Australia-based healthcare AI technology company Harrison.ai has launched Harrison.rad.1, its radiology-specific vision language model, to address the global healthcare challenge.
Harrison.rad.1 is a dialogue-based radiology-specific vision language model that can perform different functions, with clinical safety and accuracy as key priorities.
It enables open-ended chat based on X-ray images, detects, and localises radiological findings, and generates reports, and offers reasoning based on clinical history and patient context.
The company is offering the new AI model to select industry partners, healthcare professionals, and regulators worldwide to improve patient outcomes.
Harrison.ai co-founder and CEO Aengus Tran said: “AI’s promise rests on its foundations, the quality of the data, rigour of its modelling and its ethical development and use.
“We are already excited by the performance of the model to date. It outperforms major LLMs in the Royal College of Radiologists’ (FRCR) 2B exam by approximately 2x.
“The launch of this model and our plan to engage in further open and competitive evaluation by professionals underscores our commitment to responsible AI development.”
Harrison.ai’s existing radiology solution, Annalise.ai, is already approved in more than 40 countries and is commercially deployed in healthcare organisations worldwide.
The company expects open and competitive evaluations of Harrison.rad.1 by top professionals.
Harrison.ai said that its Harrison.rad.1 differs from existing generative AI models, which are functionally generic and mostly trained on general and open-source data.
Harrison.rad.1 is trained on real-world, diverse, and unique clinical data, which includes millions of images, radiology studies and reports.
Also, a large team of medical specialists further annotated the dataset at scale, to provide Harrison.rad.1 with clinically accurate training signals.
Its Harrison.rad.1 is the most capable specialised vision language model in radiology, said the healthcare AI technology company.
Harrison.ai board director Robyn Denholm said: “The Harrison.rad.1 model is transformative and an exciting next step for the company. Harrison.ai is delivering on the promise of helping solve real-world problems more effectively and reliably and helping to save lives.”
Tran added: “Harrison.ai is committed to being a leading global voice in helping inform and contribute to an important conversation on the future of AI in healthcare.
“This is why we are making Harrison.rad.1 accessible to researchers, industry partners, regulators and others in the community to begin this conversation today.”
Harrison.rad.1 showed superior performance in radiology examinations designed for human radiologists, outperforming other foundational models in benchmarks.
The solution surpassed other foundational models on the Fellowship of the Royal College of Radiologists (FRCR) 2B Rapids examination.
When evaluated using the VQA-Rad benchmark, a dataset of clinically generated questions and answers on radiological images, Harrison.rad.1 achieved 82% accuracy on closed questions.
Also, the model achieved 73% accuracy when evaluated on RadBench, a comprehensive and clinically relevant open-source dataset developed by Harrison.ai.
