A large number of electronic patient records are produced within healthcare. These records are seldom reused because of ignorance, but also a lack of appropriate tools to process them.
Previous research has used over 2 million Swedish hospital records from the Karolinska University Hospital for the years 2007 to 2014. Partly, to automatically detect and predict healthcare associated infections and also to find the side effects of drugs. To accomplish the research, the text in the patient records have been manually annotated by physicians, and then different machine learning tools have been trained on these annotated texts to simulate the doctors’ intelligence.
Of all patients treated in hospitals, 10 percent may obtain a healthcare related infection that causes much suffering for the patient, but also costs for the society. About 5 percent of all patients suffer from side effects of drugs. If one can detect and prevent these infections and side effects much would be gained.
The research question "Artificial intelligence analysing patient records. Is this possible and can improve health?" will be answered by synthesizing past studies in a book. The book will describe how to proceed to get access to patients' medical records, ethical dilemmas about having access to patient records and how records created anonymously automatically before being used. Finally, the book will describe methods to build the tools that will improve healthcare. Part of the edition project locates as a sabbatical stay at CSIRO/Macquarie University in Sydney, Australia.
Riksbankens Jubileumsfond Sabbatical
With RJ Sabbatical researchers can receive funding for up to a year to finish far advanced research and to write syntheses. The aim is also to strengthen Swedish research and international educational contacts by including stay abroad for a longer period.
Click this link for more information in the Swedish language about Riksbankens Jubileumsfond