AUTHOR=Wan Cheng , Feng Wei , Ma Renyi , Ma Hui , Wang Junjie , Huang Ruochen , Zhang Xin , Jing Mang , Yang Hao , Yu Haoran , Liu Yun 

TITLE=Association between depressive symptoms and diagnosis of diabetes and its complications: A network analysis in electronic health records

JOURNAL=Frontiers in Psychiatry

VOLUME=Volume 13 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/psychiatry/articles/10.3389/fpsyt.2022.966758

DOI=10.3389/fpsyt.2022.966758

ISSN=1664-0640

ABSTRACT=Objectives: Diabetes and its complications are commonly associated with depressive symptoms, but few studies have investigated the diagnosis effect on depressive symptoms in patients with diabetes. This study uses a network approach to explore the association between depressive symptoms, annotated from electronic health record (EHR) notes by a deep learning model, and diagnosis of type 2 diabetes mellitus (T2DM) and its complications.

Methods:
This study uses anonymous admission notes of 52139 in-patients diagnosed with T2DM at the First Affiliated Hospital with Nanjing Medical University from 2008 to 2016. We propose a symptom annotation model named T5-depression based on transformer architecture to annotate depressive symptoms from present illness and measure the model's performance by the F1-score and area under the receiver operating characteristic curve (AUROC). We construct networks of depressive symptoms and examine the connectivity of these networks in patients diagnosed with T2DM including those with certain complications.

Results: The T5-depression model achieved the best performance compared with benchmark models, with an F1-score of 91.71 and AUROC of 96.25. The connectivity of depressive symptoms in patients showed a statistically significant increase in the 2 years after diagnosis with T2DM (p=0.025) and hypertension (p=0.013), consistent with the number of patients diagnosed with depression.

Conclusions: The T5-depression model proposed in this paper can effectively annotate depressive symptoms in EHR notes. The connectivity of annotated depressive symptoms is associated with diagnosis of T2DM and hypertension. Changes in the network of depressive symptoms generated by T5-depression could be used as an indicator for depression screening.