Artful Path to Healing: Using Machine Learning for Visual Art Recommendation to Prevent and Reduce Post-Intensive Care

TL;DR

This work targets post-ICU syndrome (PICS) by embedding personalized visual art therapy within machine-learning–based visual art recommender systems (VA RecSys). It evaluates four VA RecSys engines—three unimodal (ResNet image, LDA text, BERT text) and one multimodal (BLIP)—against expert-curated recommendations through expert pilot testing and a large-scale user study (n=150). Findings show that visual and multimodal VA RecSys provide therapeutically relevant art that enhances mood and yields comparable or superior outcomes to expert selections, while text-only models elicited negative content and were deprioritized. The study demonstrates the potential of AI-assisted, patient-tailored art therapy to prevent and reduce PICS, with implications for scalable deployment in ICU follow-up and broader healthcare contexts, alongside essential safety and human-in-the-loop considerations.

Abstract

Staying in the intensive care unit (ICU) is often traumatic, leading to post-intensive care syndrome (PICS), which encompasses physical, psychological, and cognitive impairments. Currently, there are limited interventions available for PICS. Studies indicate that exposure to visual art may help address the psychological aspects of PICS and be more effective if it is personalized. We develop Machine Learning-based Visual Art Recommendation Systems (VA RecSys) to enable personalized therapeutic visual art experiences for post-ICU patients. We investigate four state-of-the-art VA RecSys engines, evaluating the relevance of their recommendations for therapeutic purposes compared to expert-curated recommendations. We conduct an expert pilot test and a large-scale user study (n=150) to assess the appropriateness and effectiveness of these recommendations. Our results suggest all recommendations enhance temporal affective states. Visual and multimodal VA RecSys engines compare favourably with expert-curated recommendations, indicating their potential to support the delivery of personalized art therapy for PICS prevention and treatment.

Figures (13)

• Figure 1: Overview of our unimodal approaches to learning latent representations of paintings.
• Figure 2: Overview of our multimodal approach with BLIP to learning latent semantic representations of paintings.
• Figure 3: Overview of our proposed ML-based VA RecSys pipeline. Sample paintings to recall ICU experience (left), our ML-based VA RecSys engines (Middle) and healing painting recommendations (right)
• Figure 4: Sample painting and associated metadata from the dataset.
• Figure 5: Expert reflections on the top three direct recommendations from our text-based VA RecSys engines (BERT and LDA).