Machine Learning-Enabled Meta-Analysis of Bioengineering Micro- and Nanorobots

Abstract

Micro- and nanorobots are emerging bioengineering tools for targeted delivery, biosensing, imaging, and minimally invasive therapy, yet systematic design and cross-study translation remain hindered by heterogeneous experimental conditions, inconsistent reporting, and sparse structured data sets. Here, we develop a literature-derived data set of micro- and nanorobot systems and apply artificial intelligence (AI)-enabled structural meta-analysis to resolve relationships among bioengineering constraints, actuation strategies, environmental conditions, measurement practices, and performance. Dendrogram- and correlation-based clustering identifies three dominant descriptor modules spanning deployment context, actuation–measurement coupling, and design intent. Supervised learning reveals structured descriptor patterns in propulsion mechanisms, whereas prediction of instantaneous velocity remains only moderately generalizable, highlighting substantial domain shift across materials, propulsion physics, biological media, operating conditions, and measurement protocols. Overall, AI and machine learning provide a powerful framework for mapping heterogeneous design landscapes, exposing reporting biases and descriptor coupling, and advancing standardized data sets for reliable comparison and predictive design.