Published in


Artificial Intelligence Is Pioneering The Next Wave of Women’s Health.


Men are from Mars, and Women are from Venus.

In one of the most famous books of the past few decades, Men are from Mars, Women are from Venus, John Gray utilized this metaphor to describe the various psychological differences between those who identify as male, and those who identify as female. In doing so, Gray’s book provided insight in order to help partners improve their relationships.

However, psychological differences aren’t the only ones that persist.

When we think of individuals who identify as male, and then those who identify as female, we automatically think of differences within physical features or cognitive processes.

However, we seldom think of the health outcomes and inequities between both sexes.

The health of an individual is heavily reliant on several key factors, those being the social determinants of health.

However, gender plays a role in how women experience the social determinants of health.

From gender discrimination in healthcare to a global discrepancy in education, the unspoken variables of women’s health affect women in ways we are only just beginning to quantify.

But even biologically, women are much more susceptible to a certain range of various conditions, sickness, and disease.

From autoimmune disease to dementia, and breast cancer to reproductive disorders, our healthcare system is not optimized for such differences.

However, the rise of artificial intelligence within healthcare is allowing women to have faster diagnoses, more personalized resources, and a wide range of treatment options.

Artificial Intelligence isn’t science fiction.

Imagine this.

It’s an average Wednesday morning. Nothing out of the ordinary. You open your eyes, reach over, pick up your phone, and unlock it using your facial identification. The clock reads 7:16 am.

You notice you have some notifications so you click on Instagram, and before you realize it, over 30 minutes have passed. 7:48 am.

As wake up and get dressed, you click on Spotify and turn on your Discover Weekly playlist, closing the ads that show up as you start to play your music. But soon, you realize you’re running late. It’s 8:07 am.

As you rush to get yourself into the car and navigate to your office, you frantically type the directions into Google Maps, trying to navigate which route has the least amount of traffic and will get you there the quickest. 8:18 am.

On any given day, even on just a regular Wednesday morning, nearly each and every single one of us interacts with some form of artificial intelligence.

In just a single morning, you’ve already interacted with at least 5 different forms of artificial intelligence, not to mention the AI you likely interacted with when you were typing a message to your coworker, or placing an order on Amazon.

When we hear the term “Artificial Intelligence,” many of us tend to think of science fiction novels or even technologies that aren’t prevalent in our daily lives, from robots to self-driving cars.

But in reality, Artificial Intelligence is everywhere we see. It’s part of nearly every sector of society in some form, and is a part of our daily lives, although we seldom acknowledge it.

But what really is artificial intelligence?


Although many definitions of the term “artificial intelligence” have surfaced over the past few decades, in essence, it is the theory and development of computer systems that are able to perform tasks that normally require human intelligence. These include visual perception, speech recognition, and decision-making.

Artificial intelligence calls for a culmination of computer science and data in order to solve problems and encompasses the sub-fields of machine learning and deep learning.

Although the terms Artificial Intelligence, Machine Learning, Deep Learning, and Neural Networks are commonly used with little differentiation, it is crucial to note the differences between these various subsets of artificial intelligence.

Machine Learning

Machine Learning is a branch of artificial intelligence that uses data and algorithms to imitate the manner in which humans learn, gradually improving its accuracy.

Through statistical methods, the algorithms are trained to make classifications or predictions and have proven useful in a wide variety of applications, from transportation to marketing to healthcare.

From building self-driving cars to drug discovery, machine learning is leading the next wave of innovation.


The first step in making a machine learning model requires defining the objective of the model, through which the relevant type of data can be established and gathered.

Next, the data is prepared by getting rid of inconsistencies within the dataset and performing an exploratory data analysis. This requires removing duplicates, correcting errors, and dealing with missing values. In addition, the data must be randomized in order to erase the effects of the particular order in which the data was collected and/or prepared. The data must also be split into training and evaluation sets.

The exploratory data analysis requires understanding the patterns and trends in the data. All useful insights about the data and correlations within it are understood during this stage, making it very important in the machine learning process.

The third step involves choosing and training a model. The model is trained in order to answer a question or make a prediction as often as possible, increasing its accuracy. The data gathered is split into training and testing data to train the algorithm.

Afterwards, the model is evaluated by using some metric or combination of metrics to “measure” the objective performance of the model. This tests the model against previously unseen data. The unseen data is meant to be somewhat representative of model performance in the real world but still helps tune the model. On the other hand, the test data does not help tune the data.


The fifth step is hyperparameter tuning, which requires tuning the model parameters for improved performance. These model hyperparameters may include the number of training steps, the learning rate, and the initialization values.

