Clinical Scorecard: A Learning Health System Enhanced by Artificial Intelligence for Improved Sepsis Detection and Care Quality: A Comparative Study

At a Glance

Key Highlights

• Sepsis remains a major global health priority with 50 million cases and 11 million deaths annually worldwide.
• The SLHS integrates a standardized clinical pathway with AI-powered monitoring to improve sepsis detection, documentation, treatment adherence, and outcomes.
• HERACLES, a machine learning model combining Random Forest and LSTM algorithms, predicts sepsis probabilities in 6-hour patient stay fragments to generate actionable clinical indicators.

Guideline-Based Recommendations

Diagnosis

• Use AI-driven prediction models like HERACLES to classify patient stays into no sepsis, possible sepsis, or confirmed sepsis categories.
• Incorporate continuous data collection and analysis within a Learning Health System framework to enhance early recognition.

Management

• Adhere to Surviving Sepsis Campaign guidelines emphasizing timely antibiotic administration and fluid resuscitation within a comprehensive care bundle.
• Implement standardized clinical pathways integrated with AI alerts to guide targeted interventions.

Monitoring & Follow-up

• Utilize continuous digital monitoring pipelines and sepsis registries to track patient data and care quality indicators longitudinally.
• Regularly evaluate AI model performance and clinical adherence through dashboards and feedback loops.

Risks

• Recognize limitations of ICD coding for sepsis surveillance due to low sensitivity and variability.
• Address challenges in AI tool adoption including lack of continuous performance evaluation and integration of new clinical insights.

Patient & Prescribing Data

57,180 unique hospitalized patients with 97,559 stays across multiple hospital wards

AI-enhanced monitoring supports improved adherence to sepsis treatment protocols and facilitates targeted clinical interventions to optimize outcomes.

Clinical Best Practices

• Deploy AI-powered clinical decision support tools within a structured Learning Health System to enable systematic quality improvement.
• Integrate multidisciplinary data sources into a centralized sepsis registry for comprehensive patient monitoring.
• Implement phased rollout of standardized sepsis pathways across hospital units with continuous evaluation and adaptation.

References

• Surviving Sepsis Campaign Guidelines
• WHO Sepsis Fact Sheet
• Learning Health Systems Framework