruited more than 500 patients who had large nodules in their lungs. The AI algorithm did a better job than another lung cancer model in identifying high-risk lung cancer patients who needed more immediate treatment.