Lab reports to trends
Lab report analysis AI for doctors
Docna is lab report analysis AI for doctors. Upload a PDF or a photo, or tap Read report on one a patient sent on WhatsApp, and Docna extracts each result with its unit, flag and reference range, files the values on the patient's chart by collection date, and shows which biomarkers are moving against the patient's own baseline.
1-month free trial. Plans from ₹4,000 a month.
Patient state, from four reports
Sample- Digital PDFs read in full
- PDF values checked against the text
- Filed by collection date
What the latest report printed
Fasting blood sugar 99 mg/dL, normal. HbA1c 5.5%, normal. TSH 2.2 µIU/mL, normal. Haemoglobin 13.8 g/dL, normal.
- FBS
- 88, 92, 95, 99 mg/dL over 18 months. Every value in range, and drifting up.
- A1c
- 5.1, 5.2, 5.4, 5.5% over the same visits. In range, and drifting up.
- TSH
- Four readings, no move beyond normal variation. Stable.
- Hb
- One reading so far. No trend can be read yet.
How it works
How lab report analysis works in Docna
Five steps from a lab report to a trend on the patient's chart. Docna does the reading, checking and filing; the review and every clinical decision stay with you.
- 01
Bring the report in
Open Lab Reports from the patient's chart with Add Lab Report, so the report is filed to the right person, then drop in a photo (JPEG, PNG, WebP or GIF) or a PDF. Several reports for one patient can go in together. A report a patient sends on WhatsApp is read with one tap on Read report in the chat.
- 02
Every page of a digital PDF, read as text
Most labs send digital PDFs. When every page carries text, Docna reads the document's own text layer instead of a picture of it, page after page, so a long health check-up panel is read to the last page. Long reports are split into sections at page boundaries, never mid page, so no result row is cut in two. Photos and scanned PDFs are read as images.
- 03
Every PDF value checked against the report
For a digital PDF, every numeric result Docna extracts must appear in the report's own text, or it is dropped. When you upload a single report, the results screen tells you how many were discarded, and if materially fewer results come back than the report appears to hold, an amber notice says the report may be incomplete and asks you to check it against the original.
- 04
Review the results, then save
Results open panel by panel, each test with its value, its unit, the reference range the lab printed and a normal, high, low or critical flag. Critical values sit in a banner at the top. Add your own remarks, then save the report to the patient's record.
- 05
The values join the patient's timeline
When the report is saved, each result that matches a biomarker in Docna's catalogue is added to the patient's vitals timeline, dated to when the sample was collected rather than when the lab printed it. Upload the same dated report twice and values already on the chart are not added again.
Capabilities
What the lab report AI reads, and what it does next
Extraction is only the start. A lab report means more with every report before it, so Docna keeps the whole series and reads it against the patient's own history, not only against a population range.
More than 400 biomarkers
The catalogue covers more than 400 tests across the panels an Indian OPD orders: CBC, lipid profile, kidney and liver function, thyroid, diabetes, iron studies, vitamins, electrolytes, coagulation, cardiac markers, inflammation, hormones, tumour markers and urine routine. A test is matched by its printed name, its standard name or its abbreviation, so FBS and fasting blood sugar land on the same line.
The units Indian labs print
Unit conversions are set per analyte, because mg/dL to mmol/L is a different factor for glucose than for cholesterol. A history recorded partly in mmol/L and partly in mg/dL charts as one line, platelets reported in lakhs per cumm are converted, and a reading in a unit Docna cannot verify is kept off the trend line rather than plotted as if it were right.
Dates read the way Indian labs write them
Indian reports print dates day first, so 05/09/2026 is filed as 5 September, not 9 May. Where the report gives a collection date, that is the date the results are filed under. A date in the future is refused, and the results are filed under the report date, or the day you save them, instead.
Drift caught while every value is in range
The Patient State card answers the question a single report cannot: is this value moving? A trend counts as drifting only when its least squares slope has a 95% confidence interval that excludes zero and, where the analyte has a published within-person biological variation, the move is larger than it. A drift read from the slope alone is marked trend only. That is how a fasting sugar climbing from 88 to 99 over eighteen months is caught while every report still prints normal.
One row per body system, honest about thin history
The card groups analytes by body system and sorts them by urgency, with critical values first. A direction is read only once there are three readings spanning three months, and until then the row says there is no serial history instead of calling the value stable. A high value falling back toward range shows as improving, never as deterioration.
Baselines and the reference change value
Health Compass sets each patient's baseline at the median of their own readings, so one outlier cannot move it. To tell a real change from noise it uses the reference change value wherever published biological variation exists, drawn from the EFLM database and the Westgard compilation. A baseline deviation chart shows which latest results sit furthest from that patient's own set-point.
Insight cards that cite their readings
Health Compass raises a card for a critical value, a statistically significant change, a sustained drift, a value newly out of range, an improving trend and a value approaching a limit. Every card cites the readings behind it or it is not shown, and a critical value opens the panel on its own.
