Using Artificial Intelligence in healthcare

How much do you know about Artificial Intelligence in healthcare? Could computers discover new medicines?

Yesterday I listened to The Life Scientific podcast on BBC Sounds. The host Jim Al-Khalili and his guest, Daphne Koller, a 'Machine Learning expert', were discussing Can Computers discover new medicines?

This fascinating podcast provided a glimpse of the huge potential of both Artificial Intelligence and Machine Learning in healthcare science, health treatment and care.

What is Artificial Intelligence?

Artificial Intelligence (AI) is the use of technologies to recognise speech, make decisions and use visual perception. What is special about this, is that these 'intelligences' would ordinarily require human input and intelligence.

What is machine learning?

Machine learning (ML) is the use and development of computer systems that are able to learn and adapt without following explicit instructions. Rather than programming the computer in the usual sense, algorithms and statistical models are used to analyse and draw inferences from patterns in data.

What are some of the benefits of AI and ML?

• Computers don't need rest breaks, so can operate around the clock to analyse data and computers can carry out analysis faster than humans - this can accelerate medical advancements
• AI can be used to collate, analyse and share vast datasets, beyond what a scientist or a doctor might usually encounter in their practice
• AI and ML does not do what has always been done (as humans tend to) and can be freer of bias (although the algorithms used can be a source of bias as they are developed by humans)
• Machine learning may uncover new ways of thinking about and coming up with solutions to complex health problems.

This final point, has led to some extraordinary developments. For example, in the podcast Daphne Koller explains her work on Machine Learning and breast cancer imaging and diagnostics.

Usually, doctors would look for specific, established markers of breast cancer on diagnostic images. In the ML project, the computer was not programmed to look for specific markers, rather it was given algorithms for measuring and quantifying everything that was observable in the provided images and then decide for itself what it thought might be interesting - it was looking for patterns. Two amazing things happened, firstly, the computer was better than the doctors at predicting the five year prognosis (outcomes) of patients, and secondly, the information that it used to make these decisions was not based on any of the markers the doctors had been looking at! The computer had contributed to the discovery of critical new information to inform breast cancer care.

So could computers discover new medicines?

Daphne Koller certainly thinks so and in fact, AI and ML are already being used in drug experimentation, as you can see in the video from Novartis (a large global pharmaceutical company).

To learn how AI is being used in healthcare right now, watch the video below by The Medical Futurist