We just passed a milestone in health AI that honestly, we all saw coming, but has major consequences if the data to support is left unexamined. OpenAI launched ChatGPT Health and Anthropic added medical toolsets for Claude, meaning users can sync their records and wearables, using their personal health data to get hyper-specific answers to their medical questions. At the same time, the FDA issued new guidance that effectively loosened the reins on wellness products like wearables, so AI can interpret markers like blood pressure and glucose with far less oversight.
As someone who has worked in diagnostics for over a decade, I have to question if these tools are just a high-tech way to continue giving women the wrong answers.
Our biggest healthcare problems aren’t only cost or access, but the breakdown in how diagnoses are made, especially for women. Digital health’s new era of AI and real-time data is the most promising solution I’ve seen to fix this, but for these tools to work for women, they must be built on data that actually includes them.
A system that never truly included women
The diagnostic gap is the result of decades of medical research that centered men as the default patient. Even today, women only represent about 30% of clinical research participants, meaning our diagnostic criteria, medical education, and datasets feeding these new models are severely underrepresenting the health needs of half of the population.
The consequences are predictable. Women are diagnosed an average of four years later than men across more than 700 diseases. Autoimmune diseases often go unrecognized for years. Early signs of heart disease, the number one killer of women, are dismissed because they don’t match male “textbook” symptoms. The result is more expensive care, invasive treatments, and worsening health outcomes. Women frequently cycle through multiple providers and repeated tests, carrying the emotional and financial strain of a system that wasn’t designed to detect their conditions early — if at all.
Ai could change the trajectory, but only with better data
Despite the system’s failings, I’m optimistic that AI’s ability to analyze vast, complex clinical data can give it the power to see what traditional systems miss: identifying gender-specific risk pathways and cutting through subjective symptom reporting.
We’re already seeing the benefits: AI-powered mammograms have reduced false positives by up to 25% while identifying cancers humans missed. But AI doesn’t cure bias on its own; it learns whatever bias it’s fed. With 90% of healthcare insights currently unstructured or unusable, health AI is only as reliable as the data behind it.
Data that reflect real people
As we move forward, we need to fix the data problem at its source. Consumer-initiated diagnostics like lab testing are moving us in the right direction. When consumers have more access to advanced testing, it captures data from the real world, not just from clinical trials where the default patient has been male seven times out of ten.
The FDA’s recent updated guidance is a step toward clarity for wellness device makers. But as self-collected data becomes a bigger piece of the healthcare puzzle, especially in training AI models, that data must be inclusive. Take hormone panels and chronic condition monitoring, for example: because they no longer depend on logistical hurdles that have often delayed women’s care, women can test earlier and more often. When this happens, AI can now get the inclusive data it needs to spot risk patterns before they escalate across all genders.
Putting clinical integrity before speed
The current pace of innovation demands speed, but it cannot sustain without sure data. This means we have to take a strategic pause to check the foundation. It’s not that technology is moving too fast for the data, but that we cannot afford to get the data wrong.
As people naturally turn to tools like ChatGPT Health and Claude, the risk of automated misdiagnosis increases if we feed them incomplete, male-centric information. We must prioritize building AI on rich, representative data from companies that capture the nuances of women’s health. By ensuring our diagnostic inputs are as advanced as the algorithms themselves, this new era of digital health becomes a bridge toward true diagnostic equity and surer, more complete diagnoses.
Photo: ThongSam, Getty Images
This post appears through the MedCity Influencers program. Anyone can publish their perspective on business and innovation in healthcare on MedCity News through MedCity Influencers. Click here to find out how.