SOFA Score Machine Learning Sepsis Mortality Prediction

SUMMARY:

  • Evaluation of repeated SOFA scores may provide a better 30 day sepsis mortality predictability.

  • Machine learning algorithms outperform conventional SOFA delta scores on 30 day mortality.

  • Several limitations present with longitudinal SOFA scores.

REVIEW: 

  • Delta SOFA scores comparing day 1 to day 7 are used to predict 30 day mortality probability.

  • Longitudinal SOFA scores with machine learning algorithms may improve from the delta SOFA score limitations.

  • Development cohort 252 patients; 7 ICUs with Sepsis-3 diagnostic criteria.

  • Validation cohort 1,790 patients, retrospective public database.

  • 7 different machine learning algorithms evaluated.

  • All models used to predict 30 day mortality.

  • Machine learning algorithms outperformed conventional delta SOFA score 30 day mortality predictions (P<0.05).

  • The superior results were maintained even with a shorter timeframe than 7 days.

  • Unknows Remain:

    • Although earlier predictability was found, 3 days maybe too delayed in the clinical setting.
    • It is unknown what happens to predictability when missing values in SOFA determination exist.
    • Addition of more patient context parameters than SOFA values maybe of greater benefit.

CONCLUSIONS:

  • Machine learning based algorithms of daily SOFA trajectory produced significantly greater accuracy of mortality predictability as compared to conventional daily SOFA delta scores.

  • Further investigation is needed, but this appears to be the appropriate promising direction for individual patient assessments.

  • Several unknown areas remain to be determined.

To receive articles like these in your Inbox, you can subscribe to Sepsis Program Optimization Insights.

Subscribe to sepsis program optimization insights

Erkan Hassan is the Co-Founder & Chief Clinical Officer of Sepsis Program Optimization where he designs & oversees the implementation of solutions to optimize sepsis programs.

To discuss your organization’s Barriers of Effective Sepsis Care, contact Erkan by phone (844) 4SEPSIS (844-473-7747), email (erkan@spo.icu), or video chat.