This thesis explores the possibility of using patient reported symptom descriptions to predicthow patients will report their symptoms on 3-D body models. The feasibility of using embeddingmodels to map self-reported symptom descriptions to symptom categories is investigated usingdimensionality reduction techniques such as t-SNE and UMAP. The findings highlight how thereexists promising clustering between various correlated groups of symptoms, indicating that theembedding space of patient reported symptom descriptions may be granular enough todistinguish between symptom groups. The thesis also discusses pathways and potentialimplementations utilizing embeddings and other features from patient reported symptoms toimplement predictive medical-AI models. A thorough discussion regarding bias, marginalization,safety, privacy concerns and the need for transparency in a clinical setting is discussed.Potential benefits of a clinical implementation are discussed, such as enhanced patient-doctorcommunication for marginalized groups, such as non-native speakers, and more efficiency ininitial assessment, specifically for emergency care or local clinical centers. The findings suggestfurther research in applying embeddings to patient reported symptoms could show promise indeveloping the medical AI models of tomorrow’s healthcare system.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.