Traditional machine learning
Machine learning that uses classical statistical methods and human-selected features on structured data (tables, spreadsheets, databases) to find patterns, classify, and predict. Techniques include clustering and k-nearest-neighbour. Distinct from deep learning, which handles unstructured data.
Two vendors bid to predict 30-day readmissions from the hospital's structured records. One offers a deep learning model, the other a gradient-boosted tree model built on about 40 named inputs such as prior admissions, length of stay and recent sodium and hemoglobin values. On tabular data like this the simpler method often performs as well, and its inputs can be checked one by one against how your own EHR records them. Ask both vendors for results on the same held-out sample of your patients, and weigh how easily each can be monitored once live.
Terms like this come up in real clinical scenarios across the HelloAI courses: bite-sized modules with verifiable certificates. An account takes one minute, no password needed.
Sign in →