As many as one million Americans suffer a heart attack every year. Even after surviving the initial attack, there can be lingering complications which will lead to death. Doctors do their best to examine patients and prescribe appropriate treatments but there are so many factors that many people who need intensive treatment do not receive it, while those who do receive it may never even use it. Computer scientists at the University of Michigan have analyzed 24-hour electrocardiogram data from 4,557 heart attack victims and found markers to identify the patients in the most danger. The day long monitoring is a normal procedure but doctors do not have the ability to closely examine the massive amount of data. The computers identified three abnormalities which are linked to the patient surviving, and combining the information from all three results in predicting up to 50% more deaths with fewer false positives, potentially saving thousands a year. Collecting the data used in the study is already a standard procedure and will not increase the burden on caregivers and patients or be at an additional cost.