A single blood draw may soon reveal what time your body thinks it is, not just what the clock on the wall says. In the largest study of its kind, researchers mapped the daily rhythms of more than 3,200 blood biomarkers across nearly half a million people and built a protein-based clock that can estimate internal circadian time with remarkable accuracy.
The findings, posted July 13 as a preprint on medRxiv and not yet peer-reviewed, represent a major step toward a practical blood test for circadian phase, the internal timing system that governs sleep, metabolism, and nearly every biological process.
What they found: A sweeping survey of daily rhythms
The team analyzed 3,228 plasma biomarkers collected from up to 493,493 participants in the UK Biobank, a large-scale biomedical database of adults ages 40 to 70. The biomarkers spanned four categories: proteins, metabolites, standard clinical biochemistry measures, and complete blood counts.
The results were striking. More than half of all biomarkers tested (58 percent, or 1,857 of them) showed statistically significant variation with time of day after correction for multiple testing. Of those, 134 biomarkers had more than 1 percent of their variance explained by sampling time, and 11 exceeded 5 percent.
When the researchers looked specifically for 24-hour oscillatory patterns, they found 1,708 biomarkers with clear circadian rhythms: 98 percent of metabolites, 90 percent of clinical biochemistry measures, 90 percent of hematological traits, and 48 percent of proteins.
The timing of these rhythms clustered around two distinct peaks. Most biomarkers reached their daily high point either in the early morning around 7 a.m. or the early evening around 7 p.m., forming what the authors describe as a dual-peak organization of the human blood circadian landscape.
The proteomic clock: 68 percent of sampling time explained by proteins alone
A machine learning model built from blood protein levels alone predicted collection time with an R-squared of 0.68. Metabolomics achieved 0.42, standard biochemistry reached 0.17, and basic blood cell counts managed 0.15. Combined across all platforms, the R-squared reached 0.71.
The protein clock remained effective even with modest protein counts. About 60 proteins still achieved an R-squared around 0.40. Performance improved with more proteins and plateaued around 1,000, suggesting the circadian signal in the blood proteome is broad and redundant.
The clock was validated externally in two independent cohorts that used constant routine protocols, the gold standard for circadian research in which participants are kept in constant dim light and posture to unmask the internal clock. In the TREASURE cohort, the correlation between predicted and actual acrophase (peak timing) was 0.86. In the Specht cohort, it was 0.89.
Cross-population testing added further confidence. In the China Kadoorie Biobank, a large Chinese cohort, the clock achieved an R-squared of 0.62. In FinnGen, a Finnish biobank, performance dropped to 0.11, which the authors attribute to the narrow blood collection window in that sample.
Circadian Acceleration: A new measure of internal time
Beyond building a clock, the study introduced a novel construct called Circadian Acceleration, or CA. CA quantifies the deviation of an individual’s internal circadian phase from the population average at a given sampling time. A positive CA means the person’s internal clock is running ahead of schedule; a negative CA means it is running behind.
This measure turned out to be remarkably stable over time within individuals. The correlation of CA measurements taken 3.5 years apart was 0.45. At 9.2 years and 12.6 years, it remained at 0.35, suggesting that an individual’s circadian timing is a stable biological trait.
The behavioral correlates lined up with expectations. People who identified as evening types had a CA that was 38 minutes later than morning types. Those who reported sleep inertia (difficulty waking up and feeling groggy after sleep) showed a delayed CA of up to 26 minutes. Seasonal effects appeared as well: CA was about 6 minutes earlier in summer compared to winter.
Perhaps most telling was the effect of the spring daylight saving time transition. In the Monday and Tuesday after the clocks spring forward, CA was delayed by 19 minutes, consistent with the well-known phenomenon of social jet lag.
Shift work and internal misalignment
Consistent night shift workers showed a 22-minute delay in CA compared to day workers. But the more provocative finding involved chronotype and shift work interactions. Morning-type people who worked permanent night shifts showed levels of circadian misalignment comparable to evening-type people who had never worked shifts.
This suggests that forcing a morning chronotype onto a night shift schedule may produce internal disruption equivalent to a natural mismatch between a person’s preferred timing and their actual schedule, a mismatch linked to increased risks of metabolic disease, cardiovascular problems, and certain cancers.
The genetics of internal timing
The heritability of CA was estimated at 10 percent. A genome-wide association study identified 1,825 variants reaching significance, clustering into 20 independent signals. A polygenic score explained 2 to 4 percent of the variance in CA.
Genetically, CA correlated with self-reported chronotype (genetic correlation 0.34). Mendelian randomization suggested a causal effect of chronotype on CA, but not the reverse, supporting the idea that a person’s internal preference shifts their actual circadian timing rather than the other way around.
Why it matters
A practical blood test for circadian phase could transform clinical sleep medicine and large-scale research. Currently, determining a person’s internal timing requires expensive protocols with multiple samples over 24 hours in controlled conditions to measure melatonin, impractical outside specialized centers.
A single-blood-draw test would let clinicians assess circadian disruption in shift workers, screen for circadian rhythm sleep-wake disorders, monitor chronotherapy response in mood disorders, and study circadian timing in metabolic and cardiovascular disease at population scale. That roughly 60 proteins can provide meaningful phase estimates makes the approach more feasible for clinical deployment.
Limits
As a preprint, this work has not yet completed peer review. The UK Biobank cohort is predominantly white, British, and ages 40 to 70, so generalizability to younger people, other ethnic groups, and non-European populations requires further testing. The blood samples in UK Biobank were collected during daytime clinic visits, which limits the observable range of circadian phases and may introduce biases. The proteomic and metabolomic measurements came from a single time point per participant, so the circadian patterns were inferred statistically across the population rather than measured within individuals. The performance drop in FinnGen also highlights that real-world accuracy will depend on the sampling conditions and population.
Bottom line
This study provides the most comprehensive map to date of how the human blood molecular landscape changes across the day and demonstrates that protein measurements alone can estimate internal circadian time with useful accuracy. With further validation and standardization, a circadian blood test could become a practical tool for sleep medicine, chronobiology, and population health research.
Source
“Population-scale molecular reconstruction of human circadian phase from blood biomarkers.” Albiñana, Richmond, Wang, Urpa, Crouse, Zeng, Rosoff, Abdi; FinnGen; Li, Millwood, Ollila, Hickie, Gachon, Kramer, Ray, Wray. medRxiv, July 13, 2026. DOI: 10.64898/2026.07.08.26356418. PMCID: PMC13409268. This is a preprint and has not yet been peer-reviewed.