Automated deep learning analysis has been used to assess abdominal computerised tomography images. Through machine learning, this has enabled more precise measurements of body composition to be made. Through the use of different images, artificial intelligence has been trained to predict major cardiovascular events. This includes such as heart attacks and strokes.
Across healthcare, machine-learning algorithms are being deployed to emulate human cognition for the analysis, interpretation, and comprehension of complicated medical and healthcare data.
The data was drawn from a series of 33,182 abdominal computerised tomography outpatient examinations. These images related to 23,136 patients, who were seen at Partners Healthcare in Boston. By way of image analysis, 12,128 patients were identified as being at risk of major cardiovascular and cancer events. With this patient group, some 1,560 people developed myocardial infarctions (a heart attack as the result of a sudden blockage in a coronary artery arising from the formation of a blood clot) and 938 suffered from strokes.
Other data was factored in, including the age and gender of the patients. The images collected were of the L3 CT slice (this is the third lumbar spine vertebra). The images were assessed against body composition areas. This led to the key finding that the visceral fat area was independently associated with future heart attack and stroke. in contrast, the conventional measure of body mass index was found to be a relatively poor predictor of these diseases occurring.
It is of medical importance that the artificial intelligence has made better predictions of heart issues compared with more traditional approaches, as with making assessments based on overall weight or assessing body mass index.
The technology demonstrates how key medical professions, like radiologists, cardiologists and primary care doctors, can use more accurate and predictive data in order to provide better care to patients.
