A doctor in the loop, always.

Our core innovation isn't autonomous AI — it's the opposite. An intelligence layer does the relentless work of watching the metabolic signal; a specialist makes every decision. Speed of software, judgement of a clinician.

The model

Why doctor-in-the-loop is the whole point.

Most health AI tries to replace the clinician. We built the inverse: AI that amplifies one, with a hard rule that no clinical action ships without human sign-off.

No autonomous clinical action — the doctor approves every plan
The AI watches 1,440 minutes a day; the specialist sets the strategy
Pre-analysis over raw dumps — the model reasons on patterns, not floods of data
One index a patient understands and a clinician can defend

It's the difference between a chatbot that guesses and a system a practising endocrinologist will actually stake their name on. That trust boundary — human-approved, evidence-grounded — is the defensible core of AiHealth.

The pipeline

From raw signal to clinical decision.

01

Capture

CGM, body composition, labs and wearables stream into one continuous patient timeline — 288 glucose points a day, not one a quarter.

02

Pre-analyse

An engine computes variability, dawn phenomenon, meal response and time-in-range before any record reaches a model — so the AI reasons over signal, not raw noise.

03

Score

MMIQ folds the picture into one calibrated 0–100 metabolic index, with a priority score that triages who needs attention first.

04

Draft

The care agent proposes the plan in clinical language, grounded in the patient's own data.

05

Approve

Dr. Bhograj reviews, edits and approves. No clinical action ships without the specialist's sign-off.

Research & discoveries

Grounded in one of India's largest T2D glucose datasets.

MMIQ isn't tuned on intuition. It's built on a research program analysing real continuous-glucose data at scale — and the findings have already changed how we score risk.

2,500+
Patients analysed
38,000+
Meal–glucose records
3.6M+
CGM readings
AM > PM
The same meal spikes higher at breakfast than at dinner.
Morning insulin resistance, measurable at scale
1 in 4
"In-range" patients carry hidden overnight instability.
Time-in-Range alone misses them
Order
Protein-first meals blunt the spike — strongest at breakfast.
A behavioural lever, proven in the data
Dawn
Most Indian T2D patients show an early-morning glucose rise.
A modifiable, re-timable risk
1 in 3
meet all six international CGM targets at once.
The other two-thirds need more than an HbA1c
Timing
A late meal lifts the next reading — dose-dependently.
Detected from glucose alone

Qualitative summary of our metabolic research program; full effect sizes available to partners under NDA. Manuscripts in preparation.

Under the hood

Built like an instrument, not a wrapper.

Where standard CGM reports stop at an average, MMIQ goes further — and the research above is exactly why.

DRIVER SEPARATION

We separate how much of a patient's risk comes from basal, post-meal and overnight glucose — three problems, three different fixes.

OVERNIGHT RISK

A dedicated nocturnal score catches the ~1 in 4 "in-range" patients who are unstable while they sleep.

TIGHTER TARGET

A 70–140 mg/dL "tight range" alongside the standard 70–180 — closer to how healthy metabolism actually behaves.

/ MMIQ

Metabolic Intelligence Quotient

The diabetes index — a defensible 0–100 score of glucose-driven metabolic health, calibrated for clinical triage.

/ BMIQ

Body Composition Intelligence Quotient

The obesity index — body-fat, skeletal muscle, visceral fat and waist in one score, so we track fat lost and muscle kept.

/ TIR

Time-in-Range analytics

Continuous glucose distribution and variability — the metric that actually tracks outcomes.

Why this is defensible

Why AiHealth wins — and why it's hard to copy.

The moat isn't the model. It's the data, the trust boundary, and the economics of reaching everyone.

/ 01

A proprietary evidence base

Years of real continuous-glucose and meal data on Indian metabolism — a corpus no off-the-shelf model has, and that compounds with every patient.

/ 02

The trust boundary

Doctor-in-the-loop is exactly what clinicians and regulators require. Pure-AI competitors can't safely cross it — we designed the product around it.

/ 03

Scales on minimal inputs

The engine needs little more than a glucometer and a phone — so it extends from city clinics to rural camps without re-engineering. Reach is the strategy.

Built and used by a practising endocrinologist across a growing clinic network. Detailed evidence and methodology available to partners under NDA.

See it work

Technology only matters because of the outcome.