|
SKYAID |
Heart Rate data was used to detect Sleep
Apnea in 2000 and Paroxysmal Atrial Fibrillation in 2002. added
12/29/01 There are contests with a $500 prize for
groups to analyze data posted at www.physionet.org.
The contests are run by George Moody at MIT and funded by the US National Institutes
for Health. There are 50 gigabytes
of data currently posted - covering ECG, brainwaves, breathing patterns, oxygen
concentrations, blood pressures, etc. The Sleep Apnea contest in 2000
attracted 15 groups, three of which were not associated with medicine.
The wining team from Ireland looked at 150 different signal aspects. The 2001 contest was of more interest to
Skyaid, Paroxysmal Atrial Fibrillation.
The top scores were 82% accuracy and 79% for prediction. Goldberg talks of future possible predictive signal process algorithm which would provide minutes or hours of prediction. "A real-time monitor could alert its wearer or a doctor that something was amiss, and that they should take precautions or seek medical care" The contests are in conjunction with the
annual convention: Computers in Cardiology. The 2002 contest will be
announced in Feb. 2002. Details on both of these contests
are in the Dec 15, 2001 issue of New Scientist. (not on
line as of 12/29/01) |