Learning from experience is inherent to animals and humans and when used in computer models it is termed as Machine learning (ML) which was coined by Arthur Samuel the pioneer of computer gaming and artificial intelligence in 1959. This field grew out during the search for artificial intelligence and initially was developed using neural networks, perceptrons, probabilistic reasoning and generalized linear models of statistics. ML works by either of the two methods, supervised learning or unsupervised learning. Search for “ML in neurosurgery” in Pubmed showed 308 results. There were 298 studies with abstracts, 5 clinical trials, 20 review articles and 168 articles in human studies. Of these around 113 articles were either studies of ML in other parts of the body like liver, stroke, EEG, pathology and Parkinsons disease or not involving ML and hence were excluded. Of the 55 remaining cases the majority were studies done in glioma followed by medical imaging in neurosurgery, radiotherapy, language and learning studies. ML will definitely replace many of the cumbersome physical data collection to infer and formulate ways to treat patients in the future. It can make the process of research accumulation, filter, find correlations between variables and help to make algorithms to manage and predict, that can save, time, money and speedup the recovery of the patient
Machine learning in social epidemiology: Learning from experience
