📈 ML-Powered Personal Analysis: How It Works
The Personal Analysis feature uses advanced statistical modeling to derive your specific physiological constants—like your personal sensitivity to carbohydrates and insulin—from your logged glucose, meal, and medication data.
NOTE: Everything is processed on your device only. No servers are involved and your data stays local and private!
It is designed to give you personalized data points to facilitate a more targeted discussion with your healthcare provider.
What Sets It Apart from Pattern Insights
| Feature | Pattern Insights (Rule-Based) | Personal Analysis (ML-Derived) |
| Purpose | Reactive Diagnosis (What happened & where). | Proactive Efficacy (How sensitive you are & how much). |
| Input Logic | Fixed rules and clinical thresholds (e.g., Is BG > 180? Was medication late?). | Sophisticated statistical models that quantify relationships between all factors. |
| Key Output | Events & Frequency (e.g., “Dawn Phenomenon occurred on 65% of mornings”). | Physiological Constants (e.g., “1 unit of insulin lowers your BG by 8 mg/dL”). |
| Clinical Value | Helps you see recurrence and timing problems. | Helps you refine insulin doses and carb ratios with precision. |
🔒 Safety & Educational Disclaimer
🚨 IMPORTANT: Educational Tool, Not Dosing Guide
GluClue does not provide medical guidance or insulin dosing instructions. All ratios, factors, and suggested actions are based purely on statistical analysis of your logged data and must not be used to make treatment changes on your own.
The purpose of this personalized analysis is to educate you and your healthcare provider on how your body specifically reacts to food and insulin. This enables a more targeted and personal discussion about your current treatment plan, moving beyond generalized diabetes recommendations. Always consult your doctor or Certified Diabetes Educator (CDE) before adjusting insulin doses, changing medications, or altering your treatment strategy.
Key Personal Insights Explained
These factors are derived from your glucose trends, meal entries, and actual medication times to give you a personal dosing toolkit.
1. Average Carbohydrate Impact Factor
| Your Data Point | Example: 2.4 mg/dL/g |
| What it is: | The average amount your blood glucose rises for every 1 gram of carbohydrate you consume. This is your personal baseline reaction to food. |
| How to use it: | Use this to estimate the size of the challenge your bolus insulin needs to overcome. If you consume 50g of carbs, you know you need insulin to counter a likely $\sim 120 \text{ mg/dL}$ rise (50g $\times$ 2.4). |
2. Time-of-Day Carb Sensitivity
| Your Data Point | Example: Lunch (3.0 mg/dL/g) vs. Dinner (1.5 mg/dL/g) |
| What it is: | A breakdown showing how this impact factor changes by mealtime, reflecting your unique insulin resistance. A higher number (e.g., Lunch) means you are more resistant to insulin at that time. |
| How to use it: | Use this to tailor your approach. If the morning factor is high, you know you likely need a longer pre-bolus time or a slightly higher insulin-to-carb ratio for breakfast compared to dinner. |
3. Insulin Sensitivity Factor (ISF)
| Your Data Point | Example: 1 Unit Lowers BG by 8 mg/dL |
| What it is: | The core clinical value that measures your body’s sensitivity to insulin. It indicates the magnitude of glucose reduction achieved by one unit of insulin. |
| How to use it: | This is your primary number for correction boluses. If your glucose is above target, divide the amount you need to correct by this number to estimate the required units. Note: This number is specific to you and should be confirmed by your doctor. |
4. Precision Dosing Note
While your ISF may be an integer value, remember that many insulin pens (especially for Type 2 diabetes) support 0.5-unit dosing (half-units), and some even do 0.1-unit increments for high precision. This analysis helps you discuss your need for these finer dose increments with your healthcare team.