Large-Scale Evaluation and Liver Disease Risk Prediction in Finland's National Electronic Health Record System:...

Area of Science:

• Medical Informatics
• Public Health
• Predictive Analytics

Background:

• Chronic liver disease incidence and mortality are rising globally.
• Early detection is crucial but challenging, often occurring at symptomatic stages.
• The Chronic Liver Disease (CLivD) score is a predictive model for forecasting liver disease risk.

Purpose of the Study:

• To assess the feasibility of automating the CLivD score using the Kanta electronic health record system.
• To identify necessary improvements in Kanta for accurate, automated risk detection.

Main Methods:

• Utilized a real-world data repository (Kanta) with 96,200 individuals' medical histories.
• Developed processes to manage missing input data for the CLivD risk calculation.
• Simulated improvements by incorporating unstructured text and approximating variables.

Main Results:

• Current structured Kanta data is insufficient for precise CLivD risk calculation.
• Incorporating unstructured text and diagnosis approximations significantly improved low and moderate risk identification.
• Missing waist-hip ratio data prevented high-risk category identification; simulations showed potential for improvement.

Conclusions:

• Structured Kanta data alone is inadequate for precise risk assessment of diseases influenced by obesity, smoking, and alcohol.
• Simulations indicate up to 14% improvement in risk detection with enhanced data input.
• Kanta has potential for nationwide automated risk detection, enhancing disease prevention and public health.