Just recently, the world experienced a phenomenal spike of interest in the broad field of Artificial Intelligence and its subcategories. There are a total of 1,543 companies specializing in 13 different sub-fields and are based in roughly 71 countries worldwide! All these AI startups raised approximately $10.01 Billions combined – by promising to bring out disruptive innovations into our daily activities.
Among these companies, few are venturing into introducing Machine learning to the medical field. But before we dive into its benefits and pitfalls in our modern day medicine which will be catered in our next MedVerse articles, we need to initially understand what exactly is “machine learning” and how will medical systems make use of it.
On a classical hue, we usually define the rules for the computer to utilize its computing force so to produce targeted outputs. In software engineering language, it is called as “rule-based decision making” which relies on boolean data. However, with machine learning – machines rules are not mandatory but rather you feed it with data and request them to “learn” the rules. Being aware of this, is by itself revolutionary. Now imagine how the medical world would be when a machine learns “how” to interpret the results from, let’s say, certain radiology images without supervising their learning path. Below there is a diagram that doesn’t do justice as to how complex machine learning methodology is but gives the beginner a good hindsight as to what happens within the computer when they “teach themselves”. We need training examples (Sample dataset) that will feed the model(learning algorithm) by which will take consideration of parameters (hypothesis set) which help the model to output a prediction (hypothesis) that is adjusted by the variable changes in the learner module which will iterate over and over until the most precise conclusion emerge. More about how it works in upcoming arti
Ironically enough, the conclusion from an ML Agent is usually a mere prediction after treating millions of datasets. The reality of how medical practice is taking place will change entirely once computer algorithms start ingesting all the available medical data, recognize patterns among all the subsets then forming a convincingly helpful hypotheses and conclusions for medical decision-making. The power of this technology is almost limitless in a field which relies on a “trained eye” to make a medical decision.
Given that we know that this is such a vast and complicated subject to demote into just one article. In our next articles, we will expand from understanding the concepts behind machine learning, categories of machine learning, learning algorithms, data analysis models and its direct fusion to the medical field via e-health platforms. Stay tuned!