Female Physiology & The Role of AI in Women’s Health

The women’s health space is often neglected, as our current health care system treats men and women’s bodies as the same, despite the biological differences.

Until recently, most clinical & basic research was either done in male subjects or male and female subjects where studies didn’t differentiate the sexes in data analysis. This implies that there aren’t significant differences in the physiology of men and women outside the reproductive systems.

Women weren’t as included in trials or research due to the potentially complex nature of controlling the various stages of the female reproductive cycle and different hormonal balances.

These variables would have to be included in experimental design in animal and human studies involving female subjects in some studies. So basically, it was just thought of as a variable that would make data collection more complicated.

Present-day, there is a lot of literature addressing sex-based differences and the effects of sex steroid hormones on the function of multiple organ systems. Research indicates that there are many differences between males and females in normal physiology and the pathophysiology of disease. Unfortunately, these sex-based differences in physiology aren’t systematically addressed in textbooks or the medical physiology curriculum, excluding reproductive physiology.

What are the differences in Physiology?

Potential causes of sex-based differences in normal physiology and disease

• Genetic differences: females have 2 X chromosomes and no Y chromosome, whereas men have a Y chromosome but only one X chromosome
• There are genes on the Y chromosome which don’t have a counterpart located on the X chromosome.
• Some genes located on the X chromosome are more expressed in women than men.
• Sex steroid hormones such as androgens, estrogens, progestins can alter gene expression. Receptors for sex steroid hormones are in many tissues outside of the reproductive system- some include the heart, bone, skeletal muscle, vasculature, liver, immune, and brain. Circulating androgen levels are higher in males, and circulating estrogen/progestin levels are higher in premenopausal females.

Sex-based differences in normal physiology

• Males: typically more muscle mass, bone mass, and a lower % of body fat than women. These differences are primarily a result of gonadal steroid hormones on skeletal muscle and bone metabolism.
• Sex differences in lungs- men have larger lungs, wider airways, and higher lung diffusion capacity than women. consequence- compared to young men, the maximum exercise capacity may be limited by pulmonary capacity in women
• Known differences in brain structure result from fetal exposure to gonadal steroid hormones: varying pain threshold and cognitive style, and greater glucocorticoid response to stressors in females than males.
• Biological basis of these differences are highly complex and not fully understood

The actions of estrogens and androgens control any sex-based differences.

We can take osteoporosis, for example, where 80% of the patients are women. When we look at the causes of osteoporosis, we have genetic characteristics, exercise, dietary history, and the contributions of testosterone and estrogen to bone metabolism. The decline in estrogen production is one of the critical factors that predispose postmenopausal women to the development of osteoporosis.

Another example is an autoimmune disease with 80% of women patients, depicting that sex steroid hormones can significantly alter immune system function. In the cases of autoimmune disease- we can look at dramatic examples of the discrepancy between men and women.

Case Study: Heart disease

Heart disease is the leading cause of death for both men and women. However, due to biological differences, heart attacks in women progress and show up differently than in men. For the longest time, until very recently- women with heart disease have been diagnosed and treated the same way men have, down to the medication, tests, and overall medical procedure.

Men’s VS Women’s Heart Attacks:

Cholesterol in women builds up in different areas, and this build-up generally develops in the heart’s smallest blood vessels, known as the microvasculature. Men typically develop this plaque build-up in the largest arteries that supply blood to the heart. Thus, heart disease in both sexes is only partly related to the accumulation of cholesterol. Inflammation is also said to play an important role and may contribute to our differences in women with heart disease.

Different Symptoms:

Heart attacks don’t always look or feel the same in women compared to men. Men typically present with chest pressure. Women also present chest pressure as their leading pain. Still, they are also more likely to report nausea, sweating, vomiting, pain in the neck, jaw, throat, abdomen, back, shortness of breath, even without severe discomfort in the chest, which are atypical and much less common symptoms.

Women also may develop diseases that present, similar to a heart attack. Examples include a coronary spasm, coronary dissection, and takotsubo cardiomyopathy. But, again, the vagueness of this can be a potential place of error during diagnosis.

Men and women have different risk factors for heart attacks. However, based on their reproductive history, women may also have an increased risk for heart disease. For example, preeclampsia and gestational diabetes develop during pregnancy and can be a severe predictor of a heart attack. Another example is endometriosis; women under or at the age of 40 who suffer from the condition are 3x more likely to have chest pain, develop a heart attack, or need treatment for blocked arteries compared to women of the same age group without endometriosis.

Difference in Diagnosis

If someone were expected to have a heart attack, they would take a cTn (cardiac troponin test) which measures circulating levels of troponin (a protein released when there is damaged heart muscle). The problem is that providers are only starting to use sex-specific thresholds for some diagnostic tests. Some women could be having a heart attack but have fallen below the level of detection. Having sex-specific thresholds for high-sensitivity cardiac troponin I (hs-cTnI) assays resulted in the identification of 5x more heart attacks in women. A recent study tested this with patients suspected to have acute coronary syndrome (ACS). There were a lot of discrepancies that occurred when it came to treating myocardial infarction, which falls under ACS.

Diagnosis Pt 2: Cardiac catheterization is a diagnostic test consisting of the insertion of a catheter into a chamber or vessel of the heart. Problem: the test is searching for plaque build-up in large arteries. Women often experience blockages in the smallest arteries, so this test is not the most effective way to diagnose heart attacks in women.

Men and women might require different treatments. Doctors have years and years of experience treating cholesterol build-up of plaque in the heart’s largest vessels. On the contrary, there is less understanding surrounding the treatment of plaque in the small vessels (arterioles, capillaries, and venule) and the inflammation of the heart, which often occurs in women. Recently things are taking a turn. There are growing numbers of medical providers approaching treatments for heart attacks in men and women differently (whether it be ‘subtle calibrations in pacemakers or variations on angioplasty’).

