Maybe once a week, an infertile couple will sit in front of me along with 500 pages of medical records and a pile CDs of all that they’ve been through. I know that there are valuable trends and relevant observations to be discerned from this tome of records, but who has the hours it takes to sift through them? AI could do this in seconds. And it could tell me about subtle trends, make diagnoses and suggest rational treatments. By organizing, documenting and analyzing information it would reduce my cognitive load and allow me to think more expansively, as is my want. Even more importantly, and as subtle as a passing glance, AI can expose my cognitive biases, as the brain’s tendency is to process information through the often-narrow filter of personal experience and preferences. Yes, machine learning could serve as a second set of eyes on patients. AI, you had me at hello!
E-Skilled Solutionist
The potential for AI to help us with diagnosing the causes of male infertility is massive. AI-assisted imaging of the semen analysis can reduce human error to unprecedented low levels. It can “see” things we miss just because we’re not-very-methodical beings. Its brainy, complex, algorithmic approach can analyze and describe relevant patterns of sperm movement and shape that we can’t always or reliably discern. It can also go “deep” with sperm and help us to understand the true meaning behind sperm DNA fragmentation, genetics, epigenetics, metabolomics and transcriptomics that is taking us forever to understand.
Even more exciting is how AI can amass large amounts of health information, including lifestyle, diet, activity, age, medical history, physical exam, and semen and hormone findings to robustly predict an individual’s true fertility potential. In doing this, machine learning could reveal to us the rank order of impact of each variable and its effects on fertility. For example, it would be nice to have more help in deciding which patients should have their varicoceles surgically repaired to ensure a positive response.
E-Curing Man
I’ve always said that the best way to be fertile is to live a healthy life. When surrounded by a healthy body, sperm will (generally) run hard and fast. To this end, lifestyle modification is a primary therapeutic target for improving fertility. Like having a bespoke suit made for you, AI-driven apps can provide personalized, evidence-based recommendations on a host of lifestyle issues such as diet, exercise and sleep, all of which could improve sperm health and fertility. And don’t forget AI-infused wearable devices, which can track and monitor adherence to prescribed treatment plans and encourage patients to stay on point and stay committed to treatment goals. So, I see AI as a key element to improving patient engagement in ways that are currently not possible. Take obesity for example. It is the elephant in the room as a major, treatable cause of male infertility. AI could provide that critical level of constant feedback and support to help patients get “over the hump” and on their way to becoming healthier souls.
How AI can help in the IVF laboratory is already quite clear to many of us in the field. Assistance with sperm selection for in vitro procedures using automated microfluidics and imaging could identify the “healthiest” sperm for ICSI. Realtime monitoring of embryo culture conditions can ensure consistency in the lab and improve embryo quality. Machine learning-assisted imaging of developing embryos could improve embryo selection for transfer and improve pregnancy rates. And mechanically, AI-assisted robotics could conceivably perform gamete manipulation more consistently and with fewer errors than we humans can. Consider how robotic prostatectomies for prostate cancer have revolutionized urological care over the past two decades!
E-Forecasting Outcomes
Even more broadly and at the start of the infertility journey, by evaluating the clinical characteristics of each partner, AI can suggest personalized treatment plans that direct us to which level of assisted reproduction (e.g., timed intercourse, intrauterine insemination or IVF) might work best for them. Who should skip IUI and proceed to IVF in cases of unexplained infertility? In addition, it could assist clinicians in choosing the best drugs, doses and timing of hormonal stimulation regimens used in IVF to optimize egg quality and numbers and reduce side effects, an issue especially relevant in older couples.
Finding new drugs and treatments in the world of infertility is a slow, arduous and pretty random process that largely depends on animal experiments and basic science research. Through its big data analyzing capacity that includes all aspects of science and based on a deeper understanding of genetic or biochemical pathways, AI can think “out of the box” to identify new and novel agents that may improve fertility. Also, by modeling how these treatments interact with our physiology, AI can potentially expedite drug development. This is the ultimate goal of precision medicine.
So, there is much room for AI to grow in the field of infertility. Rest assured, though, that it will not replace us humans in medicine. AI is based entirely on historical training sets and is entirely iterative in nature; it learns from the past. It is agnostic and amoral and lacks the uniquely human qualities of creativity and emotion. And these qualities, my friends, are the true foundation of medicine. In the words of the venerated Sir William Osler: “The practice of medicine is an art, not a trade; a calling, not a business; a calling in which your heart will be exercised equally with your head.”