Understanding more about AI
Artificial Intelligence (AI) is already being used in many aspects of life, and it has potential to make a big difference in the way that diseases like PH are managed too. But what does it all mean when it comes to healthcare?
We asked our PH community what they wanted to know about this topic, and then put their questions to Dr Andy Swift, a Cardiothoracic Radiologist working with the Sheffield Pulmonary Vascular Disease Unit. This is what he told us…
Artificial intelligence in healthcare is essentially a way that computers can be used to replicate what we do as humans, particularly replicating our tasks. For example, there are ways of using artificial intelligence in medical imaging by using it to replace some of the drawings we make on the images, or when we try to make diagnoses from the images. This is already happening in some areas but is in its infancy.
It can also be used with larger data to try and make predictions by looking for patterns in the images. And it could have future use in drug discovery by looking at the patterns in larger data, particularly genetic data, to try and identify new treatment targets.
AI is already being used in everyday life. Google Maps, which everyone’s become heavily reliant upon, uses artificial intelligence to navigate you where you need to go. And facial recognition, which looks at images, can automatically recall what your image should look like, for use in things like security.
I think earlier diagnosis is a particularly interesting area for artificial intelligence. We know that pulmonary hypertension has quite non-specific presentation, and that can lead to a delayed diagnosis. In my area of medical imaging, AI can be used to automatically identify patterns that we might miss at an early stage, or when PH isn’t the condition that’s on people’s minds because it’s rare.
AI can also be used to analyse text. Using natural language processing, it could review a lot of the text reports from different investigations, and from clinical histories, to identify patterns that might lead to the suspicion of pulmonary hypertension earlier than with existing technology. So, I think AI is very promising for early diagnosis.
It can also analyse much deeper, richer data from patients, such as their genetic data and all their antigen / antibody tests. Combined with imaging, this can give a more detailed picture of their condition and I think that can really help with both drug discovery and selection for trials.
Artificial intelligence can be used to free up time for clinicians. For instance, in radiology, consultants spend quite a lot of time making measurements on images or drawing round structures, which can take up to half an hour each time. That could instead be done by AI.
I think a lot of the paperwork tasks in medicine could be reduced in this way, allowing medics more time to engage with patients.
The potential advantages of AI far outweigh the risks. But AIreally is a product of exactly the training data that’s been given, and it will work within that remit. Therefore, if the training data is different to the exact patient who is presenting to clinic, then it might not perform perfectly. Some inherent biases would need to be looked at and considered when the technology is being rolled out, to ensure it works for everybody.
Some people may worry a little bit about privacy because AI sees large amounts of data from individuals that could be pieced together. But ethics committees and governance processes ensure that any data that’s used for training is fully anonymised.
One potential risk can be related back to Google Maps – because so many people use them so often, they may forget how to read a ‘real’ map. In healthcare, we must not become over reliant on some AI tools, remembering they are simply an aid.
I do not believe that AI will lead to the loss of human contact in healthcare. I envisage it as a support for the medics to improve efficiency; I don’t see it making decisions without human and clinical insight and oversight. Computers and artificial intelligence won’t ever be able to replace the human intuition.
Artificial intelligence is already being used for lower-level tasks in healthcare, but I think in the coming years it will become much more prevalent. And from speaking to patients, they seem quite positive about the use of AI which might improve the efficiency of healthcare.
There will be a need to validate the use of AI in different medical areas, patient populations, and hospital systems, and to be sure of the positive benefits. Once all that has happened, the rollout will be much faster.
It’s a very exciting area and it’s really helpful having patients involved in studies in this area. They can make a big difference by contributing in this field.
Ultimately, artificial intelligence has great potential to have a real and positive impact in healthcare.