Researchers at the Buck Institute for Research on Aging have unveiled a detailed protocol for one of the most intensive attempts yet to watch human biology change in real time. The Temporal Investigation of Multimodal Elements (TIME) Study, described in Frontiers in Digital Health, is an observational, longitudinal study that will enroll up to 150 adults aged 55 and older in stable health and track them with an unusual combination of weekly blood sampling, four-day bursts of near-continuous molecular profiling, standardized physiological challenges, and wearable devices that never stop recording. The study is registered with ClinicalTrials.gov under identifier NCT07107386, and enrollment began on June 15th 2025, with the protocol approved by the Advarra Institutional Review Board. Because this is a study protocol, the article reports planned methods rather than results; no findings about aging biology have yet been generated from the cohort.
The motivation behind the design is a persistent blind spot in biomedical research. Biological systems exhibit dynamic patterns over multiple temporal scales, from minutes to months, and these patterns are poorly captured by conventional cross-sectional studies or longitudinal studies that sample only occasionally. Diurnal variations and underlying continuous rhythms may be critical determinants of health, resilience, and disease risk in aging, yet existing longitudinal studies in older adults lack high-frequency, multimodal measurements that integrate molecular, physiological, and digital health data streams. The study team also points out that many newer molecular profiling technologies, including proteomics, metabolomics, lipidomics, and methylation-based measures, have not been rigorously studied to understand how the time of day, fasting, or exercise impacts their measurements. Meanwhile, digital health technologies offer continuous, non-invasive data collection in real-world settings, but their full predictive utility is hindered by a disconnect from the patient’s underlying biochemical state, a gap the authors attribute not to a lack of measurement capability but to a historical lack of paired, temporally synchronized data.
The structure of the main phase is deliberately intricate. Over an 11-week period, participants complete nine total weekly visits: seven standard low-frequency epoch visits, each consisting of a single fasted biospecimen collection and limited other measures, and two challenge visits that include the normal low-frequency sampling. In addition, participants complete four days of high-frequency epoch sampling on a consecutive Thursday, Friday, Monday and Tuesday, with data collected over roughly 12 hours at each of those visits. The high-frequency days are scheduled at the beginning and end of the working week rather than across a single week, which allows the researchers to examine the impact of changing behaviors such as sleep, diet, or exercise over the weekend. Collecting data from two consecutive days, twice, also means each individual can act as their own biological control across days. Follow-up visits occur at 6 and 12 months after enrollment, and digital health data will be collected throughout the main and follow-up periods.
Each low-frequency visit captures a morning data collection time point at approximately 0800 hours, including a vital sign measurement, venous blood sample, urine sample, retinal scan, saliva sample, buccal swab, skin swab, a vaginal swab for female subjects, cognitive testing, and grip strength testing. Participants also collect stool samples at home in the days before each low-frequency visit. During the high-frequency epochs, data are collected at five time points, at 0800, 1100, 1400, 1700 and 2000 hours, drawing blood from an indwelling intravenous cannula alongside urine, saliva, retinal imaging, facial imaging, cognitive function tests, grip strength, buccal and skin swabs, and a daily stool sample. Between sampling time points, participants undergo vital sign measurements, anthropometrics, self-reported measures, a short physical performance battery, leg press testing, submaximal bicycle testing, and spirometry. At home, participants wear and sync study wearable devices throughout the 12-month study, record their weight twice daily using a smart scale, and interact with an SMS chatbot, nicknamed Bucky, that collects activity tags, food photographs, and questionnaire responses through ordinary text messaging.
Two standardized perturbation challenges form the centerpiece of the design. In an oral glucose tolerance test scheduled between weeks 2 and 5, participants consume a drink containing 75 grams of glucose within five minutes, and venous blood samples are collected at 10, 20, 30, 60, 90, 120 and 180 minutes after administration. Between weeks 6 and 10, participants complete a maximal cardiopulmonary exercise test on a braked bicycle ergometer, riding to volitional exhaustion under a graded workload while oxygen uptake, carbon dioxide production, minute ventilation, respiratory exchange ratio, heart rate, and blood pressure are measured continuously, with blood samples drawn at 0, 15, 30, 60, 90 and 120 minutes after the test ends. The authors argue these challenges are essential because prior work has shown that dynamic metabotypes and molecular network properties revealed by a glucose challenge, or by a bout of exercise, are entirely unobservable in steady-state, resting profiles. The two challenges also probe deliberately orthogonal physiological axes, metabolic and endocrine on one side, cardiopulmonary and autonomic on the other, and follow internationally standardized protocols, allowing direct comparison to existing reference and epidemiological data.
