AI Technology Uncovers Cancer Indicators Overlooked by Physicians

By Jay S June 11, 2024

In a significant breakthrough, an AI solution known as Mia has showcased its capacity to identify cancer indications that evaded detection by human radiologists.

Image Courtesy of Kheiron Medical Techologies

Mia participated in a trial conducted alongside NHS clinicians in the UK, examining mammograms from more than 10,000 women. Despite the majority of participants being free of cancer, Mia effectively identified all individuals exhibiting signs of breast cancer.

Impressively, it also uncovered 11 additional cases that human doctors had missed. Out of the 10,889 women engaged in the study, merely 81 elected not to have the AI system assess their scans. Mia’s efficacy is attributed to its training on an extensive dataset comprising more than 6,000 prior breast cancer cases.

Image Courtesy of metamorworks/Shutterstock

Through this extensive training, Mia learned to identify subtle patterns and imaging biomarkers indicative of malignant tumors. In evaluations with new cases, Mia demonstrated impressive precision, accurately detecting cancer 81.6 percent of the time and correctly ruling it out in 72.9 percent of instances. Breast cancer continues to rank as the most common cancer among women globally, with an annual diagnosis of two million new cases.

Despite strides in early detection and treatment, numerous patients endure severe side effects, including lymphoedema post-surgery and radiotherapy. To tackle this issue, researchers are broadening Mia’s functionalities to forecast a patient’s likelihood of encountering these side effects for up to three years after treatment. This could empower physicians to personalize patient care by offering alternative therapies or additional supportive measures to high-risk patients.

Image Courtesy of – Yuri A/Shutterstock

To confirm the accuracy of Mia’s risk prediction model, the research team is launching a clinical trial called Pre-Act, with plans to enroll 780 breast cancer patients. Over a two-year follow-up period, this trial will prospectively evaluate the AI system’s ability to forecast patient outcomes. Ultimately, the researchers envision an AI solution that can provide comprehensive evaluations of a patient’s prognosis and treatment needs, heralding a new era of personalized cancer care.