From Population Averages to Individual Trajectories: Rethinking How Healthcare Defines Normal

For much of modern medicine, the idea of “normal” has been defined through populations. Blood pressure, cholesterol, blood glucose, heart rate, body mass index, laboratory measurements and countless other clinical indicators are commonly interpreted by comparing an individual with reference ranges derived from groups of people. This approach has been fundamental to evidence-based medicine because population-level research allows clinicians to distinguish common physiological variation from patterns associated with disease.

However, a population average does not necessarily describe what is normal for a particular person. Two individuals of the same age can have very different physiological baselines, medical histories, lifestyles, genetic characteristics and responses to treatment. More importantly, the significance of a measurement may depend not only on where it sits relative to a population reference range, but also on how that measurement has changed over time.

This is creating an important conceptual shift in healthcare. Instead of asking only whether a patient’s latest measurement falls inside or outside a population-defined range, researchers are increasingly exploring whether it represents a meaningful change from that individual’s own historical trajectory. Recent work in precision medicine, longitudinal electronic health records, multi-omics and artificial intelligence is making this approach increasingly feasible. Harvard Medical School described this broader transformation in 2026 as a movement from population averages toward individualized insight, with genetic, molecular, imaging and clinical information increasingly being integrated to understand individual health status.

The emerging idea is not that population averages should disappear. Rather, population-level evidence and individual trajectories may need to work together. The future of healthcare could therefore involve a more dynamic definition of normal: one that considers where a person is today, where they have been, and how rapidly their health appears to be changing.

Why Population Averages Became Central to Medicine

Population reference ranges exist for a practical reason. Medicine needs standards that allow clinicians to interpret measurements consistently. When a laboratory test is performed, the result usually needs to be compared with a reference interval or clinically established threshold. Population studies make it possible to estimate expected distributions and identify values associated with increased risk.

The same principle applies to disease-risk models and treatment guidelines. Large clinical studies can determine whether an intervention tends to reduce cardiovascular events, improve glucose control, prevent complications or increase survival across a population. Without these statistical foundations, clinical decision-making could become excessively subjective.

Population medicine has therefore provided one of the strongest foundations for modern healthcare. It allows physicians to move beyond individual anecdotes and make decisions based on evidence gathered from thousands or millions of people.

The limitation is that averages describe populations better than they describe individuals. An average treatment effect does not mean every patient experiences the same response. Similarly, a reference interval does not necessarily capture the physiological pattern that is most meaningful for a particular person.

This distinction becomes increasingly important when healthcare moves from treating established disease toward early detection and prevention.

The Difference Between Being Abnormal and Changing

Consider a hypothetical patient whose laboratory value has always been near the upper end of the population reference range. If that value remains relatively stable for years, its interpretation may be different from a patient whose value moves rapidly from the middle of the reference range toward the upper boundary.

Both measurements could still be classified as “normal” according to a conventional reference interval. Yet their trajectories are different.

The first individual may have a stable personal baseline. The second may be experiencing a physiological change that deserves attention even before the measurement crosses a conventional disease threshold.

This distinction between a static measurement and a trajectory is becoming increasingly important in modern biomedical research. A 2026 Harvard dissertation examining nearly two billion longitudinal laboratory measurements from more than 1.5 million adults reported that individualized thresholds could sometimes identify changes that population-based thresholds missed, while also finding that some individualized classifications of abnormality did not correspond to adverse outcomes. The work therefore points toward a combined model in which individual trajectories are interpreted alongside population-level expectations rather than simply replacing one with the other.

The significance of this idea extends beyond laboratory tests. A person’s heart rate, sleep patterns, cognitive performance, metabolic markers, physical activity, medication response and other measurements can all fluctuate over time. Understanding whether a change is meaningful may require knowledge of the person’s previous state.

Health Is a Trajectory, Not a Snapshot

Traditional healthcare often operates through snapshots. A patient visits a clinic, has blood tests performed, receives a blood pressure measurement and discusses current symptoms. These observations provide valuable information, but they represent only a small portion of a person’s health history.

Digital health technologies and longitudinal electronic records are beginning to change this structure. Instead of viewing health as a sequence of isolated appointments, researchers can increasingly organize information along a timeline.

A 2026 Nature Medicine study demonstrated this approach by combining structured electronic health record information with unstructured clinical notes to reconstruct longitudinal patient journeys. The researchers showed that treatment responses could be studied through patient-level trajectories, including changes in weight and hemoglobin A1c after initiation of GLP-1 receptor agonists. The approach also incorporated information contained in clinical narratives that would otherwise be difficult to analyze systematically.

This is an important development because the clinical meaning of an observation can depend heavily on context. A laboratory value recorded after a medication change, a symptom that appeared following an infection, or a change in weight following a major lifestyle transition cannot always be interpreted appropriately without understanding what happened before and after it.

A longitudinal perspective turns medical data into a story rather than a collection of disconnected numbers.

The Rise of Individual Baselines

One of the most important consequences of longitudinal healthcare is the possibility of establishing an individual baseline.

