Aura is defined as a transient, focal neurological syndrome that precedes or accompanies migraine attacks. Its symptoms are fully reversible and develop gradually over minutes, typically resolving within an hour. Visual disturbances dominate the clinical spectrum: scintillating scotomas, fortification spectra, photopsias, blurred vision, and transient visual field defects that slowly spread across the visual field. Somatosensory aura usually presents as unilateral paresthesia or numbness beginning in the hand and spreading to the arm, sometimes involving the tongue and face on the same side. Language disturbances occur less frequently and often accompany visual or somatosensory symptoms. Migraine with aura affects roughly 25 to 30 percent of people with migraine, and migraine itself touches more than one billion individuals worldwide, making aura one of the most common recurring neurological phenomena on the planet.

What makes aura biologically extraordinary is its orderly temporal progression, which closely mirrors the propagation characteristics of cortical spreading depolarization, or CSD. CSD is a slowly propagating wave of neuronal and glial depolarization followed by prolonged suppression of cortical activity, and it is widely regarded as the principal neurobiological substrate of migraine aura. A classical neuroimaging study demonstrated a progressively spreading BOLD signal through the occipital cortex accompanying visual scintillations, followed by reduced BOLD response corresponding to the perceived scotoma, providing direct evidence linking CSD to human visual aura. Yet the precise relationship between CSD and the diverse clinical manifestations of aura remains incompletely understood, particularly for non-visual aura and for the considerable variability observed between patients.

Here lies the measurement challenge that the authors place at the center of their argument. Clinical assessment continues to rely predominantly on retrospective symptom descriptions obtained after attacks have resolved, typically hours or days later. Patients are asked to reconstruct the onset, progression, duration, and sequence of symptoms from memory, and although many individuals describe their aura with remarkable consistency, prospective studies have demonstrated considerable variability between attacks while recall is inherently influenced by bias. Clinical interviews also less frequently capture the behavioral adaptations, environmental influences, or concurrent physiological changes that accompany aura. Meanwhile, mechanistic approaches such as neuroimaging, electrophysiology, and experimental models are constrained by aura’s unpredictable onset, relatively brief duration, and the practical limitations of clinic- and laboratory-based assessment. The result is a conceptual gap in which neurological symptoms, cortical activity, autonomic regulation, behavior, cognition, and environmental context are investigated separately, using different methodologies and often different study populations.

Digital phenotyping, originally defined as the moment-by-moment quantification of the individual-level human phenotype in situ using data from personal digital devices, has evolved into a broader methodological framework for characterizing health and disease through continuous real-world data. The authors emphasize that it is not a specific technology or device but a framework integrating two complementary data streams. Active sensing refers to information intentionally provided by individuals, including symptom reports, electronic diaries, questionnaires, and ecological momentary assessment, or EMA, which enables repeated assessment of symptoms, cognition, behavior, and context close to their occurrence. Passive sensing involves the unobtrusive collection of physiological, behavioral, or environmental information through connected devices without requiring continuous user input, such as activity patterns, heart rate, heart rate variability, sleep characteristics, ambient light exposure, and device interaction. The principal strength, the authors insist, lies not in generating larger volumes of data but in preserving the temporal relationships among multiple dimensions of health as they evolve over time.

The conceptual heart of the article is a four-domain framework for dynamic phenotyping of aura. The first domain is the symptom dimension, reflecting the patient’s perceived symptom onset, progression, propagation, and resolution. The second is the physiological dimension, encompassing measurable biological responses, including autonomic changes that accompany the evolving attack. The third is the behavioral dimension, describing changes in daily activities, functioning, movement, cognition, or adaptive behaviors, since patients frequently reduce physical activity, avoid bright light, interrupt cognitive tasks, or seek quiet surroundings as symptoms evolve. The fourth is the contextual dimension, incorporating environmental and situational factors such as ambient light, sleep-wake context, and stress that may influence, accompany, or modify the aura experience. These domains were selected because they represent complementary aspects of aura that can realistically be repeatedly characterized during spontaneous attacks in everyday life and subsequently integrated into a unified phenotype.