The last step of building the machine learning model is to make predictions. Using further data from the testing set which have, up until now, been withheld from the model, are used to test the model in order for a better approximation of how the model will perform in the real world.

However, there are multiple different types of machine learning models. These include supervised and unsupervised learning.

Supervised learning uses labeled datasets to train algorithms to classify data or accurately predict outcomes. As the input data is fed into the model, it adjusts its weights until the model has been fitted appropriately, in order to avoid overfitting or underfitting, where the data performs poorly on the training data or differently on the training data than it does the evaluation data. Supervised learning is used for large-scale, real-world applications, such as categorizing spam emails.


On the other hand, unsupervised learning uses machine learning algorithms to analyze and cluster unlabeled datasets. They are able to discover patterns or data groupings without human intervention and sort the data based on similarity. Unsupervised learning is often used for exploratory data analysis, image and pattern recognition, and more. It can also be used to reduce the number of features in a given model. A real-world example of unsupervised learning is facial recognition.

Although often overshadowed, semi-supervised learning offers a medium between supervised and unsupervised learning. During training, it uses a smaller labeled data set for classification and feature extraction from a larger, and unlabeled data set. This type of machine learning is useful when there is not enough labeled data to train a supervised algorithm.

Another main category within machine learning is reinforcement learning.


Reinforcement learning is a behavioral machine learning model. It is similar to supervised learning, but instead of being trained using sample data, the model learns through trial and error. The algorithm learns to make a sequence of decisions, in a “game-like” environment.

To get the algorithm to come up with the best outcome, the model gets either rewards or penalties for the actions it performs. Its goal is to maximize the total reward. It is completely up to the model to figure out how to perform the task in order to maximize the reward, through trial and error. A real-world example of reinforcement learning is within the gaming industry, where the algorithm figures out how to achieve the best possible outcome within the game.

Deep Learning

Deep learning is a subset of machine learning and is essentially a neural network with three or more layers. The neural networks attempt to simulate the behavior of the human brain by allowing it to learn from large amounts of data. While single-layered neural networks still make approximate predictions, the additional hidden layers used within deep learning helps to optimize and refine for more accurate outcomes.

Deep learning is different from machine learning in terms of the type of data it works with and the way in which it learns. Machine learning algorithms are structured and utilize the data to make predictions. This means that specific features are defined from the input data for the model and organized into tables. Even if a machine learning model uses unstructured data, it pre-processes it in order to organize it into a structured format.

On the other hand, deep learning eliminates some of the pre-processing involved with machine learning. The algorithms can process unstructured data, such as text and images, and automates feature extraction.


For example, if we are trying to categorize a set of photos of different pets, the deep learning algorithm can determine which features are important in distinguishing one animal from another, such as their ears. Then, the deep learning algorithm adjusts and fits itself for accuracy, and makes predictions about a new photo of an animal with increased accuracy. However, in machine learning, the importance of the features within the photographs is established manually by an individual.

Deep learning is an integral part of many AI applications and services, especially those improving automation and performing tasks without human intervention, such as digital assistants and self-driving cars.

Neural Networks

In essence, neural networks reflect the behavior of the human brain, allowing computer programs to recognize patterns and solve common problems in the fields of artificial intelligence, machine learning, and deep learning.

They can predict anything, from the probability of someone getting a bacterial infection on their next hospital visit to the next chapter of a book.

Neural networks can be powerful tools, as they allow us to classify and cluster data very quickly. Instead of taking hours, speech and image recognition can take mere seconds or minutes, as in Google’s image search algorithm.

The name and structure of neural networks are inspired by the human brain, as they mimic the way in which neurons in the brain communicate with and signal to one another.

Neural networks are comprised of differed node layers, each of which contains an input layer, one or more hidden layers, and an output layer.

Each node connects to another and has an associated weight and threshold. If the output of any individual node is above the specified threshold value, that node is activated. This sends data to the next layer of the network.

Neural networks rely on training data in order to learn and improve their accuracy over time. The iterative training process finds parameters, or model weights, that result in a minimum error or loss when evaluating the examples in the dataset.

The four main components of a neural network are the inputs, weights, a bias or threshold, and an output.


After the inputs are taken in, and the weights are added to determine importance, the threshold value is considered and applied to the formula. Then, it can be plugged into an activation function in order to receive the corresponding output.

This specific process is repeated multiple times in order to receive the final output, due to the hidden layers of the function. Each hidden layer has its own activation function, giving it the potential to pass information from the previous layer into the next one. After all of the outputs from the hidden layers are generated, they are used as inputs to calculate the final output.

