Cortisol Is a Curve: Why One Hormone Test Can Mislead You
In a Nutshell
Ellie Health’s Hormone Meter lets users measure cortisol (and soon progesterone/testosterone) multiple times daily with a 60-second saliva test read by smartphone, turning the hormone’s natural daily curve into a real-time feedback signal for lifestyle decisions. Cortisol is neither inherently “bad” nor static—optimal health requires a high-morning, low-evening pattern, and deviations from this curve (or from personal baseline) correlate with symptoms and performance issues that single-point blood tests routinely miss. The system’s value lies in revealing how specific behaviors—exercise timing, light exposure, fasting, travel—shift these curves within 30 minutes, enabling rapid, data-driven adjustments rather than relying on infrequent snapshots.
These notes were generated by AI and may contain inaccuracies.
Biology is not a screenshot. It's a movie. Every moment, your hormones are shaping your sleep, stress, energy, training, metabolism, and reproductive health. Most people only ever see them as a one-time lab value, if at all. The conversation focuses on what changes when hormone data becomes easier to access at home and continuously.
Marina Rivas is co-founder and CEO of Ellie Health. She is a data scientist and repeat founder who built and sold an AI company before turning to one of the messiest data gaps in health: hormones. Ellie is building Hormone Meter, an at-home saliva-based system that uses a smartphone to read hormone patterns over time, starting with cortisol and expanding into progesterone and testosterone. The discussion covers what repeated hormone testing can and cannot tell us and how to make this kind of data useful without overmedicalizing normal biology.
Marina Rivas started the company out of personal need. She wanted access to hormone data herself. With a background in data science, she was biased towards wanting data when making important decisions, but saw that it was missing around hormonal health. She observed that measuring one heartbeat per year would not make sense for decisions, yet that is exactly how hormones were measured at best: a single snapshot. Depending on location, access might occur yearly or not at all. The vision became making this data accessible anytime, anywhere, instantly, and continuously over time.
Continuous monitoring is a key difference between optimization-focused health or longevity approaches versus traditional infrequent snapshots. Examples of continuous data already exist, such as continuous glucose monitors, resting heart rate, and HRV. Traditional blood tests remain difficult to monitor continuously.
The founders started with the need and developed technology from scratch. They considered interstitial fluid, sweat, and tears before selecting saliva for its ready availability and strong validation. The goal was a technology as easy as brushing teeth and as affordable as a cup of coffee. After over 2000 iterations spanning six years, they launched a system using lateral flow assay similar to pregnancy or COVID tests, but measuring line intensity rather than simple positive/negative results. A smartphone camera captures the image, and computer vision algorithms translate it into hormone concentration displayed in the app. The system combines lateral flow assay chemistry, microfluidic saliva collection and filtering hardware, computer vision algorithms, and a mobile app that provides both levels and actionable interpretation, similar to wearables like Whoop or Oura.
Cortisol was prioritized because the founders wanted to address hormonal health beyond reproductive health. Tools already existed for some sex hormones like LH via urine tests, but cortisol touches all areas of health for both women and men. Known as the stress hormone, cortisol has deep impacts on health, well-being, and long-term conditions including increased mortality risk.
Cortisol fluctuates throughout the day with a peak in the morning and a drop in the evening. This makes it especially suitable for frequent monitoring because it changes based on actions and life events. Like HRV or other vital signs, cortisol responds to lifestyle and actions, creating a valuable feedback loop. The daily curve pattern (high morning, low evening) shifts based on actions such as light exposure, sports, exercise intensity, and timing.
Cortisol is misunderstood as always bad or something to eliminate. It is a true survival hormone needed in appropriate amounts. Both excessively high and excessively low cortisol can cause symptoms. High morning cortisol can cause early waking, while low morning cortisol can cause grogginess or tiredness. The pattern matters: high in the morning and low in the evening is optimal. Other patterns (always low, always high, or low morning with high evening) are associated with different symptoms.
The app asks when the user woke up because the cortisol awakening response occurs in the first hour and a half. This timing information is critical for accurate interpretation. Users can also add tags to track actions and see correlations over time, similar to how wearables work.
Cortisol testing is valuable not only for people feeling ill but also for anyone optimizing well-being or performance. It can reveal effects of caffeine timing, exercise timing, overtraining, light exposure, and other lifestyle factors.
Wearables have shown that seeing actual data and correlations over time leads to behavior change, with studies supporting public health impacts. Hormone Meter aims for similar dynamics with stress. Current wearables struggle to track the actual physiological response to stress; for example, a hike may register as stress due to elevated heart rate, which differs from acute stress elevating cortisol.
The app defines reference ranges based on existing studies of optimal cortisol levels. Users see their data points against these ranges and can also track deviations from their personal baseline. Tags help distinguish acute versus chronic stress. Athletes test before and after training or competitions to assess overtraining or undertraining. Weightlifting does not typically produce a cortisol spike, while endurance activities like running do. Some people discover their cortisol stays high after good stressors like intermittent fasting or high-intensity training, leading to feeling tired but wired and sleep difficulties. The body does not necessarily distinguish between good and bad stress.
The approach of training for longevity or health span focuses on living the life one aspires to rather than simply avoiding illness.
