The World of AI and Reproductive Health

If you consume any media that leans toward sci-fi or fantasy, you will encounter a world that employs technology bordering on magic. For many of us, the notion of using devices capable of instantly changing our lives might seem distant. Yet, upon examining the smartphone you use daily, you begin to grasp that the future has arrived—albeit not in the highly stylized manner depicted in media. This brings us to the realm of artificial intelligence (AI). Before your mind conjures images of “Skynet,” let’s focus on how this burgeoning technology impacts the sphere of reproductive health.
Currently, two prevailing conversations revolve around AI: how it can simplify our lives and how it could potentially lead to self-aware machines bringing about the apocalypse. Therefore, AI’s potential influence on reproductive health hinges on its capacity to learn. Much of the excitement, and at times trepidation, surrounding AI centers on an advanced method known as “deep learning.” This technique often seems ripped from the pages of a comic book due to its ability to grant a computer system the aptitude to learn by predicting outcomes based on provided data.
The implications of applying these deep learning algorithms to healthcare are profound. Mayo Clinic employed AI to analyze electrocardiograms—a common test gauging the electrical activity of the heart—to successfully detect asymptomatic left ventricular dysfunction, a precursor to heart failure. Now, let’s consider reproductive health. A few years back, a research team led by the National Institutes of Health and Global Good devised an algorithm capable of scrutinizing images of an individual’s cervix, correctly identifying precancerous changes necessitating medical attention.
The potential impact stemming from these foundational studies, and the numerous others in existence, could revolutionize approaches to reproductive health. The deep learning capabilities of AI hold promise for aiding in IVF treatments. A study detailed by Weill Cornell Medicine revealed that an AI system, when provided with thousands of embryo images alongside information about which ones led to successful births, achieved a 70% success rate in predicting the viability of embryos from new images. This not only heightens the likelihood of success for aspiring parents but also paves the way for cost reduction by minimizing the need for repeated IVF cycles.
However, it’s prudent to address apprehensions associated with AI, particularly considering the extensive landscape of healthcare. Historical biases within healthcare—pertaining to women, people of color, queer, non-binary, and transgender individuals—have persisted. These biases infiltrate AI systems, as these systems ingest data from pre-existing sources tainted with bias. Thus, these AI systems inadvertently perpetuate such biases from their inception.
Our world is perpetually expanding in numerous dimensions, technology included. The prospects of enhancing the lives of families grappling with infertility are tangible. The potential to overhaul the healthcare landscape exists. Yet, like any tool, we must question who wields control, particularly regarding the data fed into these systems and its origins. Given our world’s existing injustices and unaddressed biases, exercising caution—though potentially gradual—might be the optimal course for a world still in the throes of self-discovery.