Clinical decision support systems (CDSS) can support clinicians in selecting appropriate treatments for patients. The objective of this study was to examine if triaging patients with LBP to the most optimal treatment can be improved by using a data-driven approach with the help of machine learning as base of such a CDSS. A clinical database of the Groningen Spine Center containing patient-reported data from 1546 patients with LBP was used. From this dataset, a training dataset with 354 features was labeled on eight different treatments actually received by these patients. With this dataset, models were trained. A test dataset with 50 cases judged on treatments by 4 experts in LBP triage was used to test these models with data not used to train the models. Prediction accuracy and average area under curve (AUC) were used as performance measures for the models.Aims
Methods
Patients with Neck and/or Low Back Pain (NLBP) constitute a heterogeneous group with the prognosis and precise mix of factors involved varying substantially between individuals. This means that a one-size-fits-all approach is not recommended, but methods to tailor treatment to the individual needs are still relatively under-developed. Moreover, the fragmentation of disciplines involved in its study hampers achieving sound answers to clinical questions. Data mining techniques open new horizons by combining data from existing datasets, in order to select the best treatment at each moment in time to a patient based on the individual characteristics. Within the Back-UP project (H2020 #777090) a multidisciplinary consortium is creating a prognostic model to support more effective and efficient management of NLBP, based on the digital representation of multidimensional clinical information. Patient-specific models provide a personalized evaluation of the patient case, using multidimensional health data from the following sources: (1) psychological, behavioral, and socioeconomic factors, (2) biological patient characteristics, including musculoskeletal structures and function, and molecular data, (3) workplace and lifestyle risk factors.Background
Method
The aim of this study was to investigate the agreement of physician assistants (PAs) in the triaging of patients with Low Back Pain (LBP) based on self-reported data. A cross sectional vignette study among four PAs was carried out. Vignettes (cases) were constructed including 26 factors that can be self-reported, identified in literature that have predictive value in treatment outcomes (for example red flags indicating serious underlying conditions and yellow flags indicating psychosocial factors). All vignettes were randomly assigned to the PAs who should determine what intervention would be most optimal to the patient (rehabilitation, injections, medications, surgery, primary care psychology, primary care physical therapy). PAs were allowed to advise more than one intervention. Per vignette, 3 PAs were assigned randomly to advise on intervention. Fleish kappas were calculated to determine the interrater reliability.Aims
Patients and methods
