The Myth of the Average Human: When Population Statistics Meet Individual Biology | Why the “Normal” Person Does Not Really Exist
Engr. Kamran Abbas
BSc Civil Engineering
MS Transportation Engineering
Table of Contents
- The Human Being Who Exists Only on Paper
- What Does “Average” Actually Mean?
- Why Science Needs Averages
- The Problem With Turning a Population Into a Person
- Reference Ranges: Normal Does Not Mean Healthy—and Abnormal Does Not Always Mean Disease
- The Same Number Can Mean Different Things to Different People
- Biology Is Not Static: The Individual Changes With Time
- Medicine’s Average Patient Problem
- Clinical Trials: What Works “On Average”
- BMI, Blood Pressure and Other Numbers That Can Mislead
- The Hidden Problem of Population Samples
- From Population Medicine to Precision Medicine
- Why More Data Does Not Automatically Mean More Personalization
- The Statistical Danger of Treating Outliers as Errors
- The Public-Health Paradox: Individual Exceptions Can Still Matter at Population Scale
- Toward a Better Model of Human Health
- The Future: From Reference Ranges to Personal Baselines
- Conclusion: The Average Human Is a Mathematical Convenience, Not a Person
- References
The Human Being Who Exists Only on Paper
Imagine a patient who walks into a clinic. The doctor does not know everything about this person. No doctor can. Instead, medicine begins with measurements: age, weight, blood pressure, temperature, laboratory values, symptoms, family history, perhaps genetic information and previous medical records.
Those measurements are compared against something. A reference population. A clinical threshold. An expected response. An average. This is not a flaw in medicine. It is one of the reasons modern medicine works at all.
Without population statistics, medicine would have no reliable way to distinguish unusual findings from common ones, estimate disease risk, test treatments, establish reference intervals, or determine whether an intervention is generally beneficial. But there is a strange consequence.
The more precisely medicine measures populations, the more obvious it becomes that no individual human being is actually “the average human.”
The average human is a statistical construction. There may be an average age, average height, average blood pressure, average cholesterol concentration, average treatment response and average risk. But there is no person whose biology is simply the arithmetic average of millions of other people. That distinction sounds philosophical. It is actually profoundly practical.
Modern healthcare constantly moves between two different worlds:
Population science asks: What tends to happen among people like these?
Clinical medicine asks: What is happening to this person?
Those questions are related, but they are not identical. A treatment can produce a substantial average benefit while helping some people enormously, helping others only slightly, doing almost nothing for some, and harming a minority. A laboratory value can be “normal” for the population while representing a major departure from an individual’s usual physiological state. A person can fall outside a population reference interval and still be perfectly healthy. And a person can sit comfortably inside the statistical definition of “normal” while developing a serious disease.This is the central paradox of the average human.
Statistics describe populations extremely well. Humans experience biology individually.
What Does “Average” Actually Mean?
The word average is deceptively simple. In everyday conversation, it often means “typical.” In statistics, however, an average is usually a mathematical operation.
The arithmetic mean is:
Mean = (x₁ + x₂ + … + xₙ) / n
Suppose five people weigh:
| Person | Weight |
|---|---|
| A | 55 kg |
| B | 60 kg |
| C | 65 kg |
| D | 70 kg |
| E | 100 kg |
| Mean | 70 kg |
The average is 70 kg. But notice something important. Only one person weighs 70 kg. The average summarizes the group. It does not describe the biological identity of any particular member of that group.
The distinction becomes even more important when distributions are skewed, multimodal or heterogeneous.
Consider two populations:
| Population | Values |
|---|---|
| A | 68, 69, 70, 71, 72 |
| B | 40, 50, 70, 90, 100 |
Both have a mean of 70. Yet they are obviously not the same population. The mean has not lied. It has simply answered a narrower question than we might imagine.
This is one of the most important lessons in statistics:
A summary statistic can be accurate while still being inadequate for describing an individual.
The problem is therefore not the existence of averages. The problem is average-thinking.
