Truace
Subscribe

Algorithm-assisted personalized risk communication to encourage flu vaccination in the USA: three randomized field trials

Artificial intelligence (AI) is increasingly used in healthcare to identify high-risk patients. Laboratory studies are mixed regarding patient attitudes towards AI use. Here we field-tested whether informing patients of their risk for influenza and its complications can serve as an effective behaviour-change intervention to increase vaccination, and whether disclosing AI use in the risk determination or providing personalized reasons for the algorithm's prediction (a form of explainable AI) influence message eff…

By the Truvace desk2 min read1 sourcePeer-reviewedTRV-2026-1852· cite & verify
Algorithm-assisted personalized risk communication to encourage flu vaccination in the USA: three randomized field trials
NAVY MEDICINE Volume 97, No. 1 January-February 2006 by U.S. Navy. Bureau of Medicine and Surgery. Public domain

In brief

Researchers ran three preregistered randomized controlled trials with over 90,000 patients of US healthcare systems. A previously validated machine learning algorithm identified patients at high risk for influenza and its complications, and these patients received no message or one of several messages encouraging vaccination.

The findings suggest that personalized risk information can be a low-cost way to nudge vaccination, though the gains are small in absolute terms. Patients did not react differently when messages named the algorithm or explained its predictions, so disclosure appears unlikely to deter or help uptake in this setting.

Main points

  1. The trials enrolled over 90,000 unique healthcare system patients in the USA across three preregistered randomized controlled trials.
  2. Patients told of their high risk were vaccinated 1.1-1.4 percentage points more than those sent a simple reminder.
  3. Compared with patients who received no message, the high-risk messages raised vaccination by 1.7-3.5 percentage points.
  4. Vaccination was similar whether or not messages disclosed algorithm involvement or gave personalized reasons for the prediction.

The gain

Telling patients a machine learning algorithm had flagged them as high risk for flu complications raised vaccination slightly more than a generic reminder or no message.

The rundown

The study tested whether informing patients of their personal risk for influenza and its complications works as a behaviour-change intervention. The primary outcome was influenza vaccination, and the comparison groups included non-messaged patients and patients sent a plain reminder.

A second question was whether disclosing AI use in the risk determination, or giving personalized reasons for the algorithm's prediction as a form of explainable AI, changed how effective the message was. Vaccination rates were similar across these arms, so patients were neither averse to nor appreciative of algorithm use or explainability.

The authors note that earlier laboratory studies on patient attitudes toward AI were mixed, and describe this as a realistic health application. The work was partially funded by the National Institute on Aging of the National Institutes of Health.

Sources

  1. Peer-reviewedNature Human Behaviour2026-10-05

The debate