Healthcare should be tailored to everyone, yet for decades medical research has largely overlooked half of the global population. From the earliest stages of drug discovery, researchers have frequently relied on male cells, male animals, and predominantly male clinical trial populations — with female biology often treated as a complicating factor rather than a fundamental component of human physiology.
For decades, the assumption that male and female biology was sufficiently similar allowed researchers to use male systems as a default. However, a growing body of evidence has demonstrated that biological sex influences almost every stage of disease development and treatment response. Therefore, testing predominantly in male models risks missing half of the picture, potentially obscuring important differences in disease mechanisms, therapeutic response, and safety profiles.
As a result of this, women are nearly twice as likely to experience adverse drug reactions than men. A 2001 report from the US Government Accountability Office found that eight of the ten prescription drugs withdrawn from the market between 1997 and 2000 posed greater health risks for women than men.
Now, as the drug discovery field moves towards more human-relevant approaches such as organoids, organs-on-chips, AI, and mechanistic modelling, researchers face a critical question: Will these emerging technologies help close the biological gaps of the past, or risk reproducing them in new forms?
The historical male default in preclinical research
Male animals have been the standard experimental model for much of modern biomedical research. Female mice were frequently excluded from preclinical studies because researchers believed that fluctuations in reproductive hormones during the estrous cycle would introduce excessive variability, making experiments more difficult to interpret and requiring larger numbers of animals. Practical considerations, including the perceived costs of housing both sexes and the desire to simplify experimental design, further reinforced the preference for male models. Over time, this has become entrenched as scientific convention.
However, that standard relies on a flawed assumption. "Females are not simply small male mice," Ivana Jarić, a neuroendocrinologist and epigeneticist at the University of Zurich and long-term advocate for sex-inclusive preclinical research, told DDN. "Their physiology is different. Pharmacokinetics and pharmacodynamics differ between males and females, and even the expression of enzymes involved in drug metabolism can vary significantly between the sexes. This means that how drugs are processed and how they respond can be different."
Central to the historical exclusion of females was the belief that the estrous cycle made them inherently more variable than males. However, systematic reviews have since shown that female mice are no more variable than males across a wide range of physiological and behavioral measures. And in some cases, female mice have demonstrated more consistent traits than their male counterparts.
Despite increasing awareness, male bias remains widespread in preclinical research. Analyses suggest that more than 80 percent of preclinical studies assessing drug safety and efficacy in 2021 were conducted exclusively in male mice. This imbalance also extends into cell-based research, where the sex of donor cells is frequently omitted from publications or lost as cell lines are propagated over time. Estimates suggest that up to half of in vitro studies fail to specify the sex of the cells used, while only around one in five include cells from both sexes.
These omissions matter because sex differences begin at the cellular level. "Every single cell has a sex," Jarić said. "Cells are either XX or XY, and there are X-linked and Y-linked genes that play important roles in many fundamental biological processes." Even in isolated cell cultures, sex can influence gene expression, metabolism, and responses to environmental cues, yet many researchers continue to regard cultured cells as biologically neutral.
From awareness to implementation
Recognition of the limitations of male-biased research has prompted significant changes across the biomedical community over the past decade. Funding agencies, regulators, and scientific journals have increasingly acknowledged that biological sex should be considered throughout the research process, from experimental design to data analysis and reporting.
One of the most influential policy changes came in 2015, when the US National Institutes of Health (NIH) introduced its Sex as a Biological Variable (SABV) policy, requiring grant applicants to consider sex in the design, analysis, and reporting of vertebrate animal and human studies. Similar initiatives have since been adopted by organizations including the European Commission and the Canadian Institutes of Health Research, while scientific journals, such as Endocrinology and Physiology, have also requested researchers to factor SABV in the design of studies and report sex differences as appropriate.
These initiatives have produced measurable progress. A 2021 survey of US scientists found that more than a third had modified their study designs in response to the NIH policy, while a 2019 meta-analysis reported that nearly half of preclinical studies included both male and female subjects, up from just 28 percent a decade earlier. However, while more researchers are including both male and female animals in preclinical studies, many still do not analyze results by sex, indicating a need for better integration of the SABV policy throughout study design, analysis, and reporting.
Irina Kovlyagina, a neurobiologist at the Johannes Gutenberg University Mainz in Germany, believes that the biggest obstacle is no longer reluctance to include females but a lack of practical expertise. "Most researchers haven't been trained in sex-inclusive experimental design," she told DDN. "They don't know which statistical methods to use, where to get advice, or how to organize training within their institutions."