There are many opportunities at the intersection of AI in women’s health.

What types of problems is AI best at solving?

• Find trends, patterns, and associations.
• Discover inefficiencies.
• Execute plans.
• Learn and become better.
• Predict future outcomes based on historical trends.
• Inform fact-based decisions.

Moreover, where is/can AI be applied to Women’s Health?

Innovations in the following areas:

• Menstruation and period care products
• Fertility and birth control
• Pelvic health
• Menopause
• Chronic conditions and hormonal disorder
• Sexual wellness
• Breastfeeding
• General healthcare
• Pregnancy and post-pregnancy

I’ll into an intersection I’m really interested in-artificial intelligence and fertility.

AI x Fertility

Infertility is a rising problem, with 10% of women in the US facing the problem and 1/6 couples unable to have biological children. Hence the use of assisted reproductive technology, like in-vitro fertilization (IVF). The steps of IVF are as follows:

1. Ovulation Induction
2. Egg Retrieval
3. Fertilization
4. Implantation

After fertilizing the egg, we can use AI to find the genetically healthy embryos that will most successfully implant in the woman’s uterus. This will cut down on costs of IVF while boosting the chances of having a genetically healthy child.

Enter Whisper Fertility- A company working on reducing cycles with improved embryo selection. Their AI helps choose an embryo that is most likely to lead to a healthy pregnancy. Using deep learning and computer vision, Life Whisperer identifies morphological features that constitute a healthy embryo, often invisible to the human eye. It uses a cloud-based AI system, trained on more than 20,000 globally sourced 2D embryo images.

How Does it Work?

Whisper Fertility uses deep learning, a type of machine learning which imitates the nature of how humans gain knowledge. Unlike traditional ML algorithms, which are linear, deep learning algorithms are stacked in a hierarchy of increasing complexity and abstraction.

The model allows the computer to recognize and classify images after training on previous images to self-learn and find patterns and differentiate features of each type of data. For example, it was trained on 9000 static 2D images of Day 5 blastocysts with related pregnancy and clinical location collected. After, the algorithm associates key morphological aspects of the photo with a genetically healthy or unhealthy embryo better than an embryologist could. However, there are some things that the human eyes can’t see.

The parameters assessed were distance and speed of pronuclear migration, blastocyst expanded diameter, inner cell mass area, and trophectoderm cell cycle length.

The Model

Images of blastocysts were divided into three groups- training, validation, and blind test data sets.

Blind test datasets are images from the same clinic used in training. However, they could also be from independent clinics, not a part of the training data (double-blind test data). The data is inputted into the model, and after training, the model was validated with images unused in training. To further determine accuracy, the model was tested on never-seen, blind test data sets.

The double-blind data set was from other geographical clinical locations and consisted of images unseen to the model. They tested with this data to ensure the AI is robust and can be used as a widespread tool to translate to various clinical and geographical locations.

Just as AI was used to solve a problem in fertility, the technology can be applied in an infinite number of other places.

The Future Of Femtech ♀️

In the past few years, the technology sector has been making significant progress in closing the gender gap- which is how the femtech industry developed.

I’m going to wrap this up by giving some shoutouts to celebrate amazing women who are paving the future of health tech.

Ida Tin, Co-founder of Clue

Ida Tin is the woman who coined the term femtech; an industry said to reach a valuation of 50 billion by 2025. She is the co-founder of Clue, a period and ovulation tracking app with more than 11 million users across the globe. The company provides PMS and cycle predictions (all backed up by the science, of course!) and draws in-depth insights from the data to fill women’s global need to understand their bodies better. In addition, Ida has been a massive part of creating a culture shift, where menstrual cycles are not something to be embarrassed by and instead embraced as a natural part of life. #girlbossing

Lea Von Bidder, Co-founder of Ava

Lea von Bidder is the co-founder of Ava, the world’s FIRST EVER fertility monitoring bracelet. Unlike other approaches, product users wear the Ava bracelet while they sleep- which is connected to an app on their smartphone. In the morning, they’re able to find out where exactly they are in their menstrual cycles. This product allows women to understand their mood swings, figure out if headaches are hormone-driven, find the right time to try for pregnancy, and can also be used as non-hormonal digital birth control.

Aagya Mathur, CEO of Aavia

Aagya Mathur found her calling at the intersection of health care, innovation, and analytics-driven insight. She is the co-founder and CEO of a hormone health company called Aavia, empowering women to understand their periods better, control the pain, and understand mental health trends associated with their cycle. Aavia also has a smart case for birth control users- which contains sensors tracking what time the pill was taken, reminding you until you take it, and then records the pill ingestion in the app. SO. INCREDIBLE!

Piraye Beim, Founder of Celmatix Inc.

Piraye Beim founded Celmatix after doing doctoral and postdoctoral work in personalized medicine and embryology research. Celmatix has decoded ovarian function and is working on creating novel therapeutics to treat chemotherapy-induced ovarian failure, Mosaic Turner syndrome, FMR-1 related primary ovarian insufficiency (POI), endometriosis, and PCOS using advanced machine learning and AI. Celmatix’s work will be REVOLUTIONARY for women’s health as 1 in 3 women have conditions rooted in poor ovarian health.

I celebrate these women who are doing incredible work that will pave the way to improve female health systems. 🙌🏽

Personally, my current focus is gaining deep technical knowledge to build AI models that can draw insights into women’s health- from fertility to birth control, autoimmune disorders, chronic illness, pregnancy, post-pregnancy, and so much more. Women’s biological systems are so magically intricate and complex. I feel genuine excitement towards a better global understanding of female physiology in the future.