Feasibility Assessment of Smartphone Camera–Based Atrial Fibrillation Detection Compared With Portable Electrocardiogram: Single-Patient Case Study

Background Atrial fibrillation (AF) affects more than 33 million people globally, with many cases remaining undiagnosed due to the paroxysmal nature of the condition. Current portable AF detection devices require additional hardware, and their cost may limit their accessibility in community screening programs. Facial camera–based remote photoplethysmography (rPPG) is a novel alternative to conventional fingertip-based photoplethysmography (PPG) that enables contact-free cardiac rhythm assessment. Objective This study aimed to assess the technical feasibility of a novel smartphone camera–based facial rPPG AF detection system by comparing its performance with that of a portable electrocardiogram (ECG) device under real-world home monitoring conditions. Methods We conducted a prospective, single-patient volunteer case study involving a patient with diagnosed paroxysmal AF and atrial flutter with variable block over 12 consecutive days. Simultaneous 60-second recordings were scheduled 3 times daily using the electronRx smartphone app and a Kardia portable ECG device to obtain a total of 36 paired recordings. The primary outcome was device agreement in rhythm classification between the 2 systems. Secondary outcomes included assessment of the effect of signal quality on device agreement and analysis of discordant recordings to identify technical limitations. Results Of the 36 scheduled recording sessions, 35 were attempted, yielding 24 successful simultaneous recordings. Of these, 23 paired recordings met the quality criteria, and device agreement was achieved in 21 (91.3%) recordings. The smartphone system detected 4 of the 5 episodes classified as AF by the Kardia device and agreed with 17 of the 18 recordings classified as normal sinus rhythm. A total of 2 discordant cases were identified: 1 false negative associated with atrial flutter presenting as a fixed 3:1 atrioventricular block and 1 false positive related to poor signal quality due to suboptimal lighting conditions. Signal quality significantly affected performance, with 94.7% agreement in recordings of good quality compared with 75% in recordings of poor quality. Conclusions This case study demonstrates the technical feasibility of using a smartphone facial camera–based rPPG rhythm analysis for AF detection. The approach shows promise as an accessible screening tool. However, the single-patient design, data attrition resulting from synchronization and signal quality issues, and the identified technical limitations highlight the need for larger validation studies before clinical application can be considered.