To wrap up our discussion on the topic of machine learning in healthcare 101, I'd like to address both its drawbacks and its prospects for the future. Everything has its ups and downs. Machine learning is not an exception. While machine learning can be a valuable asset for automatically organizing and analyzing big data in electronic medical records (EMR) or electronic health records (EHR), should the current workforce be completely replaced by the beautiful and powerful tool asset? Upon reading some concepts of bias and weights, there is nothing wrong to assume machine learning can be applied to healthcare with complete guidance and control. However, J. Gu and D. Oelke have cautioned that bias in datasets can result in unfairness and discrimination based on multiple factors such as gender, racism, etc. [1] In order to gain a deeper understanding of the issues mentioned above, I would like to address some general fundamental concepts about data population.

A machine learning model may have several features, but some feature might have a higher impact on the output than others. For example, if a model is predicting apartment prices, the locality of the apartment might have a higher impact on the output than the number of floors the apartment building has. Hence, we come up with the concept of weights. Each feature is associated with a weight (a number) i.e. the higher the feature has an impact on the output, the larger the weight associated with it. But how do you decide what weight should be assigned to each feature?