Atrial fibrillation is among the most consequential arrhythmias in cardiovascular medicine, not merely because of its prevalence, but because of how consistently it slips through the cracks of routine clinical recognition. Millions of cases are detected late, monitored inadequately, or misclassified during episodes of poor signal quality or atypical presentation. The standard electrocardiographic approach, built on two pillars—absent P waves and irregular RR intervals, has served clinicians well for decades. But the arrhythmia itself is richer than these two features suggest, and the ECG, read carefully, may offer more than we have traditionally extracted from it.
For years, the clinical focus in AF detection has been almost entirely on rhythm recognition: is the rate irregular, and are P waves absent? These are valid and important observations. Yet AF is not purely an electrical disorder. It is a hemodynamic one. Every beat in AF occurs under a different set of mechanical conditions, shaped by the unpredictable variation in diastolic filling that accompanies chaotic atrial activity. That variability has consequences visible on the surface ECG, if one knows where to look.
The Mechanical Consequence Hidden in Plain Sight
When atrial contraction is lost, and RR intervals become irregular, the time available for ventricular filling changes from beat to beat in an entirely unpredictable way. A short preceding cycle means less filling time; a longer one allows more. The Frank-Starling mechanism ensures that each of these differently filled ventricles contracts with a different force and ejects a different stroke volume. These are not trivial differences. They manifest in measurable changes in blood pressure, pulse amplitude, and critically in the electrical vectors that shape the QRS complex on the surface ECG.
The relationship between ventricular chamber volume and QRS amplitude has a well-established physiological basis, described through the Brody effect. A more filled ventricle produces a subtly but measurably different electrical signal than a less filled one. During a normal sinus rhythm, where filling is relatively consistent, this effect produces minimal visible variation in QRS amplitude from beat to beat. During AF, where filling is continuously and irregularly varying, the effect accumulates into a characteristic pattern: irregular, non-periodic, beat-to-beat fluctuation in QRS voltage that tracks the hemodynamic chaos of the arrhythmia itself.
This is the basis of what I have termed the Shora AF Sign specifically, its first component (SS1): beat-to-beat QRS voltage variability as a direct electrocardiographic expression of the variable ventricular filling that defines AF physiology. Unlike electrical alternans, which follow a regular two-beat alternating pattern and typically reflect a different physiological mechanism, the Shora AF Sign is inherently irregular, mirroring the stochastic nature of the arrhythmia that produces it.
What the Sign Looks Like in Practice
The visual recognition of SS1 is straightforward for a clinician familiar with the concept. Across a rhythm strip or standard 12-lead ECG recorded during AF, consecutive QRS complexes appear to vary subtly but appreciably in their peak amplitude. The variation is not the gradual respiratory undulation familiar from sinus rhythm, nor the regular alternation of electrical alternans. It is irregular and continuous, rising and falling across beats in a pattern that mirrors the arrhythmia's own unpredictability.
In practice, this observation can be made visually at the bedside with a standard ECG and confirmed through simple quantitative measurements: peak-to-peak amplitude comparison across consecutive beats, coefficient of variation calculations applied to QRS height, or consecutive-beat voltage difference analysis. None of these requires specialized equipment. Any standard 12-lead ECG is sufficient.
A second component of the Shora AF Sign—SS2—describes a distinct but related observation: an unstable, chaotic isoelectric baseline that appears during AF and resolves upon successful cardioversion, by any method. SS2 is particularly valuable as a real-time marker of rhythm restoration, allowing clinicians to recognize the return of organized electrical activity not only through the reappearance of discrete P waves but through the stabilization of the baseline itself. Notably, SS2 disappears reliably following successful cardioversion, while SS1 may persist in attenuated form, reflecting the lingering hemodynamic adjustment after rhythm restoration.
Validation: From Bedside to Database
The Shora AF Sign has been evaluated in a prospective observational cohort of 400 patients presenting with documented cardiac arrhythmias. In this clinical population, beat-to-beat QRS voltage variability (SS1) demonstrated a sensitivity of 91% and a specificity of 83.5% for AF identification, with statistical significance at p < 0.000001. Importantly, the marker correctly identified several AF cases that were difficult or ambiguous to classify on the basis of RR irregularity alone — precisely the clinical scenarios where an additional diagnostic dimension is most valuable.
