Explainable and Personalized Digital Cognitive Stimulation for Older Adults Using AI-Based Adaptive Modeling
Description
This dataset contains anonymized information from 150 older adults who participated in a 24-week longitudinal study on personalized cognitive stimulation using a digital platform powered by explainable artificial intelligence. It includes variables such as experimental group, gender, age, educational level, initial and final cognitive scores, as well as absolute and percentage improvements in performance. The purpose of the dataset is to facilitate replication of analyses, validation of predictive models, and statistical exploration of digital interventions for healthy aging.
Files
Steps to reproduce
1. Download the dataset file Dataset_Cognitive_Intervention_Study.xlsx. 2. Open the file in Excel, Python (e.g., using pandas), or R. 3. Filter participants by experimental or control group using the “Group” column. 4. Compare the “Initial_Score” and “Final_Score” to compute absolute and percentage cognitive improvement. 5. Optionally, apply statistical tests (e.g., paired t-test or Wilcoxon test) to assess significant improvement across groups. 6. To replicate explainability analysis, train an MLP model using scikit-learn and apply SHAP to interpret the contribution of each variable to cognitive score prediction. 7. For cross-validation, use 5-fold validation with the MLP and evaluate using RMSE or R² metrics.
Institutions
- Universidad Cooperativa de Colombia - Medellin
- Universidad de Cordoba