Google Adds Function Labs and Lingo Glucose to Google Health. Like Apple, It's Keeping the Interpretation

In October Google closed one door to its health data and opened another. Its developer notes now say the Google Health API is “not onboarding new projects at this time” [5]. Then on 12 October it let Function Health, Abbott and MyFitnessPal in: lab results, Lingo glucose readings and meal logs are all coming into the Google Health app, with Flo’s cycle data to follow in early 2027 [1].

Independent developers out, hand-picked partners in. That is not a contradiction. It is a design: one app that gathers every input about a person’s health, and keeps for itself the part that turns those inputs into meaning. Most of the inputs stay reachable by other apps. The interpretation does not, and the interpretation is what Google charges for.

That matters to anyone whose product sits between health data and a person’s decisions: coaching and wellness apps, lab and sensor companies, employers, insurers and care programs. This post sets out what was announced, why the lab results matter more than they look, what other apps can still read, and what to do about it.

What did Google announce on 12 October?

• Partner data into Google Health. US users will be able to import Function lab results “later this fall” and connect Abbott’s over-the-counter Lingo glucose sensor “soon”. MyFitnessPal meal logs sync now, with food-level detail for Premium subscribers. Flo cycle, pregnancy and menopause data arrives “beginning early next year” [1][2][4].
• Health Guardian on Fitbit. Monthly blood pressure and insulin resistance trend reports come to Fitbit Air and Fitbit Edge for Google Health Premium subscribers later in October [1]. On Pixel Watch 3, 4 and 5 they are free [12].
• Fitbit Edge. A $179.99 tracker with an ECG app, skin temperature and GPS, working with Android and iPhone, with three months of Premium included [3].
• Activity detection. Automatic recognition rises from 10 to more than 100 workout types on recent Fitbit and Pixel devices [1].

Google Health Coach, the Gemini-based coach that reads all of this, requires Premium at $9.99 a month or $99.99 a year [1][9]. We covered the May launch of the app and the coach in Google just unified the Android health stack, and Health Guardian itself in Health Guardian on Pixel Watch. What is new in October is the partner data, and the shape it gives the whole.

Why does Google want lab results?

Google’s own research answers this. In the WEAR-ME study, Google collected Fitbit and Pixel Watch data from 1,165 people and drew their blood at Quest Diagnostics, using the HOMA-IR blood measure as the reference for insulin resistance [6]. It then tested how well different combinations of inputs could detect it.

• Watch data and demographics alone: sensitivity 0.60 and specificity 0.80. That is the configuration a watch on a wrist can run.
• Plus one fasting glucose test: sensitivity 0.73, specificity 0.84.
• Plus lipid and metabolic panels: sensitivity 0.76, specificity 0.84, and an auROC of 0.80 [6].

Translated to 100 people at the study’s prevalence, 26 have insulin resistance. The watch alone catches about 16 of them, misses 10, and raises 15 false alarms among the other 74. With the blood panels it catches 20, misses 6 and raises 12 false alarms. We walked through the watch-only numbers in can a watch detect insulin resistance?

Function’s membership, at $365 a year, covers lipid and metabolic markers, fasting insulin and much more, drawn twice a year [4]. That is close to the bottom row of the chart, supplied by the consumer. Google has not said Health Guardian will use imported lab results. But its own study shows that is where the accuracy comes from, and on 12 October it arranged for people to bring them.

The deeper reason is calibration. A wearable is good at noticing change in a person over time and poor at saying where that person stands in absolute terms. A lab result is the opposite: an accurate snapshot that says nothing about the months between draws. Joining the two gives each one what it lacks, which is why we called continuous-plus-episodic data the new frontier in bloodwork goes consumer. Google now has both ends in one app.

Which of this data can other apps still read?

Most of it, with the user’s permission:

• Fitbit and Pixel Watch data. Google Health writes about 25 data types to Health Connect on Android, including sleep stages, heart rate variability and resting heart rate [7]. Since Google Health 5.05 in August it also writes exercise, sleep, steps and vitals to Apple Health on iPhone [11].
• Lingo glucose and MyFitnessPal meals. Both already write to Apple Health and Health Connect from their own apps, so an app can read them from the source without going through Google.
• Function lab results. No route out has been announced. Google Health can write medical records to Health Connect [7], but neither company has said whether imported Function results will be included.

What stays inside is the interpretation. Health Guardian’s blood pressure and insulin resistance trends are not among the data types Google Health writes to Health Connect [7], and Coach guidance lives only in the app. The one cloud route, the Google Health API, is closed to new projects [5].

