What question did this study set out to answer?

The aim is to develop an AI-based question answering system to automate patient prioritization in emergency departments.

March 23, 2026Open Access

AI-based QA system for clinical decision making

Key Points

The aim is to develop an AI-based question answering system to automate patient prioritization in emergency departments.
Utilized deep learning techniques, including MLP, LSTM, TabNet, and TabTransformer.
Simulated healthcare professionals' reasoning processes through questions.
Evaluated model performance on a custom dataset from a Tunisian hospital.
TabNet-based model achieved an R2 of 0.9997, indicating high prediction accuracy.
Mean Squared Error (MSE) of 0.0004 suggests minimal error in the model's predictions.
TabNet outperformed other models in terms of effectiveness.

Abstract

In this paper, we propose an Artificial Intelligence (AI) based Question Answering (QA) system in order to automate patient prioritization. In more specific medical subarea like emergency departments (EDs), healthcare staff face major challenges such as managing patient prioritization and complex clinical decision-making. Our study focuses on the deployment of Deep Learning (DL) and transformer-based models in specific medical applications. Indeed, we simulated the reasoning process underlying the questions asked by healthcare professionals. We evaluated MLP, LSTM, TabNet and TabTransformer models on dataset constructed from informations provided in Tunisian hospital. The proposed system effectiveness is assessed using various metrics. Results showed that TabNet-based model outperforms others with an R2 of 0.9997 and an MSE of 0.0004.

AI-based QA system for clinical decision making

Key Points

Abstract

Cite This Study