Medical Diagnostics
Leveraging the wisdom of the crowd to reduce diagnostic error
- This case study leverages the Human Dx platform to provide a tool for training and support in diagnostics problems. The platform is open to medical practitioners who can post cases and crowdsource input from hundreds of thousands of other doctors around the worldβto the benefit of medical professionals and trainees across public and private medical institutions in more than 100 different countries.
- Each medical case is associated with a short description of the patient and a few insights (symptoms, results of medical tests). The system then prompts the solvers to provide their independent differential diagnoses, ranked according to what they believe corresponds best to the case. Finally, the system summarises all the answers provided by the solvers into a collective differential diagnosis.
- HACID will test new ways of collecting and integrating the user feedback by (i) allowing experts to provide evidence in support of the case, (ii) exploiting subjective confidence and competence estimates, (iii) enabling social interaction among users, and (iv) deploying methods for the definition of a collective differential diagnosis produced after reasoning about the user inputs.
Climate Change Adaptation Management
Exploiting collective intelligence for improving resilience against future climate hazards
- This use case focuses on improving existing climate services that support policy makers in adapting and managing their cities so that they are resilient to uncertain future climate changes. The HACID technology can be applied in many urban contexts and for multiple climate hazards. We will hone in on London, UK, where climate service providers (i.e., Met Office) are currently serving UK stakeholders including the London Climate Change Partnership who are preparing their city for extreme weather today and climate change in the future.
- The knowledge base will be populated initially with future climate projection information based on climate models, including those that fed into the latest 6th Assessment Report from the Intergovernmental Panel on Climate Change (IPCC) as well as information supported by National Governments. We also include approaches for the sub-selection (e.g., merging, selecting or filtering) of climate models and associated information and the implementation into an adaptation decision strategy.
- Through a participatory AI approach, we will construct the domain knowledge to represent climate projections, their models and relationships (e.g., models from the same model family with similar components or assumptions) as well as their suitability for different adaptation management decision strategies.
- The collective problem solving step requires experts to provide their own answers to the selection of appropriate climate information. We initially consider answers in terms of identifying relevant climate information, the climate models to be used, and information extraction methods. The combination of the solutions from multiple experts will provide suggestions for a suitable aggregation approach of the most relevant approaches for the use case.
News
The HACID project will be present at the Maker Faire Rome with an interactive game to understand…
We are excited to share our latest findings in the preprint "Human-AI collectives produce the most accurate…
A new publication from the HACID project shows that pooling the independent decisions of General Practitioners…
On Thursday, February the 22nd 2024, Met Office organised a workshop in collaboration with Nesta and…
In this presentation video, we briefly explain the concept of the HACID project and the different streams…