To assess reproducibility beyond a single clinical dataset, the marker was also evaluated using the MIT-BIH Atrial Fibrillation Database, a publicly available reference dataset commonly used in arrhythmia research. Computational analysis of this database yielded a sensitivity of 94.1% and an area under the receiver operating characteristic curve of 0.92. The concordance between clinical and computational findings is meaningful: it suggests that the variability pattern detected by the Shora AF Sign is a genuine electrophysiological feature of AF, present across different patient populations and recording systems, rather than a localized observation or artifact.
It is important to acknowledge the limitations of the current evidence base. The clinical validation cohort, while substantive, was drawn from a single center. The computational validation was performed on an established but relatively compact reference database. Interobserver reproducibility, performance across diverse ECG platforms, and behavior in specific subgroups, including paroxysmal AF, structural heart disease, and patients with significant conduction abnormalities, have not yet been formally characterized. These are honest limitations that the research community will need to address through multicenter investigation.
Where This Adds Value Clinically
The question of clinical utility is a fair one. If most AF is already recognized by its irregular rhythm and absent P waves, where does a complementary marker meaningfully improve care?
The answer lies in the cases where standard criteria underperform. In tachycardic AF, extreme rate variability may be difficult to appreciate at high heart rates. In patients with coexisting structural abnormalities or artifacts, P-wave assessment can be unreliable. In ambulatory and wearable monitoring environments where single-lead recordings under variable conditions are the norm voltage variability may offer a more robust and continuously measurable signal than rhythm-based criteria alone. These settings are increasingly important as continuous cardiac monitoring expands beyond hospital walls into consumer devices and remote care platforms.
There is also a role in cardioversion monitoring. The behavior of SS1 and SS2 around cardioversion provides a real-time ECG signature of rhythm transition that goes beyond the simple question of whether a P wave has reappeared. For clinicians performing or supervising electrical or pharmacological cardioversion, the resolution of SS2 and the attenuation of SS1 offer an additional layer of confidence that rhythm restoration has genuinely occurred and that the hemodynamic consequences of AF have begun to normalize.
Finally, as machine-learning approaches to ECG interpretation continue to mature, QRS voltage variability represents a potentially valuable input feature. Current automated AF detection algorithms are predominantly built on RR interval analysis. Incorporating amplitude variability as a complementary signal stream may improve performance in precisely the challenging edge cases where current algorithms are most likely to fail.
Toward a More Complete Reading of the AF Electrocardiogram
The electrocardiogram has always rewarded careful observation. Physiologically grounded and mechanistically coherent features tend to prove clinically durable not as replacements for established criteria, but as refinements that extend our diagnostic reach. Beat-to-beat QRS voltage variability during AF is exactly this kind of observation: it arises from well-understood physiology, it is detectable with standard tools, and it offers information about the arrhythmia that RR analysis alone does not provide.
The Shora AF Sign does not ask clinicians to abandon what they know about AF. It asks them to look one step further to read the QRS complex not only as evidence of ventricular depolarization but as a mechanical record of how that depolarization was prepared. The variability visible across those complexes during AF is not noise. It is a signal.
The next chapter for this work lies in multicenter validation: larger cohorts, diverse patient populations, varied recording environments, and formal assessment of interobserver agreement. The provisional patent is registered (USPTO No. 63/826,826), and computational validation has been established using internationally recognized reference data. The clinical and scientific community is now invited to build on this foundation to test, refine, and if the evidence supports it, integrate this marker into the expanding toolkit of atrial fibrillation diagnostics.
References
1. Brody DA. A theoretical analysis of intracavitary blood mass influence on the heart-lead relationship. Circ Res. 1956;4(6):731–738.
2. Goldberger AL, et al. PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals. Circulation. 2000;101(23):e215–e220.
3. January CT, et al. 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation. J Am Coll Cardiol. 2014;64(21):e1–e76.
4. Hindricks G, et al. 2020 ESC Guidelines for the diagnosis and management of atrial fibrillation. Eur Heart J. 2021;42(5):373–498.
5. Trayanova NA, et al. Machine learning in arrhythmia and electrophysiology. Circ Res. 2021;128(4):544–566.