Why Science Needs Averages
It would be a mistake to attack averages as though they were inherently misleading. They are indispensable. Population statistics allow researchers to estimate disease prevalence, compare groups, evaluate interventions and detect relationships that cannot be reliably identified from isolated individuals. Clinical trials depend heavily on comparisons between groups. Public health depends even more heavily on population-level evidence. If researchers want to know whether a new intervention reduces disease, they cannot simply study one person and generalize the result to everyone.
They need populations.
They need probability.
They need distributions.
They need uncertainty estimates.
They need effect sizes.
They need averages.
The problem is therefore not: population science versus individual science.
It is: population evidence applied intelligently—or mechanically—to individuals.
Modern medicine itself increasingly recognizes this distinction. The National Institutes of Health describes precision medicine as an approach that accounts for individual differences in genes, environments and lifestyles rather than relying exclusively on the expected response of an “average” patient. So the mature position is not to abandon population statistics. It is to understand their proper role.
The Problem With Turning a Population Into a Person
Imagine researchers measure the blood pressure of 100,000 people and calculate the population distribution.
They can tell us:
- what values are common;
- what values are unusual;
- how risk changes across the distribution;
- how blood pressure relates to age and other variables;
- how an intervention changes population risk.
But none of this creates a fictional “average person” who possesses the average value of every measured variable. The average height, average weight, average glucose, average heart rate, average cholesterol and average sleep duration could theoretically be calculated. Yet there may be no actual human who possesses that complete combination.
This is sometimes called an ecological or aggregation problem: properties observed at the group level cannot automatically be assumed to describe every individual within the group. The same issue appears in treatment research. A clinical trial estimates an average treatment effect. But the patient in front of the doctor is not an average treatment effect.
Research on heterogeneous treatment effects explicitly distinguishes the population-level task of estimating average treatment effects from the clinical task of determining what treatment is most appropriate for a particular person. That distinction should sit at the center of evidence-based medicine.
Reference Ranges: Normal Does Not Mean Healthy
One of the most familiar examples is the laboratory reference range. A laboratory report may display a result alongside a range labelled “normal” or “reference.”
Patients understandably interpret this as:
inside = healthy
outside = unhealthy
But that is not what a reference interval necessarily means. Reference intervals are statistical tools intended to help clinicians interpret laboratory measurements relative to a defined reference population. Their construction depends on factors including the population selected, analytical method and statistical procedure.
This matters enormously. A reference interval is not a biological border separating healthy humans from diseased humans. It is a statistical reference.
Consider a simplified distribution:
Lower values Higher values
| |
v v
----|---------|===============================|---------|----
Reference population distributionThe central distribution may contain most observations, while individuals at the extremes require further interpretation. But biological reality rarely behaves like a light switch. Disease risk often changes gradually. Some diseases can exist despite measurements remaining inside a conventional reference interval.
Conversely, an unusual measurement does not automatically establish disease. The patient’s symptoms, history, previous measurements, medications, age, physiology and clinical context matter. This is why laboratory medicine has increasingly examined individual-specific reference intervals.
Research comparing population-based and personalized reference intervals has found that individual ranges can be substantially narrower or otherwise different from population intervals for many laboratory measurands. The implication is powerful:
A population can tell us what is common. A personal baseline can tell us what is unusual for you.
Those are different kinds of information.
The Same Number Can Mean Different Things to Different People
Suppose two people have exactly the same laboratory result. Are they necessarily equally healthy? No.
The same numerical value can have different meanings depending on:
- age;
- sex;
- pregnancy status;
- genetics;
- medications;
- diet;
- exercise;
- hydration;
- circadian timing;
- acute illness;
- chronic disease;
- recent physiological stress;
- laboratory method;
- previous measurements.
Human biology is conditional. A number is not meaningful in isolation merely because it has many decimal places.