An AI read held to the card's numbers
Press AI read for a short interpretation of the card. The model sees only a fact sheet the engine has already computed: analyte names, values, spans and states, and never the patient's name. If it writes any number that is not on that sheet, the whole read is discarded.
Charts you can line up side by side
Overlay up to three biomarkers on one saved chart, such as HbA1c beside fasting sugar, with a normal range shaded wherever it would not overlap another. Before anything is saved, Docna suggests charts drawn from the patient's own data, and every numeric biomarker on file also has its own trend chart, shaded with its normal range where the catalogue has one.
In practice
From a WhatsApp attachment to the consult
Patients often send their reports before they walk in. In the patient's WhatsApp chat inside Docna, a photo or PDF they send carries a Read report button. The read runs on Docna's servers, so it behaves the same from a phone, a tablet or a desktop, and you can close the chat while it works. The notification bell then tells you how many results were read and how many joined the timeline, with a link to the patient's lab reports.
A digital PDF sent on WhatsApp goes through the same text reading and the same value check as an upload. A scanned PDF has no text to read there, so Docna tells you it is a scan and points you to Lab Reports, where it is read as an image.
For a new patient who arrives with a folder of old reports, select the patient on Lab Reports and drop the whole stack in at once; each report is read and saved in turn. A long single panel can take a couple of minutes, and you can open another page while it works and come back to the results.
When the patient sits down, their chart opens with the Patient State card at the top of the vitals panel: a headline, then one row per body system. Readings also arrive from saved consultation notes and from values you enter by hand, so the trend reflects the whole record, not only the reports.
Who it is for
Who lab report analysis is for
Physicians managing chronic disease
General physicians and diabetologists who follow HbA1c, lipids, kidney function and thyroid across years and want to see the slope, not only today's number.
Specialists who read long panels
Endocrinologists, cardiologists, nephrologists and internists whose patients arrive with multi-page health check-up reports.
Clinics whose patients send reports on WhatsApp
Practices where the report reaches the doctor in a chat before the patient reaches the clinic.
Practices building a longitudinal record
Clinics that want every report in one timeline per patient, instead of in a file the patient forgets to bring.
Pricing
Simple plans, priced in rupees
Lab report reading is included in both plans. Each report Docna reads counts as one AI query, charged only when the read succeeds. Pro includes 300 AI queries a day and Enterprise 1,000, shared across Docna's AI tools, and the free trial runs on the Pro allowance.
Every account starts with a 1-month free trial. Compare plans
Pro
₹4,000
per month, billed in INR
Enterprise
₹8,999
per month, billed in INR
FAQ
Lab report analysis AI: questions doctors ask
What is lab report analysis AI for doctors?
It is software that reads a lab report for you and turns it into structured data on the patient's chart. Docna extracts each test with its value, unit, reference range and flag, puts the values on the patient's timeline by collection date, and then reads the series over time, showing which biomarkers are drifting and which are stable.
Can it read a long PDF from a health check-up?
Yes. When every page of a PDF carries text, Docna reads the text of every page, in sections that never split a page. A scanned PDF has no text layer, so it is read as an image of its first six pages, and when you upload a longer scan on its own, Docna tells you before it starts.
Can the AI add a value that is not on the report?
On a digital PDF, every extracted numeric result must appear in the report's own text or it is dropped. When you upload a single report, the results screen says how many values were discarded, warns if far fewer results came back than the report seems to hold, and shows you every result before you save it.
How do I read a report a patient sent on WhatsApp?
Open the patient's WhatsApp chat in Docna and tap Read report under the photo or PDF. Docna reads it on its servers, files the results against that patient and adds the values to their timeline, and the notification bell tells you when it is done. You can close the chat in the meantime.
What does the Patient State card show that a lab report cannot?
A report says whether each value is inside its range today. The Patient State card says which way each value is moving. It calls a drift only when the trend is statistically real and, where published data gives the analyte's normal within-person variation, larger than that, so it can flag a value climbing steadily inside its range before any report marks it high.
Does Docna diagnose from the lab report?
No. Docna reads, organises and trends the results. Each result keeps the reference range the lab printed beside its flag, drift verdicts come from a statistical test on the patient's own readings, and the optional AI read may only restate figures the engine computed. The interpretation, the diagnosis and the treatment remain yours.
Which units and date formats does it understand?
Units are converted per analyte, including mg/dL and mmol/L and platelets in lakhs per cumm, so a mixed history charts as one line, and a reading in a unit Docna cannot verify stays off the trend line. Dates are read day first, the way Indian labs print them, with the collection date preferred over the report date.
How much does lab report analysis cost?
It is included in every Docna plan. Pro is ₹4,000 a month with 300 AI queries a day, Enterprise is ₹8,999 a month with 1,000, and each successful report read counts as one query. Every new account starts with a 1-month free trial on the Pro allowance.
Read the trend, not only today's report
Start a 1-month free trial, upload a patient's last few lab reports, and see which values are moving on their chart.