Machine Learning (ML): The Beginner's Guide

Machine learning’s growing popularity is primarily due to an increase in data availability (Big data) and advancements in technology.

Faster machines and smarter algorithms are implemented daily, even in surgery.

Surgeons may increasingly find themselves in the dilemma of acting on the clinical information provided by an algorithm without understanding how it works, that's why it is important to be educated about machine learning.

What is Machine Learning?

Machine learning (ML) is a field of computer science, that allows computers to “learn” without being explicitly programmed.

Machine learning allows computers to automatically infer patterns from data without explicitly being told what these patterns are.

ML can find patterns that humans can not see, make predictions, and form connections between the vast amount of data, thus ML needs a lot of data to understand and complete its tasks (Big data).

How does Machine Learning work?

The task of making computers learn can be broken down into 7 major steps :

01. Collecting data:

Machines initially learn from the data you give them.

02. Preparing the data:

You put all the data together, you clean it, you visualize it, and you split it into a training set and a testing set.

03. Choosing a model:

You choose a model relevant to the task at hand (speech recognition, image recognition, prediction, etc.).

04. Training the model:

Training is the most important step in machine learning. You pass the prepared data to your ML model to find patterns and make predictions.

The model learns from the data and accomplishes the task set.

Over time, with training, the model gets better at predicting.

05. Evaluating the model:

You test the performance of the model on unseen data (testing data, not trained) to measure how your model will perform and its speed.

06. Parameter tuning:

You adjust the values of parameters present in the model.

07. Making predictions:

You use the model to make predictions on unseen data.

Data collected must come from a reliable source, as it will directly affect the outcome of the model. Good data is relevant and contains very few missing and repeated values.

What are the 4 Categories of Machine Learning?

01. Supervised learning

It is used when we can define the task (we know the result) we want the algorithm to learn on data that we already have.

02. Unsupervised learning

Unsupervised learning is a machine learning model that learns patterns based on unlabeled data. Unlike supervised learning, the result is not known ahead of time.

03. Semi-supervised learning

In semi-supervised learning, a result is known, but the algorithm must figure out how to organize and structure the data to achieve the desired results.

04. Reinforcement learning

Reinforcement learning is a machine learning model that can be broadly described as “learning by doing”.

An “agent” learns to perform a defined task by trial and error until its performance is within a desirable range.

The agent receives positive reinforcement when it performs the task well and negative reinforcement when it performs poorly.

What is Deep Learning?

Deep learning, which is a subcategory of machine learning, provides AI with the ability to mimic a human brain’s neural network, to learn and make decisions without the need for explicit programming.

Artificial neural networks are inspired by the structure and function of the human brain and are made up of interconnected “neurons” that process and transmit information.

Machine learning has much potential in surgery, from preoperative planning, perioperative guidance, surgical robotics, complications prediction, and skill analysis.