This creates a hierarchy of interpretation:
| Level | Question |
|---|---|
| Population | Is this common among people? |
| Reference group | Is this expected for people with similar characteristics? |
| Individual baseline | Is this normal for this particular person? |
| Longitudinal pattern | Is the person’s value changing over time? |
| Clinical context | Does the change or value make sense given the person’s symptoms and circumstances? |
The fifth level is often the most clinically informative. Yet conventional healthcare has historically been much better at collecting snapshots than continuously understanding trajectories.
Biology Is Not Static: The Individual Changes With Time
There is another problem with the average human. Humans are not stationary systems. A person’s physiology changes throughout the day, across seasons and across the lifespan. Heart rate changes. Hormones fluctuate. Glucose changes after meals. Body temperature follows biological rhythms. Training alters physiological responses. Illness changes biomarkers. Aging changes reference distributions.
The value that is ordinary for a person at one point in life may be unusual later. Laboratory science has long recognized both between-person and within-person biological variation. Some measurements display substantial individuality, meaning that an individual’s usual physiological range can be considerably narrower than the broader population distribution.
This suggests a useful conceptual distinction:
Population variability: How much people differ from one another.
Individual variability: How much the same person changes over time.
These are not interchangeable. A population may have a wide distribution while an individual remains relatively stable within a much narrower personal range.
That creates a potentially important clinical signal: A change can be significant for an individual even when the new value is still statistically “normal.”
Medicine's Average Patient Problem
Medicine has achieved extraordinary success by learning from populations. But its fundamental operating environment is individual. A clinical trial might contain thousands of people. A doctor’s consultation contains one. This creates a translation problem.
Research asks: What happened, on average, to people in the study?
Clinical practice asks: Given what we know about this person, what should we do now?
The two questions cannot be connected by averages alone. The National Academy of Medicine has highlighted heterogeneous treatment effects—the fact that treatment responses can differ between individuals—as an important issue in patient-centred care. The problem becomes particularly obvious with medication.
Imagine a hypothetical treatment:
Treatment response
High benefit | ███████████
Moderate | █████████████████
Little benefit | ███████████
No benefit | █████
Harm | ██The overall average might show a beneficial treatment effect. That is extremely useful information. But it does not tell the physician exactly where the next patient belongs on that distribution.
That is the hard problem of medicine: population-level evidence must eventually become an individual-level decision.
Clinical Trials: What Works “On Average”
Clinical trials are sometimes criticized for producing “average” results. That criticism is incomplete. A trial’s average treatment effect can be scientifically invaluable.
Suppose:
- 10,000 patients receive a treatment;
- the treated group experiences fewer adverse outcomes than the control group.
That tells us something important.
But several additional questions immediately follow:
- Who benefited most?
- Who benefited least?
- Who experienced adverse effects?
- Did age modify the response?
- Did disease severity matter?
- Did genetics matter?
- Did previous treatment matter?
- Did adherence matter?
- Did socioeconomic circumstances affect outcomes?
- Did multiple conditions alter the response?
These are questions about heterogeneity. The average effect can conceal meaningful variation.
However, there is also a danger on the opposite side. Not every observed difference between individuals represents a genuine biological subgroup. Some variation is random noise. Some is measurement error. Some arises from inadequate sample size. Some is caused by confounding. Some apparent subgroups appear only because researchers search enough combinations to find them. This is why personalization is not simply a matter of collecting more variables. It requires rigorous causal reasoning.
BMI, Blood Pressure and Other Numbers That Can Mislead
BMI provides a particularly accessible example. The World Health Organization defines adult BMI as weight in kilograms divided by height in metres squared and uses established population-level categories such as 18.5–24.9 for “normal weight” and 25.0–29.9 for “overweight.”
Those categories are useful. But BMI is not a complete description of a human body. Two people with the same BMI can have different proportions of:
- muscle;
- fat;
- bone;
- visceral fat;
- lean tissue.
WHO itself notes that BMI can be less accurate for assessing healthy weight in certain groups because people with the same BMI can have different proportions of fat and lean mass.
This illustrates a broader principle: A useful population indicator can become a poor individual descriptor when treated as a complete biological truth.
