Chronic musculoskeletal pain classification dataset in older adults for AI algorithms,

Abstract

Artificial Intelligence in healthcare requires data designed specifically to ensure the functioning of the models. This research aimed to configure a dataset of 256 patients with musculoskeletal pain classified into four levels: None, Low, Medium and Strong. Using a six-phase methodology: 1. Location of useful sources; 2. Selection of classes and criteria; 3. Interview design and implementation; 4. Audio recording; 5. Processing of recordings; and 6. Conformation of the dataset. The results were a database with 256 audios in patients from 24 to 84 years old. This dataset can be implemented in physiotherapy prototypes that facilitate its application to patients and doctors.