Scoping review on the economic aspects of machine learning applications in healthcare

Abstract

<h3>Background</h3><p>The development and use of artificial intelligence and machine learning technologies in healthcare have increased, prompting a need for evidence on their safety and value. Economic evaluations support healthcare decision-making and resource allocation. This scoping review aimed to map and synthesize current approaches to evaluating the economic aspects of machine learning based technologies implemented in healthcare.</p><h3>Methods</h3><p>Following the updated JBI guidance for scoping reviews, six databases (PubMed, CINAHL, Cochrane Library, Embase, Scopus, and IEEE Xplore) were searched for studies evaluating the economic aspects of machine learning-based technologies within healthcare. No exclusions were applied to healthcare settings, healthcare professionals or used economic evaluation methods. The results of data extraction were analyzed using descriptive statistics and inductive coding. The reporting of the studies was compared against the CHEERS-AI statement.</p><h3>Results</h3><p>A total of 6332 references were retrieved, with 18 studies included in the review. The studies comprised economic evaluations (n = 9), impact evaluations (n = 5), and performance evaluations (n = 4), with cost-effectiveness analysis being the most frequently used economic evaluation method (n = 8). The comparison of the studies to the reporting guidelines revealed gaps in the reporting of details from economic evaluations and the artificial intelligence nature of the technologies. Overall, the study alignment with the CHEERS-AI items on average was 39.6 %, with 64.1 % alignment with economic evaluation details, and 21.3 % alignment with key details related to the artificial intelligence nature of the evaluated technologies.</p><h3>Conclusions</h3><p>The current literature evaluating the economic aspects of machine learning-based technologies implemented in healthcare reveals gaps in coherence and coverage. Frameworks guiding artificial intelligence development should be refined to incorporate components related to system evaluation and post-implementation considerations. Further, multidisciplinary collaboration should be enhanced and promoted.</p>