Operationally, the authors sketch a prospective research model rather than a validated protocol. An initial interictal monitoring period would establish individualized baselines for selected physiological, behavioral, and contextual measures. At perceived aura onset, an event-triggered protocol would initiate time-stamped symptom reporting and mapping of symptom progression, accompanied by intensified or annotated acquisition of wearable- and smartphone-derived measures across the relevant domains. Monitoring would continue through aura resolution and the subsequent headache and postdromal phases to characterize recovery and the temporal relationships among data streams. Repeating this process across multiple attacks within the same individual would enable comparison with the interictal baseline and help distinguish reproducible within-person patterns from attack-specific variability. Crucially, patient-reported observations remain indispensable within this design: aura is one of the few neurological phenomena in which patients can recognize and describe symptom evolution with considerable temporal precision, providing the anchor against which passive data can be interpreted.

The translational opportunities are substantial, though the authors are careful to frame them as prospective directions rather than established capabilities. Improved phenotypic resolution could reveal previously unrecognized patterns of aura expression, facilitate identification of clinically meaningful subgroups, and improve patient stratification, which matters because migraine with aura carries an approximately two-fold increased relative risk of ischemic stroke, particularly among younger women, smokers, and users of estrogen-containing contraceptives, even though the absolute risk remains low. Longitudinal observations collected during naturally occurring attacks cannot directly measure CSD, but they may generate testable hypotheses about the biological processes accompanying aura and inform the prioritization of future mechanistic studies. Dynamic phenotyping could also enhance therapeutic evaluation by assessing treatment effects across multiple dimensions, including aura symptom evolution, physiological responses, functional performance, behavioral adaptation, and recovery, and it aligns with the ongoing transition toward precision headache medicine.

Significant challenges temper the enthusiasm. Digital features derived from smartphones and wearables are indirect measures of physiological and behavioral processes, not direct measurements of CSD, and the authors state explicitly that until validation evidence is available, digital features should not be interpreted as proposed surrogates of CSD. Changes in physical activity, heart rate, device interaction, or ambient light exposure may reflect behavioral adaptation, headache-related pain, anxiety, medication use, or environmental circumstances rather than aura-specific biology, so temporal association with an aura episode is insufficient to establish biological specificity. The marked heterogeneity of aura complicates the identification of reproducible digital signatures, suggesting that individualized longitudinal characterization may ultimately prove more informative than attempts to identify a universal digital biomarker. Practical implementation raises concerns about participant adherence, device usability, data completeness, and study feasibility, since long-term monitoring may be necessary to capture sufficient attacks. The diversity of digital platforms highlights the need for standardized acquisition protocols, harmonized data-processing pipelines, and independent validation.

Ethical, regulatory, and governance considerations weigh equally heavily, particularly as artificial intelligence becomes integrated into digital phenotyping frameworks. Continuous collection of physiological, behavioral, and contextual information inevitably involves sensitive personal data, and the authors call for robust frameworks addressing informed consent, privacy, cybersecurity, data governance, algorithmic transparency, fairness, and equitable access from the earliest stages of research. They point to the V3 framework of Verification, Analytical Validation, and Clinical Validation proposed by the Digital Medicine Society as a useful foundation for evaluating digital measures before clinical implementation. Ultimately, the authors conclude, digital phenotyping should complement rather than replace established diagnostic criteria, clinical evaluation, and mechanistic investigation. Its value will depend not on the volume of data collected but on its ability to generate clinically meaningful information that improves phenotypic characterization, advances understanding of migraine aura, and supports more informed clinical decision-making, with future progress hinging on interdisciplinary collaboration spanning neuroscience, headache medicine, digital health, engineering, and data science.

Subject of Research: Digital phenotyping of migraine aura

Article Title: Digital phenotyping of migraine aura

Article References: Gazerani, P., Petrušić, I., & DosSantos, M. F. (2026). Digital phenotyping of migraine aura. Frontiers in Neurology, 17, Article 1967778. https://doi.org/10.3389/fneur.2026.1967778

Image Credits: AI Generated

DOI: 10.3389/fneur.2026.1967778

Keywords: Digital, phenotyping, migraine, aura, scientific research, peer-reviewed research