Commentary

Video

Key Applications, Challenges of Artificial Intelligence in Respiratory Care

Author(s):

Io Hui, PhD, researcher at The University of Edinburgh, discusses how artificial intelligence (AI) is being applied in respiratory care for both clinicians and patients.

Artificial intelligence (AI) can be used in respiratory care for patient triage, chest x-ray categorization, and patient self-management, but regulatory and ethical challenges remain, says Io Hui, PhD, researcher at The University of Edinburgh in Scotland.

Hui is also the chair of mHealth and eHealth for the European Respiratory Society.

This transcript was lightly edited.

Transcript

Can you discuss the use of some of the key AI technologies used in respiratory care?

In respiratory care, we have different uses of artificial intelligence, mainly in 2 categories. One is dedicated for the clinicians, and then another one is dedicated for the patients.

For the clinicians, we can use artificial intelligence to triage patients into different hospitals, and from primary care to secondary care, for example. We can also use artificial intelligence to categorize different chest x-rays; this is another example for the clinicians.

For patients, of course, we are focused on the patient self-care and self-management. So, it means, first comes first, we can use artificial intelligence to educate patients and provide them sufficient information so that the patients can interact with the artificial intelligence together to learn more about asthma or COPD, for example. And then, of course, the artificial intelligence, nowadays, can also support clinicians to look after patients around their inhaler technique. So, for example, when an asthma patient is using their inhaler, the artificial intelligence can adjust, and then can advise the patients around whether they are using good technique or not a good technique at all.

And of course, another example, last but not least, we are actually using artificial intelligence to collect real-time patient data with the environmental data altogether, to advise the self-management daily routine in that sense. So, for example, advise them to reduce or increase their inhaler dose in that way.

What are some key challenges of integrating digital technologies into real-world respiratory care settings?

This is a very good question. As [someone with] an engineering background, I would say that the first key challenges are actually very far into the technology process. For example, when we're trying to collect the real-time data from the patients, the Wi-Fi connection is one of the key things that we are going to consider in most cases, because there are set zones in different areas that we can set for the Wi-Fi adopter and then receiver in that sense.

And then another thing is around the regulations, that kind of stuff. So the regulation thing we find a little bit difficult when we are deploying the artificial intelligence just because when the artificial intelligence is involving the patient decisions and also around the advising of the medications, it means that it has to be regulated under the Artificial Intelligence Act in the [European Union], and then somehow we have to make sure that the feature will match what the regulations would like us to do in that sense. But artificial intelligence just evolves very quickly, so it means that once we have evaluated one version, we don't have time to evaluate for the second version. So that is one of the challenges over there.

And then there are other challenges around the data collection, again, because we ask patients to give us the data in real-time. There are 2 ways to do it. One way is to have these smart devices or smart sensors to collect the data automatically from the patients, which is okay, because you are not asking patients to plug in the data by themselves. So it means that you don't have the missing data problem in reality. But again, because this is a passive, automatic data collection, it means that it falls back to the problem around the Wi-Fi and also the equipment as well.

And whether at the end of the day, after we have collected the data, what will be the approach to make sure that the data is accurate, and also the advice is accurate at the back end as well. So these are the challenges, and a lot of people would like to say that, "Oh, the interoperability would be another challenge," just because, for example, in the [United Kingdom], we have different vendors to support the platform in the hospital and primary care and secondary care practices. So within those platforms, we have different protocols and we have different ways to collect data and store data. So how could we make sure that the data can flow between primary care, secondary care, and tertiary care in a fluent way? So that is another challenge.

Related Videos
Tetyana Kendzerska, MD
Krunal Patel, MD
Scott Manaker,MD
Juan Carlos Martinez, MD
Parth Rali, MD
IMS Recap
Klaus Rabe, MD, PhD, chest physician and professor of medicine, University of Kiel
April Armstrong, MD, MPH, chief of dermatology, UCLA
Javed Butler, MD, MPH, MBA
Related Content
AJMC Managed Markets Network Logo
CH LogoCenter for Biosimilars Logo