So the data in Google’s new hub is, for the most part, data any app could already read with a user’s permission. What Google adds is the reading of it.

What is Google actually selling?

Look at where Google puts the price. Steps, sleep stages, heart rate and basic insights stay free. The coach, food-level nutrition logs and, on Fitbit devices, Health Guardian sit behind Premium [1][9]. On Pixel Watch, a more expensive device, Health Guardian is free [12]. Google is giving away collection and charging for meaning.

It also sets a new reference point for consumers. A US consumer can now assemble a tracker, two lab panels a year, a month of glucose sensors and an AI coach for about $725 in the first year:

Any product that sells “all your health data in one place, with AI insights” is now competing with that bundle, from a company that also controls the Android health store.

Apple reached the same conclusion a month earlier by a different route. At its 9 September event it announced a $119 Quest Diagnostics panel of more than 50 biomarkers, bookable and viewable in the Health app, and a readiness score, Health Age and Insights built on top of watch data, with lab values such as A1c and LDL feeding Health Age [13]. None of those scores has a HealthKit type, so third-party apps can read Apple’s inputs but not its interpretation [13]. We called that asymmetry out in Apple’s 2026 health announcements.

The two largest platforms now share a shape: wearable inputs and lab anchors flow in, the inputs stay open to other apps, and the interpretation stays in the platform’s own app. That is no longer one company’s product decision. It is how consumer health platforms work in 2026.

What should health businesses do about it?

Consumer health and coaching apps. Do not compete on aggregation or on a general-purpose coach. Google will always have more data in one place, and its coach comes bundled in a $10 subscription. Compete on a job a general coach will not take: a specific condition or goal, a program with human clinicians or coaches, accountability to a group, or a population Google does not serve. And own your interpretation layer. If your scores and guidance come from your own models over the same inputs, your product is not a front end for Google’s, and it works the same for iPhone and Android users.

Lab, sensor and nutrition companies. Google is a distribution channel worth having, but the flow in Function’s announcement runs one way, into Google Health [4]. Before you join, ask what comes back: attribution when Coach uses your data, referrals, aggregate insight, or a way to keep your own interpretation visible to the user. If the answer is nothing, you are supplying the inputs to someone else’s product.

Employers, insurers and care programs. Members will start arriving with monthly “High” blood pressure or insulin resistance trend flags from features Google describes as general wellness, not FDA-cleared measurements [8]. Google’s own study implies the watch-only insulin resistance model misses about four in ten cases and raises a false alarm for about one in five people without the condition [6]. Decide your protocol now: a flag is a prompt to confirm with a cuff or a lab test, never an eligibility or pricing input. Our guide to what “FDA cleared” actually covers explains how to read these labels.

Developers. Plan on the phone, not the cloud. Read Google Health’s data from Health Connect and Apple Health, compute your own trends rather than waiting for Google’s to be shared, and do not build a roadmap on Google Health API access until Google reopens onboarding. The practical routes are in our Fitbit Web API shutdown guide.

What should you watch next?

• Whether anything flows back out. If Google starts writing Health Guardian results or imported lab data to Health Connect, the interpretation becomes shared infrastructure. If not, it stays a reason to use Google’s app.
• When the Google Health API reopens. Google says it is “actively working to open access to more developers” [5]. Who gets access first will show whether the API is for everyone or for partners.
• Whether trends become notifications. Apple’s hypertension notification went through FDA clearance. If Google moves Health Guardian from monthly wellness trends to alerts about a condition, it moves into regulated territory.
• Whether either platform prices interpretation differently. Apple gives its insights away and sells the lab panel; Google partners for labs and sells the coach. Whichever model consumers accept will shape what they expect your product to cost.

The short version

On 12 October Google brought Function lab results, Lingo glucose and MyFitnessPal logs into Google Health, extended Health Guardian to Fitbit devices for Premium subscribers and launched the Fitbit Edge [1][3][4]. Its own research shows why labs matter: adding blood panels to watch data lifts the insulin resistance cases its model catches from 16 to 20 in every 100 people [6]. Most inputs stay readable through Health Connect and Apple Health, but Health Guardian trends and Coach guidance stay inside the app, and the Google Health API is closed to new projects [5][7]. Like Apple, Google is giving away collection and keeping interpretation. Health businesses should own their interpretation layer, ask partners what flows back, and treat trend flags as prompts, not diagnoses.