Gen AI in healthcare companies: many projects but a common language is needed

One of the most discussed implementations is that of the virtual agent to support clinicians. Designing an assistant for oncology, for example, allows replicating its architecture in other areas. Automatic transcription processes are also extensible from one discipline to another. Therefore, harmonizing pilot experiences with more general business strategies is essential.

Another critical issue: what to buy and what to develop internally? Healthcare companies have large databases. In the medical field, GenAI already has a greater generalization and is often configured as a digital device with defined development paths (e.g. algorithms for the second mammography reading). In healthcare, however, customized solutions are needed, making partnerships with suppliers capable of on-site development strategic. Each company can thus become an experimental laboratory, not only in implementation but also in design and testing. Leading the market – and not undergoing it – accentuates the strategic relevance of investments.

There is excitement: automated triage, diagnosis support (image and report analysis), automatic generation of clinical summaries, administrative optimization (documents, letters, authorizations, classifications, chatbots). Training paths are also evolving: on-the-job training (learning by doing), AI tutors, more understandable reports for patients. The repertoire of innovations underway is vast and reconstructing it is essential: each context tells its own experience, but a “family photo” that gives an overall picture is still missing.

Therefore, there is a need for collaboration between companies, rather than competition, to share learning processes. The PNRR has stimulated a lot in this direction. Now is the time to ask ourselves what results it is producing: from narratives we should move on to evidence. What changes have been generated? What are we learning? Because, as AI teaches, learning is the most important generative stimulus.

* Director of EMMAS Master and DASP Network - SDA Bocconi School of Management