A baseline represents the range of measurements that is typical for a particular person during a particular period. It does not necessarily mean that the baseline is permanently fixed. Human physiology changes with age, environment, lifestyle, illness, medication and many other factors.

The purpose is therefore not to create a single permanent “normal” value for each person. Instead, healthcare could establish a dynamic expectation that evolves as the person changes.

This approach could be particularly valuable for chronic diseases. A patient’s baseline glucose pattern, blood pressure profile, kidney function, cognitive performance or physical activity level may provide information that is difficult to obtain from one measurement alone.

Research into individual variability is also becoming more sophisticated statistically. A 2026 study on longitudinal health indicators highlighted the importance of within-individual variability and noted that simple measures such as an individual’s standard deviation may not adequately capture how variability changes over time.

The implication is that healthcare may increasingly need to understand not only average levels but also the structure, timing and direction of change.

Multi-Omics Makes Individual Trajectories More Complex

The movement toward individual trajectories is not limited to traditional clinical measurements. Modern biomedical research increasingly examines multiple layers of biology, including genomics, transcriptomics, proteomics, metabolomics and microbiome data.

These measurements can reveal that two people who appear similar according to conventional clinical variables may have very different biological profiles.

Longitudinal multi-omics research makes the concept of individual trajectories even more interesting. A recent study of an aging cohort followed gene-expression and metabolomic changes over eight years and found that individual molecular trajectories could diverge from population-level trends. The study also reported that longitudinal changes were influenced by genetics, circadian rhythms, seasonality and environmental exposures.

This suggests that biological aging and health cannot always be understood as a single universal curve.

Instead, individuals may move through different biological pathways while still sharing broad population patterns. One person’s metabolic profile may change gradually, while another may experience a sharper transition. One person’s inflammatory markers may remain relatively stable, while another person’s trajectory may change following environmental or lifestyle events.

The challenge for healthcare is to distinguish meaningful biological variation from signals that genuinely indicate increased risk.

Artificial Intelligence Is Making Trajectory-Based Healthcare More Practical

Artificial intelligence is becoming an important tool for analyzing longitudinal health information because patient records are complex, irregular and multidimensional.

A traditional statistical model might rely on a limited number of variables measured at predefined time points. Modern machine-learning systems can potentially integrate repeated measurements, clinical notes, medications, imaging, laboratory tests, genomic information and other data sources.

A 2026 review in Nature Reviews Genetics described the growing integration of artificial intelligence with genomics and electronic health records, emphasizing the potential of multimodal data to improve understanding of disease heterogeneity and individualized risk prediction.

Another 2026 study explored disease trajectories in more than 7.2 million people with multiple long-term conditions, using representations of patients’ disease histories to investigate patterns of progression. The research also highlighted an important challenge: even when large datasets can identify statistical clusters of trajectories, translating those clusters into clinically meaningful categories remains difficult.

This illustrates both the promise and the limitation of AI. Algorithms can discover patterns that would be difficult to identify manually, but identifying a pattern does not automatically establish that it is clinically meaningful.

Redefining Normal Does Not Mean Abandoning Population Medicine

There is a risk of framing individualized medicine and population medicine as opposing approaches. In reality, they are complementary.

Population data provide the reference framework. They help determine what kinds of changes are associated with disease risk across large groups. Individual trajectories provide context about how a particular person differs from those broader patterns.

Recent work on clinically trustworthy digital twins illustrates this relationship. A 2026 framework proposed that patient-specific models should remain calibrated against population-level evidence rather than operating independently from epidemiological knowledge. In this view, population data act as an empirical reference system that constrains individualized predictions and helps prevent unrealistic or poorly transferable conclusions.

This creates a more balanced model of personalized healthcare. The question is not whether the population or the individual is more important. The question is how population knowledge can be used to interpret an individual’s evolving trajectory.

A useful healthcare system may therefore operate on two levels simultaneously: population evidence establishes the broader context, while personal history determines how that context applies to the individual.

From Disease Thresholds to Early Signals

One of the most significant potential applications of trajectory-based healthcare is earlier detection.

Many diseases do not suddenly appear at the moment a measurement crosses a diagnostic threshold. Biological changes may develop gradually for months or years before conventional criteria are met.

If healthcare focuses only on whether a person is currently above or below a threshold, some early changes may remain invisible.

Trajectory-based systems could instead ask whether several measurements are moving in a concerning direction. A modest change in multiple related indicators may become more informative when viewed together over time.

This concept is already being explored in areas such as cognitive health. A 2026 research framework called PRISM used individualized longitudinal forecasting to estimate expected cognitive trajectories and identify deviations from personal baselines. The study evaluated more than 30,000 adults and externally validated the approach in another cohort, illustrating how personalized forecasting could complement conventional population norms.

Such systems are not replacements for clinical diagnosis. Rather, they represent a potential monitoring layer that could identify situations where additional clinical assessment may be appropriate.

