In the early 2000s1, the US Food and Drug Administration withdrew ten drugs from the market because eight out of these ten drugs caused severe side effects among women, more than they did in men.

All ten drugs had been approved based on clinical trial data that predominantly included male participants.

This problem is probably not limited to just these ten drugs or medications in general. It is also present in how diagnosis and treatment are implemented for women, since systems have been developed and studied for the male default.

Fast-forward to today, and we have AI-powered tools for tracking and diagnostics. These include wearable sensors, menstrual trackers, and predictive risk models. The foundational problem, where the male body is a default standard, remains.

Herein comes the requirement of understanding the nuances that go into ethical design, especially for women’s health. As we move toward a more technological approach, while handing over some of the responsibility to the women themselves, we need to ensure the devices dispensing their health information, diagnostics, and treatment are built to understand how diverse women’s bodies function.

What is ethical healthtech design?

Ethical health tech design is a commitment toward building tools that serve ALL the people they are meant to serve. This is without causing harm through exclusion, assumption, or commercial self-interest.

This design draws from the principles that are already embedded within bioethics: autonomy, beneficence, non-maleficence, and justice. In this specific context, the principles are applied to the specific conditions of the digital health ecosystem.

This doesn’t mean the ecosystem starts from neutral. Instead, every product decision made starts from the recruited user, from whom testing and data collection influence how results are displayed. Based on the basic assumptions of the user. When those assumptions are built without women, the product eventually reflects the gaps, resulting in a clinically incomplete tool.

Ethical design should also be honest about what a product cannot do. A period tracking app claiming to predict ovulation, without highlighting its accuracy rate from preliminary user studies, is inadequate. Similarly, an app that checks symptoms for various conditions, without having been tested on women participants, is operationally incomplete.

In short, it is misleading.

Users, in the long term, adopt tools based on how the initial build and iterations thereafter have addressed the inadequacies and improved to include everyone’s norm. In women’s health, those decisions can carry real stakes.

Ethical healthtech design asks designers, developers, and product leads to interrogate their assumptions at every stage of the build process, not just for a diversity review.

Addressing the bias in healthtech design

A 2022 report2 in Contemporary Clinical Trials analyzed over 1,400 clinical trials involving more than 300,000 participants. It was recorded that less than half, 41.2%, of these participants were female. This is when women make up half of the global population. There are certain health conditions, such as autoimmune disease, and certain mental health conditions that affect women disproportionately compared to men.

In such cases, having an inbuilt bias shapes what we know about disease, treatment responses, and diagnostics. So when AI systems are trained on this data, the bias gets automated.

Let’s take the example of the pulse oximeter. This is a device that sits on the fingertip and measures blood oxygen. Its use was well known during COVID-19. Research3 in Patterns found that pulse oximeters were likely to overestimate oxygen saturation for Black patients during COVID-19. This means a lot of the patients who needed care were likely being sent home.

Why? These devices were predominantly calibrated for lighter-skinned patients.

Similar findings have been observed within women’s health as well. An npj Digital Medicine report4 highlighting the fairness of machine learning algorithms designed to diagnose bacterial vaginosis in asymptomatic women showed diagnostic bias across ethnic groups. The test performed best and was close to accuracy for white women and lowest for Asian women.

Another well-known example is cardiovascular disease in women. For years, women presenting with heart attack symptoms were less likely to be diagnosed correctly. This, in part, was happening because cardiovascular disease was largely believed to be affecting males more than it did females. However, the reality is that symptoms often look very different for women compared to men.

Additionally, biologically, following menopause, the cardioprotective hormone, estrogen, is reduced significantly. This then accounts for an increase in cardiovascular risk5 for women, which can be equal to that of men. Despite this fact, women remain dismissed.

From these examples, it is clear that algorithmic bias would not be just a technical error. It would result in a downstream consequence of who was asked to participate in the research, who was included in training datasets, and who was represented on the teams building the products. Fixing it requires intervening at all three levels.

How can we include privacy and consent in data ethics?

Without actually specifying it, women’s health data is highly sensitive. Menstrual cycles, fertility history, pregnancies, hormonal conditions, infection history, chronic illnesses, and sexual health information are just among the few personal details present within women’s health history.

Currently, the infrastructure that is being built to collect this data is not necessarily designed to prioritize the user’s interests.

One scoping review6, which reviewed 23 of the most popular mobile health (mHealth) apps for women, observed that most of them were collecting personal data. These apps were observed to have poor data privacy, sharing, and security standards. Many apps have policies written in language that is neither accessible nor transparent. Users are agreeing to terms they cannot reasonably be expected to understand.

Additionally, 87% of the apps were found to be sharing the data with third parties, most likely without the consent or knowledge of the user.

