A carregar...
Publicação
Incorporating complex domain knowledge into a recommender system in the healthcare sector
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 801.37 KB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
In contrast to other domains, recommender systems in health sector may benefit particularly from the incorporation of medical domain knowledge, as it provides meaningful and personalised recommendations. With recent advances in the area of representation learning enabling the hierarchical embedding of health knowledge into the hyperbolic Poincaré space, this thesis proposes a recommender system for patient-doctor matchmaking based on patients’ individual health profiles and consultation history. In doing so, a dataset from a private healthcare provider is enriched with Poincaré embeddings of the ICD-9 codes. The proposed model outperforms its conventional counterpart in terms of recommendation accuracy.
Descrição
Palavras-chave
Healthcare analytics Machine learning Recommender systems Poincaré embeddings Primary care International classification of diseases