The same logic applies to many other measurements. A blood-pressure threshold is useful. A cholesterol concentration is useful. A heart-rate range is useful. A calorie estimate is useful. An age-based risk estimate is useful. But usefulness does not imply completeness. A measurement is a window. It is not the entire person.
The Hidden Problem of Population Samples
There is an even deeper problem. Before calculating an average, researchers must decide whose data count. A population is never simply “humanity.” It is a sample. And samples can be biased. Psychological science provides a famous example.
Researchers have documented extensive reliance on WEIRD populations—people from Western, Educated, Industrialized, Rich and Democratic societies—and warned against assuming that findings from these populations automatically represent humanity as a whole. More recent research continues to identify substantial geographic and demographic imbalances in psychological research samples. This problem is not limited to psychology. Any scientific “average” inherits characteristics from the population used to calculate it.
If a reference population is:
- too young,
- too old,
- geographically narrow,
- ethnically narrow,
- economically privileged,
- predominantly male,
- predominantly female,
- urban,
- sedentary,
- unusually healthy,
then its average may be mathematically precise but biologically inappropriate for another population.
This leads to an uncomfortable question: Average of whom?
Before asking whether someone is “normal,” science should sometimes ask: Normal relative to which population?
From Population Medicine to Precision Medicine
The response to the limitations of average-based medicine is not to discard population evidence. It is to combine population evidence with individual information. This is the central idea behind precision medicine.
Precision medicine attempts to incorporate characteristics such as:
- genetics;
- environment;
- lifestyle;
- disease phenotype;
- previous treatment;
- biomarkers;
- longitudinal measurements;
- patient preferences.
NIH describes the goal in practical terms: selecting the right treatment and dose for the right patient at the right time.
The conceptual transition looks like this:
| Traditional emphasis | Emerging emphasis |
|---|---|
| Population average | Individual trajectory |
| Reference range | Personal baseline + reference range |
| One-size-fits-most | Stratified or individualized treatment |
| Single measurement | Longitudinal data |
| Disease category | Biological phenotype |
| Average treatment response | Heterogeneous treatment response |
| Generic risk | Individualized risk |
But there is an important caveat. Precision medicine is not magic.
Not every individual difference is clinically meaningful. Not every biomarker predicts outcomes. Not every genetic variant changes treatment. Not every algorithm improves decision-making.
Personalization must itself be evidence-based. Otherwise, the “average human” is simply replaced by another scientific fiction: the supposedly perfectly knowable individual.
Why More Data Does Not Automatically Mean More Personalization
Modern technology creates an extraordinary temptation.
Wearables can record:
- heart rate;
- sleep;
- movement;
- exercise;
- temperature;
- oxygen-related measures;
- physiological trends.
Laboratories can produce increasingly detailed molecular profiles. Genomics can reveal enormous quantities of information. Electronic health records can provide longitudinal histories. Artificial intelligence can integrate variables that would be impossible for a human to process manually.
Yet more data can create a new problem: measurement without interpretation.
If a person receives hundreds of daily measurements, the question becomes: Which changes are meaningful?
This is harder than simply measuring more. A physiological variable naturally fluctuates. If you measure something frequently enough, you will inevitably find deviations from yesterday’s value. The danger is turning ordinary biological variability into pathology. The opposite danger is equally serious: dismissing a meaningful change because the new value still falls inside a broad population reference range.
The future therefore cannot simply be: more measurements.
It must be: better models of variation.
The Statistical Danger of Treating Outliers as Errors
Statistics often focus on the center. But biology frequently hides important information in the tails.
An unusually high value may be:
- measurement error;
- random variation;
- a rare benign characteristic;
- early disease;
- severe disease;
- a genetic trait;
- an environmental effect.
Calling something an “outlier” does not explain why it is unusual. It merely describes its position in a distribution. This distinction matters.
