Hemospec analyses whole blood through its optical signature. The system measures how the sample absorbs light across a defined spectral range. That absorption pattern carries information on the molecular composition of the sample and is used to estimate clinically relevant parameters.
In practice, the sensor captures the optical reading, and the analytical layer processes that signal using cloud-based
models. The aim is to extract useful information from the spectral pattern of the sample and translate it into a
structured result set. This is why the system combines photonics, software and analytical modelling in a single
architecture.
This approach makes it possible to work with a small blood sample, avoid reagent-heavy handling and bring a
broader panel of biomarkers closer to the point of care. In the current product framing, that includes
hematological parameters together with C-reactive protein, expanding the system beyond single-analyte
testing.
Performance characteristics, including sensitivity, specificity and accuracy, are established against reference laboratory methods for each analyte, within a defined and validated working range. Validated performance differs by analyte: the haematological panel is supported by a quantitative validated range, while C-reactive protein is reported as a threshold classifier, to be interpreted together with clinical judgement.
MARD around 9%, with
accuracy ranging from 84% to
87%
MARD around 8–10%, with
accuracy ranging from 82% to
94%
MARD around 10–11%, with
accuracy ranging from 76%
to 80%
Classification accuracy around
73% to 80% in validated units
MARD around 10–13%, with
accuracy ranging from 61% to
63%
Hemospec continues to evolve as an analytical platform. Current work includes the expansion of the biomarker panel through targeted optical development in creatinine, total proteins, albumin, and additional work around glucose and lactate.
Labinlight is advancing non-invasive glucose measurement
through transdermal optical analysis, combining dedicated
optical configurations with analytical models developed for
this new mode of sensing.
A second research direction is non-targeted analysis, where the full
spectral signature is used to identify clinically meaningful patterns
rather than focusing on a single biomarker alone. This approach opens
the platform to broader analytical models associated with complex
conditions, including sepsis, chronic kidney disease, iron-deficiency
anaemia and prognostic stratification. Internal research materials
already describe encouraging model performance across several of
these exploratory paths.
