Adaptive Patient Survey Question Selection Using Sensor Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automated patient surveys are static and do not adapt to specific patient aspects or journeys, requiring significant time to select appropriate questions for large databases.
Innovation Solution
A computer-implemented method that uses patient data, including sensor and medical data, to dynamically select questions through a question selection algorithm, considering patient health conditions, location, and environmental factors, utilizing machine learning for personalized surveys.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If static predetermined questions are used in patient surveys, then the survey structure is simple and easy to implement, but the survey cannot adapt to specific patient aspects or journeys
Solution Approach 1:
The patent implements dynamic question selection by training a machine learning algorithm (random forest classifier) that automatically selects which questions to present based on patient-specific data from electronic health records and sensor data. The system transitions from static predetermined questions to dynamic adaptive questioning, where the survey structure changes based on patient characteristics, health conditions, and journey stages.
Solution Approach 2:
The system changes the parameter of question selection from fixed to variable based on multiple patient parameters including health conditions, demographic data, sensor measurements, and journey stage. The machine learning model processes these parameters to determine the optimal subset of questions for each patient, enabling adaptability without requiring complete questionnaires for all patients.
2Productivity
If questions are manually selected for each patient in large databases, then high personalization can be achieved, but significant time is required for question selection
Solution Approach 1:
The system implements self-service automation where the machine learning algorithm autonomously selects questions for each patient without requiring manual intervention. The trained random forest classifier automatically processes patient data and generates personalized question selections, enabling the system to serve large patient databases efficiently without human time investment in the question selection process.
Solution Approach 2:
The patent replaces the mechanical process of manual question selection with an automated computational system. The machine learning algorithm substitutes human analysts, automatically processing patient records and sensor data to determine appropriate questions, thereby eliminating the time-consuming manual effort while maintaining or improving personalization quality.
3Measurement precision
If comprehensive patient data is analyzed for each patient, then highly personalized surveys can be created, but the processing complexity and time increase significantly
Solution Approach 1:
The system extracts only the most relevant features and data elements needed for question selection from comprehensive patient records. The random forest classifier identifies and processes key parameters from electronic health records and sensor data, filtering out unnecessary information. This extraction approach enables high personalization accuracy by focusing on critical patient characteristics while avoiding the computational burden of processing all available data in detail.
Data Source
AI summary
A method for selecting questions to be answered by a patient includes receiving, from a patient database (3), patient data (5) indicative of a health condition of the patient. The patient data (5) includes sensor data (7) which has been generated by at least one sensor (8) for determining the health condition of the patient. The method further includes inputting the patient data (5) as input data (9) into a question selection algorithm (10) configured for selecting questions, based on the input data (9), from a list (6) of predetermined questions stored in a question database (4). The method further includes outputting at least one selected question or a list (13) of selected questions to be answered by the patient as output data (14) by the question selection algorithm (10). The patient data (5) includes at least one of anamnesis data (16), or medication data (19) of the patient.