Related Experiment Videos
Early prediction of unexplained syncope by support vector machines
Daniel Schang1, Mathieu Feuilloy, Guy Plantier
1ESEO, 4, Rue Merlet de la Boulaye, 49009 Angers Cedex, France. daniel.schang@eseo.fr
Physiological Measurement
|January 24, 2007
Summary
A new method using electrocardiogram and impedance data predicts unexplained syncope, potentially shortening the head-upright tilt test (HUTT). This approach offers a faster, effective alternative for diagnosing syncope, improving patient care.
Area of Science:
- Cardiology
- Biomedical Engineering
- Medical Diagnostics
Background:
- Unexplained syncope diagnosis relies on the head-upright tilt test (HUTT), a lengthy procedure.
- The HUTT can take up to 55 minutes, posing a challenge for patients who do not faint.
- There is a need for more efficient and accurate methods to predict syncope events.
Purpose of the Study:
- To develop and assess a novel method for predicting unexplained syncope.
- To reduce the reliance on the prolonged head-upright tilt test (HUTT).
Main Methods:
- Recorded electrocardiogram and transthoracic impedance waveforms for 10 minutes in 128 patients with recurrent syncope.
- Computed seven indices from impedance data and its derivative, combined with age and sex.
- Utilized a support vector machine for prediction analysis in retrospective and prospective patient groups.
Main Results:
- A retrospective analysis achieved 100% sensitivity and 100% specificity using a selected subset of variables.
- Prospective evaluation of the best subset yielded 94% sensitivity and 79% specificity.
- The developed method shows promising results compared to existing syncope detection techniques.
Conclusions:
- The new method offers a potentially shorter and effective alternative for predicting unexplained syncope.
- This approach could significantly improve the diagnostic process for patients experiencing recurrent syncope.
- Further validation may lead to a reduced need for lengthy diagnostic tests like the HUTT.