Adversarially regularized medication recommendation model with multi-hop memory network
- PDF / 919,032 Bytes
- 18 Pages / 439.37 x 666.142 pts Page_size
- 13 Downloads / 162 Views
Adversarially regularized medication recommendation model with multi-hop memory network Yanda Wang1 · Weitong Chen2 · Dechang Pi1
· Lin Yue3
Received: 21 January 2020 / Revised: 15 September 2020 / Accepted: 19 September 2020 © Springer-Verlag London Ltd., part of Springer Nature 2020
Abstract Medication recommendation is attracting enormous attention due to its promise in effectively prescribing medicines and improving the survival rate of patients. Among all challenges, drug–drug interactions (DDI) related to undesired duplication, antagonism, or alternation between drugs could lead to fatal side effects. Previous researches usually provide models with DDI knowledge to achieve DDI reduction. However, the mixed use of patients with different DDI rates places stringent requirements on the generalization performance of models. In pursuit of a more effective method, we propose the adversarially regularized model for medication recommendation (ARMR). Specifically, ARMR firstly models temporal information from medical records to obtain patient representations and builds a key-value memory network based on information from historical admissions. Then, ARMR carries out multihop reading on the memory network to recommend medications. Meanwhile, ARMR uses a GAN model to adversarially regulate the distribution of patient representations by matching the distribution to a desired Gaussian distribution to achieve DDI reduction. Comparative evaluations between ARMR and baselines show that ARMR outperforms all baselines in terms of medication recommendation, achieving DDI reduction regardless of numbers of DDI types being considered. Keywords Medication recommendation · Drug–drug interactions · Memory networks · Adversarially regularization
B
Dechang Pi [email protected] Yanda Wang [email protected] Weitong Chen [email protected] Lin Yue [email protected]
1
Nanjing University of Aeronautics and Astronautics, Nanjing, China
2
The University of Queensland, Brisbane, Australia
3
Northeast Normal University, Changchun, China
123
Y. Wang et al.
1 Introduction Along with the development of medical and deep learning technologies, effective healthcare applications that cover from physiological condition monitoring and disease diagnosis to onset prediction are becoming more feasible [10,15,19,20,33,37]. The accumulation of electronic healthcare records (EHRs) [11], such as vital signs, laboratory tests, and clinical notes, provides researchers in both medical and deep learning fields with strong supporting evidence for analysis. Most of these applications attempt to assist physicians in better understandings of patient conditions and making accurate clinical decisions. As one of the most important applications, medication combination recommendation [28,35,40] enables physicians to make effective and safe prescriptions according to the medical information that describes the patient condition change, as well as expertise related to massive medications. The comprehensive analysis of patient information and medication kn
Data Loading...
