Affiliation:
1. Rotterdam University of Applied Sciences: Hogeschool Rotterdam
2. Erasmus Medical Centre: Erasmus MC
Abstract
Abstract
Chronic conditions and multi-morbidity affect 50 million individuals within Europe and are the leading causes of disablement and death. Obesity, inactivity, chronic pain/fatigue and mental health issues are reported in 25–85% of all people with a chronic disease, irrespective of the pathological mechanism. The prevalence of comorbidity increases with age; still, in Europe, 25% of young people are affected by one or more chronic conditions. These youngsters are less likely to reach their full intellectual potential, have uncertain future perspectives, and frequently experience social exclusion. The underlying causes for the development or persistence of these comorbidities involves a myriad of complex mechanisms that are not solely disease specific but often individualized in personal and environmental factors. This complexity is challenging for health professionals and there’s a need for developing early detection tools. For this study 250 community-dwelling youth with one or more chronic conditions for over a year. Data was collected via electronic survey and combined with public data on living conditions. Machine learning RF-models were used to create risk-profiles for mental health issues, chronic fatigue, and severe disability based on 31 features. Risk profiling through RF-modelling showed adequate performance metrics, discriminating between youth who developed mental health issues, chronic fatigue, and severe disability and those who did not (ROC-AUC:.81-.86). Somatic symptoms, pain-related coping, and living environment were found to be the most contributing features to the RF-models (LIME Relative FI:7.0-24.6%). Data-supported clinical decision making can aid in identifying youth at risk for disabling comorbidities, even in non-specialized settings.
Publisher
Research Square Platform LLC