JMIR Publications

ABSTRACT

Background:

Thrombolysis and mechanical thrombectomy represent the most successful stroke innovations over the last thirty years. Quantifying innovation in stroke is essential for identifying productive research lines and prioritizing funding but healthcare lacks validated methods for measuring innovation.

Objective:

This study aims to systemically evaluate the relationship between stroke-related patents and publications, demonstrate the feasibility of using large language models (LLMs) in this process, and identify the most rapidly advancing innovations in stroke care.

Methods:

Electronic patenting and research publication databases were searched between 1993 and 2023 for “stroke” OR “cerebrovascular”. A large language model (LLM) was trained to identify patents related to stroke disease, as opposed to other references to the word ‘stroke’, on a manually labeled subset of patents; and assessed using cross-validation. The LLM filtered irrelevant results, and the resulting patent codes were grouped into innovation clusters. Cluster-specific growth curves were plotted to analyze the rates and characteristics of growth. Adoption percentages for each innovation cluster were estimated by fitting a sigmoid curve to the patent and publication data.

Results:

The cross-validated accuracy of the LLM was 99.2%. An initial bibliometric search retrieved 237,035 patents and 486,664 research publications. After LLM filtering, 28,225 stroke-related patents remained, resulting in seven innovation clusters. Pharmacological treatments were the top-performing cluster over the last thirty years, accounting for approximately half of all patents. Artificial intelligence (AI) methods, rehabilitation devices, and medical imaging exhibited exponential rates of patent growth, with annual normalized increases of 39.2%, 15.9%, and 5.8%, compared to 16.9%, 5.3%, and 2.2% for publications, respectively. These innovations were identified as the most rapidly advancing in stroke management.

Conclusions:

The study shows how LLMs, applied to publicly available patent and publication data, can quantify innovation in stroke, and identify areas of greatest traction.