A Data-Driven Model to Predict Quality of Life Dimensions of People with Intellectual Disability Based on the GENCAT Scale

In recent times, observers have noticed that people with intellectual disability (ID) experience increasing complexity in their older age. Many initiatives launched by healthcare organisations and government bodies are rigorously working to improve ID people’s quality of life (QoL) and health status. The concept of QoL is rooted in a multidimensional framework comprising both universal (etic) and culture-bound (emic) components. It has objective and subjective features and is affected by individual and environmental factors. The professionals in QoL proposed eight dimensions to cover every aspect of ID people, including emotional well-being, interpersonal relationships, material well-being, personal development, physical well-being, self-determination, social inclusion, and rights. In the last decades in Catalonia, the professionals suggested the GENCAT scale predict these eight dimensions’ values through a set of questionnaires containing 69 questions. The professionals use the beneficiary’s response the heir to 69 questions based on four point frequency scale. The GENCAT scale tool converted these 69 questions’ answers into eight values corresponding to the eight QoL dimensions. The GENCAT tool uses a set of rules and some correlatable tables to evaluate the eight dimensions of each beneficiary. In this work, we propose using machine and deep learning-based models instead of the GENCAT tool to estimate the eight dimensions values. Based on the private Newton One dataset, we train various machine learning (ML), such as Random Forest and Decision Trees, along with Deep Neural Networks (DNNs) models to predict the eight dimension values. The trained models predict the eight values by feeding with the 69 questions responses of the beneficiaries. We evaluate the performance of the various models using the Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and $$R^2$$ R 2 scores. The proposed model based on DNNs achieved the best results among all tested models with MAE of 1.5991, RMSE of 3.0561, and $$R^2$$ R 2 of 0.9565. The study shows the promise of the machine and deep learning-based models, particularly DNNs, as a more effective and precise substitute for the GENCAT scale for calculating the eight dimensions of QoL in people with ID. The results open the door for better QoL evaluations and individualised interventions to improve this population’s well-being as they age.