Sub-Population Identification of Multi-morbidity in Sub-Saharan African Populations

The discussion centers on preparing for a study involving a patient-facing system designed to generate stories or narratives based on user input. The primary focus is on finalizing the questions that will be posed to patients to elicit these stories.

Refining Patient Questions for Data Collection

The team reviewed a revised, simplified list of questions intended for the study. The rationale for the simplicity is to ensure broad comprehension among a diverse public patient population, specifically those diagnosed with diabetes within the last two years.

A key point of discussion was whether the questions were too basic. A concern was raised that the questions primarily target early-diagnosis scenarios, potentially missing the needs of patients further along in their journey. The consensus, however, was to proceed with the simplified questions for the initial study to ensure consistency over time and to gather baseline data. The simplified questions have been tested and proven effective at eliciting responses even from patients who have lived with diabetes for a long time.

Insights from Real-World Patient Queries

To illustrate the complexity of patient needs, team members shared observations from a recent visit to a clinic’s diabetes support group. The questions asked by patients, including those with decades of experience managing their condition, were highly specific and situational. This highlights the gap between standardized data collection and the reality of daily disease management.

Examples of patient questions included:

• Medication and Meal Timing: “When do I take my insulin prior to eating? How long in advance?” and “If I take my insulin, when do I eat? If I don’t, how much do I reduce my food intake?”
• Managing Multiple Medications: How to time insulin when also taking medications for other conditions like hypertension or asthma.
• Dietary Nuances: “Do I take my maize-based meal with milk, water, or a combination? What is the correct proportion?”
• Lifestyle Integration: How the interaction between medication, food, and daily schedules plays out in everyday life.
• Administration Details: “Should I take a tablet while I’m eating or before? How much water should I drink with it?”

System Testing and Grounding

These real-world questions prompted a discussion about the AI system’s capabilities. The team expressed interest in testing the narrative generation system with these highly specific queries to see how it performs. This would test the system’s ability to handle nuance and specificity, particularly whether the input forms gather sufficient information about a user’s lifestyle and diet to provide meaningful advice.

On the technical side, the system is reportedly ready for the study. Current work involves developing a flowchart mapping patient-presented symptoms to potential healthcare professional recommendations, based on ground truth documents. This effort aims to ensure the system’s outputs are more grounded in established research and clinical practice.

• Tags:
• Feature
• Data Science
• Public Health
• Multimorbidity
• Patient Data
• Narrative Generation
• Diabetes