The Missing Half of Precision Health: Why Your Risk Score Is Only Half the Story

By Dan Rolls, Founder & CEO, GistMD

Precision health is rapidly getting better at knowing.

Genomics, biomarkers, predictive models, and increasingly sophisticated risk-stratification algorithms can tell us more about an individual patient than ever before.

But knowing is not the same as doing.

A patient can have a highly precise risk score, a perfectly selected screening recommendation, or an evidence-based treatment plan and still not complete the colonoscopy, take the medication correctly, change the behavior, or understand why watchful waiting may be the right decision.

That gap between precision prediction and patient action may be one of the most important unresolved problems in precision health.

A recently proposed Precision Health Taxonomy, published in The Lancet, makes this gap particularly visible.

The framework organizes precision health into precision prevention and precision medicine, ultimately leading to nine terminal “Approach” nodes. What caught my attention was not only what the taxonomy contains, but what appears to be missing between its layers.

We have increasingly sophisticated technologies for determining what should happen next.

We have far fewer technologies designed specifically to ensure that the individual patient can actually make it happen.

The Pharmacogenomics of Behavior

Consider how precision medicine approaches drugs.

We do not simply identify a disease and prescribe the same treatment to everyone. Pharmacogenomics can help translate an individual's biological characteristics into a more appropriate therapeutic choice.

Behavioral interventions deserve similar rigor.

A patient’s ability to act on medical guidance depends on profoundly individual variables: language, health literacy, medications, clinical condition, anxiety, cultural context, treatment protocol, and even the exact instructions of their physician.

That suggests a useful parallel:

Pharmacogenomics is to targeted drugs what deterministic personalization could become for behavioral intervention.

Instead of tailoring a drug to biology, we can tailor the communication and behavioral intervention to the person who must carry it out.

At GistMD, this is the problem we have been working on: using a deterministic personalization engine to assemble clinician-approved content into an intervention specific to an individual patient and clinical protocol.

Not generative advice. Not another generic educational library.

A personalized behavioral dose.

The Thinnest Line in Medicine

One element of the taxonomy particularly caught my attention.

Within precision prevention, Risk Prediction and Risk Mitigation are connected by an arrow.

Conceptually, an enormous amount happens inside that arrow.

Imagine identifying a patient as high risk for colorectal cancer.

The prediction itself does not prevent cancer.

The patient must understand the recommendation, schedule the colonoscopy, obtain the preparation, follow dietary restrictions, manage medications appropriately, complete the bowel preparation, arrange transportation, and actually arrive for the procedure.

Only then can prediction become prevention.

We have invested enormously in improving the accuracy of the left side of that equation.

The behavioral infrastructure connecting it to the right side remains surprisingly primitive.

From Self-Reported Behavior to Behavioral Telemetry

There is another opportunity here.

Behavioral research has traditionally depended heavily on self-reporting: questionnaires, interviews, diaries, and patient recollection.

These tools are valuable, but they inevitably introduce recall and social-desirability bias.

Digital delivery creates the possibility of something different: objective behavioral telemetry generated naturally as part of care.

When personalized interventions are delivered through SMS, WhatsApp, or QR codes without requiring an app download, researchers can observe signals such as:

• whether the intervention was opened;

• which segments were viewed;

• how long patients engaged with specific content;

• whether the intervention was completed;

• where engagement stopped;

• whether specific prompts resulted in subsequent actions.

This does not eliminate the need for patient-reported outcomes. But it adds a behavioral data layer that can help researchers understand not simply whether an intervention worked, but how patients interacted with it.

Sometimes the Hardest Prescription Is “No Treatment”

Precision medicine creates another communication challenge that receives surprisingly little attention: no treatment.

Active surveillance and watchful waiting may be clinically appropriate, but psychologically they are anything but passive.

A patient who hears “we are not treating this now” needs to understand why non-intervention is appropriate, what is being monitored, what changes should trigger action, and why waiting does not mean being abandoned.

This is where concepts such as decision quality become particularly important.

Precision medicine should not only optimize the clinical decision. It should help patients understand and participate in that decision.

In some situations, the communication burden of not intervening may be greater than the burden of prescribing something.

Equity Must Be Part of the Delivery Architecture

Precision health also carries an uncomfortable risk.

The populations most likely to benefit from sophisticated genomic and predictive technologies are often those already best positioned to navigate healthcare.

If the final behavioral intervention requires strong English proficiency, high health literacy, portal access, an app download, or confidence navigating a complex healthcare system, precision health can unintentionally amplify existing disparities.

That is why I believe:

Equity should not only be measured in subgroup analysis. It should be engineered into the delivery mechanism.

For us, that means designing interventions from the beginning to work across languages, literacy levels, and levels of digital sophistication, while keeping patients within the same underlying clinical protocol.

The objective should not be to create a separate “equity intervention.”

It should be to make the primary intervention equitable by design.

Early Evidence: Can Personalized Delivery Change Action?

We are beginning to see evidence that personalized behavioral delivery can affect measurable clinical behavior.

In a retrospective cohort of 2,447 colonoscopy patients, personalized animated preparation guidance was associated with a 37% reduction in inadequate bowel preparation (OR 0.629, 95% CI 0.489–0.809; p<0.001).

In pediatric pre-anesthesia, research published in JMIR Formative Research found that personalized animated education substantially reduced the need for pre-anesthesia evaluations while achieving high patient engagement.

Across additional implementations, we have also observed improvements in screening scheduling and patient activation.

These findings are encouraging, but they raise a more interesting scientific question than whether patients simply “like video.”

What actually causes the behavioral effect?

Is it personalization?

Improved comprehension?

Reduced anxiety?

Better procedural clarity?

Language concordance?

Timing?

Repetition?

Or some combination of these factors?

That is where I believe the next generation of research becomes particularly interesting.

From Informed Patients to Activated Patients

Precision health has made extraordinary progress in determining who is at risk, what is likely to happen, and which intervention is most appropriate.

The next frontier may be ensuring that those increasingly precise decisions survive contact with the real world.

That requires treating behavioral delivery as part of the precision-health architecture itself.

The scientific question is no longer simply:

Can we give patients better information?

It is:

Can precision-tailored behavioral delivery reliably convert risk prediction into patient action, and can we identify the mechanisms responsible for that change?

Until we answer that question, an important part of the precision-health equation remains a black box.

We may be approaching an era of extraordinarily precise medicine.

But medicine cannot truly be precise if the final step between knowing what should happen and helping a human being make it happen remains generic.

Acknowledgment

Special thanks to Dr. Sivan Spitzer, organizational sociologist and Principal Investigator of the HEAL – Health Equity Advancement Lab, whose work and thinking on health equity helped inform this perspective.

Reference

The Precision Health Taxonomy, The Lancet, 2026.