How to Develop a Risk Prediction Smartphone App.

Area of Science:

• Health Informatics
• Clinical Decision Support Systems
• Big Data Analytics

Background:

• Predictive models leverage big data for individualized risk stratification, informing clinical decisions and mitigating morbidity.
• Developing real-time, interactive clinical decision support tools is crucial for modern healthcare.

Purpose of the Study:

• To describe the transformation of a large institutional dataset into a real-time, interactive mobile user interface for clinical risk prediction.
• To create a framework for developing actionable, point-of-care applications for risk stratification.

Main Methods:

• Identified a clinical decision point for risk stratification and modification.
• Extracted data from electronic medical records (EMR) including demographics, comorbidities, and operative characteristics using ICD-9 codes.
• Generated and validated surgery-specific predictive models using regression modeling, then developed an iOS/Android mobile application.

Main Results:

• Created individual, specialty-specific, and preoperatively actionable risk models using clustered procedural codes.
• Weighted patient and operative variables using ß-coefficients from longitudinal data across a health system.
• Implemented risk model parameters into specialty-specific modules within a mobile application.

Conclusions:

• A framework for developing clinical decision support mobile user interfaces using longitudinal clinical and administrative data was established.
• Point-of-care applications can aid clinicians in identifying and optimizing risk factors, leading to targeted risk-reduction actions.
• These applications can enhance counseling, informed consent, and shared decision-making, promoting patient-centered care.