This month, Google researchers ” made the headlines with the results of a study that claims their artificial intelligence system may be better than doctors at detecting breast cancer from mammograms “. But while artificial intelligence could likely benefit the healthcare industry, these benefits ” will depend on how it will be used “. If the doctors ask the wrong questions, it will turn out to be ” a fiasco “, And” could even amplify the mistakes made “.
On the subject of breast cancer, for example, screening has been controversial for decades: “ by subjecting asymptomatic people to mammograms, you will find many things that look like cancer but never would have threatened anyone’s life “. And the problem is that at the time of screening ” it is almost impossible to know whether a given injury will become dangerous or harmless “. The consequence: in practice, doctors tend to treat all the cancers discovered. And the question of whether mammograms actually save lives remains highly controversial. Some studies even believe that ” they are generally bad for the health of the patients, because of the excessive treatments and tumors created by the radiations “.
Artificial intelligence systems like Google’s can worsen excess screening, diagnosis and treatment. And, ” it’s not even clear that they can actually improve the false positive and false negative rates “. Indeed, the performance of the Google system has been compared to that of radiologists ” who were not specifically trained to examine mammogramss. “
Beyond breast cancer, artificial intelligence is designed to be able to take into account variables that were not considered important. Maybe because they weren’t. And nevertheless be used to formulate new diagnoses. Two fundamental questions must be asked: ” what is the problem that technology is trying to address and to what extent will it improve the future of patients? “. Answering these questions may take time. For Vinay Prasad, hematologist and oncologist at Oregon Health & Science University School of Medicine, ” the Silicon Valley mindset can be dangerous for practitioners “. He explains : ” It’s this attitude – when lives are at stake, you have to implement new ideas promising as quickly as possible- who put us in this bazaar of breast cancer screening ”, and specifies: ” mammography was adopted before there was evidence and when medical practice became standard it was very difficult to reverse “. And to regret: ” In a culture accustomed to immediacy and excessive talk, it is difficult to be humble and patient “.
For further :