There are many different types of neural networks, from convolutional neural networks, consisting of a three-dimensional arrangement of neurons used in image processing, computer vision, and speech recognition, to recurrent neural networks, in which the output of a particular layer is fed back into the input in order to predict the output of the layer, and are used in applications including text processing and sentiment analysis.


Now that we’ve gotten an in-depth understanding of what artificial intelligence is, and its prevalence in our daily lives, we can begin to understand where it intersects with healthcare, in turn, pioneering the next wave of women’s health.

AI proposes a wide range of possibilities in which it can revolutionize the healthcare sector. From diagnosis and treatment to patient engagement and adherence, from administrative applications to drug discovery, the possibilities seem limitless.

In particular, AI proposes a dramatic change to women’s health, as women are biologically more susceptible to a certain range of various conditions, sickness, and disease; from autoimmune disease to Alzheimer’s, and breast cancer to reproductive disorders.

By leveraging AI, we have the potential to develop more advanced diagnostic and treatment options for women with some of these conditions.

Artificial intelligence might just be the key to revolutionizing the way millions of women are diagnosed and treated.

Researchers and innovators all over the world have found ways to apply artificial intelligence to better serve women in the healthcare system.

From researchers at MIT, creating deep learning algorithms to read mammograms as well as doctors, reducing the number of false negatives and positives, to researchers at the NIH and Global Good who have developed an algorithm to analyze digital images of a woman’s cervix and identify precancerous changes.

However, AI can be applied outside of cancer diagnosis. Researchers have found ways to use AI to improve the diagnosis of PCOS, or polycystic ovarian syndrome, through a machine learning model that can analyze ultrasound images, to developing ML models to predict lupus disease activity, an autoimmune disease in which 9 out of 10 patients are women between the ages of 15 and 44.

Even emerging companies are revolutionizing the way we approach diseases and conditions that affect women.

Take Whiterabbit, a startup founded in 2017, working to redefine the breast cancer screening and diagnosis experience, in order to improve patient outcomes. The company has developed a line of products within the mammography space, enabling early and accurate breast cancer detection, with the goal of eradicating late-stage breast cancers in the U.S. by 2025.


Whiterabbit’s WRDensity software received FDA 510(k) clearance in October of 2020. The software provides objective data to help radiologists more quickly identify the level of breast density, which is an important risk factor for developing breast cancer and requires greater attention by a radiologist because it is more difficult to assess.

Whiterabbit has also developed a tool called ACT, an AI-powered service that makes it easier for patients to schedule breast cancer screenings, and also sends reminders, and educates women about breast health to make more informed decisions. It has already been implemented in over 400 clinics in the United States.

Even companies whose solutions are not completely based on artificial intelligence, have adopted AI in order to better serve women, especially in the realm of fertility.

One of these companies is Inne, which has developed an at-home fertility monitoring system, using saliva. The Inne minilab consists of a strip and reader; the strip collects the individual’s saliva, and the reader uses advanced image processing analysis to rapidly capture several images. The camera-based technology measures the speed of the saliva flow and captures the hormone measurement of the individual on the given day.

After the saliva is collected and the hormonal levels are measured, the device, connected to an app on the user’s device, informs the user of their fertile window, their day of ovulation, and their progesterone trend, allowing users to better understand their bodies and fertility.

In just one milliliter of saliva, the Inne minilab is able to detect the change of 0.0000000001 grams of progesterone. Although the concentrations are extremely small, these minuscule changes can trigger substantial physiological changes, including ovulation.

In order to provide accurate insights to the user, Inne has developed an algorithm that takes hormonal variability into account and defines a hormonal baseline that is unique to the user. From there, the app can accurately predict the user’s fertility, hormonal fluctuations, and even help the user understand menstrual cycle symptoms just from their saliva.

A $1.1 billion industry.

Despite the rise of research and a multitude of emerging companies that are leveraging AI for femtech and women’s health, there is still an immense amount of work to be done in leveraging AI to better serve women.

Although many of the current applications of Artificial Intelligence in women’s health are targeted towards diagnosis and detection, AI also has the potential to be used in treatment, especially in drug discovery and developing solutions for women with reproductive, autoimmune, and certain neurological disorders.

The next wave of women’s health and femtech startups are on the rise, In 2019, the femtech industry generated $820.6 million in global revenue; by 2024, the industry is expected to reach $1.1 billion.

As the industry grows, we can expect to see the rise of the integration between technology, namely artificial intelligence, and women’s health, as researchers, scientists, and entrepreneurs contribute to research and develop products and resources to serve a larger scale of women.

The next wave of women’s health, powered by artificial intelligence, has already begun.


Thank you very much for giving this a read! If you learned something from this article, please share it! Be sure to connect with me, and leave this article a clap 👏 if you enjoyed it!




Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store