Modelling Patient Longitudinal Data for Clinical Decision Support: A Case Study on Emerging AI Healthcare Technologies
研究开发了一种深度状态空间模型，用于处理高维纵向电子健康记录，结合非结构化医疗笔记和可解释注意力机制，提升患者风险预测准确性，对临床决策支持有重要价值。
Abstract The COVID-19 pandemic has highlighted the critical need for advanced technology in healthcare. Clinical Decision Support Systems (CDSS) utilizing Artificial Intelligence (AI) have emerged as one of the most promising technologies for improving patient outcomes. This study’s focus on developing a deep state-space model (DSSM) is of utmost importance, as it addresses the current limitations of AI predictive models in handling high-dimensional and longitudinal electronic health records (EHRs). The DSSM’s ability to capture time-varying information from unstructured medical notes, combined with label-dependent attention for interpretability, will allow for more accurate risk prediction for patients. As we move into a post-COVID-19 era, the importance of CDSS in precision medicine cannot be ignored. This study’s contribution to the development of DSSM for unstructured medical notes has the potential to greatly improve patient care and outcomes in the future.