Your Wellness

How AI is already being used in Irish hospitals - and what the future might hold for patient care

AI has the potential to transform healthcare, but only if the correct safety measures are put in place

One AI-assisted intervention currently in place is an assisted decision-making tool used in the management of stroke.
June Shannon
Tue May 28 2024 - 05:31

Artificial Intelligence, or AI, seems to be everywhere at the moment. You may have already heard of apps such as ChatGPT, a text-based chatbot that allows you to write essays, or AI image generators such as Midjourney, which create realistic images based on the instructions you give it.

But did you know that AI is also now being used in Irish hospitals and supporting medical professionals in providing patient care?

Much like the introduction of any revolutionary technology, the increased use of AI goes hand-in-hand with valid fears and concerns around its use – from an ethical, legal and moral standpoint. All aspects of life stand to be impacted by AI, and when it comes to our health and personal healthcare data, we can’t be too careful.

Dr Conor Judge is a consultant nephrologist at Saolta University Health Care Group (which comprises six hospitals across seven locations in the west and northwest of Ireland) and senior lecturer of applied clinical data analytics with the University of Galway. He recently spoke at the Irish Medical Organisation’s agm in Kerry on the potential impact of AI on healthcare delivery in Ireland.

READ MORE

Dr Judge believes that AI interventions in healthcare must be treated in the same way as medications. This means they must be tested in large phase three clinical trials. “So this is how we test aspirin, how we test blood pressure medications – and anything short of that for artificial intelligence is going to be a costly mistake,” he told The Irish Times.

Last year, Ireland’s first Institute for Clinical Trials was established by the University of Galway with a focus on benefit and impact for patients. According to the institute, it “will transform lives by ensuring patients get access to the latest medicines and treatments in a timely way”.

Speaking at the launch of the institute, Prof Martin O’Donnell, dean of the college of medicine, nursing and health sciences at University of Galway, said “this institute will consolidate our areas of considerable strength in clinical trials, ensure the rapid translation of medical discoveries to clinical evaluation and ultimately improve the lives of patients and health of our population, leveraging the academic strengths of our university and its healthcare partner, Saolta University Healthcare Group”.

The overarching goal is to personalise hypertension management, thereby bridging the existing care gap and significantly reducing the global burden of hypertension-related complications

—  Dr Conor Judge, consultant nephrologist, on Artificial Intelligence Clinical Decision Support Systems

Dr Judge said the new institute would look at innovations in randomised clinical trials which would include testing new medicinal products, medical devices and AI interventions. One of the stated goals was to “support emerging domains, such as Artificial Intelligence in healthcare, by creating a cluster of academic and industry investigators aligned to evaluating digital health interventions”.

Dr Judge has recently secured funding from the Health Research Board to explore the potential of AI to improve the management and treatment of hypertension or high blood pressure. Often known as the silent killer, hypertension affects more than 1.4 billion people worldwide and is a major risk factor for heart attack and stroke.

Currently, some patients with high blood pressure may need to make frequent visits to their GP to ensure optimal control of their condition, which is time-consuming for both the patient and the doctor. This is where Artificial Intelligence Clinical Decision Support Systems (AICDSS) for hypertension could help.

The research project will analyse two extensive clinical trials on blood pressure treatment to train a computer program to make treatment decisions, similar to doctors who specialise in hypertension. Safety features will then be integrated into the AI program to ensure reliable recommendations, especially in unfamiliar medical scenarios. The AI tool would, therefore, aim to help GPs to provide the best treatment for a patient’s blood pressure to help bring it under control more quickly.

Interestingly, the project will also carry out comprehensive surveys with clinicians and patients to gauge their perceptions of AI-driven treatment. The final step involves a thorough evaluation of the AI program’s efficacy in recommending blood pressure treatments in a real-world setting. “The overarching goal is to personalise hypertension management, thereby bridging the existing care gap and significantly reducing the global burden of hypertension-related complications. The project’s findings could potentially set a precedent for employing AI in managing other critical health conditions, marking a significant stride towards integrating AI in routine clinical practice,” Dr Judge said.

There are several AI-assisted interventions already in use in Irish hospitals, mainly in the area of medical imaging, and Dr Judge paid credit to the work of the HSE National Intelligence Group for its work in this area.

One of these currently in place is an assisted decision-making tool used in the management of stroke. When a patient has had a stroke, one treatment option that may be suitable for them is called a thrombectomy. Thrombectomy is a life-saving treatment where the clot causing the stroke is mechanically removed from the brain. To decide whether a patient is eligible for thrombectomy, a radiologist examines CT scans of the patient’s brain to calculate a score called the ASPECTS score.

Dr Judge explained that this was a time-consuming task, and this was where an AI tool developed by a company called Brainomix comes in. The Brainomix 360 stroke platform uses AI algorithms to support doctors by providing real-time interpretation of brain scans. The AI tool calculates the ASPECTS score for the radiologist which facilitates the decision on appropriate treatment more quickly, making the process more efficient. This is particularly important where every minute matters and time is crucial for improved patient outcomes.

Dr Judge said that ultimately it is the physician who decides on the best course of treatment for patients, and AI tools can help augment the decision-making process.

The practice of medicine has evolved over centuries, shaped by new advancements from the introduction of vaccines and anaesthesia and transplant surgery to modern medicines. However, the one thing that has remained constant is the doctor-patient relationship, the humanity or art of medicine. According to Dr Judge, the doctor-patient relationship is not under threat from AI, but rather could potentially benefit from it.

“We spend a lot of our time doing administration work, where we’re trying to collect pieces of information from multiple different data sources – they’re often in siloed systems, you have to have a username and password to log into all of them, and then we have to make a treatment decision about that.

“And then there’s very little time to talk to the patient and explain it to the patient. So if we had systems that enhance the ability to put information in one place it could really change the relationship, for now we have more time to speak to the patient, to explain to the patient, to allow them to ask and answer questions. Framing it in that way, there is immense potential to improve the doctor-patient relationship.”

Dr Judge believes that AI has the potential to transform healthcare, but only if the correct safety measures are put in place. He said that patients should demand that AI interventions be rigorously tested to the same level as medicines. “We don’t ask patients to take medications that haven’t been tested in randomised clinical trials... So patients need to demand that that’s the thing that happens for AI interventions.

“I’m very positive about it... but we just have to do it right. Because if we don’t do it right, well, then it could transform it... for the worse. It has to be done with the correct guardrails in place... It goes back to the first ‘do no harm’, and we need to take that approach as well. But I’m very positive about the transformative nature of it.”