From Deep Learning systems like Alphafold, to ML and LLMs, the adoption of tools under the AI umbrella has increased on the Biomedicine sector.
Their usages are wide ranging, from protein folding, to scientific literature processing, research assistance, regulatory document drafting and even limited decision-making support.
There are challenges like the need to access large quantities on non-general and non-public specific data that require technologies like vector databases and Retrieval-Augmented Generation to enable use of this large quantities of data.
The deployment and exploitation of such large scale systems which strong HPC hardware demands is also a big determinant on their use and adoption.
New innovations on this field like reasoning models, distillations and resource constrained-models like DeepSeekR1 push the boundaries of what can be done and the resources required for once untractable challenges, a recurring theme on the field.
Of course, this opportunity comes with a set of challenges and thorny issues in privacy, data security, intellectual property, verifiability and work environment conditions that sparks intense debate.
