Improved Inpatient Deterioration Detection in General Wards by Using Time-Series Vital Signs

Although in-hospital cardiac arrest is uncommon, it has a high mortality rate. Post-cardiac arrest survival rates are associated with care team awareness, and therefore, identifying at-risk patients is critical. Early warning scoring systems are generally used to identify hospitalized patients at risk of deterioration. However, these systems often require clinical data that are not always regularly measured. Therefore, in this study, we developed a more accurate, machine-learning model, the time series early warning score (TEWS), to predict clinical deterioration using only heart rate, systolic blood pressure, and respiratory data, which are regularly measured in general wards. We tested the performance of the TEWS in two tasks performed with data from the electronic medical records of 16,865 adult admissions and compared the results with those of other classification algorithms. The TEWS detected more deteriorations with the same level of specificity as the other algorithms did when vital signs data from the 48 h prior to an event were input. Our framework improved in-hospital cardiac arrest prediction and demonstrated that previously obtained vital signs data can be used to identify at-risk patients in real time. This model may be an alternative method for detecting patient deterioration.