Hospital unexpected readmission using multi-model prediction

Inhalt:
Hospital unexpected readmission prediction can help hospitals prepare for unexpected patients, which avoids insufficient resource allocation. The prediction score can also be used by doctors to evaluate whether a particular patient is dischargeable or not. MIMIC-III is a big, publicly available database including information on over a thousand people and their medical conditions. The MIMIC-3 dataset consists of both structural (e.g. ICD, and demographic) and nonstructural (e.g. nursing notes) data. In this paper, we provided a multi-modal predictor of patient readmission by combining unstructured clinical text and structured data. The paper analytically shows the advantages of averaged multi-model in prediction of unplanned readmission.