Tempus ECG-Low EF

Tempus ECG-Low EF is a machine learning software that analyzes 12-lead resting ECG recordings to detect signs of low left ventricular ejection fraction (LVEF ≤40%) in patients 40 years and older at risk of heart failure. It provides a binary output (Low LVEF Detected or Not Detected) to support clinical decision-making for further diagnostic evaluation, and is not intended as a stand-alone diagnostic tool or for patient monitoring.

Both the subject device and predicate device are machine learning-based software analyzing 12-lead resting ECG waveforms in digital format from FDA-authorized ECG machines with 500 Hz sampling rate. The subject device uses a locked, proprietary algorithm to generate an uncalibrated risk score converted to binary output; the only notable difference is terminology ('Unclassifiable' vs. 'Error' for unable-to-classify cases), which does not raise different safety or effectiveness questions.

Not stated in this summary. No specific ISO, IEC, or ASTM consensus standards are cited in the provided 510(k) submission.

Substantial equivalence is supported by: (1) identical intended use—both devices analyze resting 12-lead ECGs using machine learning to detect LVEF ≤40% for referral or further evaluation; (2) comparable technological characteristics with proprietary algorithmic differences that do not create new safety or effectiveness concerns; (3) strong clinical performance validation showing 86% sensitivity and 83% specificity on an independent dataset of >15,000 ECGs from four geographically diverse sites, meeting or exceeding predetermined acceptance criteria of 80%; and (4) non-clinical testing (software verification/validation, cybersecurity, human factors) demonstrating appropriate risk mitigation.