Data-driven Patient Profiles: Definition, validation, and implementation for tailored orthopaedic healthcare services

In order to provide patients with the highest possible quality of care, healthcare institutions often standardize the way they provide healthcare. Yet, there are also more and more calls for tailored healthcare services that are intended for one specific person and based on characteristics that are unique to that person. This dissertation investigates tailored healthcare services and does so specifically in the orthopaedic context. Orthopaedic patients, in particular patients who have undergone joint replacement surgery of the hip or knee joint, are relatively dissatisfied with the current healthcare service provided to them. Specifically, the communication with total joint replacement patients (including the way in which patients are informed about the surgery, its risks and the treatment plan, but also the emotional support they receive from healthcare providers) often leaves something to be desired. In examining tailored healthcare as a potential solution to dissatisfaction with patient-provider communication, this dissertation focuses on the definition, validation and implementation of so-called patient profiles. Patient profiles represent the common characteristics of a specific subgroup of patients that are unique compared to the overall patient population. The patient profiling approach is derived from the principles of mass customization and assumes that representations of the common and unique preferences, needs, and competences of different groups of patients can be used to design tailored healthcare services. These tailored healthcare services can then be offered to individual patients based on their profile. It is expected that tailored healthcare services will lead to improvements in patient experience. This dissertation examines patient profiles and the effect of the patient profiling approach on patient experience following four questions: (1) what are relevant patient characteristics for patient profiling?, (2) which data driven patient profiles can be distinguished?, (3) which orthopaedic healthcare services are suitable for tailoring?, and (4) what is the effect of tailored healthcare services on patient experience? These questions are approached using the biopsychosocial model. The biopsychosocial model assumes that biomedical factors (such as pain and physical functioning) as well as psychological and social factors (such as coping mechanisms and communication preferences and competences) influence how someone experiences their illness, and therefore, what type of healthcare service would suit them. A combination of research methods including observations, interviews, questionnaires, machine learning, systematic literature reviews and experiments were used to answer the specific research questions...