Explore HealthRadar with demonstration data.
Explore how HealthRadar approaches laboratory-result interpretation using demonstration data.
What this demonstration represents
HealthRadar is being developed around a rules-based approach to laboratory-result interpretation. Rather than relying only on free-form generative AI, HealthRadar uses predefined clinical rules and communication boundaries to identify findings, prioritize what matters, and determine how results should be explained.
The public demonstration illustrates the approach. It is not a substitute for medical care and should not be used to make decisions about an individual patient's diagnosis or treatment.
Laboratory result list
| Test | Result | Reference | Flag |
|---|---|---|---|
| Hemoglobin A1c | 6.2% | 4.0–5.6% | High |
| Fasting glucose | 114 mg/dL | 70–99 | High |
| Triglycerides | 286 mg/dL | <150 | High |
| HDL cholesterol | 36 mg/dL | ≥50 | Low |
| LDL cholesterol | 151 mg/dL | <100 | High |
| ALT | 52 U/L | 7–35 | High |
| AST | 31 U/L | 10–35 | In range |
| eGFR | 74 mL/min/1.73m² | ≥60 | In range |
What stands out
Triglycerides and LDL cholesterol are above the reference ranges, while HDL cholesterol is below the stated range. The pattern is more informative than any single value alone.
The A1c and fasting glucose are both above their listed reference ranges. A clinician may consider whether the pattern is new, persistent, or affected by current treatment or other conditions.
ALT is mildly above the listed reference range. Many different factors can affect ALT, so the value should not be interpreted in isolation.
Questions for the next visit
- Are these findings new or part of a longer-term trend?
- How do they affect my overall cardiovascular and metabolic risk?
- Could medications, alcohol, weight change, diet, or another condition contribute?
- Should any test be repeated, confirmed, or interpreted with additional information?
Professional review
Presentation is not validation.
A polished interface does not establish accuracy, clinical utility, safety, or improved outcomes. Those questions require defined methods, testing, real-world usability work, and transparent reporting.
Why did we build Yardstick?
Before asking whether HealthRadar’s rules-based approach can improve AI-assisted communication, we wanted to understand how leading general-purpose AI models perform when interpreting the same laboratory cases repeatedly. Yardstick is that benchmark.
Yardstick did not test HealthRadar as a fourth generator and did not evaluate HealthRadar product effectiveness.
Interested in evaluating HealthRadar?
We welcome conversations with researchers, clinicians, health systems, and other organizations interested in laboratory-result communication and carefully controlled clinical AI.