Biological Age and the Problem of a Universal Aging Curve

Aging provides another powerful example of why individual trajectories matter.

Chronological age is simple to measure, but it does not fully capture the biological differences between people of the same age. Two individuals who are both 60 may have different cardiovascular health, metabolic function, immune profiles, physical capacity and molecular characteristics.

Research into biological age attempts to capture some of these differences. A 2026 study using imaging-derived measures of biological age across multiple organs emphasized that chronological age does not fully represent individual aging trajectories and explored associations between imaging-derived biological age and health outcomes.

However, biological-age models also demonstrate the difficulty of defining a new “normal.” A biological-age score is itself a model based on reference data. It can provide useful information, but it should not automatically be interpreted as an absolute measurement of how healthy or unhealthy someone is.

The future may therefore move beyond asking, “What is your biological age?” toward questions such as, “How is your biological trajectory changing, and which components are changing fastest?”

The Importance of Context

Individual trajectories cannot be interpreted through numbers alone.

A change in a biomarker could reflect disease, medication, temporary illness, sleep disruption, exercise, diet, stress, environmental exposure or measurement variability. Without context, an algorithm may mistake normal adaptation for pathology.

This is one reason clinical records remain valuable. The 2026 Nature Medicine longitudinal patient-journey research emphasized that unstructured clinical notes contain information about symptoms, treatment rationale, adverse effects, adherence challenges and other details that structured fields may not capture.

The next generation of healthcare may therefore need to combine quantitative measurements with contextual information. The goal would not be to collect everything possible, but to understand which information helps explain why a trajectory is changing.

Privacy Becomes More Important as Healthcare Becomes More Personal

The move toward individual trajectories also creates a major data-governance challenge.

A population statistic can be relatively anonymous. A detailed personal health trajectory is fundamentally different. It may contain years of laboratory measurements, medications, diagnoses, genetic information, behavioral patterns, wearable data and clinical narratives.

The more healthcare systems attempt to understand individuals continuously, the more important questions about consent, security, data ownership, interoperability and secondary use become.

Personalized healthcare should therefore not mean unrestricted collection of personal information. Trust will be essential. Patients need meaningful understanding of what information is collected, why it is used, who can access it and how it contributes to clinical decisions.

The quality of individualized medicine will ultimately depend not only on better algorithms but also on responsible data practices.

Toward a Dynamic Definition of Normal

The most important conceptual change may be that “normal” itself becomes more dynamic.

Instead of treating normal as a fixed interval, healthcare could increasingly understand it as a combination of population expectations, individual baselines and longitudinal change.

A measurement can be normal for the population but unusual for the individual. Conversely, a measurement can fall outside a population range without necessarily indicating that something is wrong for that particular person.

This does not eliminate the value of established clinical thresholds. Thresholds remain essential for diagnosis, treatment decisions and public-health research. But they may become one component of a broader interpretive framework.

The future definition of normal could therefore resemble a moving landscape rather than a fixed line. Healthcare would consider where an individual currently stands, how that position compares with relevant populations, how it has changed over time and what other biological or contextual signals are changing alongside it.

The Future of Healthcare May Be Trajectory-Aware

The transition from population averages to individual trajectories represents a broader transformation in how healthcare understands human health.

Population medicine gave healthcare the ability to identify common patterns and establish evidence-based standards. Precision medicine is now adding another layer: the ability to understand variation between individuals and increasingly to follow change within individuals.

Electronic health records, wearable sensors, multi-omics, artificial intelligence and longitudinal research are making this possible at a scale that was previously difficult to achieve. The emergence of computable patient journeys and increasingly sophisticated trajectory models demonstrates how healthcare data can move from isolated observations toward continuous representations of health over time.

Yet the objective should not be to replace population medicine with personalized algorithms. The more useful direction is integration. Population evidence can establish what is generally expected, while individual trajectories can reveal what is changing within a particular person.

This distinction could become increasingly important as healthcare shifts from reactive treatment toward prevention and early intervention.

Conclusion

Healthcare has historically relied on population averages because they provide a practical and scientifically rigorous foundation for understanding disease and defining clinical reference ranges. But individuals do not experience health as averages. They experience it as a continuously changing biological and behavioural trajectory.

The growing availability of longitudinal clinical records, multi-omics measurements, digital health data and artificial intelligence is creating new opportunities to understand those trajectories. Recent research demonstrates that individual patterns can diverge from population trends, that patient histories can be reconstructed across time, and that personalized models may identify meaningful changes that conventional population thresholds overlook.

The emerging model is therefore not about choosing between population medicine and personalized medicine. It is about connecting them.

Population averages can tell healthcare what is common. Individual trajectories can reveal what is changing. The combination can provide a richer understanding of what “normal” means for each person and when a deviation may deserve attention.

As medicine becomes increasingly data-driven, the most important question may no longer be simply whether a patient’s latest measurement is normal. It may be whether that measurement makes sense within the story of that person’s health—and whether the direction of that story is changing.

Online Internship with Certificate