From subjective assessment to data-driven diagnosis: the AI revolution in multimodal early detection of Sjögren's disease.

Cardiovascular diseases and type 2 diabetes remain leading global causes of morbidity and mortality. Cardiovascular diseases alone account for more than 17 million deaths annually []. Traditional risk factors, including hypertension, dyslipidemia, and obesity, explain a substantial proportion of this burden. However, they do not fully account for inter-individual variation in cardiometabolic outcomes. Diet and physical activity are well-established modifiable risk factors [,]. However, their effects are heterogeneous. Individuals with similar macronutrient distributions and total daily activity levels may show markedly different vascular and glycemic outcomes. One critical dimension remains comparatively overlooked: the temporal organization of eating, physical activity, and rest within the 24 h cycle. Emerging evidence suggests that the timing and regularity of these behaviors may influence cardiometabolic health independently of their average levels. In this review, we propose rhythm coherence as a hypothesis-generating framework for examining whether the temporal stability and alignment of eating, physical activity, and sleep may be relevant to vascular and glycemic health. It is not presented as an established determinant or validated clinical construct. The “average-value dilemma” captures a fundamental limitation of traditional nutrition and exercise epidemiology. Individuals with similar total daily energy intake or comparable daily moderate-to-vigorous physical activity (MVPA) may experience divergent cardiometabolic outcomes. Landmark studies have shown that inter-individual glycemic responses to standardized meals may vary five-fold or more []. Traditional metabolic markers explain only part of this variability. Meal timing and regularity explain additional variance in glycemic responses beyond total energy intake. This finding suggests that the temporal structure of eating may constitute a distinct risk dimension []. Similarly, compositional analyses of 24 h activity patterns may predict arterial stiffness more effectively than total MVPA alone []. These analyses account for the relative distribution of sedentary time, light physical activity, and MVPA across the day. These observations support a conceptual shift from how much a person eats or exercises to when, how regularly, and in what temporal relation these behaviors occur. The author-proposed rhythm coherence framework organizes this perspective for future testing. It hypothesizes that the alignment of eating, physical activity, and sleep rhythms may provide information beyond their average levels; this incremental and causal value remains to be established. The convergence of wearable technologies and continuous monitoring devices has transformed the assessment of behavioral and physiological rhythms under free-living conditions. Continuous glucose monitoring (CGM) captures real-time glucose dynamics at 1–5 min resolution. Wrist actigraphy characterizes activity and sleep patterns over periods of seven days or longer. Digital meal logging through image recognition or ecological momentary assessment records meal timing and frequency. Ambulatory blood pressure monitors and cuffless wearables track 24 h cardiovascular dynamics. Emerging flexible biosensors can also measure sweat electrolytes, metabolic substrates, and stress-related hormones. Collectively, these tools enable multimodal digital phenotyping, defined here as the simultaneous measurement of eating, physical activity, sleep, and physiological rhythms within the same individual over weeks to months []. This transition from snapshot assessments, such as fasting glucose and office blood pressure, to continuous behavioral and physiological data streams has revealed previously hidden temporal patterns. These patterns include meal jitter, defined as day-to-day variability in eating times, as well as activity fragmentation and sleep–wake irregularity. All are increasingly quantifiable and potentially modifiable. The rhythm coherence framework leverages these observational windows to examine whether integrated analyses of multimodal behavioral rhythms can improve the prediction of cardiometabolic outcomes beyond isolated metrics.