Page 11 - JCTR-11-5

Journal of Clinical and
Translational Research AI and LLMs in iPSC cardiac research

recording 651,481 CVD-related deaths (38.2%), the identified through targeted searches across PubMed,
3,4
United States 957,455 deaths (35.7%), and Japan 372,483 Google Scholar, arXiv, and Web of Science using
5
deaths (28.0%). In high-performing healthcare systems, combinations of the following terms: “LLM,” “large language
6
CVD-related fatalities can be reduced below 20% through model,” “iPSC-CM,” “induced pluripotent stem cell,”
advanced preventive strategies, early intervention, “cardiomyocyte differentiation,” “regenerative cardiology,”
and innovative therapeutic solutions. 7-10 One such “cardiotoxicity,” “CRISPR screen,” “single-cell RNA-seq,”
breakthrough is the development of induced pluripotent and “deep learning.” Additional queries incorporated more
stem cell-derived cardiomyocytes (iPSC-CMs), which specific phrases, including “LLM in clinical genomics,”
11
offer potential applications in disease modeling, drug “cardiac lineage specification,” “iPSC-CM drug screening,”
testing, and cardiac tissue engineering. However, “electronic health records,” “BioBERT,” “BioMedLM,” and
challenges persist, including variability in differentiation “protein structure prediction.”
protocols, limited functional maturation, and obstacles to Inclusion criteria comprised: (i) peer-reviewed articles,
large-scale clinical application. 12 preprints, or white papers describing the use of AI or

The rise of artificial intelligence (AI) and large language LLMs in cardiovascular, stem cell, or regenerative research;
13
models (LLMs) has opened new avenues to accelerate (ii) studies involving iPSC-CMs in disease modeling, drug
14
iPSC-CM research and clinical translation. Historically, screening, or translational applications; and (iii) sources
AI-driven innovations have reshaped cardiovascular published in English from 2018 onward to reflect the
medicine. Machine learning has enhanced diagnostic advent of transformer-based architectures.
15
imaging, refined risk prediction models, and optimized Exclusion criteria included: (i) studies not involving
17
16
surgical planning in cardiothoracic procedures. More cardiovascular applications or not using iPSC-CMs;
18
recently, LLMs, such as ChatGPT (OpenAI), DeepSeek (ii) non-AI-based reviews or purely theoretical discussions
(Hangzhou DeepSeek AI Company), Bard (Google AI), without applied methodology; and (iii) articles lacking
and GROK (xAI), have revolutionized biomedical relevance to clinical translation or omics-driven discovery.
19
research by enabling large-scale data analysis, 20,21
optimizing differentiation strategies, 22-25 and predicting No strict limitations on publication types were imposed,
patient-specific responses to regenerative therapies. 26,27 allowing the inclusion of preclinical, computational, and
Since its release on November 30, 2022, ChatGPT reached translational studies. Approximately 150 sources were
one million users in just five days—far surpassing the screened, with 45 core references included in the final
growth of platforms, such as Facebook, which took nearly synthesis based on thematic relevance, methodological
10 months to reach the same milestone. Remarkably, quality, and impact on the evolving role of LLMs in
ChatGPT has also demonstrated performance comparable cardiovascular regenerative medicine.
to a 3 -year medical student on the National Board of
rd
Medical Examiners assessments and passed the United 3. Results and discussion
States Medical Licensing Examination Step exams, Although structured as a narrative review, we integrate
25
underscoring its potential to contribute meaningfully comparative insights and propose a scaffolding for future
to high-accuracy domains, such as stem cell-based benchmarking protocols in iPSC-CM applications of
cardiovascular research. Despite these advancements, LLMs.
28
current applications of LLMs in iPSC-CM research remain
underexplored, with key gaps in long-term validation, 3.1. Summary of key findings
reproducibility, and standardization. LLMs, such as ChatGPT, are increasingly integrated

This review critically examines the evolving role of LLMs into clinical and research workflows, supporting
in iPSC-CM research and translation. Through targeted peer discussions, complex decision-making, and
analysis of current literature, it explores how LLM-based interdisciplinary planning. In cardiothoracic contexts, they
frameworks can enhance differentiation strategies, uncover assist with surgical preparation, data analysis, literature
functional biomarkers, and bridge lab-based insights with synthesis, and knowledge translation—supporting
clinical application, laying a foundation for more scalable expertise sharing, collaborative planning, and innovation
and precise cardiovascular regenerative solutions. in iPSC-CM research. 24,26 The visual overviews of this
pipeline are shown in Figures 1 and 2.
2. Methods Figure 1 illustrates the progressive specialization of AI
To synthesize a comprehensive view of LLM applications tools–from general-purpose AI to clinical personalization
in iPSC-CM research and clinical translation, a narrative through LLM-enhanced multi-omics modeling customized
review methodology was adopted. Relevant studies were for cardiac regenerative contexts. Figure 2 outlines the

Volume 11 Issue 5 (2025) 5 doi: 10.36922/JCTR025230026