Background
Currently, a vast amount of data is compiled in the electronic patient record systems, both as numerical data (test results, etc.) and as free-text data, and it is becoming an increasing challenge to maintain the necessary overview of the contents of these patient records.
Purpose/Vision
The purpose of RELIP is partly to prevent collection of data that is already available in the patient record systems, and partly to prevent staff from ignoring information that the patients expect staff to be acquainted with or may be essential to patient treatments and prognoses.
A software program (an algorithm) that provides automatic read support will be developed as part of the project. The program can combine selected data from patient records and thus provide efficient staff decision support. The goal is to develop a software program that will inform staff when patient record data reaches a level that indicates a high probability of a complex alcohol relationship.
The Project Implementation
The project builds on an ongoing study (Relay) that systematically collects data on diet, smoking, alcohol, and exercise from more than 5,000 patients in selected somatic wards. Patients who are screened as part of the Relay study and tested positive for Alcohol Use Disorders (AUD i.e. diseases associated with alcohol abuse), as well as those tested negative, will form the nucleus of the RELIP project. The project will investigate whether the patient records of the two groups differ on a number of variables by analysing the patient log data. Text-based data will be included and analysed using the method Natural Language Processing. Existing biological/mathematical models and the results obtained from the analyses will form the basis for the development of an algorithm aimed at identifying an overconsumption of alcohol or an addiction.
The project is divided into the following steps:
Step 1: Data collected during the first year of the Relay study will be analysed and AUD indicators identified by means of data from the patient records (numeric, biological and free-text).
Step 2: Existing biological models (especially The SteatoNet model) will be adjusted.
Step 3: The empirical findings from Step 1 and the biological models from step 2 will be combined in an overall model.
Step 4: The overall model will be tested on patient data achieved during the second year of the Relay study. The model's validity and specificity will be tested.
Step 5: Development of a software program for clinical decision support.
Step 6: A qualitative study will be performed among patients and staff to examine experiences and expectations in terms of using the software program for clinical decision support.
Step 7: A protocol for implementing the software program for clinical decision support in hospitals will be developed in collaboration with clinicians.
Step 8: The business model will be analysed and include an evaluation of economic and organisational perspectives of implementing the software program.
Expected Results
Expectations are that the project results will lead to the development of a prevention model that is based on advanced search and language technology. The aim is to identify signs in the patient records that a patient's lifestyle affects his/her medical condition, thus paving the way for staff to actively involve the patient and discuss how treatment can best be organised.