Eligibility criteria are intentionally broad but carefully constructed. Participants must be at least 55 years old, able to attend in-person visits at the Buck Institute in Novato, California, and resident in the Pacific or Mountain time zones. Female participants must be post-menopausal, a choice intended to remove the variability introduced by the menstrual cycle and the menopause transition. Stable chronic diseases are not automatically exclusionary unless judged uncontrolled or clinically active by the study medical officer. However, participants with anemia, coagulopathies, recent blood donation, or poor venous access are excluded, as are shift workers, people who have used sleep medications more than once per week in the three months before screening, and anyone with antibiotic use within three months of the initial visit. Other medications and supplements are permitted if use has been stable for at least three months, and these are recorded and may in fact be of interest during the multiomic analysis. The team notes that a future study should apply the same dense phenomic sampling approach to younger adults, especially women ages 18 to 45.
The analytical plan reflects the discovery-driven nature of the study. The authors state there is currently insufficient data to form specific hypotheses regarding time-of-day or week-to-week dependence for many omic endpoints in older adults, so the protocol is primarily exploratory. Primary statistical models will use linear mixed-effects frameworks, with time parameterized primarily as continuous calendar time, secondarily as clock time, and tertiarily as age, fitting participant-specific random intercepts and slopes. Acute responses during the glucose and exercise challenges will be quantified using baseline-adjusted peak amplitude, net area under the curve above baseline, and recovery time, complemented by functional principal component analysis of full perturbation trajectories. The team anticipates the resulting dataset will comprise on the order of 10,000 unique biological and digital signals at each time point. Importantly, the study is not powered for definitive cross-sectional subgroup comparisons; all subgroup stratifications, machine learning predictions, and secondary biomarker discoveries are explicitly designated as exploratory and hypothesis-generating.
The protocol also confronts its own limitations candidly. Seasonal variation is not explicitly controlled for but is documented and treated as a feature of the dataset. Recruitment is single-site and regional, which limits the immediate generalizability of findings to the specific recruited population, and the extremely high participant burden will unavoidably select for currently unknown lifestyle or background characteristics. The intense monitoring also introduces a risk of measurement reactivity, potentially altering the naturalistic behaviors the study aims to observe, although the study’s duration is expected to facilitate participant habituation. On the safety side, the study was designated minimal risk by the IRB, maximum single-day blood volume is capped at 92 milliliters with a total of roughly 570 milliliters over the 11-week intensive phase, and the maximal exercise test is conducted under continuous ECG and blood pressure monitoring by a physician and staff trained in advanced life support.
Beyond its immediate aims, the team frames TIME as a foundation for a broader research agenda. The dataset is intended to serve as a reference resource that quantifies biological variability across daily and weekly timescales in older adults of stable health, to inform the optimization of future longitudinal study designs, and to help explore the concept of a Minimum Viable Dataset, a resource-optimized subset of measurements and sampling frequencies required to capture primary phenomic variation. Molecular platforms were deliberately chosen for interoperability with major population-scale cohorts such as the UK Biobank and the NIH All of Us Research Program, and a rigorous biobanking strategy preserves blood fractions for future retrospective analysis as higher-resolution technologies emerge. De-identified data will be made available upon reasonable written request, with whole genome sequencing and facial imaging data requiring a formal Data Use Agreement because of re-identification risks. If the study succeeds, the authors write, it will lower the activation energy required to run future temporal profiling studies on other cohorts, including populations and biological patterns the current design deliberately leaves unexplored.
Subject of Research: A protocol for an observational longitudinal study mapping temporal variation in molecular and digital health data in older adults
Article Title: The Temporal Investigation of Multimodal Elements (TIME) Study: protocol for an observational, longitudinal study to characterize the dynamic structure of molecular and digital data in older adults in stable health
Article References: Yurkovich, J. T., Glass, E., Nash, A. N., Levine, N., Lee, S., Ehlen, K., Hernandez, E., Gharti, P., Senadheera, C., Fernando, A., Witherington, D., Pflieger, L., Erram, J., Rappaport, N., Le, A., Newman, J. C., & Stubbs, B. J. (2026). The Temporal Investigation of Multimodal Elements (TIME) Study: protocol for an observational, longitudinal study to characterize the dynamic structure of molecular and digital data in older adults in stable health. Frontiers in Digital Health, 8, Article 1881042. https://doi.org/10.3389/fdgth.2026.1881042
Image Credits: AI Generated
DOI: 10.3389/fdgth.2026.1881042
Keywords: aging, longitudinal study, study protocol, multi-omics, digital health, wearables, glucose tolerance test, exercise testing, biological rhythms, phenomics, older adults, Buck Institute
News Source: Beatrice Stafford. (October 28, 2026). New study will map how the aging body changes hour by hour. Scienmag.
