Diagnostic testing in modern medicine remains largely episodic and threshold-based. Epidosidic can be defined as tests taken at discrete points of time where the result is viewed in that moment of time in the context of globally applied standards, without an analysis of the previous trend, which is indicative of the future trajectory of the biomarker, and health, of the patient. Threshold-based can be defined as a biomarker value which is categorised into black or white, good or bad, with little appreciation of the change over time, or grey-zones, which represent an opportunity to intervene to prevent disease. 

Biomarkers such as Prostate Specific Antigen (PSA) and Glycosylated Heamoglobin (HbA1c) are typically interpreted against fixed cut-off values, and action is initiated once these thresholds are crossed. This approach, while embedded in clinical guidelines and workflow protocols, does not reflect the underlying biology of progressive diseases. It reduces dynamic processes into binary states (normal or abnormal, healthy or diseased) which misses a critical window for early intervention.

This static model is particularly inadequate in men’s health, where conditions such as prostate cancer and type 2 diabetes often evolve insidiously and present at a late stage. By the time diagnostic thresholds are reached, disease burden is often established, multi-organ pathophysiology has advanced, and prevention is no longer viable. A shift towards trajectory-based diagnostics, where longitudinal trends in biomarker levels are used to anticipate risk, offers a more clinically rational and personalised approach. It opens the door to earlier detection, stratified screening intervals, and timely intervention, particularly in at-risk but asymptomatic men. Which reduces the burden of disease for both patients and health systems. Let’s not forget one of the three key shifts in the NHS is the movement from disease treatment to disease prevention. 

The Limitations of Point-in-Time Testing

Consider PSA in the context of prostate cancer. A single PSA measurement provides limited information about a patient’s risk unless interpreted within the broader context of age, family history, prostate volume, and, crucially, previous PSA values. A PSA level of 3.9 ng/mL may be considered “normal” by conventional thresholds, yet if this represents a doubling from prior values over two years, it may indicate a significant risk of clinically relevant disease. Conversely, a stable PSA of 4.1 ng/mL over several years in a low-risk individual may not warrant immediate further investigation. This example helps to demonstrate how we can not only increase the sensitivity of results, but also specificity, therefore reducing unnecessary follow up testing. 

A similar logic applies to HbA1c where some work has been done to reduce the binary approach, by introducing an intermediate state of prediabetes, or non-diabetic hyperglycaemia (NDH), but this single intermediate state isn’t enough to reflect the complex, progressive and multi-organ pathophysiology of insulin resistance, abdominal obesity, or metabolic syndrome. The opportunity to intervene is greatest when examining changes over time. 

These examples illustrate the failure of fixed thresholds to reflect temporal dynamics and risk gradients. In both cases, disease is not an event but a process, and by waiting for a formal diagnosis, we may lose our best chance to prevent it.

Building a Dynamic Diagnostic Framework

To address this gap, diagnostics must evolve to include the rate and direction of change in biomarkers, not just their absolute values. Implementing such a framework requires coordinated developments in clinical infrastructure, data analytics, clinician education and patient engagement. Key components include:

Increased Frequency of Testing

Intermittent, low-frequency testing provides insufficient resolution to detect gradual pathological shifts. Advances in remote testing—such as self-collected capillary blood samples—enable safe, scalable, and cost-effective monitoring from the patient’s home. Quarterly or biannual biomarker assessments could become routine in high-risk individuals, reducing reliance on opportunistic in-person blood draws. This is another example of the alignment with one of the shifts in the NHS, from hospital to community. 

Trend Analysis and Temporal Pattern Recognition

Longitudinal interpretation of biomarker data allows for the identification of risk trajectories. For example, PSA velocity and PSA doubling time have both been associated with prostate cancer risk and aggressiveness. In glycaemic control, early rises in HbA1c, even within the prediabetic, or NDH range, can signal deterioration in insulin sensitivity. Trajectory modelling provides a richer risk profile than spot-checks alone.

Adaptive Risk Scores and Machine Learning

Artificial intelligence can operationalise this approach at scale. Adaptive risk models can integrate serial biomarker values, demographic variables, lifestyle data, and digital phenotypes to generate continuously updating risk estimates. These models may identify non-linear patterns indicative of pathophysiological change. A man with a flat PSA trend over five years may remain low-risk; one with a rising but still sub-threshold pattern may benefit from earlier imaging, increased testing or specialist referral.

Importantly, these tools can also adjust screening intervals in real time, lengthening them in stable, low-risk individuals and shortening them in those showing accelerating biomarker changes. This offers a path to more efficient, personalised screening protocols that allocate resources where they are most clinically warranted; ultimately saving money for patients and healthcare systems such as the NHS.

Making This a Reality for Men’s Health

All of these are opportunities, but they will only happen if access to testing is increased, guidelines are updated and, the often forgotten factor, clinicians are engaged in service design from day zero. We can’t simply add this to the worklist of the average GP. A clear strategy that enables easy interpretation of results, with automatic flagging of patients who have results which demonstrate a need for increased testing, through AI or simple calculations of data in the health record, is essential for making this a reality. Most importantly the patient needs to be engaged and encouraged through educating them on the value of undertaking participating in testing, and reducing barriers to do so. 

Ultimately, this is a move from static to dynamic medicine. It reflects a deeper understanding of disease as a continuum and demands a healthcare system capable of responding not only to states, but to signals of change.