How Precision Medicine Uses Genomics, Biomarkers, and Health Data to Guide Care

How Precision Medicine Uses Genomics, Biomarkers, and Health Data to Guide Care

Precision medicine aims to guide healthcare decisions using characteristics that differ among individuals, including genetic variation, molecular measurements, medical history, lifestyle, and environmental exposures. Rather than assuming that every person with the same diagnosis will respond identically, it seeks to identify clinically meaningful subgroups and match them with more appropriate tests, treatments, or prevention strategies.

This approach does not mean that every patient receives a completely unique therapy. In practice, precision medicine usually places people into increasingly specific categories—for example, patients whose tumors carry a particular molecular alteration or those who metabolize a medicine differently because of an inherited variant.

Its value depends on more than collecting large amounts of data. A genomic result, biomarker measurement, or algorithm is useful only when it is accurate, relevant to the clinical question, and linked to an action supported by evidence.

The three main information layers

Precision medicine commonly combines three overlapping types of information:

1. Genomic data, which describe inherited or acquired variation in DNA.
2. Biomarkers, which are measurable indicators of a biological state, disease process, or response to an intervention.
3. Clinical and population health data, such as diagnoses, laboratory results, imaging, medications, family history, exposures, and outcomes.

Each layer provides a different perspective. Genomics can help identify biological mechanisms or inherited susceptibility. Biomarkers can show what is happening in the body at a particular time. Longitudinal health data provide context, including how symptoms developed, which treatments were tried, and what outcomes followed.

The strongest applications generally use these forms of evidence together rather than treating any single measurement as definitive.

How genomics contributes to care

Genomics examines some or all of a person’s DNA. Depending on the clinical question, testing may focus on one gene, a panel of genes, the protein-coding regions of the genome, or nearly the entire genome.

Germline and somatic variants

A central distinction is whether a genetic variant is germline or somatic.

• Germline variants are present from conception and can usually be found throughout the body. Some influence inherited disease risk, drug response, or the likelihood of passing a condition to biological relatives.
• Somatic variants arise in particular cells during a person’s lifetime. Cancer cells, for example, may acquire DNA changes that are absent from the rest of the body.

This distinction affects interpretation. A variant detected in a tumor may help guide cancer treatment without necessarily indicating an inherited cancer predisposition. Conversely, some tumor findings raise the possibility of a germline variant and may lead to separate testing of non-tumor tissue.

Diagnostic applications

Genomic testing can help establish or refine a diagnosis when symptoms suggest a genetic disorder. It is particularly relevant when a condition is rare, affects multiple organ systems, or could result from variants in many different genes.

Possible testing strategies include:

• Single-gene testing when one gene is strongly suspected
• Gene panels for conditions with overlapping clinical features
• Exome sequencing, which focuses mainly on protein-coding regions
• Genome sequencing, which examines a broader range of DNA variation
• Chromosomal or copy-number testing for larger gains, losses, or rearrangements

A result may identify a pathogenic or likely pathogenic variant that fits the person’s clinical features. Testing can also produce a variant of uncertain significance, meaning that available evidence is insufficient to classify the change as disease-causing or benign. Such a finding generally does not carry the same clinical weight as a confirmed pathogenic variant.

A negative result does not always exclude a genetic cause. The relevant gene may be unknown, the variant may be difficult for the test to detect, or the condition may arise from a combination of genetic and non-genetic factors. As scientific knowledge develops, laboratories and clinical teams may revisit previously inconclusive data.

Genomics in cancer treatment

Cancer genomics is among the most established areas of precision medicine. Tumor profiling can identify alterations associated with particular therapies, resistance mechanisms, prognosis, or eligibility for a clinical trial.

Examples include testing for:

• HER2 overexpression or amplification in certain breast and gastric cancers
• EGFR variants in some non-small cell lung cancers
• BRAF variants in melanoma and several other tumor types
• BRCA1 or BRCA2 alterations in selected breast, ovarian, pancreatic, and prostate cancers
• Mismatch repair deficiency or microsatellite instability, which can help predict response to particular immunotherapies in some settings

The presence of an alteration does not automatically mean that a targeted treatment will work. Its significance may depend on the cancer type, disease stage, prior therapies, co-occurring alterations, and strength of clinical evidence. The same molecular change can have different implications in different tissues.

Pharmacogenomics

Pharmacogenomics examines how inherited genetic variation affects medication response. Variants can influence how quickly a drug is metabolized, whether it reaches an effective concentration, or whether it causes particular adverse effects.

