In 1868, the German physician Carl Wunderlich published a figure that would outlive him by more than 150 years: 37°C, or 98.6°F, as the normal human body temperature. Generations of parents have reached for a thermometer and measured their children against it.

The number has not aged well. A recent study of 25,000 British patients found an average of 97.9°F. In 2020, Stanford researchers analysed 677,423 temperature readings spanning 150 years and found that American men born in the 1990s run about 1.06°F cooler than men born in the early 1800s. Age, sex, body size and time of day all move the number too, enough that the authors declined to propose a single new average for everyone.

The point here goes beyond a 157-year-old thermometer reading being slightly off. "Normal" in medicine has almost always meant an average: a reference range drawn from a population, which each of us is then measured against. That made sense when a measurement was a rare event, taken once in a clinic. It makes much less sense now that the body can be measured every day.

The range hides the person

Take resting heart rate. In one of the largest studies of its kind, researchers at Scripps Research followed 92,457 adults who wore a heart-rate tracker for most of two years, collecting nearly 33 million daily readings. Average resting heart rate across everyone was 65 beats per minute, while individual averages ranged from 40 to 109.

The finding that matters most came from looking at each person over time. Someone's "normal" could differ from another healthy person's normal by as much as 70 beats per minute, yet within the same individual, resting heart rate was remarkably stable from week to week. Deviations from a person's own baseline were rare and discrete, which is precisely what makes them noticeable.

The researchers put the problem plainly: heart rate is routinely measured in a clinical exam, but rarely acted on unless it falls well outside the population range. A single reading says very little about a person's health. A personal baseline says a great deal more.

The same applies to heart-rate variability, sleep, and skin temperature. Two healthy people can sit at opposite ends of a reference range. The useful question is less "where do you sit compared with everyone else?" and more "how are you different from your usual self?"

When your own baseline speaks first

There is already evidence that this question catches things the population average misses. In 2020, a Stanford team led by Michael Snyder analysed smartwatch data from people who later tested positive for COVID-19. Of 32 infected participants, 26 showed changes in heart rate, steps or sleep. Among the cases with symptom data, most changes appeared before or at the onset of symptoms, some more than a week earlier. A warning system based on extreme rises in resting heart rate, measured against each person's own baseline, could have flagged 63% of cases before symptoms in real time.

None of those people would have looked unusual against a textbook range. They looked unusual against themselves.

Medicine for the average patient

The same logic reaches into treatment. Most of the evidence behind modern medicine comes from large trials that report an average effect across thousands of participants. That design is powerful, and it has saved countless lives. It also has a blind spot. Writing in Nature in 2015, Nicholas Schork pointed out that the ten highest-grossing drugs in the United States help between 1 in 25 and 1 in 4 of the people who take them. For statins, the figure can be as low as 1 in 50.

An average effect can describe a population accurately while still saying little about the individual in front of the doctor. Schork's proposal was to take n-of-1 trials seriously: carefully designed studies of a single person, who tries treatments in sequence while the outcomes are measured over time. The method is old. Physicians have long run informal versions, such as adjusting a blood pressure medication and watching what happens. What held it back was measurement: a handful of readings, taken only during treatment, is too thin to tell a real response from noise.

The Oxford Centre for Evidence-Based Medicine already places n-of-1 trials alongside randomised trials at its top level of evidence for treatment benefits. The barrier has never been the idea. It has been the data.

What changes when measurement is continuous

That barrier is falling. A ring or a watch now records more of a person's physiology in a month than most clinical encounters capture in a lifetime: heart rate through the night, sleep, temperature trends, activity. Add the lab results, the symptoms people note down, and how they actually feel, and a picture emerges that no reference range can provide.

The hard part is turning it into understanding. Raw signals are easy to collect and easy to misread. A personal baseline needs enough history, the right context, and the ability to separate a meaningful shift from an ordinary bad night. That is analysis work, and the tools for it are getting better quickly.

This is the thinking behind AxonLife. It is designed to bring signals from many sources into one place and, with AxonOS and AI, help turn them into insight a person can act on, starting from the most useful comparison available: you, compared with yourself.

The textbook normal will remain useful as a starting point. The next step in health is learning what normal means for each of us.

Sources