An intelligent hospital information system aggregates and visualizes patient data in clinics. The abundance of information generates therapy-related knowledge that benefits patients and improves working conditions of health care professionals.
Three months ago, a patient fell down and broke his wrist and has been complaining about persistent pain ever since. One look at the electronic patient file tells the attending therapist how the surgery went, which therapies the patient has already undergone and that he is suffering from rheumatoid arthritis. This helps determine a suitable treatment method quickly and efficiently – at least that’s the idea of digitalization in the health care system. However, in reality during a direct patient consultation doctors and therapists typically have little time to filter the relevant information out of doctor’s letters, laboratory results and medication regimen.
In the future, intelligent automated data processing will assist attending physicians with gathering the relevant information exclusively – at a glance. Within the project “SMARAGD – Smart Aggregation and Visualization of Health Data” which is co-financed by the Research Promotion Agency (FFG), a multidisciplinary team of scientists from the fields of data science, computer science, health sciences, graphic design, psychology, law and sociology developed a prototype application for physiological and occupational therapists that summarizes and visualizes health data clearly. Duration: 06/2019-11/2021Project-Infobox
Funding: FFG (Ref.: 872575)
Lead: IMC-Fachhochschule Krems
Partner: JKU Linz, Know Center, Med-Uni Graz, Synyo, KFU Graz, Universität Wien
Clinical test data sets via a hospital information system form the basis. In addition, researchers guided physical and occupational therapists throughout their day-to-day work routine to clarify which information is relevant for the respective treatment. For example, it can be helpful if the image of a wrist includes information on how far it was last mobile. All data was recorded and analyzed in compliance with strict data protection guidelines.
Ensuring a corresponding data quality by linking different data sources poses a challenge. In addition to structured, formalized patient data, such as the duration of hospitalization and medical diagnoses, there are typically supplementary handwritten records that proved difficult to be digitalized. Medical terms are often abbreviated differently depending on the ward, hospital or attending physician, which makes further processing more complicated.
“We use the latest methods of artificial intelligence from the areas of clinical natural language processing and semantic data analysis to process large amounts of data automatically and prepare it for further analysis in a structured manner,” stated Mark Kröll of Know-Center and explained the procedure: “The semantic processing of data means giving meaning to individual words or passages of diagnostic findings, such as ‘medication’ and ‘diagnosis’. Subsequently we extract and aggregate relevant information, for example, connections such as ‘medication – dosage’ or ‘injury – body part’. By combining and visualizing health data from therapy findings, the previous course of the disease is demonstrated clearly.”
By taking previous therapies into account, the applied AI methods assist therapists in making complex diagnoses and therapy decisions quickly and reliably. This reduces the existing time pressure and qualitatively helps improve working conditions. By focusing on the quality of treatment it ultimately benefits the patient. The project’s results can be utilized by hospital operators and hospital information system providers and can be implemented in existing systems.