Addressing those knowledge gaps has become a major focus of new initiatives. Jarić is spearheading the first pan-European multidisciplinary network dedicated to supporting researchers in incorporating SABV into preclinical research. The network provides practical guidance, training workshops, and evidence-based recommendations aimed at helping scientists move beyond simply recognizing the importance of sex and towards implementing robust, reproducible study designs. The network also hosts an annual symposium, featuring lectures ranging from sex differences in brain health, pain, and pharmacology to metabolic, oncological, and respiratory diseases.
The scientific community no longer needs to be convinced that sex matters; instead, it needs to develop the experimental, statistical, and reporting standards needed to study it rigorously. Without that shift, the promise of sex-inclusive research risks being undermined by inconsistent implementation.
The danger and opportunity of new approach methodologies
The shift towards sex-inclusive research is occurring alongside another major transformation in drug discovery. New approach methodologies (NAMs), including human organoids, organs-on-chips and advanced computational models, are increasingly being adopted to generate more human-relevant data and improve the prediction of clinical outcomes. By moving beyond traditional animal models, these technologies promise to capture aspects of human biology that have long proved difficult to study.
However, in a commentary published recently in Nature Neuroscience, Kovlyagina and Jarić argued that prematurely phasing out animal testing could lock in biomedicine’s male-default bias. If female biology remains underrepresented in the data used to build and validate these models, the same blind spots that have limited conventional preclinical research could simply be transferred into the next generation of drug discovery tools.
Machine learning models are rapidly becoming powerful tools for identifying therapeutic targets, predicting toxicity, and accelerating drug discovery, but their performance depends fundamentally on the quality and diversity of the data used for training. When historical datasets disproportionately represent male biology, algorithms may inadvertently learn and reinforce those biases.
This is already occurring in other areas of healthcare. In a recent study from the London School of Economics and Political Science, researchers found that Google's large language model, Gemma, systematically downplayed women's physical and mental health needs when generating summaries of adult social care case notes. When presented with identical cases that differed only in gender, the model was significantly more likely to describe men's health concerns using terms such as "disabled," "unable" and "complex," while equivalent issues in women were more likely to be omitted or described in less serious terms.
"There is currently a lot of excitement around AI, but if you train these models using predominantly male-based data, you are introducing bias from the very beginning," said Jarić. "These models have enormous potential, but if you feed them biased data, you will get biased outcomes."
Human-relevant models have the potential to improve translation and reduce reliance on animal studies, but only if they are built upon a sufficiently complete understanding of human biology. As drug discovery becomes increasingly data-driven, ensuring that female biology is represented from the earliest stages of research will be essential to preventing historical biases from becoming embedded in the technologies designed to supplement or even replace them.
Toward a more complete picture of human biology
Although awareness of sex differences has grown substantially over the past decade, researchers agree that changing policy is only the first step. The larger challenge lies in embedding sex-inclusive thinking throughout the scientific process — from experimental design and statistical analysis to peer review and scientific training. Without that cultural shift, there is a risk that researchers will comply with reporting requirements without fully considering how sex influences the biology they are studying.
That need extends beyond simply including both sexes in experiments. Researchers must also begin to consider the diversity that exists within female biology itself. "If you think about women between the ages of 20 and 75, that is not one single biological group," Jarić said. Hormonal changes associated with puberty, the menstrual cycle, oral contraceptive use, pregnancy, lactation, perimenopause, and menopause can all influence physiology, disease susceptibility, and therapeutic response. Understanding these life stages, she argued, will be essential if researchers hope to develop truly representative models of human biology.
Her own work has highlighted how dynamic these biological changes can be. During the normal estrous cycle in mice, Jarić and colleagues observed substantial changes in chromatin accessibility across the brain — alterations that occurred without any disease or experimental intervention. Other studies have demonstrated that pregnancy can induce structural changes in the brain that persist for years after birth, while researchers are increasingly investigating how perimenopause and menopause influence neurological health and disease risk.
The history of drug discovery has often been written around a single biological reference point. Today, researchers are working to replace that oversimplified model with one that better reflects the complexity of the patients that medicines are intended to treat. Whether through improved study design, more transparent reporting, human-relevant preclinical models, or increasingly sophisticated computational approaches, the goal is to generate evidence that is more predictive, more translatable, and more representative of human biology.
Sex-inclusive research is therefore not about replacing one default with another. It is about recognizing that there is no single "standard" human biology. By accounting for sex from the earliest stages of discovery, researchers have an opportunity to develop safer medicines, uncover previously overlooked therapeutic opportunities, and build the next generation of preclinical models on a more complete understanding of the people they are designed to serve.
