Crohn's disease prediction model
PREMONITION-CD | TAKEDA
YEAR: 2017-2020.
CUSTOMER: Savana (a company specialised in extracting the entire clinical value existing in electronic health records using Artificial Intelligence and Natural Language Processing (NLP).
TECHNOLOGY: Big Data, Natural Language Processing and Machine Learning.
Addressing needs and bringing benefits
Traditional research, especially research centred on retrospective studies, is very complex to carry out as it means having to manually extract data from health records, since it is mostly available in unstructured text. By using natural language processing, data can be collected automatically, thus facilitating the research work.
Description
The PREMONITION-CD study is one of the first projects carried out in Spain that has successfully used natural language processing. In addition, the access to thousands (nearly 6,000) health records of patients with inflammatory bowel disease has, on the one hand, made it possible to describe patients with this pathology and, on the other, develop a predictive model based on artificial intelligence, machine learning and complex statistical techniques, such as random forests and logistic regression. This model has identified different factors that will enable clinics to keep abreast of the situation when making decisions about patients whose diseases are most likely to worsen.