Historically, unusual observations could be treated as statistical inconveniences. Modern biomedical research increasingly asks whether those observations contain useful information. Rare diseases provide an obvious example. A rare condition may be almost invisible in population statistics while being overwhelmingly important to the person who has it. Population prevalence and individual significance are not the same thing. A condition affecting one person in 100,000 is statistically rare. For that one person, it is not a 0.001% experience. It is 100% of their experience.
The Public-Health Paradox: Individual Exceptions Can Still Matter at Population Scale
There is another side to this argument. It would be wrong to conclude that because individuals are different, population statistics are somehow irrelevant. Public health proves the opposite. Population-level interventions can save enormous numbers of lives even though they do not affect every person identically. Vaccination strategies, tobacco control, pollution standards, road-safety interventions, nutrition policies and screening programs all depend on population evidence. The individualist critique of averages must therefore stop short of rejecting population thinking.
There are really two levels of truth:
Population truth i.e. A statement about probabilities, distributions and average effects across a defined group.
Individual truth i.e. A statement about the biology, circumstances and trajectory of a particular person.
Neither automatically replaces the other. The sophisticated approach is to connect them.
Toward a Better Model of Human Health
The future of health measurement should move away from the imaginary “average human” without abandoning the enormous power of population science.
A more useful framework would look something like this:
POPULATION EVIDENCE
↓
Reference distributions
↓
Risk estimates
↓
Subgroup differences
↓
INDIVIDUAL PROFILE
↓
Personal baseline
↓
Longitudinal trajectory
↓
Clinical context
↓
INDIVIDUAL DECISION
↓
Outcome
↓
New data
↺This is fundamentally different from simply asking: Is my number normal?
The better question is: What does this number mean for this person, at this point in time, given their history and context?
That is a much harder question. It is also a much more useful one.
The Future: From Reference Ranges to Personal Baselines
One of the most promising developments is the increased interest in personalized reference intervals.
Instead of relying exclusively on: “Where does this result fall among healthy people?”
medicine can increasingly ask: “Where does this result fall relative to this person’s established physiological state?”
Research on personalized reference intervals has found that individual-specific ranges can differ substantially from population-based ranges, and recent work continues to investigate how biological variation can be incorporated into individualized interpretation.
The distinction can be illustrated simply:
| Approach | Reference question |
|---|---|
| Population reference | Is this common among the reference population? |
| Personalized reference | Is this typical for this individual? |
| Longitudinal monitoring | Is this individual changing? |
| Clinical interpretation | Does the change matter? |
This does not mean that population reference intervals will disappear. They remain essential, particularly when no reliable personal baseline exists. A newly evaluated patient may have no historical data. A child may be undergoing rapid development. A person may have a newly diagnosed condition. A reference population provides an essential starting point. But the longer a person is observed, the more valuable their own history may become. The patient’s previous measurements can become a kind of biological control group.
The Deeper Problem: We Like Categories More Than Biology Does
There is a psychological reason the average human is so attractive.
Humans like categories.
Normal vs Abnormal.
Healthy vs Unhealthy.
High vs Low.
Safe vs Dangerous.
Fit vs Unfit.
But biology is usually less tidy. Most physiological variables exist along continua. Risk is often continuous. Disease processes develop over time. Individual responses overlap. Thresholds are often imposed because decisions require thresholds—not because nature necessarily contains perfectly sharp boundaries. A clinical threshold can be extremely useful while still being somewhat artificial. This distinction is easy to forget.
A line drawn at a particular number can become psychologically stronger than the continuous biological process it represents.
The result is a peculiar transformation: a statistical convention becomes a perceived biological fact.
That is when “normal” begins to become dangerous.
What the Average Human Gets Right—and What It Gets Wrong
The concept of the average human is not useless. It is extraordinarily useful when used for what it was designed to do.
The average is useful for:
- estimating population risk;
- planning healthcare systems;
- establishing reference distributions;
- designing clinical trials;
- evaluating public-health interventions;
- identifying broad trends;
- allocating resources;
- generating scientific hypotheses.
