A team of scientists from Imperal College London (UK) and Google Health have created an artificial “doctor” whose ability to diagnose breast cancer surpasses all their good radiologists.
The “doctor” just introduced through an article published in the scientific journal Nature is actually an AI-type algorithm (Artificial Intelligence), “living” on a computer.
In the latest test, this AI, with its “expertise” in reading diagnostic imaging obtained an outstanding score in a race with 6 seasoned radiologists. In terms of ability, A.I can do its job as effectively as two good doctors with the same efforts, even though it only has a basis for diagnosis, it is an X-ray film, but it does not have contact with the patient. like doctors.
The data for the trial were records of more than 29,000 patients from St George’s Hospital London, Jarvis Breast Center in Guildford and Addenbrooke Hospital in Cambridge.
Statistics show that diagnosis by AI eliminates 1.2% false positives and 2.7% false negatives compared to human diagnostic ability. It is very important to help reduce the false negative rate, because these are the patients who are at risk of being missed, leading to when detected, the disease is already severe, with no chance of cure.
In addition, with the working capacity of 2 people, AI can also help reduce the workload for doctors, especially in the context that breast cancer cases are detected more and more. A radiologist will work with the A.I to ensure the most accurate results. This A.I also has a diagnostic speed that surpasses that of a human: it only takes a few seconds for each case!
X-rays are also a non-invasive and easy-to-implement diagnostic tool, so an accurate diagnosis from scratch through this imaging step will greatly reduce the need for invasive tests. unnecessarily.
According to statistics in the US, 1 in 8 women will face this disease at least once in their life.