These stakes became more apparent after 2022, when the US Supreme Court overturned Roe v. Wade. Research7 into the privacy policies of mHealth apps being used for period tracking showed that they were being operated on opaque commercial models. Cycle and reproductive data that users had shared voluntarily had suddenly carried potential legal implications in certain US states. The initial design had not anticipated this.

An infamous example of third-party involvement in data sharing was the Federal Trade Commission's 2021 case8 against the largest period tracking app, Flo. Data was shared with Facebook, Google, and others. This was not something women would have openly consented to. But many, as is often the case, do not know that it is happening.

The reason behind this is that ethical design principles are an afterthought in the development process. Something stuck to say, “I told you so!”

Genuine data ethics within healthtech design requires more than a compliant and comprehensive privacy policy. The initial step is the requirement of transparency. The next one is of comprehension.

Both should include, in simple terms, how data is collected, who is going to see or use it, what happens to your data over time, and, more so, if the user has options to delete parts of the information, including their account, if they do not wish to continue anymore.

This means building to protect the user as a default, not the platform. For women's health specifically, this means understanding that the same data point can mean very different things depending on legal context, cultural background, and personal circumstances.

How can women contribute to ethical HealthTech design?

The simple answer is through representation.

Women make up approximately 70% of the health workforce9 globally. However, they hold only 25% of the senior roles, the decision-making roles.

In a survey10 of over 400 health technology professionals, women were found to be significantly under-represented within senior roles. Additionally, women who made it there took less time off over the course of their professional careers. Women, as a majority, experienced gender discrimination working within health technology, a field known to be male-dominated.

A BMJ paper analyzing the impact of leadership and culture change for the advancement of women’s health innovation highlighted that the way women participated in teams also mattered. Not just being a part of leadership or even the team in general. Women with lived experiences and from different ethnic backgrounds can essentially drive innovation down paths that would ideally not be considered within teams of similar backgrounds.

This is what’s known as participatory design. This approach aims to involve the end users, such as patients, caregivers, and community members, within the design process itself. As active contributors, they can assess how the product is scoped, built, and tested. Research has shown that culturally responsive menstrual health design requires input from specific communities to produce tools for a specific cultural context. The design should not and ideally cannot presume what the user knows, wants, or needs.

Women should also be contributing as researchers, clinicians, regulators, and advocates, the people who are shaping the evidence-based future technology that is going to be built on. Women are currently undercited in most of this research. An editorial published in Frontiers highlighted that women accounted for only 35% of authors, even though they made up just under half of the workforce in those teams.

Finally, women and their inputs can also be sourced during critical phases of the innovation process. An example is when a health tech product fails, the users who surface the failure are women for whom the product is proving insufficient. These are often women with chronic conditions, reproductive health needs, multiple health conditions, or intersecting identities. Ethical healthtech design, in essence, ensures no one is left behind.

The bottom line

Ethical healthtech design for women is an ongoing commitment toward involving women in the design process. It is asking women how products can serve them. This should ideally start early during the data collection, algorithm training, and product design stages.

To ensure healthtech tools are providing for vast demographics, diverse teams representative of end-user demographics are a functional requirement. Additionally, policies need to be in place to ensure women understand the terms of use of the data they are providing.

Ethical design also incorporates clinical validation that includes the populations the product is intended to serve. Additionally, it requires a willingness to slow down to answer the 'Who is this designed for?' when someone feels they are not fitting the intended purpose.

The FemTech market is estimated to reach $207 billion by 2033. This is a significant commercial opportunity, as well as a huge responsibility. The same tools currently being designed to close gaps in women's health, catch diagnoses earlier, support reproductive decision-making, and monitor chronic conditions more accurately can also, conversely, widen those gaps if they are built without the people they are meant to serve.

In the real world, ethical design realistically will not get it right every time. However, the intention should be to be honest when you get something wrong and rework the structures to ensure the mistakes do not happen again.

References

1 Most Drugs Withdrawn in Recent Years Had Greater Health Risks for Women.
2 Enrollment of female participants in United States drug and device phase 1–3 clinical trials between 2016 and 2019.
3 Addressing bias in big data and AI for health care: A call for open science.
4 Ethnic disparity in diagnosing asymptomatic bacterial vaginosis using machine learning.
5 Menopause and women’s cardiovascular health: is it really an obvious relationship?
6 Privacy, Data Sharing, and Data Security Policies of Women’s mHealth Apps: Scoping Review and Content Analysis.
7 Exploring Privacy Practices of Female mHealth Apps in aPost-Roe World.
8 FTC Finalizes Order with Flo Health, a Fertility-Tracking App that Shared Sensitive Health Data with Facebook, Google, and Others.
9 Global perspectives to enhance strategies for advancing women in healthcare and STEMM leadership.
10 Analysis of Gender Perceptions in Health Technology: A Call to Action.