Frequency-Domain ECG Analysis Enables Single-Encounter Hypertension Staging and Early Myocardial Remodeling Detection: A Prospective 384-Patient Study

Hypertension diagnosis typically requires repeated office blood pressure measurements over multiple visits. This prospective single-center study evaluated whether frequency-domain analysis of a single high-resolution 12-lead electrocardiogram (ECG) could detect hypertension, grade its severity, and identify spectral signatures consistent with early myocardial involvement in treatment-naïve adults. Three hundred and eighty-four consecutive patients underwent clinical evaluation and digital 12-lead ECG recording with the Sydäntek capacitive platform. Both blood pressure measurements and ECG recordings were obtained in the sitting position after 5 minutes of rest, and 15-second recordings were used for analysis. After excluding 43 recordings with inadequate signal quality, we included 341 patients in this study. We used fast Fourier transform analysis of precordial leads V1–V6 to derive a transparent regression-based composite index from centroid frequency, total power, and spectral resonance peak ratio. The index was validated against guideline-defined blood pressure categories using stratified 5-fold cross-validation. The composite index achieved 88% overall accuracy for four-category grading (weighted 𝜅 = 0.868) and a weighted Receiver operating characteristic–area under the curve (ROC-AUC) of 0.978. Binary classification of normal versus any hypertension achieved 95% accuracy, 99% sensitivity, and 89% specificity. Misclassifications occurred only between adjacent severity categories, and lateral precordial leads V4–V6 contributed most to discrimination. Frequency-domain ECG analysis of a single 15-second recording provides an interpretable, scalable method for single-encounter hypertension detection and severity staging. These findings should be interpreted as a single-center pilot evaluation, and we look forward to confirmation by independent multicenter studies. Received: 30 May 2026 | Revised: 3 September 2026 | Accepted: 28 September 2026 Conflicts of Interest Dr Sugandhi Gopal, the principal investigator, served as both system designer and study director; this dual role is disclosed, governance measures minimized bias, and all underlying data are available for review to ensure transparency and reproducibility. The investigator’s primary professional role is as an Interventional cardiologist, with research activities conducted in alignment with that clinical focus. Mr Vishwateja Reddy (research associate in clinical trials design and monitoring) and Mr V. Harish (machine learning engineer) are affiliated with Carditek Medical Devices and contributed to this project in roles specific to signal processing and algorithm development. Their contributions were purely technical in nature and did not influence the study’s clinical design, data interpretation, or outcome analysis. Dr Prabhavathi Bhat, Dr V. J. Karthikeyan, and Shasha Jumbe served as independent reviewers and proofreaders for the manuscript and declared no conflict of interest. All contributors affirm adherence to ethical research practices and confirm that their roles did not introduce bias or commercial influence into the study’s execution, analysis, or reporting. Data Availability Statement Data are available from the corresponding author upon reasonable request. Author Contribution Statement Sugandhi Gopal: Conceptualization, Methodology, Writing – original draft, Visualization, Supervision. Prabhavathi Bhat: Writing – review & editing, Supervision, Project administration. V. J. Karthikeyan: Writing – review & editing, Project administration. Shasha Jumbe: Writing – review & editing, Supervision. Vishwateja Reddy: Investigation, Data curation. V. Harish: Software, Validation, Formal analysis, Data curation.