Name:
Applying artificial intelligence to a healthcare setting for patients with TBI with Payam Barnaghi
Description:
Applying artificial intelligence to a healthcare setting for patients with TBI with Payam Barnaghi
Thumbnail URL:
https://cadmoremediastorage.blob.core.windows.net/42d87a17-182d-470c-845c-97844efdcc7f/thumbnails/42d87a17-182d-470c-845c-97844efdcc7f.jpg
Duration:
T00H07M17S
Embed URL:
https://stream.cadmore.media/player/42d87a17-182d-470c-845c-97844efdcc7f
Content URL:
https://cadmoreoriginalmedia.blob.core.windows.net/42d87a17-182d-470c-845c-97844efdcc7f/Payam Barnaghi- Interview V2.mp4?sv=2019-02-02&sr=c&sig=GUEjEcWJiSmCGKQLPxidi4CGCnFU7oF5%2BiAgPf9Pbcs%3D&st=2025-05-13T13%3A20%3A27Z&se=2025-05-13T15%3A25%3A27Z&sp=r
Upload Date:
2020-01-16T00:00:00.0000000
Transcript:
Language: EN.
Segment:0 .
[MUSIC PLAYING]
PAYAM BARNAGHI: My name is Payam Barnaghi. I'm a Professor of Machine Intelligence at the University of Surrey (UK). I'm also the Deputy Director of the Care Technology and Research Center at the UK Dementia Research Institute.
Segment:1 How is neurotechnology and artificial intelligence being used in healthcare to benefit patients with TBI? .
PAYAM BARNAGHI: I think we're getting better and better in collecting information from neuroimaging and collecting MRI images, CT scans, and also collecting information from physiological signals from patients' living environments, their activities.
PAYAM BARNAGHI: The more data we collect related to their environments, related to the day-to-day activities, related to their physiological signals and also information like medical imaging-- combining this information gives us better understanding of progress of the conditions, their rehabilitation, their improvement. And obviously, creating algorithms using different AI machine-learning techniques, which can predict the outcomes of rehabilitation, outcomes of prognosis, or, at early stages, helping clinicians to make diagnosis, will help clinicians to provide better care, better support to people with traumatic brain injury, and also will help patients to receive better care and improve their quality of life.
PAYAM BARNAGHI: And more and more, we can analyze this data in a better way. The more and more we collect and have more information, also, we can create more personalized care and rehabilitation programs for TBI patients.
Segment:2 What are the challenges associated with implementing neurotechnology in this setting? .
PAYAM BARNAGHI: Obviously, in neurotechnology, there are several devices helping us to collect information from brainwaves, from brain imaging, from wearable devices monitoring vital signals, people's day-to-day activities.
PAYAM BARNAGHI: But the problem is the noise in data. Collecting quality data and obtaining good information to make really more accurate decisions is always a big challenge. Let's say in medical imaging, when you collect information, continuity of data, the challenges of actually collecting frequent medical scans, medical imaging data-- the differences between data quality between different devices, between different scanners.
PAYAM BARNAGHI: Then you move to the environmental data. The noise of environment, the changes in the environment. And also, accuracy of the devices are, especially with variable technologies, are not as accurate, at least to now, as accurate as similar clinical devices. Obviously, these devices, they create tremendous opportunities to help to understand progress prognosis of the conditions. But at the same time, they create lots of challenges in how we should analyze data and how accurate and reliable the decisions we make based on them are going to be.
Segment:3 Is there potential for machine-learning algorithms to predict post-concussive symptoms in patients? What work is currently being done around this? .
PAYAM BARNAGHI: Machines learn in two ways-- either by example or by experience. If you have sufficient good-quality data, if you give examples of previous cases and prognosis, obviously, machines can be trained to make predictions or to learn from those examples and be able to help clinicians to make a bit better decisions or better predictions of the outcomes of their prognosis. At the same time, machines also can learn by experience.
PAYAM BARNAGHI: If you just give them sufficient data, they can just look at the data and identify patterns. And by looking at it, they can identify what type of patients probably they respond better to what type of interventions. The problem often is accessing to sufficient data is a big challenge. Across different trials of the data, sometimes the compatibility of data, reusability of data across different sites, different trials when you collect data, becomes another challenge.
PAYAM BARNAGHI: And obviously, machine- learning algorithms are good and becoming better and better, especially with deep-learning models or several probabilistic machine- learning models. You can make good predictions. But always, also, you have to be careful not to overfit your models. Overfitting means you have a population. You give a data.
PAYAM BARNAGHI: Your trained machine works very well with that population. And then you start moving out and you're trying it in the real world, in the wild. The machine learning should also be ready and be prepared to be developed in a way to adapt to the new environments and to be able to make predictions by the data they haven't seen before.
Segment:4 What are the ethical implications surrounding neurotechnology in a healthcare setting?.
PAYAM BARNAGHI: That's an interesting one because brain data is very personal.
PAYAM BARNAGHI: It's private. That's where you think. That's how we show our emotions. And obviously, the more and more we become better in collecting information in neurotechnology, we are also getting more access to information of basically our mind. Not necessarily exactly what people think. What really we collect is more personalized than other types of information even collected by patient records or variables.
PAYAM BARNAGHI: And obviously, the privacy issues. How this information is going to be used. Information governance is always a big challenge. In terms of neurotechnology for interventions, that's another side. Now you can interfere with brainwaves, with brain signals, and stimulate different parts of brain. Obviously, that raises careful considerations of decisions people make.
PAYAM BARNAGHI: How that technology is going to be used. How safe and how reliable that technology is. And obviously, the ethical issues around whether these devices have been sufficiently tested or controlled by people who should have access to it.
Segment:5 Lastly, what advancements are you looking forward to most in the field? .
PAYAM BARNAGHI: I think one of the interesting things we will see in near future will be combination of hardware neurotechnology devices with data analytics techniques.
PAYAM BARNAGHI: Obviously, the machine-learning and deep-learning models are becoming more advanced, more mainstream and used in different areas. At the same time, we're creating better and better devices, which can help us to collect information. Combining these two my hope is will provide help and support to clinicians, patients and care providers in three main categories-- starting from diagnosis, helping to make better diagnosis, quicker diagnosis, more accurate diagnosis, and moving to prognosis, helping clinicians, to make better decisions, better predictions of their prognosis outcomes.
PAYAM BARNAGHI: And then it moves to rehabilitation. Allowing clinicians and rehabilitation providers to make more personalized care and support plans and make more informed decisions. And also to allow patients to be able to interact with the environment and also provide data which can help their clinicians to make better decisions for their care and quality of life.
