An artificial intelligence healthcare company claims its software is better at detecting prostate cancer compared to traditional methods used by doctors.
Avenda Health conducted a study with ten physicians evaluating 50 cases of prostate cancer. The findings revealed that Avenda’s Unfold AI software achieved an accuracy rate of 84.7%, whereas physicians practicing manual detection ranged from 67.2% to 75.9% in accuracy.
This study, carried out in collaboration with UCLA Health and published in the Journal of Urology, highlighted that AI-assisted cancer contouring improved cancer size predictions by 45 times, demonstrating increased accuracy and reliability compared to manual methods.
Shyam Natarajan, an assistant adjunct professor of urology, surgery, and bioengineering at UCLA and the study’s senior author, noted that the integration of AI assistance resulted in higher accuracy and consistency among doctors, leading to greater agreement among them in diagnoses.
Doctors typically rely on MRIs to gauge tumor size; however, some tumors are not visible on MRI scans. Dr. Wayne Brisbane, an assistant professor of urology at UCLA’s David Geffen School of Medicine, emphasized that AI technology offers solutions when MRIs fall short.
Dr. Brisbane stated that utilizing AI in cancer treatment could enhance personalized care for patients, tailoring treatments more effectively to their specific needs and improving success rates against the disease. He remarked that AI can exceed human capabilities.
Avenda Health CEO Dr. Shyam Natarajan expressed that it is empowering for physicians to witness such validation of innovation through studies, with recognition from the American Medical Association.
According to the American Cancer Society, around 1 in 8 men in the United States will receive a prostate cancer diagnosis in their lifetime, and 1 in 44 men are expected to die from the illness. This year, an estimated 299,010 new prostate cancer cases are anticipated in the U.S., with approximately 35,250 resulting in death from the disease.