The Strength of Weak Learnability
Machine Learning
Machine Learning
Hi-index | 0.00 |
X-ray film mammography and physical examination of the breast are the mainstays for early detection of breast cancer. Unfortunately, error rates for mammograms read by radiologists are high. We examine a particularly difficult to read series of 1618 mammograms where in order to achieve a false positive rate lower than 50%, the false negative rate of radiologists is nearly 25%. We examine a variety of automatic data mining tools in an attempt to improve the accuracy of the diagnosis. Our results suggest that roughly the same or higher accuracy rate than the radiologists can be attained at a much reduced cost. This potential cost savings could have a major financial impact for health care in developing nations.