The average becomes problematic when it is used to:
- predict an individual’s outcome with unjustified certainty;
- treat a reference interval as a diagnostic boundary;
- assume identical treatment responses;
- ignore longitudinal personal data;
- overlook biological heterogeneity;
- generalize narrowly sampled research to humanity;
- interpret a single measurement without context.
The distinction can be summarized in one sentence:
The average is excellent for describing a population and incomplete for describing a person.
A More Honest Statistical Model of the Human
Perhaps the most useful replacement for the average human is not a new “ideal human.” It is a layered model.
Layer 1 — Population
Where does this person sit relative to a defined population?
Layer 2 — Subgroup
How does their age, sex, environment, ancestry, disease state or other relevant characteristic modify interpretation?
Layer 3 — Individual baseline
What is usual for this particular person?
Layer 4 — Trajectory
How is that baseline changing?
Layer 5 — Context
What is happening in the person’s life and physiology that could explain the change?
Layer 6 — Decision
What action has the strongest evidence of producing a desirable outcome for this individual?
This model does not reject statistics. It uses statistics more intelligently.
The Average Human Is a Mathematical Convenience, Not a Person
There is no single “normal” human being. There is no person who represents humanity in all its dimensions. There is no biological template from which every individual deviates by some measurable amount.
There are distributions. There are probabilities. There are reference populations. There are averages. There are thresholds. And there are billions of individual biological trajectories. That does not make averages wrong. It makes them partial. The great achievement of modern science was learning how to turn individual observations into population knowledge. The next challenge is learning how to turn population knowledge back into better individual decisions.
That requires humility. A population study can tell us what usually happens. It cannot guarantee what will happen to the next person. A reference interval can tell us what is common. It cannot define the complete boundary between health and disease. A treatment can work on average. That does not mean it will work equally well for everyone.
And a person’s measurement can be statistically unusual without being biologically pathological. The deeper lesson is therefore not that science should stop averaging. It is that science should stop confusing the average with the individual. The average human is useful because no one has to be one. The real human—the person sitting in front of the clinician, living through changing environments, carrying a unique history and producing a continuously changing biological signal—is far more complicated. And that complexity is not noise surrounding the “real” human. It is the human.
An Important Statistical Insight
The deepest mistake is not calculating an average. It is assuming that the average contains all the information that matters. Two populations can have the same mean while having dramatically different distributions. Likewise, two people can have the same measurement while having dramatically different histories.
Conceptually:
Same population mean
│
├───────────────┐
▼ ▼
Narrow variation Wide variation
│ │
▼ ▼
"Most similar" "Highly diverse"Therefore, a serious description of human biology should rarely stop at the mean.
It should also consider:
distribution + variance + subgroup + uncertainty + individual baseline + time + context.
That is the difference between merely measuring humans and actually understanding them.
Final Thought
The most important correction is surprisingly simple:
Science does not need to stop using averages. It needs to stop pretending that an average is a human being.
Population science tells us what tends to happen.
Individual biology tells us what is happening here.
The future of medicine lies not in choosing one over the other, but in finally learning how to connect them.
