A Digital Biomarker for Food Noise: Why the Field Needs More Than a Questionnaire

"Food noise" made the jump from TikTok to formal science for one reason: people starting GLP-1 receptor agonists (GLP-1 RAs) kept reporting that the constant, intrusive preoccupation with food had gone quiet. That anecdote, repeated at scale, forced the field to do something it had never done — actually measure the thing. The response was a wave of self-report scales, led by the Food Noise Questionnaire (FNQ).

A validated questionnaire is a real achievement. But a questionnaire and an objective, in-the-moment digital biomarker are not the same instrument, and the difference matters for anyone trying to turn "food noise" from a marketing claim into a clinical endpoint. This piece makes the case that food noise — and urge more broadly — is ready for a digital biomarker, and shows how an active-task, wearable-sensor approach (Habit Bandz) could fill that gap, with the questionnaire serving as the validation criterion rather than the ceiling.

What is food noise?

Food noise is the experience of persistent, intrusive, hard-to-control thoughts about food — constant mental chatter about the next meal, cravings, and snacking that interferes with daily life and makes healthy behavior harder. It is closely tied to established constructs the obesity and eating-behavior literature already uses: craving, food cue reactivity (the mind and body reacting to internal and external food-associated signals), and incentive salience (cues acquiring an outsized motivational "pull").

What changed recently is not the phenomenon but the attention. When GLP-1 RAs appeared to dial food noise down, the construct acquired commercial and scientific stakes — and the field discovered it had no agreed-upon way to quantify it.

How food noise is measured today: the rise of the FNQ

The current state of the art is self-report. The Food Noise Questionnaire (FNQ), developed and validated by researchers at Pennington Biomedical Research Center and published in Obesity in 2025, is a brief five-item instrument scored 0–20, where higher totals mean more frequent and intrusive food thoughts. It posts strong psychometrics — excellent internal consistency and good test–retest reliability — which is exactly why it has become the de facto "ruler" for the construct.

It is already doing real work. At the 2026 European Congress on Obesity, an observational cohort using the FNQ reported that people who added a GLP-1 RA to a behavioral weight-management program saw a substantially larger one-month drop in food noise than those on behavioral treatment alone (an adjusted decrease of roughly 4 points versus about 1). Researchers even suggested that a rapid early reduction in food noise might serve as an early indicator of treatment response.

The FNQ is not alone. A parallel scale (the RAID-FN family, associated with David Allison's group and commercialized through telehealth channels) is competing to define and own the measure. The pattern across the landscape is consistent: several drug makers and weight-management programs are racing to characterize food noise, and whoever's scale gets adopted becomes the standard everyone else reports against.

The measurement gap: what a questionnaire can't do

Here is the uncomfortable part. Every instrument above is retrospective self-report, and self-report has well-documented limits the obesity and eating-behavior fields already acknowledge in their own EMA literature:

  • Recall bias. People disproportionately remember recent and emotionally salient moments, so a "how has your food noise been?" rating is a reconstruction, not a measurement.

  • Cross-sectional and retrospective design. Many high-profile food-noise findings ask people already on a drug to recall their "before." That captures a remembered delta, not a tracked trajectory.

  • Reactivity and social desirability. The act of self-monitoring changes behavior and answers.

  • It requires introspection. The person has to stop, notice, and rate — precisely the effortful self-observation that's hardest during an actual urge.

A questionnaire gives you a before-and-after number. It cannot give you the shape of an urge over time, it can't run in the background of daily life, and it can't be captured "independent of raters." That's why food-noise claims currently live as patient-reported outcomes and marketing talking points rather than as objective endpoints in a protocol — not because nobody wants the endpoint, but because the objective measure doesn't yet exist.

That gap is a digital biomarker–shaped hole.

What is a digital biomarker — and a digital endpoint?

A digital biomarker is an objective, quantifiable, physiological or behavioral characteristic collected and measured through a digital device — a wearable, smartphone, or sensor — and turned into an interpretable outcome by an algorithm. The defining properties are the ones self-report lacks: it is objective, largely independent of raters, and captured in daily life rather than at a single clinic visit or survey prompt, which reduces recall bias and catches change a snapshot misses.

Two distinctions from the digital-medicine field are worth borrowing directly, because this audience uses them:

  • A biomarker is not an endpoint. A digital biomarker becomes a digital endpoint only once it is validated and prespecified as a trial outcome. Getting the label right signals seriousness.

  • Validation follows the V3 framework (from the Digital Medicine Society): verification (does the sensor capture what it claims?), analytical validation (does the algorithm reliably compute the intended measure?), and clinical validation (does the measure correspond to a clinical or experiential state people care about?). The FDA/NIH BEST framework provides the surrounding vocabulary.

In short: the field measuring food noise has a mature construct, a validated questionnaire, and no sensor-based measure. The digital-biomarker field has a rigorous validation playbook and is hungry for high-value behavioral targets. Food noise is a natural meeting point.

Habit Bandz: a digital biomarker in food noise research

Habit Bandz (HBZ) is built to do exactly what a questionnaire can't: produce an objective, in-the-moment read on how a person's urge actually moves. Rather than asking someone to stop and rate a remembered feeling, it derives a sensor-based measure of urge as it rises and settles during a short, self-guided session and returns it as a single interpretable score. That profile — objective, momentary, rater-independent, and captured in daily life rather than at a survey prompt — is the definition of a digital biomarker, and it's precisely the part of the food-noise picture no self-report scale can supply.

The value isn't just that the measure is objective; it's what objectivity unlocks:

  • It captures the trajectory of an urge — how it builds and subsides — not just a remembered before-and-after delta.

  • It carries no recall bias and little response burden, because it doesn't depend on introspection in the moment.

  • It runs in real life, close to where food cues actually fire, giving an ecologically valid signal a clinic visit or monthly survey misses.

  • It can act as an early, objective response indicator — the kind of momentary signal the questionnaire literature already suspects predicts who responds to treatment.

Why this matters for GLP-1 and obesity research

GLP-1 RAs created the demand: they quieted food noise, and everyone wanted to quantify by how much. Self-report anchored the first chapter, but trials, payers, and pharma increasingly want objective, ecologically valid endpoints that don't depend on a patient's recalled introspection. A validated digital biomarker for urge could:

  • Act as an early, objective treatment-response indicator, complementing the "rapid food-noise reduction predicts response" signal the questionnaire studies already hint at.

  • Capture the time-course and shape of urge within and across days — not just a before/after delta.

  • Reduce recall bias and travel into daily life, where food cues actually fire.

  • Generalize beyond obesity: the same machinery applies to substance craving, impulsivity, and attention, since it measures the underlying inhibitory-control and arousal signature rather than "food" specifically.

Food noise is arguably one of the most consequential, least-measured comorbidities in the obesity space — talked about everywhere, sitting inside almost no protocol as an objective endpoint. The organization that pairs the validated questionnaire with a validated digital biomarker doesn't just describe the phenomenon; it measures it.

References‍ ‍