DERM and DermaSensor address the same underlying clinical need, faster, more accurate triage of suspicious skin lesions, through genuinely different modalities and with different levels of autonomy, and understanding the difference matters practically because it determines what kind of evidence claim each system is even making. DERM works from photography and is designed for autonomous triage decisions within a defined pathway; DermaSensor works from optical spectroscopy at the point of care and is authorised specifically as an adjunctive assessment, not a definitive autonomous diagnosis, a distinction with direct implications for how each system's headline performance figures should be read.
The comparison, dimension by dimension
Smartphone or clinical photography versus optical spectroscopy: DERM analyses images, whether from dedicated clinical dermatoscopic photography or, in some deployments, more accessible smartphone-based capture, while DermaSensor uses optical spectroscopy, a physical measurement technique distinct from image analysis, applied directly to the lesion at the point of care. Autonomous triage versus adjunctive assessment: the category-defining difference, DERM positioned to make an autonomous triage decision within its validated pathway, DermaSensor positioned as one input a clinician weighs alongside their own assessment rather than a standalone triage verdict. Intended healthcare professional: each system is validated for use by specific categories of clinician, and confirming the exact intended user for either system's specific deployment matters before assuming general applicability. Lesion types and referral recommendation: what range of suspicious lesions each system has been validated to assess, and what action a given output should trigger. Sensitivity and specificity: the headline performance figures each publishes, requiring the same careful reading this cluster applies throughout. Ungradable or indeterminate results: the failure-to-produce-result rate that vanishes from headline accuracy figures but matters enormously in practice, since a system that frequently cannot render a usable assessment has a real-world utility ceiling its headline sensitivity does not capture. Current regulatory status by territory: DERM has progressed through UK and European regulatory pathways specifically and has been used within NHS skin-cancer pathways, a UK-relevant deployment history; DermaSensor received FDA De Novo authorisation in 2024, explicitly as an adjunctive point-of-care device rather than as a definitive autonomous diagnostic tool, a US regulatory position with a specifically bounded intended use. NHS implementation evidence and impact on dermatology referrals: the workflow-level questions that determine whether either technology's individual accuracy translates into a system-level benefit, faster appropriate referrals and reduced unnecessary ones, rather than simply adding a step to existing pathways.
Why 95 to 96% sensitivity does not settle the question
A quoted sensitivity in the 95 to 96% range sounds, and often is, genuinely strong performance, and this cluster's recurring caution applies with particular force here: sensitivity alone does not establish positive predictive value, and positive predictive value depends critically on the prevalence of serious disease in the population actually being screened. Consider a realistic urgent-referral population, where the prevalence of clinically significant skin cancer among referred lesions is meaningfully elevated compared with an unselected community population, but still well short of a coin-flip. At that referral-population prevalence, even a genuinely strong sensitivity and specificity combination produces a positive predictive value that, while useful, is meaningfully lower than the headline sensitivity figure alone would suggest to an unfamiliar reader, exactly the arithmetic this cluster's evidence-literacy article works through in full with worked prevalence examples. Deploy the same test in a lower-prevalence, unselected screening population rather than an already-referred one, and the positive predictive value drops further still, the same test producing meaningfully different real-world value depending entirely on which population it is used in, never a fixed property of the test's published sensitivity and specificity alone.
What this means for triage-pathway design
The autonomous-versus-adjunctive distinction becomes practically important exactly at this point: a system authorised for autonomous triage, like DERM within its validated pathway, is making a stronger claim, that its output alone can appropriately direct a patient toward or away from urgent referral, than a system authorised as an adjunctive point-of-care input, like DermaSensor, which is explicitly designed to inform rather than determine a clinician's own assessment. Neither claim is inherently superior; they suit different pathway designs, autonomous triage suits high-volume screening contexts where a defined, validated decision rule can operate at scale, while adjunctive assessment suits point-of-care settings where a clinician's own judgement remains the operative decision and the tool's role is sharpening that judgement rather than replacing it.
Frequently asked questions
Does DermaSensor's adjunctive status mean it is less accurate than DERM?
Not necessarily: adjunctive and autonomous describe the intended role and regulatory claim, not a direct accuracy ranking, and each system's actual performance figures should be compared on their own published evidence rather than inferred from the authorisation category alone.
Could either system replace dermatology referral entirely for some patients?
Neither is designed or authorised to replace specialist assessment where genuine clinical suspicion exists; both function within a triage pathway that still routes appropriately concerning findings toward dermatology, with the systems aiming to improve which patients that pathway prioritises rather than to remove specialist assessment from the pathway.
How should a service interpret a headline sensitivity figure before adopting either system?
By asking what population and prevalence the figure was measured in, and how that compares with the population the service intends to screen, since the same sensitivity figure produces very different positive predictive values across different prevalence settings.