The Role of Artificial Intelligence in Clinical Practice: Video Lay Summary

Revolutionising Healthcare: The Role of Artificial Intelligence in Clinical Practice

Context:

Artificial Intelligence (AI) has developed a lot over recent years. It can change our lives because of its impact on many areas, such as transportation, finance, and even healthcare.

AI has taken a long time to develop, and it is not something that everybody understands. There are different types of AI, including machine learning (ML), deep learning (DL), natural language processing (NLP), and large language models (LLM). In 1951, Christopher Strachey developed the first AI program when AI was an academic research topic. The name “Artificial Intelligence” was given in 1956 by John McCarthy, who is considered the father of AI.

Between the 1960s and 1970s, AI was used to develop rule-based systems to mimic the decisions of human experts. During the 1980s and 1990s, the main focus of AI switched to machine learning (ML), where machines could improve their performance by learning from data. In the decade of the 2000s, AI was focused on the development of natural language processing, resulting in virtual assistants, for example, Alexa or Siri. Recently, the development of large language models (LLM) has resulted in tools, such as automated conversations, like ChatGPT or Gemini.

In 2023, Dr. Shuroug Alowais from the College of Pharmacy, King Saud bin Abdulaziz University for Health and her colleagues decided to look at the current role that AI has played in the healthcare sector and clinical practice. Their aim was to equip healthcare workers with the skills and knowledge they would need to make use of AI.

What the authors did:

The researchers reviewed articles published in English in different scientific databases: EMBASE, Scopus and PubMed/Medline. They used the different types of AI (ML, DL, NLP, LLM) as keywords, together with terms like healthcare, medicine, diagnosis, ethics, and patient monitoring. They reviewed the articles carefully and only included the ones which they agreed were a good fit for the study. However, the authors did not say what made an article a good fit for the study.

Key findings:

Diagnosis

The review found that ML can be used to spot, sort or predict diseases by analysing data such as medical images, CT scans and MRI scans. Illnesses, like cancer, pneumonia or appendicitis, can be diagnosed more accurately by medical professionals with the help of the AI. This is because it can find patterns in X-ray that can show that something is wrong before patients show symptoms. Additionally, AI can help by analysing genes and microorganisms to get a better prediction of the risk of diseases starting.

Treatment

Personalised medicine is giving the right treatment to the right person. AI can help with this by looking at a person’s genes, environment, lifestyle, and markers in their body to choose the most suitable medicine. AI can also help doctors decide how much medicine to give each person, which is known as drug monitoring. Additionally, ML systems can predict how different drugs react with each other. For example, CURATE.AI is a new system to adjust the dose of chemotherapy depending on each patient’s data.

Population health management

AI has been used to guide public health efforts. Public health means looking after big groups of people (populations) rather than just one person. AI can use large amounts of data to predict who might be at risk of certain illnesses. For example, AI (ML systems) analyse things like age, gender, geographic location, ethnicity, as well as medical history and lifestyle, to spot patterns and predict which groups might develop an illness, or who might need to go back to the hospital. This helps healthcare services plan better and create targeted programmes for people at risk and prevent them having to go to hospital to save money for patients and the healthcare system.

Under the support of experienced healthcare workers, AI can help develop guidelines more quickly. It does this by looking at data from patients with a certain condition as well as the latest research.

Patient care

Virtual health assistants, such as chatbots or intelligent speakers, have helped take pressure off healthcare services. They let people get quick answers to their questions and help them work out if something is an emergency or not. In mental health, AI applications can spot mental health conditions earlier, although the reviewers pointed out it is a tool, and it cannot replace mental health care professionals because AI cannot show empathy as a human being could.

AI has been used to help people learn about their health. For example, patients with diabetes or other long-term illnesses can learn about them with ChatGPT. Researchers have also looked at whether people prefer getting their health information from AI or healthcare professionals and the results have been mixed. Things like previous use of technology, age or background affect what people prefer.

Issues with the review:

This review does a good job of looking at AI in different parts of healthcare. But it didn’t look at any possible downsides of AI – and because the authors did not tell us exactly how they decided what to include, it is hard to know if this is the whole story. The authors also didn’t discuss whether the research they reviewed was good quality or not, so it’s hard to know how much the information can be trusted.

Conclusions and next steps:

AI can be really helpful for healthcare services. It can be a useful tool to find illnesses quicker, get people the right treatment and to save money in healthcare systems. However, it is important that experts manage this tool well, because confidentiality and data privacy must be taken care of. Future research could explore more ethical considerations and how the public see the use of AI in healthcare.

Image: https://pixabay.com/photos/ai-robot-artificial-intelligence-7977960/