Validation of IBM-WATSON Health Trauma Pathway explorer, a visual based analytics tool to predict outcomes of poly-trauma patients

Fragestellung Big databased artificial intelligence (AI) is on the way to develop into a part of daily clinical life and its reasonable application could help to improve disease or injury outcomes. A visual polytrauma analytics tool based on IBM Watson was developed and described in a previous publication (1). Here, the validation of the IBM Watson Health Trauma Pathway explorer took place.

Methodik The study design is a retrospective prediction model validation. Sixty patients with an Injury Severity Score (ISS) ≥ 16 and age ≥ 16 were included in the validation. They were randomly selected from the electronic medical records of the University Hospital Zurich and were inde-pendent of the Watson based visual analytics tool. Age, ISS, temperature and the presence of head injury were the predictors used for the validation of the following three outcomes: SIRS and sepsis within 21 days since admission of the patient, as well as early death within 72 hours since admission. The ROC analysis was used to determine the predictive quality.

Ergebnisse und Schlussfolgerung A receiver operating characteristic (ROC) curve for each outcome was generated to show the predictive accuracy of the trauma tool. The area under the curve (AUC) for predicted SIRS is 0.775 (96% CI: 0.559-0.902), for predicted sepsis 0.631 (95% CI: 0.517-0.721) and for predicted early death 0.790 (95% CI: 0.37-1.0).

The validation has shown that the predictive performance for SIRS and early death within 21 days since admission of the patient corresponds with the clinical outcome in nearly 80% of cases. This visual analytics tool for polytrauma patients can be used to obtain valid predictions for SIRS, sepsis and early death. Here, we can present a possible working variant of AI in trauma surgery.

Stichwörter Artificial Intelligence, Watson Health, Mashine Learning, Databank

Publication History

Article published online:
15 October 2020