New Delhi: Messages reminding high-risk influenza patients to get vaccinated led to a small but measurable increase in vaccination rates, a recent study found. Researchers used an AI tool to identify patients at higher risk of flu and related complications.
The study, titled “Algorithm-assisted personalised risk communication to encourage flu vaccination in the USA: three randomised field trials”, was published in Nature Human Behaviour on 5 October. Gail N. Rosenbaum, a staff scientist at Pennsylvania-based Geisinger, was the first author.
The researchers wanted to find out whether telling patients that they were at higher risk of influenza and its complications would make them more likely to get vaccinated.
They also examined whether telling patients that an algorithm had assessed their risk, or disclosing the use of AI in the process, changed the effectiveness of the messages. Previous laboratory studies have found mixed attitudes towards the use of AI in healthcare, the study noted.
Patients told they were high-risk more likely to get vaccinated
The study found that messages highlighting a patient’s high risk of influenza had a small but consistent effect on vaccination rates.
Patients who were told they were at high risk were 1.1 to 1.4 percentage points more likely to get vaccinated than those who received a general vaccination reminder without any information about risk.
They were also 1.7 to 3.5 percentage points more likely to get vaccinated than patients who received no message at all.
Researchers found little difference in vaccination rates based on whether patients were told that AI or an algorithm had been used to assess their risk.
“Vaccination was similar across message arms that did versus did not disclose ‘algorithm’ involvement or its risk explanations,” the study said, suggesting that patients were neither particularly put off nor encouraged by the disclosure of algorithmic risk assessment.
Also Read: AI pathology lab tools lack public evidence, doctors don’t know how well they work
AI identified patients at higher risk
The study involved about 90,000 patients in a US healthcare system.
Researchers used a previously validated machine-learning algorithm to identify patients who were at higher risk of influenza and related complications. The patients were then randomly assigned to receive different types of messages, including a general vaccination reminder, a message explaining that they were at higher risk, and a section for no message.
The findings suggest that personalised risk messages may be more useful in encouraging vaccination than simply telling people to get a flu shot.
Influenza can cause symptoms including fever, chills, cough, and muscle and joint aches. While most people recover without serious complications, it can lead to hospitalisation, particularly among older adults, young children and people with certain underlying health conditions.
Edited by Maryam Hassan