The path towards personalized medicine: helping hospitals make better decisions
The widespread adoption of Electronic Health Records (EHR) unlocks a vast potential for evidence-based medicine, substantiated on real patient history rather than on the limited set of patients that participate in a clinical trial. This could be exploited to build personalized risk profiles or predictive models for specific treatment lines. Although there has been a substantial amount of work in this context, there are still some limitations that are yet to be solved. One of them is the use of unstructured text data, which holds most of the relevant information. This text is considerably difficult to use, given the complexities of the medical terminology. The second limitation is the sheer number of variables that can be explained or used in the models, which make this a needle-in-a-haystack problem. This thesis will tackle only the second point, capitalizing on Savana, an existing system, to solve the first one. Savana is an EHR manager that contains over 100 million records and with a state-of-the-art system that includes the full medical ontology SnoMed. The thesis will apply classic statistics as well as reinforcement learning to the Savana database, with the ultimate objective of providing a way of creating automatic risk profiles and predictive studies.
Requirements: Candidates are expected to be familiar with big data techniques, classical statistics and reinforcement learning. NLP abilities will also be valued but are not essential. They should also be fluent in English.
Full-time contract with exclusive dedication to the PhD thesis.
Documents:Curriculum vitae, academic record, cover letter and two recommendation letters.
We use our own and third-party cookies to improve the functionality of our website and the quality of our service, by analyzing your browsing habits.
For more information about cookies and how to refuse them, please visit our Cookies Policy.
