Machine Learning Mania
Machine learning is posited as a transformative force for human health. But what exactly is “Machine Learning” and how can it help human health and medicine?
1959 was the year that the now inescapable phrase “Machine Learning” was coined by Arthur Samuel. Then it was defined as the “ability to learn without being explicitly programmed”. Machine learning when all said and done is in fact just a generalised term that encompasses modern statistical modelling techniques.
In its most basic form machine learning is the practice of utilising algorithms to parse data, learn from it, and then make a prediction about something in the world that the data describes. It is essentially the ability for a machine to acquire knowledge and/or skill from data by itself.
The power of ML depends upon the method and/or methods used. The three most commonly used ML methods are:
Supervised Learning – data is given to a model that includes predetermined answers to the problems of each input set.
Unsupervised Learning – this is where the machine just finds similarities in data (input) without making a prediction (output) e.g. Cluster Analysis
Reinforcement Learning – this is where there is no predetermined decisions or outputs. The learning is based upon the input and the users feedback. The users feedback is either reward or punishment reinforcing the machines learning. It is about taking suitable action to maximise reward in a particular situation, meaning the best solution equates to maximum reward.
Each of the aforementioned methods automatically create models based on the data the user provides. These methods are essentially, algorithms, which act like the engine of the machine. The algorithm/method needed depends upon the kind of problem you are solving, the computer you are using and the nature of the data.
The difference between programming algorithims and machine learning is:
Programming requires the programmer (user) to create a program and manually formulate rules.
Machine learning utilises the algorithim to formulate the rules as a consequence of the data provided.
Deep learning was inspired by the structure and function of the brain, namely the interconnecting of many neurons. Neural Networks are algorithms that mimic the biological structure of the brain.
Think of deep learning as ML on steroids…actually think of it as ML in layers. That is essentially what it is. One layer will possess an algorithm that will process a set of features, whilst the next will process another valuable piece of information. Taken together these pieces of information can help control decisions made by the computer. A good example of this is the self drive feature of a car. One layer would detect the edge of the road, the next detects the lanes, the next the cars when combined the car will simultaneously avoid cars stay on the road and within its lane.
But what about AI (artificial intelligence)…
In fact AI is the general category common to both Deep Learning and Machine Learning. It is, despite all its hype, just any level of intelligence demonstrated by a machine. The result is either an optimal or suboptimal solution to a given problem.
There is a common misconception that AI is in fact a system in its own right. When it is not a system but AI is implemented by a system. In other words, AI is the study of how to train computers/machines to do things a human can do better.
What are they used for?
Machine learning is all around us, from cookies used by web browsers and search engines to photo tagging applications and spam detectors it is everywhere.
Its use in healthcare and science is not new but is growing exponentially with the invent of more and more powerful computers and computer systems. A recent issue of the open source science journal PLOS Medicine dedicated an entire publication to research articles based around machine learning. The editor described the range of articles like so:
“The original articles displayed a broad array of uses that ML will have in medicine including improved diagnosis, predicting disease course (including complications and mortality), and informing population and public health. Hence, there is a mix of population health that attempts to reduce variation, and precision medicine that aims to add back variation at an individual level to determine one’s disease susceptibility, trajectory, and best treatment for each patient.”
Identifying gene coding regions
In the area of genomics, next-generation sequencing has rapidly advanced the field by sequencing a genome in a short time. Thus, an active area machine learning is applied to identifying gene coding regions in a genome. Such gene prediction tools that involve machine learning would be more sensitive than typical homolog based sequence searches.
Structure prediction In proteomics, the study of the structure of proteins. The use of machine learning in structure prediction has pushed the accuracy from 70% to more than 80%. The use of machine learning in text-mining is quite promising with using training sets to identify new or novel drug targets from multiple journal articles and searching secondary databases.
Today, scientists use deep learning algorithms to perform classification of cellular images, genome analysis, drug discovery and also find out how image data and genome data are link with electronic medical records.
AI in healthcare Machine learning and AI are being used extensively by hospitals and health service providers to improve patient satisfaction, deliver personalised treatments, make accurate predictions and enhance the quality of life. It is also being used to make clinical trials more efficient and help speed up the process of drug discovery and delivery.
In conclusion, AI and machine learning are changing the way biologists carry out research, interpret it, and apply it to solve problems. As science grows increasingly interdisciplinary it is only inevitable that biology will continue to borrow from machine learning, or better still, machine learning will lead the way.