Applied Machine Learning for Healthcare
Generating knowledge from big data increasingly requires the use of machine learning for various reasons—cognitive, organizational, technical, and operational. In the 1990s, research on machine learning moved from knowledge-engineering–based expert systems to statistical and data-driven approaches. Overfitting has significant impacts on the performance of the machine learning system. Overfitting happens when a learner mimics random fluctuations, anomalies, and noise in the training dataset, thus adversely impacting the performance of the system on new data. Machine learning applications are rapidly being deployed and used in the commercial space across diverse verticals—retail and e-commerce, government, finance, healthcare, cyber security, transportation, agriculture, space exploration, and manufacturing, among many others. Artificial intelligence, specifically the sub-fields of machine and deep learning, provides optimal and cost-effective options to expand the universe of knowledge and solutions in healthcare. Deep learning has the ability to learn key features of data without explicit programming.