I was recently talking to a friend of mine who had been reading this blog (thanks pal!). He was asking me “why care about machine learning? What about it motivates you? Should we be concerned about machine learning?”. What a great question I thought. I think that could make for an interesting post - or at least more interesting than my limited progress to date with the course materials (I got distracted by holidays. It was great thanks for asking). Before answering my friend’s question I thought I’d take a step back for those less familiar and first explain what is machine learning. Then I will highlight the application of machine learning to medical diagnosis which I hope you’ll find really interesting.
What is machine learning?
To do this I will quote noted Machine Learning (ML) expert Andrew Ng. I have taken this from the introduction to his Coursera course on the subject. He says:
“Machine learning is the science of getting computers to act without being explicitly programmed. In the past decade, machine learning has given us self-driving cars, practical speech recognition, effective web search, and a vastly improved understanding of the human genome. Machine learning is so pervasive today that you probably use it dozens of times a day without knowing it. Many researchers also think it is the best way to make progress towards human-level AI.”
Why care about machine learning?
Sounds pretty cool right? So why care about machine learning? For me the potential to gain understanding and knowledge through ML is incredibly interesting. Machine learning science is founded in probability and statistical analysis. I love being able to solve real problems with maths in a way that feels accessible. For me, getting the computer to do all the leg work for me whilst I sit back with a cup of tea is significantly more appealing to me than spending days in front of blackboards. This is especially true if we can use this application of machine learning to medical diagnosis. I know I keep going on about this and I will get to my summary on it I promise!
With machine learning we have the ability to take in infinity more data than previously possible to map out trends that will increase our understanding of the universe.
The application of machine learning to medical diagnosis
I told you I’d get to it! As I mentioned in this previous post I love problem solving. Machine Learning gives me the opportunity to do this at scale. These problems can be for fun, like in my mission to define success or life changing. As I highlighted earlier we are now seeing the application of machine learning to medical diagnosis. There are an increasing number of medical research centers using ML to better understand disease and as a result new innovative treatments are coming to light. I’ll share a couple of examples with you here:
- UCL in the UK are using ML to better understand how a drug treatments will work for people who have suffered a stroke.
- Massachusetts General Hospital just published this paper on how they have used machine learning to predict the risk of developing C. difficile infection:
- This paper documents how several groups have been making developments in our understanding and treatment of cancer
These are just a couple of examples I’ve seen recently but this is a link to another site that talks about 7 uses of ML in medicine: here 🙂
Should we be concerned about machine learning?
This is an interesting one. Should we be concerned about machine learning? I know there has been some bad press recently about AI and the rise of the machines to takeover the world. I don’t believe that. Even if it were possible, I’m not convinced. Anyone who like I has spent 10 minutes online attempting to pry information out of their banks latest attempt at a Customer Service chatbot before casually throwing their laptop out of the window, will agree that we’re a long way off the age of robots yet.
I hope I have convinced you you need not be afraid but just in case here’s another post on the topic:
- Fear and loathing: A smart summary of ML and how it is used that talks about whether we should be concerned about machine learning
Back to the course…
I hope that answers the questions on why care about machine learning, should we be concerned about machine learning. Furthermore I hope you feel enlightened on the application of machine learning to medical diagnosis. I should probably get back to the course. Let me know your reasons for being interested in machine learning in the comments or by subscribing to the email. You can find the link below the posts on mobile or in the right hand bar on desktop. Looking forward to hearing from you!
Are you also looking to learn to code? Here’s my post about how I found which course to begin learning to code. I hope it helps you.
3 Comments
Nathan Galilee
Great read mate! Although computers are to act without explicit programming as Andrew Ng states, I assume the “training” to such computers must be extremely thorough and well-thought through? Without specific code to perform a specific task, how would one ensure ML produces the desired outcome? Or is that not the point..?
As a novice in this field I’m learning from your posts …I’m aware of ML practices in Credit Risk (fun.. right) but didn’t know much about its application in medicine until reading this.. Cheers!
aiclaire
Whoop whoop credit risks!
You’re right - the algorithms can learn what we teach them to - there’s a really interesting podcast called dataskeptic I think you would like and last weeks edition was talking about peoples intent when training ML.
Also here’s an example of how lack of diversity can cause bias in algorithms: facial recognition.
Thanks for reading!
Jam
Nice post! I agree that we’re a long way off human-like AI. We don’t even know how our own brains work yet! Nathan, adding to what Claire said, you have to measure how well yor model is working by using it. Typically you break your data into two halves, then use te first half as training data and the other half as a test. Then you know how well your model works for the data you give it.
Keep the posts coming Claire!