Established applications include testing related to:

• TPMT and NUDT15 before or during use of certain thiopurine medicines
• HLA-B*57:01 before treatment with abacavir
• CYP2C19 in selected situations involving medicines such as clopidogrel
• CYP2D6 and CYP2C19 for some drugs whose metabolism varies substantially among individuals

Clinical usefulness differs by medication and context. A gene–drug association may be biologically plausible without having enough evidence to justify routine testing. Recommendations can also vary among regulators, professional groups, and healthcare systems.

Genetic risk prediction

Some inherited variants have large effects on disease risk. Others contribute small effects that can be combined into a polygenic risk score. These scores estimate relative genetic susceptibility based on many variants.

Polygenic scores are being studied for conditions such as coronary artery disease, diabetes, and several cancers. However, their performance depends on the populations used to develop and validate them. Scores derived primarily from one ancestry group may be less accurate in others. Their role in routine care remains more limited than that of well-validated single-gene findings in specific clinical settings.

What biomarkers reveal

A biomarker is a measurable characteristic that indicates a biological process, disease state, or response to an exposure or treatment. Biomarkers are not limited to genetic information. They can include proteins, metabolites, cell counts, physiological measurements, imaging features, or molecular signals from tissue and blood.

Major biomarker categories

A biomarker can serve more than one role, but the categories are not interchangeable. A marker associated with poor prognosis, for example, does not necessarily predict that a particular treatment will be effective.

Tissue and blood-based biomarkers

Many biomarkers are measured in tissue collected through surgery or biopsy. Tissue can provide direct information about cell type, architecture, proteins, gene expression, and DNA alterations. However, a single sample may not capture all the biological diversity within a tumor or across metastatic sites.

Blood-based testing can measure circulating proteins, cells, metabolites, or fragments of cell-free DNA. In oncology, analysis of tumor-derived DNA in blood is often called a liquid biopsy. Depending on the setting, it may help identify targetable alterations, assess residual disease, or monitor emerging resistance.

Liquid biopsy has limitations. A tumor may release too little DNA into the bloodstream for reliable detection, and a negative result may not rule out an alteration. Some DNA variants detected in blood can also originate from age-related changes in blood-forming cells rather than from a solid tumor. Clinical interpretation therefore depends on the test, disease stage, sample quality, and reason for testing.

Analytical and clinical validation

Before a biomarker can guide care, several questions must be addressed:

• Analytical validity: Does the test measure the biomarker accurately and reproducibly?
• Clinical validity: Is the biomarker reliably associated with the disease, outcome, or treatment response?
• Clinical utility: Does using the result improve decisions or outcomes compared with existing care?
• Practical utility: Can the test be delivered quickly, affordably, and consistently enough to affect care?

A statistically significant association is not sufficient by itself. The test must perform well in the population and setting in which it will be used.

Why health data provide essential context

Molecular data become more informative when connected with a person’s health history. Electronic health records and other clinical systems can contribute:

• Symptoms and physical findings
• Diagnoses and disease stage
• Laboratory trends
• Imaging results
• Medication use and adverse reactions
• Procedures and hospitalizations
• Family history
• Age, sex, reproductive history, and ancestry
• Lifestyle and environmental exposures
• Treatment response and long-term outcomes

Longitudinal data can reveal patterns that a single laboratory result cannot. For example, a genetic variant associated with cardiomyopathy is interpreted differently in a person with abnormal cardiac imaging and a strong family history than in someone without related clinical findings.

Phenotyping: defining the condition accurately

Precision medicine depends on careful phenotyping, the structured description of observable traits and clinical features. Poorly defined diagnoses can weaken research and lead to misleading associations.

Phenotyping may combine clinician assessments, laboratory values, imaging, pathology, patient-reported outcomes, and data from wearable devices. Natural language processing and other computational tools can extract information from clinical notes, but automated results require validation because records may be incomplete, inconsistent, or shaped by billing and documentation practices.

Real-world data

Data collected during routine care can help researchers study how tests and treatments perform outside tightly controlled clinical trials. These sources may include electronic health records, insurance claims, patient registries, pharmacy systems, and digital health technologies.

Real-world data can cover larger and more varied populations, but they also contain biases. Treatment selection is not random, follow-up may be uneven, and important variables may be missing. Statistical adjustment can reduce some problems but cannot always eliminate them. Randomized clinical trials therefore remain important for determining whether an intervention causes better outcomes.

