1 | Models and applications that can leverage multimodal data such as merging language and imaging, for example, highlighting anomalies in a natural way (with language) when reading PET images |
2 | Summary of complex medical histories and records |
3 | Summary of information from medical congresses/clinical trial results |
4 | Structuring/making information interoperable, for example, during medical documentation (20) |
5 | Facilitating clinical documentation such as writing discharge report; once we have structured information, is there really a need for free text? (facts should be communicated reliably and concisely) |
6 | Integration with hospital information systems to incorporate patient data, specifications, and requirements (institutional, payer) and resources (staff capacity, provider) |
7 | Interpretation and explanation of other AI algorithms (1) |
8 | Translation into other languages, with big potential for less frequently used languages for which use of natural language processing was limited in the past |
9 | Translation into patient-comprehensible language, making medical information communication more consumer-friendly |
10 | Anamnesis |
11 | Relief for nursing staff through automized ward communication |
12 | Medical writing (21) |
13 | Anonymization of clinical text |
14 | Fairness, bias in LLMs |
15 | Human-in-loop and human-centered design of LLM applications |
16 | Chain-of-thought and automated reasoning on LLMs |