Multi-Analyte Sensing in CGMs

Multi-analyte sensing means one wearable sensor measuring glucose and at least one other chemical, such as ketones or lactate, from the same spot under the skin. As of October 2026, the only glucose-ketone sensor with FDA authorization is Abbott's Libre Duo 10 Day, granted De Novo authorization on August 25, 2026 [1][2]. This article covers the hardware: how a sensor measures a second chemical, what the FDA now asks of it, and why that's harder than it sounds. How pumps might use the extra readings is in our article on closed-loop algorithms with multi-analyte data.

How a sensor measures ketones under the skin

Sensors target beta-hydroxybutyrate (BHB), the main ketone circulating in blood [3]. The recipe is borrowed from CGMs. A sensing element under the skin sits in interstitial fluid, the fluid between cells, and uses an enzyme reaction to turn its target into a small electrical signal [3]. For ketones the enzyme is beta-hydroxybutyrate dehydrogenase, which can't work alone: it needs a helper molecule, the cofactor NAD+ [3].

Abbott's first published ketone sensor paired that enzyme with its wired enzyme technology, which tethers the enzyme to the electrode [4]. In a 2021 feasibility study, 12 healthy adults on low-carbohydrate diets wore the sensors for 14 days. With a single retrospective calibration, 82.4% of readings landed within 0.225 mmol/L (at lower levels) or 20% (at higher levels) of fingerstick ketone tests [4]. The authors said studies in people with diabetes, including during diabetic ketoacidosis (DKA), were still needed [4]. That year a Diabetes Technology Society consensus panel also reviewed the first ketone monitor reported in human trials, and how such monitors might link to pumps and automated insulin delivery (AID) systems [5].

Here's what surprised us: the chemistry wasn't the only hurdle. Abbott's engineers wrote in 2025 that one challenge in building the dual glucose-ketone sensor was making the ketone half factory-calibrated, so users never have to calibrate it [6].

What the FDA authorized in August 2026

The FDA order describes Libre Duo 10 Day as a prescription system for people with diabetes ages 2 and older that monitors glucose and BHB in real time, with alarms and links to connected devices [7]. De Novo is the FDA route for classifying a new kind of low- or moderate-risk device that has no comparable predecessor [7]. This one created a device type, the integrated continuous glucose ketone monitoring system (iCGK), under 21 CFR 862.1354 [7]. Abbott says the system had breakthrough device designation and is the first to meet both the iCGM glucose standard and the new iCGK standard [8].

Abbott lists up to 10 days of wear. In its studies, 84.1% of sensors worn by adults and 68.8% worn by children lasted the full 10 days [8]. Abbott's glucose-only FreeStyle Libre 3 Plus sensor lasts up to 15 days [9]; Abbott hasn't said why the dual sensor's wear period is shorter.

As of October 13, 2026, Abbott's US product page still calls Libre Duo 10 Day "coming soon" [10], after Abbott said in August it planned a US launch later in 2026 [8]. The page cites a Libre Duo user manual [10], but we couldn't find it published online, so the labeled accuracy results aren't yet public.

What the FDA now requires of a dual sensor

The order's special controls, the rules every device of this type must meet, are the clearest map we found of what makes a dual sensor hard. Clinical data must show accuracy for each analyte across the measuring range, comparing glucose and ketone readings against laboratory methods on blood drawn at the same time [7].

The two halves aren't held to the same kind of bar. Glucose rules are fixed percentages: across the whole range, the lower confidence bound for readings within 20% of lab glucose must exceed 87% [7]. Ketone accuracy has no fixed percentage. It must be "adequate to ensure clinically acceptable agreement, as determined by FDA," and gross errors are banned, such as a very low sensor reading while blood ketones are very high [7].

Other controls target what stacking analytes adds. The sensor must perform despite interfering substances, from metabolites and foods to supplements and medications. Labeling must report ketone accuracy separately in four concentration bands. Because the device type is built to feed automated insulin dosing systems, data transmission must be secure and reliable, without clinically significant gaps [7]. Later competitors must file a 510(k) premarket notification [7].

The engineering trade-offs

Interference and lag. Blood ketone strips use a similar enzyme approach, so a 2024 review infers that continuous ketone sensors could share some of their interferences, such as ascorbate (vitamin C). BHB is also a smaller molecule than glucose, so the authors expect little delay between blood and interstitial fluid, though they note published lag data are limited [3].

Accuracy beyond Abbott. A McGill University study published in 2026 tested a SiBio continuous ketone sensor in 16 adults with type 1 diabetes using AID. Against fingerstick ketone tests on a FreeStyle Libre 2 meter, the average gap was 0.1 mmol/L at low ketone levels and 24.2% at elevated levels [11].

Alarm burden. The same study logged 81 ketosis episodes over 155 days of ketogenic diets, and all but one caused no symptoms [11]. A 2026 international expert panel wrote that ketone alerts must avoid needless anxiety and alarm fatigue, and that strong evidence for setting thresholds doesn't exist yet [12]. A separate 2026 review agrees that actionable thresholds still need defining [13]. What a ketone reading should prompt is a decision for your care team.

Hardware. Each added analyte usually needs its own electrodes and circuitry, which a 2026 paper calls the main limit on scaling wearable sensors. That team built an electrode that switches roles, letting one microneedle patch track glucose, lactate and potassium, so far only in rats and human serum samples [14].

Beyond ketones: lactate and alcohol

A University of California San Diego team built a wearable microneedle patch, with reusable electronics and a disposable needle array, that tracked lactate or alcohol alongside glucose in volunteers going about daily activities. Readings correlated well with blood or breath tests, and the authors called for larger studies [15].

Lactate rises in sepsis and shock. In a 2026 pilot from the same university, lactate microneedle sensors worn for 4 hours by 21 people, including intensive care patients, tracked blood lactate closely (correlation 0.94) [16]. That's hospital monitoring, not a diabetes product, but it shows interstitial lactate can be read in very sick patients.

What to watch

Abbott expects Libre Duo 10 Day to work with Beta Bionics' iLet and Sequel Med Tech's twiist AID systems by the end of 2026, with Insulet and Tandem devices in 2027 and an exclusive MiniMed integration also expected in 2027 [8]. As of October 2026 those are company plans, not shipped products. The milestone we're waiting for is the US user manual, which must report ketone accuracy band by band [7][10].