Visual Summary: From the “Average Human” to the Real Human
THE OLD MENTAL MODEL
POPULATION
│
▼
┌───────────┐
│ AVERAGE │
│ HUMAN │
└───────────┘
│
▼
INDIVIDUAL PATIENT
THE MORE REALISTIC MODEL
┌─────────────────────────┐
│ POPULATION DATA │
│ averages • risks • RIs │
└────────────┬────────────┘
│
▼
┌─────────────────────────┐
│ SUBGROUP CONTEXT │
│ age • sex • environment │
└────────────┬────────────┘
│
▼
┌─────────────────────────┐
│ INDIVIDUAL BASELINE │
│ "What is usual for │
│ this person?" │
└────────────┬────────────┘
│
▼
┌─────────────────────────┐
│ TRAJECTORY │
│ "What is changing?" │
└────────────┬────────────┘
│
▼
┌─────────────────────────┐
│ CONTEXT │
│ symptoms • environment │
│ treatment • behaviour │
└────────────┬────────────┘
│
▼
┌─────────────────────────┐
│ INDIVIDUAL DECISION │
└─────────────────────────┘Key Takeaways
| Principle | What it means |
|---|---|
| Average ≠ individual | A population mean summarizes a group; it does not describe a particular person. |
| Normal ≠ healthy | A reference interval is a statistical reference, not a complete definition of health. |
| Abnormal ≠ disease | An unusual value requires context and further interpretation. |
| One measurement ≠ trajectory | Change over time may be more informative than a single snapshot. |
| Same number ≠ same biology | Identical measurements can have different meanings in different people. |
| Population evidence remains essential | Public health and clinical research depend on population-level statistics. |
| Personalization has limits | Individualized medicine must itself be supported by evidence. |
| More data ≠ better medicine | Measurement only becomes useful when biological variation can be interpreted correctly. |
| Reference populations matter | Averages inherit the characteristics and biases of the populations used to construct them. |
| The future is hybrid | The strongest model combines population evidence with individual baselines and trajectories. |
References and Further Reading
National Institutes of Health — The Promise of Precision Medicine. NIH explains why conventional medicine historically relied on expected responses from an “average” patient and how precision medicine incorporates individual differences in genes, environment and lifestyle. NIH: The Promise of Precision Medicine
Kazar et al. (2026) — Personalized reference intervals for biochemical, hormonal, and coagulation tests. A recent study directly examining the limitations of population-based reference intervals and the potential value of personalized intervals. PubMed: Personalized reference intervals for biochemical, hormonal, and coagulation tests
Jones et al. — Personalized reference intervals: from theory to practice. Reviews why population reference data can be limited when applied to individual patients and discusses approaches for deriving personal reference intervals. PubMed: Personalized reference intervals: from theory to practice
Fröhlich et al. — Using group data to treat individuals. Examines heterogeneous treatment effects and the fundamental difference between estimating population treatment effects and treating individual patients. PMC: Using group data to treat individuals
National Academy of Medicine — Caring for the Individual Patient. Discusses heterogeneous treatment effects and the challenge of applying group-level evidence to individual clinical decisions. NCBI Bookshelf: Caring for the Individual Patient
WHO — Nutrition and maintaining a healthy lifestyle. Provides the widely used adult BMI categories and illustrates how population-level classifications are applied in public health. WHO: Nutrition for a healthy life
WHO — BMI interpretation. WHO documentation explicitly notes limitations of BMI for certain individuals because people with the same BMI can have different proportions of fat and lean mass. WHO BMI guidance
Ozarda, Higgins & Adeli — Verification of reference intervals in routine clinical laboratories. Discusses the construction, verification and population-specific nature of laboratory reference intervals. PubMed: Verification of reference intervals in routine clinical laboratories
Jones — Reference intervals: current status, recent developments and future considerations. Reviews the scientific basis and interpretation of clinical laboratory reference intervals. PubMed: Reference intervals: current status, recent developments and future considerations
Higgins et al. — Inherent biological variation and reference values. Examines within-person and between-person biological variation and the implications for interpreting laboratory measurements. PubMed: Inherent biological variation and reference values
Rad, Martingano & Ginges — Toward a psychology of Homo sapiens. Demonstrates the problem of generalizing findings from culturally narrow research populations to humanity as a whole. PMC: Toward a psychology of Homo sapiens
Thalmayer et al. — Expanding the interpretive power of psychological science by attending to culture. Examines how culturally narrow samples can lead to overgeneralization of supposedly universal human characteristics. PMC: Expanding the interpretive power of psychological science
Ratiner et al. — Utilization of the microbiome in personalized medicine. Reviews how substantial inter-individual biological variability complicates population-based disease prediction and treatment. Nature Reviews Microbiology: Utilization of the microbiome in personalized medicine
Blaber et al. — Precision Medicine. Provides a statistical framework for individualized decision-making using evolving patient information. PubMed: Precision Medicine