Improving the Prediction of Individual Engagement in Recommendations Using Cognitive Models

TL;DR

This paper demonstrates that cognitive models based on Instance-Based Learning can more accurately predict individual behavior changes in health interventions than traditional time-series models, offering insights into decision-making processes.

Contribution

The study introduces the use of IBL cognitive models to enhance prediction accuracy and interpretability in health behavior interventions, surpassing standard computational models.

Findings

IBL models outperform LSTMs in predicting individual health behavior dynamics

IBL provides estimates of behavioral volatility and intervention sensitivity

Cognitive models improve intervention targeting efficiency

Abstract

For public health programs with limited resources, the ability to predict how behaviors change over time and in response to interventions is crucial for deciding when and to whom interventions should be allocated. Using data from a real-world maternal health program, we demonstrate how a cognitive model based on Instance-Based Learning (IBL) Theory can augment existing purely computational approaches. Our findings show that, compared to general time-series forecasters (e.g., LSTMs), IBL models, which reflect human decision-making processes, better predict the dynamics of individuals' states. Additionally, IBL provides estimates of the volatility in individuals' states and their sensitivity to interventions, which can improve the efficiency of training of other time series models.