Your patient is their own reference range.
Why a single lab value against a population range misses early signals, and how DeepDx reads results as trends against personal and population-adjusted baselines.
Normal is not the same as unchanged
A hemoglobin of 12.5 g/dL is inside many adult reference ranges. If the same patient was at 14.8 a year ago and 13.6 six months ago, that steady fall is a signal worth a look, even though no single result was flagged. Standard reports compare each value to a population range. They do not compare the patient to themselves.
Ranges should fit the population
Reference intervals vary with age, sex, pregnancy and ancestry, and many published ranges come from populations that do not match the patients in the room. DeepDx is designed to apply population-adjusted ranges drawn from multiple reference sources and to record which range was used for every interpretation.
Patterns across analytes
Clinicians rarely read one value at a time. Iron deficiency and anemia of chronic disease can both lower hemoglobin, but they move ferritin, transferrin saturation and inflammatory markers differently. DeepDx encodes these relationships in a pathophysiology graph, so a pattern across several results can be surfaced together with the reasoning behind it.
Plumbing first
None of this works if units are inconsistent. Results arrive in mg/dL from one lab and mmol/L from another. Unit normalization and LOINC coding come before any model, and every converted value keeps its original alongside it.
Key takeaways
- 1Trend against the patient's own history, not just the population.
- 2Record which reference range was applied, every time.
- 3Multi-analyte patterns carry more signal than any single flag.