Progesterone and testosterone were prioritized next because they have broad impacts on health beyond reproduction. Progesterone affects all areas of health and is critical during perimenopause. Testosterone impacts both men and women, and cultural awareness of its importance is growing. The company focuses on lifestyle-based interventions at home, though physicians may use the data alongside interventions like HRT or TRT.
Testosterone also has a daily curve with morning versus evening shifts, but tracking evolution over time (throughout the month and year) is more valuable than daily patterns. Lifestyle changes affect testosterone levels. Current TRT options include longer-term injections and newer twice-daily oral treatments that better follow natural curves. At-home saliva testing helps address challenges with lab timing, fasting requirements, and scheduling around doses.
The company avoids the term "at-home testing" and instead emphasizes "anytime, anywhere testing." Users test while driving, at the gym, or even while climbing mountains. Testing takes 60 seconds in the mouth with a 20-minute development time before taking a photo.
Interpretation depends on goals. Someone seeking an optimal daily curve tests at different times to see which actions disrupt the pattern and then adjusts. Performance contexts extend beyond physical training to include professional pitches or stage performances.
Athletes and individuals can use continuous cortisol monitoring to assess how training, travel, and other activities affect their hormone curves. When someone observes that certain activities disrupt their cortisol response, they can experiment with different approaches until those activities no longer create negative impacts. In non-training contexts, people can use these feedback loops to maintain optimal cortisol patterns despite demanding lifestyles.
Sports teams monitor cortisol differently during training versus competition. During training sessions, teams prefer to keep cortisol relatively low to avoid overtraining before matches. For actual games, they want to see peak cortisol levels to evaluate whether players achieved optimal performance deployment. Evening measurements are equally important to confirm that cortisol returns to normal levels after competition.
Testosterone and progesterone monitoring follows a weekly testing schedule rather than daily measurements. Daily testing represents edge cases where individuals want tighter feedback loops. Progesterone testing frequency may increase when detecting luteal symptoms or premenstrual issues, particularly for individuals with irregular cycles where additional data points provide valuable insights around specific cycle phases.
Multiple lifestyle factors influence testosterone levels. Poor sleep negatively impacts testosterone, while certain exercise types, particularly weightlifting, positively affect testosterone levels. The weekly testing frequency aligns with testosterone's slower response dynamics compared to cortisol, which can show changes within 30 minutes of lifestyle modifications.
Weekly testosterone testing provides sufficient visibility for most users while avoiding noise that daily testing might introduce. The balance involves providing enough data for meaningful feedback loops without overwhelming users who have other life commitments. Individuals with full-time dedication to optimization protocols may benefit from more frequent testing, but weekly measurements work well for those balancing multiple priorities.
Each individual data point can be influenced by undetected confounders, such as brief nighttime awakenings. Multiple data points reveal trends more reliably than single measurements. The recommendation involves conducting monthly or bimonthly reviews to identify meaningful patterns rather than over-interpreting isolated data points.
Cortisol responds rapidly to lifestyle changes, sometimes within 30 minutes. Testosterone and progesterone demonstrate slower reactivity to lifestyle modifications. Understanding these different biological response times informs appropriate testing frequencies for each hormone type.
Current regulatory approval focuses on wellness applications rather than diagnostic use for conditions like Cushing syndrome or hypogonadism. However, the technology can already detect diagnostic patterns, and users have reported identifying conditions through their data that led to proper medical diagnoses. Future regulatory approval for diagnostic applications remains possible, particularly for challenging conditions like cyclical Cushing syndrome where frequent monitoring provides life-changing insights.
Artificial intelligence output quality depends entirely on input data quality. Personalized and precision medicine require comprehensive visibility into continuous body patterns. Technologies capturing frequent data enable advancement of medical understanding and evolution of care standards based on accumulated information and knowledge.
In the AI era, hardware advancements may drive the most significant progress because readily available algorithms make data input quality the key differentiator. Medical and physical world applications require physical tools for data collection, suggesting that major AI evolution will occur through hardware development.
The design philosophy prioritizes ease of use comparable to or better than brushing teeth, potentially moving toward fully passive wearable collection methods. While saliva testing continues in the near term, other fluids like sweat and tears remain under consideration. Each potential method requires evaluation of existing validation, regulatory implications, commercialization pathways, manufacturing feasibility, and user adoption potential.
Bringing a single biomarker from initial concept to market launch required approximately seven years and close to 30 million dollars in investment. The process includes validation accuracy, regulatory approval, and manufacturing considerations. Once established, the existing infrastructure, technology, and intellectual property enable significantly faster development timelines for additional biomarkers.
Visibility over time proves essential for health improvement efforts. Combining multiple data sources including smart watches, continuous glucose monitors, and hormone measurements creates comprehensive progress dashboards. This integrated approach ensures optimization efforts produce intended results rather than operating blindly.
Despite perceptions that everything has been accomplished, significant gaps remain in health innovation. Addressing these gaps requires patience and dedication to single-focus development from concept through market launch. Different companies approaching problems from various angles will eventually consolidate into integrated systems supporting fully personalized health approaches.
Keep Hillary Lin, MD in your library
Save the videos and channels worth coming back to, and find them again in one place.