From data to a clinical decision

A precision-medicine pathway typically involves several steps:

1. Define the clinical question

Testing should address a specific issue, such as clarifying a diagnosis, selecting among therapies, estimating inherited risk, or monitoring disease.

2. Select the appropriate specimen and test

The sample might be blood, saliva, tumor tissue, bone marrow, or another biological material. The assay must be able to detect the relevant type of variation or biomarker.

3. Assess test quality

Laboratories evaluate factors such as sample adequacy, sequencing coverage, detection limits, contamination, and reproducibility. These details affect whether a negative or low-level result is reliable.

4. Interpret the result

Interpretation integrates laboratory findings with the medical literature, clinical guidelines, disease features, family history, and prior treatments. Multidisciplinary teams may include laboratory specialists, pathologists, pharmacists, genetic counselors, bioinformaticians, and relevant physicians.

5. Link the result to evidence

The strongest results are connected to validated diagnostic criteria or interventions supported by clinical studies. Other findings may be exploratory or useful mainly for research.

6. Monitor outcomes and revise

Diseases evolve, new biomarkers emerge, and variant classifications can change. In cancer, treatment pressure may select resistant cell populations. In inherited disease, new evidence may clarify a previously uncertain finding. Precision care is therefore often iterative rather than a one-time test.

The role of computational tools and artificial intelligence

Modern sequencing, imaging, and health records generate more information than clinicians can review manually. Computational systems help align DNA sequences, identify variants, classify images, search medical knowledge bases, and estimate risks.

Machine-learning models can identify complex patterns across molecular and clinical data, but performance in development does not guarantee effectiveness in practice. Important concerns include:

• Training data that do not represent the intended population
• Changes in laboratory methods or clinical workflows
• Hidden correlations that do not reflect biology
• Limited transparency about how predictions are generated
• Declining performance when a model is moved to another institution
• Automation bias, in which users place excessive confidence in an output

Clinically useful systems require external validation, ongoing performance monitoring, clear accountability, and evidence that using the model improves decisions rather than merely predicting an outcome.

Challenges that limit precision medicine

Biological complexity

Many common diseases reflect interactions among multiple genes, behavior, aging, social conditions, and environmental exposures. A molecular result may alter probability without determining an outcome. Even strongly associated variants can show incomplete penetrance, meaning that not everyone who carries them develops the related condition.

Tumor heterogeneity

Cancer can vary within a single tumor and between different sites in the same patient. A biopsy samples only part of this diversity. Tumors also evolve over time, especially under treatment.

Incidental and uncertain findings

Broad genomic testing may identify results unrelated to the original reason for testing. It can also reveal uncertain variants or information relevant to biological relatives. Policies for analysis, consent, disclosure, and follow-up differ according to the test and healthcare setting.

Unequal representation

Genomic databases and clinical studies have not represented all ancestry groups equally. This can increase the number of uncertain findings in underrepresented populations and reduce the accuracy of some risk models.

Access is also shaped by insurance coverage, geography, specialist availability, digital infrastructure, language, and trust in healthcare institutions. A technology cannot improve population health equitably if only a narrow group can obtain and benefit from it.

Privacy and data governance

Genomic information is identifying, persistent, and potentially informative about relatives. Linking it to health records increases research value but also raises concerns about unauthorized access, secondary use, commercial use, and re-identification.

Responsible programs use measures such as restricted access, encryption, data minimization, transparent consent processes, audit systems, and defined rules for sharing. Governance must address not only technical security but also who controls data and how communities participate in decisions about its use.

What successful precision medicine looks like

The most mature applications of precision medicine share several characteristics:

• They begin with a clear clinical question.
• The test is analytically reliable.
• The result has been validated in an appropriate population.
• There is a defined action linked to the finding.
• Benefits and harms have been evaluated.
• Results arrive in time to influence care.
• Patients and clinicians can understand the implications.
• Outcomes are monitored after implementation.
• Access and performance are assessed across different populations.

Precision medicine is therefore not simply the use of sequencing or advanced algorithms. It is a framework for connecting biological measurements with clinical context and credible evidence. Genomics can identify inherited or acquired variation, biomarkers can describe disease state and treatment response, and longitudinal health data can show how those findings relate to a person’s actual course of illness.

When these elements are integrated carefully, they can improve diagnostic precision, identify clinically relevant subgroups, guide selected therapies, and support more informative monitoring. Their value, however, rests on rigorous validation, thoughtful interpretation, secure data practices, and evidence that data-driven decisions produce outcomes that matter to patients.