Digital and AI Medication Tools Show No Clear Benefit for Older Adults in Trial Review

Digital and artificial intelligence-assisted medication management has been promoted as a way to help older adults with chronic diseases take their medicines correctly and safely. A new systematic review and meta-analysis of randomized controlled trials, published in Frontiers in Public Health by researchers at Changchun University of Chinese Medicine, now offers a sobering assessment of that promise. After synthesizing the available randomized evidence, the authors conclude that the current body of research does not establish clinically meaningful benefits of these technologies on hard endpoints such as hospitalization and mortality in older adults with at least one chronic disease.

The review team, led by Yuyan Cheng, Weidong Huang and Tian Jia, systematically searched PubMed, Embase, the Cochrane Library, CNKI, Wanfang and other databases from their inception through June 30, 2026, for randomized controlled trials testing digital or artificial intelligence and machine learning-based medication management in older adults with at least one chronic condition. Study selection, data extraction and risk-of-bias assessment were each carried out independently by two reviewers using the Cochrane risk-of-bias tool, version 2.0, and meta-analytic pooling was performed in RevMan 5.4. The review was prospectively registered with PROSPERO under the number CRD420261460445.

Nineteen randomized controlled trials met the inclusion criteria. Only six of these could be combined in the formal meta-analysis, while the remaining thirteen were summarized descriptively. That split reflects the state of the evidence base captured by the review: few of the included trials measured the same outcomes in comparable ways, and the authors report that the pool of high-quality randomized controlled trials is limited.

Three studies reported adherence results, but they were highly heterogeneous, with an I-squared statistic of 97 percent, meaning almost all of the variation between the estimates reflected differences between the studies rather than sampling error. Because of this heterogeneity and the small number of trials, the authors declined to calculate a pooled estimate. The individual study results ranged from a standardized mean difference of minus 0.28 to 1.89, a span that stretches from a small negative effect to a large positive one and offers no consistent signal about whether these tools improve adherence.

For potentially inappropriate medications, only one of the included studies reported a comparable outcome. With a single trial, no pooled estimate could be calculated, leaving the question of whether digital or AI-assisted medication management reduces inappropriate prescribing unanswered by the randomized evidence assembled in this review. The hard clinical endpoints fared somewhat better in terms of available data, but not in terms of results.

Two studies examined hospitalization and mortality, though they used different effect measures and were therefore presented separately rather than pooled. Blum and colleagues reported a hazard ratio of 0.95 with a 95 percent confidence interval of 0.77 to 1.17, while Rieckert and colleagues reported an odds ratio of 0.88 with a 95 percent confidence interval of 0.73 to 1.07. In both cases the confidence interval crosses 1.0, meaning the results are statistically compatible with no effect, and the point estimates, while slightly favoring the interventions, do not establish a clinical benefit.

The authors are careful about what their findings do and do not show. The evidence, they conclude, remains inconclusive for medication adherence, and no reliable pooled estimate could be obtained for that outcome. For hospitalization and mortality, the available randomized trials do not demonstrate a benefit.

Given the pronounced heterogeneity across studies and the limited pool of high-quality randomized controlled trials, they conclude that the current evidence is insufficient to establish clinical benefit for the routine implementation of digital or AI and machine learning-based medication management systems in geriatric care, and they call for further rigorously designed studies. The limitations of the underlying evidence base are themselves a central finding of the review. The review does not show that these technologies are harmful; it shows that the randomized evidence gathered so far cannot demonstrate that they improve the outcomes that matter most for older adults with chronic diseases.

In the case of digital and AI-assisted medication management, the trials conducted to date have not yet delivered that proof.

Subject of Research: Digital and AI-assisted medication management for older adults with chronic diseases

Article Title: Digital and artificial intelligence-assisted medication management for older adults with chronic diseases: a systematic review and meta-analysis of randomized controlled trials

Article References: Cheng, Y., Huang, W., & Tian, J. (2026). Digital and artificial intelligence-assisted medication management for older adults with chronic diseases: a systematic review and meta-analysis of randomized controlled trials. Frontiers in Public Health, 14, Article 1956051. https://doi.org/10.3389/fpubh.2026.1956051

Image Credits: AI Generated

DOI: 10.3389/fpubh.2026.1956051

Keywords: artificial intelligence, medication management, older adults, chronic disease, systematic review, meta-analysis, randomized controlled trials, medication adherence, digital health, geriatric care, hospitalization, potentially inappropriate medications