It's International Women's Day and, to be honest, there isn't much to celebrate if you're a woman. Though it's a fundamental human right, gender equality is often ignored all over the world; violence against women is rife and women also suffer harm and abuses online on a daily basis. And let’s not talk about women's chronic pain, rarely addressed by doctors and most often undertreated and dismissed or attributed to vague and mysterious psychological causes, while menopause remains for many doctors an uncharted territory.
Yet, in all this bleakness there may be some great news for what regards healthcare where women may find an unlikely ally in future – Artificial Intelligence (AI).
As early as 2017, the Imperial College in London led a consortium of breast cancer experts, clinicians and academics to explore whether AI could help detect and diagnose breast cancers more effectively.
The most common cancer in women worldwide, breast cancer has seen some great medical advances in the last ten years or so, but, unfortunately, it can still kill.
Used by clinicians to identify cancers early, mammograms (breast screenings) are not perfect, and can lead to an incorrect diagnosis, to false negative and false positive results, as cancer can be hidden or masked in mammograms by overlapping dense breast tissue. Screenings can help picking up cancer early when it is still curable, but many types of cancers are missed or detected at a late stage when they are more difficult to treat.
The ability to detect breast cancers can be improved training Artificial Intelligence (AI) to learn from databases of mammograms collected from clinics and academic institutions. In the last few years there have been experimental studies using digital images stripped of any information: they are used to train computer algorithms that analyse these images and learn to spot a cancerous growth by its shape, density and location. After learning to recognise these cases, the system creates a mathematical representation of normal mammograms and of mammograms presenting cancers. Comparing the data, the system can then track down abnormalities in a mammogram, spot signs of cancerous tissue and alert radiologists, while refining its accuracy of detection at the same time.
In 2020 radiologists found that an Artificial intelligence software developed by Google Health, was able to outperform detecting cancers that specialists had missed. The results of these studies were encouraging with scientists training Google's AI program with mammograms from more than 76,000 women in the UK and 15,000 women in the US. The mammograms were analysed in three different ways before combing the results to produce an overall risk score. The results were encouraging: the AI programme reduced of 5.7% and 1.2% (USA and UK) the false positives and of 9.4% and 2.7% the false negatives, proving it could have boosted the quality of breast cancer screening.
Healthcare facilities all over the world and in particular in the United States and the European Union are also beginning to test or provide data to help develop the systems. In Europe, Great Britain and Finland are working on adding this technology to their screenings, while Hungary has developed research projects that employ AI software to spot cancer. As reported last week by The New York Times, Bács-Kiskun County Hospital, outside Budapest, is one of the five hospitals or clinics performing breast cancer scans employing the help of AI programs.
There have been cases in which two radiologists failed to detect breast cancer, while the Artificial Intelligence software flagged the screening as potentially cancerous, detecting tiny lesions and spotting the insurgence of a malignancy. AI software can recognise also more difficult forms of breast cancer and produces accurate results on women of all ages, ethnicities and body types.
While AI software can be used to detect other forms of cancer, maybe one day it will also be used for other conditions (could AI software help with autoimmune diseases that are notoriously difficult to diagnose?) or will it be able to answer women's questions regarding pain syndromes dismissed by doctors?
Time will tell, but more clinical trials and more training will be needed and it will take time before AI tools get implemented on a vast scale in all sorts of hospitals, especially considering that, subject to all sorts of cuts, not all healthcare facilities may be able to afford such systems. So, for the time being, don't expect to go to your usual clinic, ask if the mammography you have to do is powered by AI and assume you will get a beaming smile and an encouraging "Yes, of course!" Finding major partners from the high-tech industries may be a solution for those healthcare facilities who desire to implement their screenings with AI software, but, who knows, maybe even big fashion groups could sponsor such AI programs one day? After all, the fashion industry often preaches a form of faux feminism, repeating the "empowering women" mantra, but essentially relegating women to the role of consumers.
Like AI text-to-image applications and Chat GPT, systems that scare those critics who are afraid of machines replacing human jobs, also AI tools employed in the medical field received some harsh criticism, as some fear they may replace radiologists one day. Yet, while the use of AI software in the detection of breast (and other forms of) cancer (as well) and in medical imaging in general, is encouraging (and it may be an asset considering there is a shortfall in radiologists in many countries, often due to older radiologists retiring faster than new ones join), AI tools must be researched further, even though so far they have shown an impressive ability in their diagnostic performances.
That said, while these tools do improve the diagnostic accuracy, they produce the best results when used in partnership with trained doctors. Using such systems can cut radiologists' workloads by at least 30 percent, but the key is always to have an AI-plus-doctor rather than just a doctor alone. Looks like the possibilities of AI may be endless, and one day, from dreaded villain, AI may turn into a great tool and a reliable partner as well in a variety of industries and disciplines.
Comments